a method for functional mapping of protein–protein binding domain by preferential amplification of...

Analytical Biochemistry 303, 34–41 (2002)

A Method for Functional Mapping of Protein–ProteinBinding Domain by Preferential Amplification of theShortest Amplicon Using PCR

Yasuaki Kawarasaki,1 Yoko Sasaki, Akinori Ikeuchi, Shoji Yamamoto, and Tsuneo YamaneLaboratory of Molecular Biotechnology, Division of Molecular Cell Mechanisms, Department of Biological Mechanismsand Functions, Graduate School of Biological and Agricultural Sciences, Nagoya University,Furo-cho, Chikusa-ku, Nagoya 464-8601, Japan

We have developed a novel method for rapid andempirical mapping of the protein interaction domainusing a unique and atypical PCR-based amplificationand a conventional yeast two-hybrid system. The mod-ified PCR, designated as PASA-PCR, enables preferen-tial amplification of the shortest amplicon from acomplex expression library. PASA-PCR consists of re-iterative cycles of denaturation of template DNAs andextremely abbreviated annealing/extension of primersto prevent their complete extension in a single cycle,followed by conventional amplification cycles. InPASA-PCR, the shortest (ranging from 400 to 1000 bp)amplicon is amplified almost exclusively from tem-plates of various amplicon sizes. In addition, the fre-quency of in vitro recombination can be increasedusing low cooling rates (<0.5°C/s) between the dena-turation and annealing/extension steps, which washelpful in generating precisely trimmed protein-cod-ing regions. Identification of Spc19-binding region ofSpc34, which is a component of yeast’s spindle polebody, was achieved by a combination of the yeast two-hybrid system and PASA-PCR. © 2002 Elsevier Science (USA)

Key Words: preferential amplification of the shortestamplicon; polymerase chain reaction; protein-bindingregion; yeast two-hybrid assay.

Comprehensive and hierarchic research studies onprotein–protein interactions using the yeast two-hy-brid system have been carried out (1, 2). A number offunctionally unclassified proteins have been annotated

1 To whom correspondence and reprint requests should be ad-dressed. Fax: �81-52-789-4145. E-mail: [email protected].
ac.jp.
34

by means of the genome-wide two-hybrid interactionscreening. One of the next important steps in proteininteraction research will be the systematic identifica-tion of domains required for functional interactionsbetween their partners. Since protein interactions of-ten tend to be evolutionary conserved (3), the system-atic identification of binding domains would provide adifferent insight into the functional annotation ofgenes.

Mapping of functional domains or regions, however,always requires either preliminary information on pro-tein structure or conserved amino acid sequences orcumbersome trial-and-error analyses of a series ofnested deletion clones. Moreover, the functional map-ping by these approaches are difficult to apply to, forexample, the systematic identification of protein-bind-ing domains in the genome-scale protein interactionanalyses, where numerous interactions should be ana-lyzed in one, or a few, formatted procedures.

As a first step in establishing such methodology, wenoted that shorter fragments tended to be amplifiedmore preferentially than much longer ones during aconventional polymerase chain reaction (PCR). If theshortest DNA fragment encoding the protein-bindingdomain is amplified exclusively from a large pool of afunctionally assessed repertoire of DNA fragments, itis expected that the amplified fragment carries theinsert DNA critically trimmed into the functional re-gion. The boundaries are then easily mapped by DNAsequencing of the amplified fragment (Fig. 1). This ideais a kind of PCR adaptation of early research of in vitroevolution using Q� RNA replicase, where the shorterduplicating RNA molecules were propagated more ef-ficiently by iterative cycles of in vitro selection and

Received July 18, 2001; published online February 22, 2002

doi:10.1006/abio.2001.5569, available online at http://www.idealibra
om on ry.c
enzymatic amplification (4).

0003-2697/02 $35.00© 2002 Elsevier Science (USA)

All rights reserved.

