GENOMICS

25,674-681

(1995)

Thermal Asymmetric Interlaced PCR: Automatable Amplification and Sequencing of Insert End Fragments from PI and YAC Clones for Chromosome Walking
YAO-GUANG LIU AND ROBERTF. WHITTIER
Mitsui Plant Biotechnology Research Institute, RlTf Tsukuba Laboratory 1, TCI-D21, Sengen 2-l-6, Tsukuba 305, Japan
ReceivedFebruary22, 1994; revised September 21, 1994

Isolation of DNA segments adjacent to known sequences is a tedious task in genome-related research. We have developed an efficient PCR strategy that overcomes the shortcomings of existing methods and can be automated. This strategy, thermal asymmetric interlaced (TAIL)-PCR, utilizes nested sequence-specific primers together with a shorter arbitrary degenerate primer so that the relative amplification efficiencies of specific and nonspecific products can be thermally controlled. One low-stringency PCR cycle is carried out to create annealing site(s) adapted for the arbitrary primer within the unknown target sequence bordering the known segment. This sequence is then preferentially and geometrically amplified over nontarget ones by interspersion of high-stringency PCR cycles with reduced-stringency PCR cycles. We have exploited the efficiency of this method to expedite amplification and sequencing of insert end segments from Pl and YAC clones for chromosome walking. In this study we present protocols that are amenable to automation of amplification and sequencing of insert end sequences directly from cells of Pl and YAC clones. Q1996 Academic Press, Inc.

INTRODUCTION

Genome-related research frequently requires isolation of DNA from an unsequenced segment bordering a known sequence. This need arises when isolating insert end fragments of large clones such as Pl and yeast artificial chromosomes, cloning insertion tagged genes, obtaining regulatory sequences corresponding to cloned cDNAs, or studying oncogenic retroviral insertions. A number of PCR methods have been described for this purpose, including inverse PCR (Ochman et al., 1988; Triglia et al., 1988; Silver and Keerikatte, 1989) and hemispecific or one-sided PCR methods (Frohman
1To whom correspondence should be addressed at the Mitsui Plant Biotechnology Research Institute, TCI D21, Sengen 2-l-6, Tsukuba 305, Japan. Telephone: 81-298-58-6235. Fax: 81-298-58-6234. Email: tsuO1135@koryu.statci.go.jp.
0888-7543195 $6.00 Copyright 8 1995 by Academic Press, Inc. All rights of reproduction in any form reserved. 674

1988; Loh et al., 1989; Ohara et al., 1989; Riley 1990; Mueller et al., 1989; Parker et al., 1991; Isegawa et al., 1992). Many of these methods require special steps before PCR such as Southern analysis to determine suitable restriction sites, followed by manipulations such as restriction cutting and ligation or tailing. Targeted gene walking PCR (Parker et al., 1991) and single primer PCR (Parks et al., 1991) do not require such manipulations prior to PCR. They rely upon specific priming within the known sequence together with arbitrary priming within the flanking sequence (hence, hemispecific). However, arbitrary priming also creates nontarget molecules, and these constitute the bulk of the final product, even when the starting template sample is quite simple. The desired product must then be identified by hybridization or elongation from an end-labeled internal primer. Thus, methods that omit manipulations before PCR require more laborious screening afterward. We present here a hemispecific PCR method that overcomes these drawbacks. It requires no DNA manipulation before PCR, yet efficiently amplifies targeted segments, usually without visible background. The basis for this strategy is thermal asymmetric PCR, which was described for producing single-stranded DNA templates for sequencing (Mazars et al., 1991). Using two primers differing in length and hence thermal annealing stability, PCR cycles carried out with high annealing temperatures favor the longer primer, while annealing at lower temperatures allows both primers to function with near equal efficiency. We have developed a strategy interspersing asymmetric and symmetric PCR cycles so as to geometrically favor amplification of target molecules over nonspecific products. This strategy, thermal asymmetric interlaced (TAIL)-PCR, entails consecutive reactions with nested sequence-specific primers and a shorter arbitrary degenerate primer. Generation of insert end-specific probes from YAC or Pl clones is an essential step in genome mapping and map-based cloning programs. In this paper we describe the methodology of TAIL-PCR and its application to Pl
et al., et al.,