In conventional PCR, however, the difference in am-plification efficiency between amplicons with similarmolecular sizes is not sufficiently large to achieve theexclusive amplification of the shortest amplicon fromthe fragment library. It is likely due to insufficiency ofselective pressure on short amplicons. In this study, wesuccessfully developed a novel PCR-based method(named preferential amplification of the shortest am-plicon using PCR, PASA-PCR)2 that can amplify theshortest amplicon among a mixture of complex tem-plates by introducing selective pressure on rapid am-plification. In addition, we demonstrated that thismethod is applicable to the systematic determinationof protein-binding domains using the yeast two-hybridsystem.

MATERIALS AND METHODS

Materials

Thermostable DNA polymerases such as ExTaq, PfuTurbo, and KOD were purchased from TaKaRa, Strat-agene, and Toyobo, respectively. The thermocyclers usedin this study were PTC200 (M. J. Research) and Gene-Amp Model 2400 (Perkin–Elmer). All primers used inthis study were designed by using a computer program(http://genome-www2.stanford.edu/cgi-bin/SGD/web-primer).

Methods

Template mixture preparation. The templates usedin the characterization of PASA-PCR were constructedby ligating EcoRV-digested pBluescript II SK� withHincII-digested �x174 phage DNA. After the transfor-mation of Escherichia coli XL1-blue, plasmids withadequate insert sizes were selected. The ampliconsinvolved in the constructed plasmids consisted ofprimer annealing sites (T7 and T3 promoters), multi-cloning sites of the vector that was divided by theinsert, and an insert sequence. The sum of the primerannealing sites and multicloning sites without the in-sert sequence was approximately 160 bp in length.

PASA-PCR. The composition of the reaction mix-ture of PASA-PCR was the same with manufacturer’srecommendation for conventional PCR. A typical reac-tion mixture consisted of 0.01–1 ng/�l of template mix-ture, commercial 1� reaction buffer, and 0.2 mM eachdNTP, 0.5 pmol/�l each primer, and commerciallyavailable DNA polymerase. The sequences of the prim-ers were as follows: T7pr, GTAATACGACTCACTAT-AGGGC; T3pr, GCAATTAACCCTCACTAAAGGGA. Atypical temperature program of PASA-PCR was as fol-lows: 94°C for 2 min, followed by 30 cycles of 94°C for30 s and 58°C for 5 s, and then 10 cycles of 94°C for 30 sand 58°C for 2 min. For the determination of the inter-action region of Spc34, pGAD424 vector-specific-primers (pasa-adf, ATACCCCACCAAACCCAAAAA;pasa-adr, ACGATTCATAGATCTCTGCAGG) wereused instead of T3pr and T7pr. Unless otherwisestated, the cooling rate between denaturation and an-nealing/extension was about 3°C/s, which was the de-fault setting of the thermocyclers.

Preparation of randomly digested DNA fragment li-brary. Five-micrograms of the PCR-amplified DNAfragment encoding yeast SPC34 ORF was incubated ina reaction mixture consisting of 20 mM Tris–HCl (pH8.0), 150 mM NaCl, 1 mM MnCl2, and an appropriateamount of DNase I (Ambion) for 3 min at room tem-perature to give one to two breaks per double-strandedDNA on average. The reaction was stopped by theaddition of an equal amount of phenol/chloroform so-lution. After ethanol precipitation, the digested frag-

2 Abbreviations used: PASA-PCR, preferential amplification of theshortest amplicon using PCR; SPB, spindle pole body.

FIG. 1. An overview of the rapid determination of protein func-tional region of proteins using PASA-PCR. Our proposal for rapidmapping of the protein functional region (in particular, protein-binding domain) is illustrated. (A) Fragment library construction. Anopen reading frame (ORF) for the protein of interest is amplified byconventional PCR, followed by partial random digestion by DNase I.After blunt-ended ligation with an appropriate expression vector,host cells are transformed using the fragment library. After screen-ing the clones exhibiting functionally positive phenotype, thepositive clones are pooled in a tube (B). Step C represents thepreferential amplification of the shortest amplicon. A PCR-basedamplification technique (PASA-PCR) developed in this study enablesof the preferential amplification of the shortest amplicon. If thenumber of the functionally assessed clones is sufficiently large, thefragment amplified by this technique is expected to carry the pre-cisely trimmed functional region (depicted by dark boxes). The arrowin broken line indicates the second round PASA-PCR. In the finalstep, the preferentially amplified clones are recloned into the expres-sion vector, then the boundary sequences are determined by DNAsequencing of the plasmids after the reassessment of the phenotypeof the transformants.