INSERT

END AMPLIFICATION

BY TAIL-PCR

675
TABLE 1

and YAC systems. Detailed protocols are presented for efficient amplification of insert end sequences directly from cells harboring Pl or YAC clones and subsequent direct-sequencing of unpurified TAIL-PCR products using an automated DNA sequencer. Since no manipulation apart from PCR is involved, the process for PCR amplification and sequencing can be automated using an automated laboratory workstation or expedited using multiple-channel pipets.
MATERIALS AND METHODS
Pl and YAC clones. Pl and YAC clones containing Arabidopsis DNA were selected for TAIL-PCR from a Pl library (Liu et al., 1995) and the EG YAC library of Grill and Somerville (1991), respectively. Oligonucleotide primers. Specific primers that are complementary to the Pl and YAC vector sequences, respectively, were synthesized (Fig. 3). In addition, four arbitrary degenerate (AD) primers were used: TG(Afl~GNAG&I)ANCA(G/C)AGA-3 (ADl). AGfAl T,GNAG&T,ANCA&T,AGG-3 (ADS), CA(A/T)CGICNGAIA;G/ C)GAA-3 (AD3, I indicates inosine), and TC(G/C)TICGNACIT(A/ T)GGA-3 (AD4). These AD primers have average T,,,s of 47-48C as calculated with the formula 69.3 + 0.41 (%GC) -650/L (Mazars et al., 1991), where L is primer length. PCR procedure. This procedure was designed so that all pipettings after removal of culture medium from cell pellets can be automated in 96-well microtiter plates and MicroAmp reaction tubes (Perkin-Elmer) by a Biomek 1000 laboratory workstation (Beckmann). Alternatively, pipettings can be carried out efficiently with multiple-channel manual pipets. Thermocycling was carried out using a GeneAmp System 9600 (Perkin-Elmer). Cells harboring Pl or YAC clones were cultured overnight with gentle shaking in 96well microtiter plates (round bottom type) in 75 ~1 LB (containing 25 pg/ ml kanamycin and 1 m&f IPTG) or 100 ~1 YPD, respectively. After pelleting by centrifugation, Escherichia coli cells were resuspended in 2 vol (150 ~1) of 0.5~ PCR buffer (see below). Yeast cells were resuspended in 10 ,~l of spheroplasting solution (2 mM EDTA, pH 7.5, 1 mg/ml zymolyase) without sorbitol and incubated at 37C for 1 h. Without removing the spheroplasting solution, 1 vol (100 ~1) of 0.5~ PCR buffer was added to each well. E. coli cells or yeast spheroplasts were incubated in an air oven at 92C for 15 min. Aliquots (1 ~1) of the cell lysate ware added to MicroAmp reaction tubes containing 15 ~1 of primary TAIL-PCR mixture. The PCR mixture consisted of lx PCR buffer (10 m&f Tris-HCI, pH 8.3, 50 m&f KCl, 1.5 or 2.0 m&f MgClz, 0.001% gelatin), 0.2 mM each dNTPs, 0.15 fi specific primer (PSl, PTl, YLl, or YRl) and an AD primer (5 $kf for AD1 and AD2 or 2.5 @f for AD3 and AD4, which contain inosine residues), and 0.8 units of AmpliTaq polymerase (PerkinElmer). The thermal cycling conditions are summarized in Table 1. For secondary reactions, l-p1 aliquots of the primary PCR products were transferred to microtiter plate(s) containing 100 ~1 Hz0 in each well and mixed. Note that the Biomek 1000 workstation can pipet from MicroAmp reaction tubes containing as little as 10 ~1 solution, but can only pipet from microtiter plates containing about 100 ~1 or more solution. Dilution aliquots (2 ~1) were added to 18 ~1 secondary PCR mixtures containing 1.1~ PCR buffer, 25 fl each dNTPs, 0.8 units of AmpliTaq polymerase, 0.2 fl internal specific primer (PS2, PT2, YL2, or YR21, and the same arbitrary primer used in the primary reaction (3 @f for AD1 or AD2, 1.5 /.&f for AD3 or AD4). After amplification, l-,~l aliquots of the secondary PCR products were diluted in 100 ~1 HzO, and 10 ~1 of dilutions were added to 90 ~1 tertiary PCR mixtures containing 1.1~ PCR buffer, 25 p,V each dNTPs, 3.5 units of AmpliTaq polymerase, and 0.2 &the innermost specific primer (PS3, PT3, YL3, or YR3) and AD primer as in the preceding reaction. The PCR products (7 ~1) were transferred to microtiter plates containing 3 ~1 of 3~ loading buffer in each well and run on 1.5% agarose gels.

Cycling Conditions Used for TAIL-PCR on the GeneAmp System 9600

File no. 1 2 3 4 5 Cycle no. 1 5 1 10 12

Reaction Primary

Thermal condition 92C (2 min), 95C (1 min) 94C (15 s), 63C (1 min), 72C (2 min) 94C (15 s), 30 C (3 mm), ramping to 72C over 3 min, 72C (2 min) 94C (5 s), 44C (1 min), 72C (2 min) 94C (5 s), 63C (1 min), 72C (2 min) 94C (5 s), 63C (1 mini, 72C (2 mini 94C (5 s), 44C (1 min), 72C (2 min) 72C (5 min) 94C (5 s), 63C (1 min), 94C (5 s), 63C (1 mini, 94C (5 81, 44C (1 min), 72C 72C 72C 72C (2 (2 (2 (5 min) mini min) min)

6 Secondary 7

1 10

6 Tertiary 8 6

1 20 1

94C (10 s), 44C (1 mini, 72C (2 min) 72C (5 min)

Note. The program files in each reaction were linked automatically. n These are nine-segment super cycles each consisting of two highstringency and one reduced-stringency cycle (see Fig. 1).

Direct sequencing. For sequencing of TAIL-PCR products with the PRISM Ready Reaction DyeDeoxy Terminator Cycle sequencing kit (ABI), 5 ~1 of unpurified secondary (or tertiary) reaction products were added directly to a 17-~1 volume of sequencing mixtures containing 7 ~1 (1 pmol/pl) of the same specific primer used for the PCR amplification and 10 ~1 of sequencing mix from the kit. Cycle sequencing was carried out on a GeneAmp System 9600 thermocycler using 25 cycles of 96C for 15 s, 60C for 5 s, and 65C for 4 min. Sequencing using an automated DNA sequencer 373A (ABI) was carried out according to the manufacturers protocol. DNA hybridization. Conditions of dot and colony hybridizations for Pl and YAC library screening were as described (Liu et al., 1992). Hybridization signals were detected using a BAS-2000 image analyzer (Fuji Film).

RESULTS

Principle of TAIL-PCR PCR methods using a specific primer and an arbitrary primer or a primer that pairs with a binding site attached by ligation or tailing are known as hemispecific or one-sided PCR. In a hemispecific PCR, three types of products may form: those primed by both primers (type I), those primed by the specific primer alone (type II), and those primed by the nonspecific primer alone (type III). Type II products as well as any nonspecific type I products can be eliminated simply by carrying out successive reactions with nested specific primer(s). The type III nonspecific products, which are the major source of background, however, cannot be eliminated with nested specific primers using normal PCR cycling (see Fig. 50. The TAIL-PCR strategy is designed to favor amplification of the desired type I specific products and suppress amplification of the type

676
short arbitrary degenerate (AD) primer

LIU AND WHITTIER

Primary PCR with SPI and AD

5 high stringency cycles +

1 low stringency cycle

10 reducedstringency cycles

(A-> 1 reduced stringency cycle (thermal symmetric) 2 high stringency cycles (thermal asymmetric)

specific product (type I) -====q product yield: high or middle (detectable or undetectable)

v nonspecific product (type 11) --_-high (detectable) lo-fold 1

nonspecificproduct

(type III) D--_ --low (undetectable)

the AD primer within the unknown target sequence to create annealing site(s) adapted for the AD primer. Amplification is then carried out by interlacing highstringency with reduced-stringency PCR cycles. Since only the long specific primer can efficiently anneal to DNA template during high-stringency cycles, target sequence (type I product) is amplified linearly, and little or no amplification occurs for nontarget sequences (type III products) that are primed at both ends by the AD primer. In the following reduced-stringency cycle both primers can anneal to the template. The singlestranded target DNA produced during high-stringency cycles is replicated to double-stranded form, providing a severalfold increase of target template for the next round of linear amplification. By repeating this process (TAIL-cycling), it is possible to amplify target preferentially over nontarget sequences. Type II products primed at both ends by the long specific primer can also arise through mispriming, and these are amplified with even higher efficiency (see Fig. 2). Such undesired products are diluted out, however, in subsequent sec-

dilution 102 ; 100 + Type I product primed by both the long and short primers +- Type II product primed by the long primer only + Type III product primed by the short primer on

Secondary PCR withSP2andAD (10 super cycles)