35EXPERIMENTAL DETERMINATION OF PROTEIN INTERACTION DOMAIN USING PCR

ments were treated by pyrobest DNA polymerase(TaKaRa) with 0.2 mM each dNTPs at 72°C for 5 minto form blunt-ended DNA fragments. A portion of theresulting DNA fragments were analyzed by agarose gelelectrophoresis, and it was confirmed that nearly 30%of the fragments were uncut DNA. The blunt-endedDNAs were then ligated with SmaI digested pGAD424and used for E. coli DH5� transformation (namedpGAD-SPC34frag library).

Yeast two-hybrid screening of interacting fragments.The yeast two-hybrid assay was carried out accordingto the manufacturer’s protocol (Clontech). Briefly, Sac-charomyces cerevisiae strain HF7c carrying pGBT-SPC19 was transformed using the pGAD-SPC34frag li-brary and then plated on agar plates of syntheticmedium that lacks leucine, tryptophan, and histidine.The colonies obtained were scraped with a spreaderand pooled in a tube. Plasmids were purified from thispool and used as a template mixture for PASA-PCR.The amplified fragments were recloned in the samevector and then retransformed HF7c that carriespGBT-SPC19 to confirm its phenotype.

RESULTS

Principle and Characteristics of PASA-PCR

We first attempted to increase the differences in theamplification efficiency between amplicons using anequimolar mixture of templates (respective ampliconswere 800–1100 bp in length, see Materials and Meth-ods) of a conventional PCR. A preferable result wasobtained when we employed reiterative two-step tem-perature cycles including an extremely abbreviated an-nealing/extension period (�5 s). In practice, additionalcycles of conventional PCR after the two-step cycleswere required to obtain sufficient yield (data notshown). Any further optimization such as compositionof the reaction mixture was not required. Moreover, asfar as we tested, factors such as primer sequence(50°C � Tm � 56°C), template concentration, and tem-plate sequence did not significantly affect the prefer-ential amplification. We named this amplificationmethod PASA-PCR.

PASA-PCR using Ex Taq DNA polymerase showedhigh efficiency in biased amplifications using severalseries of template mixtures (Fig. 2A, lanes A–F). Inthese lanes, the fragments that were preferentiallyamplified from the shortest amplicon among the tem-plates were clearly detected as a major product. Theintensities of the amplified fragments were similar.For example, the yield of the amplified fragmentfrom the shortest amplicon in lane B (1100 bp) wascomparable to that in lane F (510 bp). This indicatesthat PASA-PCR enables preferential amplification ina broad range (from 500 to more than 1000 bp) with-out any changes in the cycling program. On the other

hand, PASA-PCR using the template mixtureswhich included amplicons of less than 500 bp (lanesG and H) lost its stringency in biased amplification,while some preference for shorter fragments wasobserved.

FIG. 2. Characterization of PASA-PCR. (A) PASA-PCR using ExTaq DNA polymerase. The effective range and resolution of thepreferential amplification of the shortest amplicon was analyzedusing a series of template mixtures. After PASA-PCR, 5 �l of reac-tion mixture was subjected to 1.5% agarose gel electrophoresis, fol-lowed by ethidium bromide staining. Letters on the lanes indicatethe template mixture used for PASA-PCR. The respective ampliconsizes (bp) included in the template mixture were as follows: A,2070 � 1500 � 1200 � 1100 bp; B, A � 1000 bp; C, B � 850 bp; D,C � 730 bp; E, D � 590 bp; F, E � 510 bp; G, F � 390 bp; H, G � 270bp. Lane M indicates the molecular size marker. The numbers on theright side of the gel indicate base pairs. (B) PASA-PCR using Pfu-Turbo DNA polymerase. The reaction condition and template mix-tures were the same as those of (A), except that the volume loaded onthe gel was doubled (10 �l). (C) Effect of DNA polymerase on theefficiency of preferential amplification. Five microliters of the reac-tion mixture was loaded on a gel, and the products were detected asdescribed above. Lane M, molecular size marker. Lanes P, E, and K,DNA polymerase used in PASA-PCR, i.e., Pfu Turbo, Ex Taq, andKOD DNA polymerases, respectively.