------; product yield:

specific product
high (detectable)

nonspecific product c)_----very low (undetectable) lWO.fold dilution

E s 108 !z106 i

Tertiary PCR with SP3 ad AD (20 normal cycles) t Agarose gel analysis i Direct sequencing FIG. 1. Schematic diagram of TAIL-PCR contrasting the amplification of target with nontarget products. Boldface segments denote the specific primer (SP), and small open rectangles denote the arbitrary degenerate primer (AD). Diluted type II nonspecific product is not shown after the secondary reaction. In this study standard annealing temperature settings were 30C in the low-stringency cycle, 63C in the high-stringency cycles, and 44C in the reducedstringency cycles (see Table 1). The 10 cycles of reduced-stringency normal PCR between the single low-stringency cycle and the TAILcycling are optional (when omitting these cycles the supercycle number of the TAIL-cycling is increased to 15). Carrying out these cycles prior to TAIL-cycling is helpful in reducing the amplification competition between specific and type II nonspecific products during the initial cycles. As an alternate procedure to speed processing, two secondary reactions can be carried out simultaneously using SP2 and SP3, respectively.

104 T*IL-cycling

III nonspecific products. As shown in Fig. 1, the key points of this strategy are the use of a set of nested long specific primers and a relatively short arbitrary degenerate (AD) primer having a lower T,,, (melting temperature). One low-stringency PCR cycle is carried out to facilitate the initial base-mismatch annealing of

FIG. 2. Theoretical amplification rates of type I, type II, and type III products during the primary and secondary TAIL-PCRs. For clarity and simplicity, all priming reactions are depicted as occurring with either 100 or 0% efficiency, according to annealing stringency. Calculations indicate that for a fixed number of basic cycles during TAIL-cycling, interspersing two high-stringency cycles between each reduced-stringency cycle yields the highest relative amplification of target sequences. During TAIL-cycling the theoretical amplification rates of type I, type II, and type III products are 2, 2, and 2, respectively, where n is the number of supercycles.

INSERT

END AMPLIFICATION

BY TAIL-PCR

677

A PSl(Tm=61W ~ 5'-CTGTATGTACTGTTTTTTGCGATCTGCCGTTTCGATCCT

~xxx?.K.......... GAG&&TCACTCAGCA

WTCGAGCTGGCTACGGGAACTCGGAAGTTGGGkAGT-5' mlf.Tm=63W

YLl(Tm=63W 5'-cACTCTGAACCATCTTGGAGGA?CGGTAATTATTTC~ ATCTCTTTTTCAATTGTATATGTGTTATGTTATGTAGTATAC YL2(Tm=60C) TCTTTCTTCAACAATTAAATACTCTCGGTAGCCAAGTTGGTT ~CGCAAGATGTAATTTATCACTACGGAATTCGCGGCE _EamHIcloning site 3fT 58 Cl cAATT~CC&TA~~ AAATCACTCCCAATTA - YR3 (Tm=58C) &&$GG@CTTAAGGC~CCCGCAAGCTGA YR2(Tm=59'C) YRI(Tm=63'C) AAAAAAATACAGAGGTAA-5' GCGGGGGCCCTCTAAAAAAAC FIG. 3. Specific primers used for TAIL-PCR that are complementary to the Pl pAdlOsacBI1 (Pierce et al., 1992) and YAC pYAC41 (Grill and Somerville, 1991) vectors, respectively. The primer sets for the Pl vector are designated PSl, PS2, and PS3 on the SP6 promoter side and PTl, PT2, and PT3 on the T7 promoter side (A); for the YAC vector, primer set YLl, YL2, and YL3 is specific to the left side and primer set YRl, YR2, and YR3 is specific to the right side (B). The calculated T,,, of each primer is indicated.

have not observed any type II products in the tertiary PCR. By the outset of this reaction the original source DNA has been diluted 106-fold relative to the primary reaction, leaving less than 1 original cell genome equivalent per reaction. The 20 amplification cycles in the tertiary reaction are insufficient for any newly arising type II nontarget products to reach visually detectable levels (see control reactions shown in lanes designated C2 in Fig. 5). The product specificity was verified simply by comparing the sizes of the secondary and tertiary PCR products; target products in the tertiary reactions were slightly smaller than those in the secondary reactions in agreement with the nested positions of the primers. Since specific products were not always amplified to detectable levels in the primary TAIL-PCR, we routinely omitted the agarose gel analysis of the pri-

B
ondary and tertiary reactions using internally nested specific primers. TAIL-cycling is performed in the secondary PCR as well to lower background further. TAILcycling is usually unnecessary in the tertiary PCR, but may be applied if type III products still emerge. Amplification of Insert End Sequences YAC Clones from PI and C

M 1 II 111 I II III I II III I II III I II III

For amplifying insert end sequences from Pl and YAC clones by TAIL-PCR, three nested specific primers were synthesized to each bordering vector arm (Fig. 3). For amplifying Pl insert ends, we used E. coli cells directly as the template source. Although YAC insert ends could also be amplified directly from yeast cells without spheroplasting treatment (data not shown), this simple treatment facilitated release of DNA from cells and improved the reliability of amplification. Figure 4 shows some examples of amplification of insert end sequences from Pl and YAC clones by the TAIL-PCR strategy. In most cases product bands from the primary TAIL-PCR were visible. Many such bands disappeared after the secondary TAIL-PCR, indicating that these bands were nonspecific type II and/or type I products. Specific products were not always seen in the primary reactions due to their low concentration. However, these specific products became visible after the subsequent secondary reaction. In a few cases the secondary reaction produced type II products. Such products could be identified by their failure to undergo reamplification in the tertiary reaction (e.g., the upper bands in lane II of the last clone shown in Fig. 4A). We

FIG. 4. Agarose gel analysis of TAIL-PCR products for amplification of insert end sequences from the SP6 side (A) and T7 side (B) of Pl clones and the left side (C) and right side (D) of YAC clones. Each set of 3 lanes contains products from consecutive primary (I), secondary (II), and tertiary (III) reactions for a given clone. The arbitrary primers AD4, AD3, AD2, and AD1 were used for reactions shown in A, B, C, and D, respectively. Size marker (M) is a mixture of X-Hind111 and 4X174-Hue111 digests.

678

LIU AND WHITTIER

mary reactions and checked only the secondary and tertiary reaction products. Since very short products have limited usefulness, reactions producing no products over 250 bp were considered failures and repeated using a different AD primer. With any one AD primer, successful amplification of specific products were obtained in about 7080% of cases for Pl clones and 60-70% for YACs. Most of the amplified fragments ranged from ca. 300 bp to ca. 1 kb in size, and some even exceeded 2 kb. In some cases more than one specific fragment was produced (see Figs. 4 and 5). Despite multiple product bands, sequencing of the total product using the specific primer yielded clear sequencing profiles. This observation confirmed our supposition that these multiple product bands constituted nested fragments derived from annealing of the AD primer at more than one site along the target sequence molecules.
Effects of TAIL-Cycling Products on Selection of Specific

#I
I 11 III CI C2 M

#2 I II III Cl C2

To assess the effect of TAIL-cycling in suppressing amplification of type III nonspecific products, we compared reactions for two Pl clones with altered annealing temperatures in the high-stringency cycles (Fig. 5). Reactions under standard thermal conditions produced no type III products in any reactions, including the control tertiary reactions using the AD primer alone (lanes designated Cl in Fig. 5A). This indicates that TAIL-cycling in the primary and secondary reactions had sufficiently suppressed type III product amplification. In the reactions with decreased annealing temperature (55C) in the high-stringency cycles, type III products were amplified to detectable levels for one sample (No. 2) but not for another (No. 1) (Fig. 5B). The annealing temperature was set for all cycles to 44C so that both specific and arbitrary primers would function with more nearly equal efficiency, and type III products were amplified to high levels in both samples (Fig. 50. These results clearly demonstrate that TAILcycling is very effective in suppressing nonspecific target amplification.
Sequencing TAIZ-PCR Products Directly

FIG. 5. Comparison of PCRs for two Pl clones with high (A), decreased (B), and no (C) thermal asymmetric priming during cycling. The T7 side primer set was used in combination with AD2 for clone 1 and with AD4 for clone 2. Control tertiary reactions using AD primer (Cl) or PT3 (C2) alone are shown. (A) TAIL-PCR with standard thermal conditions as shown in Table 1. (B) TAIL-PCR with annealing temperature in all high-stringency cycles decreased from 63 to 55C. The supercycle number in file 5 (see Table 1) in the primary reaction was reduced to 10 and that in file 7 in the secondary reaction reduced to 8 to compensate for higher annealing efficiency and consequent amplification rate. (Cl PCR with an annealing temperature of 44C in all cycles (with no change in file 3). The cycle number in file 4 in the primary reaction was increased to 38, and file 5 was not executed. For the secondary reaction, file 8 was used. All of the tertiary reactions in A, B, and C were carried out with normal cycling using file 8. The type III nonspecific products amplified in B and C are identified as bands of identical size in lanes I, II, III, and Cl.

Unpurified PCR products can be directly sequenced by the thermal asymmetric cycle sequencing strategy using unlabeled sequencing primers (Liu et al., 1993). In this study we adapted this method to sequence rapidly the amplified clone insert end fragments on an automated DNA sequencer using the DyeDeoxy Terminator Cycle sequencing kit (ABI). With these kits an unlabeled sequencing primer is used, and the fluorescent label is attached to the chain terminating dideoxy nucleotides. A small amount (5 ~1) of unpurified secondary (or tertiary) PCR product was applied directly to sequencing reactions. Even when using the same specific primer for PCR as for sequencing, unpurified template yielded clear sequencing profiles, reflecting the

high specificity of the products. Figure 6 shows an example of direct sequencing. Using a high annealing temperature (SOC>,interference by the carried-over AD primer was avoided. A very low concentration (22.5 /..&I of dNTPs used in the secondary and tertiary PCR reactions was also designed so that carried-over dNTPs would not unduly affect sequencing reactions.
Automation of the PCR Process

One of the advantages of the TAIL-PCR method is that no special manipulations apart from PCR are required to obtain specific products, and even template DNA isolation can be omitted when amplifying clone insert ends. This simplicity enabled us to design assembly-line style protocols suitable for large-scale amplification and sequencing using the GeneAmp System

INSERT

END AMPLIFICATION

BY TAIL-PCR

679

FIG. 6. Direct sequencing is underlined.

of secondary TAIL-PCR

product amplified from the SP6 side of a Pl clone insert. The vector border sequence

9600 or other thermocyclers that are able to process

96 samples simultaneously. With these protocols the manipulations can be automated using a Biomek 1000 laboratory workstation (Beckmann) or expedited using multiple-channel pipets when doing manually. Use of Insert End Probes for Chromosome
Walking

Amplified insert end sequences from Pl and YAC clones were used to screen Pl and YAC libraries by hybridization to find overlapping clones for chromosome walking. Figure 7 shows that the hybridization backgrounds were very low, demonstrating the high specificity of the PCR products.
DISCUSSION

We have developed a novel PCR strategy to amplify specifically segments of unknown sequence that flank known sequences. Compared to other PCR methods for the same purpose, TAIL-PCR has the following advantages. (1) Simplicity: TAIL-PCR entails neither special DNA manipulations before PCR nor laborious screening afterward. Product specificity is effectively confirmed by simple agarose gel analysis. This makes TAIL-PCR especially suitable for assembly-line style amplification allowing simultaneous manipulation of a

large number of samples. In contrast, all other PCR methods require numerous manipulations prior to PCR (restriction digestion, ligation, tailing, etc.) or manipulations afterward (Southern hybridization, primer labeling and extension, autoradiograms and gel excision, etc.). These additional steps are cumbersome, laborious, and time-consuming. In addition, TAIL-PCRs requirement for template DNA quantity (-ng) and purity (cell lysate or crude DNA) are extremely modest. In contrast, ligation- or tailing-dependent PCR methods require more DNA ( - pg> of high purity for various manipulations. (2) High specificity: the proportion of coamplified nonspecific products is so low that TAILPCR products can be used directly as either hybridization probes or sequencing templates. This is its major advantage over targeted gene walking PCR (Parker et al., 1991) and single primer PCR (Parks et al., 19911, which normally produce high backgrounds of nonspecific products. (3) High efficiency: 60430% of reactions yielded specific products with any given AD primer. In ligation-dependent PCR methods, finding suitable restriction sites and ligating them is often a problem. For targeted gene walking PCR or single primer PCR, there is a tradeoff between success rate and low background; carrying out cycling at lower annealing temperatures raises the success rate but results in higher

680

LIU AND WHITTIER

(- 1.5 X 10 bp) is about fivefold larger than humans. Therefore, this method should be readily applicable to complex genomes, including those of mammals. Taken together, these advantages make TAIL-PCR a powerful tool for chromosome walking, genome physical mapping, development of sequence-tagged sites (STS), serial gene-walking, and analysis of genomic sequences flanking T-DNA, transposon, or retrovirus insertions. TAIL-PCR requires a disparity in T, between the specific primers and the AD primer. To achieve adequate thermal asymmetric priming, the T,s of the specific primers should be at least 10C higher than the average T,s of the AD primers, and the annealing temperature in the high-stringency cycles should be set as high as possible (usually l-5C higher than the calculated T, of the specific primer). Other rules in selecting specific primers for TAIL-PCR are generally the same as those for normal PCR. Selection of an optimal specific primer for the primary reaction is important for successful amplification. Therefore, if no satisfactory results are obtained, another specific primer for the . primary reaction should be tested. For example, in a preliminary experiment we used YL2 (see Fig. 3B) as the specific primer for the primary reaction of TAILPCR but the success rate was low (data not shown). * We then selected another primer (YLl) for the primary reaction. Mispriming by the specific primer generates unwanted type II products (Fig. 2). Although these are diluted out in subsequent reactions with nested prim-_.ers, these products compete for enzyme and substrates FIG. 7. Library screening using TAIL-PCR products as probes. and can interfere with target product amplification if Before labeling, the residual dNTPs in the PCR reactions were retoo abundant. Mispriming is more common for primers moved by spin dialysis using Suprec-02 spin-filters (Takara). (A) Dot with very GC-rich 3 ends (e.g., Crameri and Stemmer, hybridization to an Arubidopsis Pl clone filter using an insert end 19931, and so we attempt to avoid these. As an addiprobe amplified from the SP6 side of Pl clone 17A6. Each spot on the filter contained DNA pooled from 96 clones (Liu et al., 19951. (B) tional precaution, specific primers should be used at Colony hybridization to a filter prepared from one plate (EGlO) of low concentration (about 0.15-0.2 ,uLM)to reduce misthe Arubidopsis EG YAC library using an insert end probe amplified priming. from the right side of YAC clone EGlOF3. The colony position third To increase the probability of annealing between AD from the bottom and third from the left is EGlOF3. primers and target sequences, we utilized degenerate oligonucleotides. Functional priming may require, in backgrounds. (4) Speed: the successive amplification most cases, at least a 3-base perfect match at the primreactions can all be completed in 1 day. (5) TAIL-PCR ers 3 end (Parker et al., 1991; Parks et al., 1991), and involves no ligation step, a process that risks chimeric this requirement should be met on average every 64 artifacts. (6) Direct sequencing: aliquots of the reaction (43) bases. After meeting this requirement, a 256-fold products are added directly to the sequencing reaction, degenerate 16-mer population will contain a match to streamlining the process from amplification to seany given sequence with an average homology of 57.8% quence determination. Based on the sequencing data, C(3 + 4 + 9 x 25%)/16], assuming random base distribuspecific primers can be prepared for PCR-based library tion. This value is much higher than the 39.1% 1(3 + screening. (7) High sensitivity: although we presented 13 x 25%)/16] expected for a simple 16-mer, and thus only the application of this method to Pl and YAC sysless base-mismatch is required for adaptation priming. tems in this paper, this method has been successfully Examining adapted sites in their targeted gene walkused to recover genomic sequences flanking T-DNA and ing strategy, Parker et al. (1991) observed priming with transposon insertions from Arabidopsis and rice gean average 45% homology between primers and temnomes (Liu et al., unpublished work; N. Fedoroff, plate sequences. Therefore, the combination of both low Washington, pers. comm., 1994; S. Ishiguro, Okazaki, stringency and primer degeneracy in TAIL-PCR should further increase the probability for successful adaptaJapan, pers. comm., 1994; K. Shimamoto, Nara, Japan, tion priming. This high level of promiscuous priming, experiments indipers. comm., 1994). Reconstruction however, does not result in high backgrounds of noncated that single-copy sequences can be amplified from specific products in our TAIL-PCR strategy. The factors hexaploid wheat (data not shown), whose genome size
---

INSERT

END AMPLIFICATION

BY TAIL-PCR

681
library of Arabidopsis which Mol. Gen. Genet. 226: 484-

determining the suitability of an AD primer may include its degeneracy level, length, and nucleotide sequence. In this study we used 128- and 256-fold degenerate arbitrary primers, and satisfactory results were obtained. The degeneracy of the arbitrary primers can be created either through inclusion of multiple bases at one position or through inosine incorporation. For example, AD3 and AD4 used in this study were prepared with inosine for a portion of the degenerate positions, and they can be used at half the concentration required for AD1 and AD2. Overly high degeneracy levels in AD primers may lead to problems in control of priming efficiency, production of undesirably short products, and generation of primer-dimer artifacts. It is important to design AD and specific primers so that no primer dimers are formed even during low-stringency annealing. The TAIL-PCR strategy described here will greatly facilitate isolation of DNA segments flanking known sequences and increase the efficiency of gene isolation by positional cloning. Using this method for rapid generation of insert end probes, we have generated a genomic contig of 37 Pl clones in Arabidopsis (Liu et al., 1995). We have also isolated genomic sequences flanking T-DNA and transposon insertions from transgenic Arabidopsis lines (Liu et al., unpublished). This method is robust and has been successfully used with various plant genomes in several laboratories. The basic utility, simplicity, and power of this method adapts it for wide application in various fields of molecular biology research.
ACKNOWLEDGMENTS We thank N. Fedoroff for helpful comments on the manuscript, H. Kishida of the RIKEN Institute for setting and use of the automated Biomek 1000 laboratory workstation, and N. Hayashida for use of YAC clones. This research is conducted as a part of the Industrial Technology Development Promotion Program of the Research Institute of Innovative Technology for the Earth (RITE) for global environmental problems, supported by the Ministry of International Trade and Industry of Japan. REFERENCES Cramer-i, A., and Stemmer, W. P. C. (1993). 10-Fold aptamer library amplification without gel purification. Nucleic Acids Res. 21: 4410. Frohman, M. A., Dush, M. K., and Martin, G. R. (19881. Rapid production of full-length cDNAs from rare transcripts: Amplification using a single gene-specific oligonucleotide primer. Proc. Nutl. Acad. Sci. USA 85: 8998-9002. Grill, E., and Somerville, C. (1991). Construction and characteriza-

tion of a yeast artificial chromosomes is suitable for chromosome walking. 490.

Isegawa, Y., Sheng, J., Sokawa, Y., Yamanishi, K., Nakagami, O., and Ueda, S. (1992). Selective amplification of cDNA sequence from total RNA by cassette-ligation mediated polymerase chain reaction (PCR): Application to sequencing 6.5 kb genome segment of hantavirus strain B-l. Mol. Cell. Probes 6: 467-475. Liu, Y.-G., Ikeda, T. M., and Tsunewaki, K. (1992). Moderately repeated, dispersed, and highly variable (MRDHV) genomic sequences of common wheat usable for cultivar identification. Theor. Appl. Genet. 84: 535-543. Liu, Y.-G., Mitsukawa, N., and Whittier, R. F. (1993). Rapid sequencing of unpurified PCR products by thermal asymmetric PCR cycle sequencing using unlabeled sequencing primers. Nucleic Acids Res. 21: 3333-3334. Liu, Y.-G., Mitsukawa, N., Vazquez-Tello, A., and Whittier, R. F. (1995). Generation of a high quality Pl library OfArabidopsis suitable for chromosome walking. Plant J., in press. Loh, E. Y., Elliot, J. F., Cwirla, S., Lanier, L., and Davis, M. M. (19891. Polymerase chain reaction with single-sided specificity: Analysis of T cell receptor 6 chain. Science 243: 217-220. Mazars, G.-R., Moyret, C., Jeanteur, P., and Theillet, C.-G. (1991). Direct sequencing by thermal asymmetric PCR. Nucleic Acids Res. 19: 4783. Mueller, P. R., and Wold, B. (1989). In viuo footprinting of a muscle specific enhancer by ligation mediated PCR. Science 246: 780-786. Ochman, H., Gerber, A. S., and Hart& D. L. (1988). Genetic applications of an inverse polymerase chain reaction. Genetics 120: 621623. Ohara, O., Dorit, R. L., and Gilbert, W. (1989). One-sided polymerase chain reaction: The amplification of cDNA. Proc. Natl. Acad. Sci. USA 86: 5673-5677. Parker, J. D., Rabinovitch, P. S., and Burmer, G. C. (19911. Targeted gene walking polymerase chain reaction. Nucleic Acids Res. 19: 3055-3060. Parks, C. L., Chang, L.-S., and Shenk, T. (1991). A polymerase chain reaction mediated by a single primer: Cloning of genomic sequences adjacent to a serotonin receptor protein coding region. Nucleic Acids Res. 19: 7155-7160. Pierce, J. C., Sauer, B., and Sternberg, N. (1992). A positive selection vector for cloning high molecular weight DNA by the bacteriophage Pl system: Improved cloning efficacy. Proc. Natl. Acad. Sci. USA 89: 2056-2060. Riley, J., Butler, R., Ogilvie, D., Finniear, R., Jenner, D., Powell, S., Anand, R., Smith, J. C., and Markham, A. F. (1990). A novel, rapid method for the isolation of terminal sequences from yeast artificial chromosome (YAC) clones. Nucleic Acids Res. 18: 2887-2890. Rosenthal, A., and Jones, D. S. (1990). Genomic walking and sequencing by oligo-cassette mediated polymerase chain reaction. Nucleic Acids Res. 18: 3095-3096. Silver, J., and Keerikatte, V. (1989). Novel use of polymerase chain reaction to amplify cellular DNA adjacent to an integrated provirus. J. Viral. 63: 1924-1928. Triglia, T., Peterson, M. G., and Kemp, D. J. (1988). A procedure for in vitro amplification of DNA segments that lie outside the boundaries of known sequences. Nucleic Acids Res. 16: 8186.