36 KAWARASAKI ET AL.

Comparison of the amplification profiles shown inlanes E and F in Fig. 2 indicates that the resolution ofthe preferential amplification of shorter amplicons be-tween two close sizes was nearly 80 bp. The resolutionof PASA-PCR was also confirmed to be 50–100 bpthroughout the effective range by other experimentsusing different sets of templates (data not shown).

Next, we tested the effect of DNA polymerase onpreferential amplification. PASA-PCR using Pfu TurboDNA polymerase, the elongation rate of which wasabout one-half of that of Ex Taq DNA polymerase, alsoshowed the preferential amplification of the shortestamplicon (Fig. 2B). In particular, the efficiency andyield of amplicons smaller than 500 bp was improvedby the use of Pfu Turbo DNA polymerase (lane G). Onthe other hand, PASA-PCR using KOD DNA polymer-ase, whose elongation rate was twice that of Ex TaqDNA polymerase, had very low efficiency in the pref-erential amplification. As an example, comparison ofthe amplification profile of PASA-PCR using KODDNA polymerase with those using other DNA poly-merases is shown in Fig. 2C. The results indicate thatthe elongation rate of DNA polymerase significantlyinfluenced the efficiency of biased amplification. It alsoindicates that the step-by-step extension of primersacross cycles, which resulted in increased differencesin the number of cycles required for each amplicondoubling, was the main driving force for the preferen-tial amplification of the shortest amplicon.

In addition to the increased difference in the numberof cycles, the generation of prematurely extended prim-ers possibly played an important role in preferentialamplification. Circumstantial evidence for the hypoth-esis was obtained when we tested the effect of coolingrate between denaturation and annealing/extension onpreferential amplification (Fig. 3).

As the prematurely extended primers have higherannealing temperature than the unextended ones, weexpected that the slow cooling facilitated the anteced-

ent annealing of the extended primers on their tem-plates, which possibly affected the amplification pro-files. In fact, the disparity in the amplificationefficiency between the shortest amplicon and the sec-ond shortest one increased as the cooling rate de-creased (Fig. 3). In the reactions using the templatemixture G, the shortest amplicon (390 bp in length)was amplified more efficiently than the second shortestone (510 bp) at lower cooling rates (�0.5°C/s).

Facilitation of in Vitro Recombination during PASA-PCR

For a practical use of PASA-PCR, a fragment libraryis prepared by random digestion of ORF with DNase Ifollowed by ligation with an appropriate expressionvector (Fig. 1). The template carrying an insert withthe crucial end points for the N-boundary and theC-boundary of the functional region, however, rarelyappears in the pool. Therefore, a prodigious number ofpositive templates must be input into the pool to coverthe exhaustiveness of the repertoire, which would be aserious drawback.

In vitro recombination between amplicons can cir-cumvent this problem. If there is a template carryingthe N-boundary, and there is another one carrying theC-boundary in the same pool, the sequence that isideally trimmed on both sides is generated by in vitrorecombination. In this case, it will be preferentiallyamplified by PASA-PCR, because the recombinantthus generated should be the shortest amplicon in thereaction mixture.

Many examples of in vitro recombination duringPCR have been reported (5–7). Among them, the man-ner of thermocycling reported in StEP recombination(7), by which efficient recombination between two vari-ants was performed by the “staggered extension pro-cess,” is quite similar to that of PASA-PCR. Therefore,it was expected that the recombination between two

FIG. 3. Effect of cooling rate decrease on the biased amplification profile. A series of cooling rates (°C/s) were applied between the heatdenaturation and abbreviated annealing/extension steps. Reactions were carried out with template mixture F or G (see Fig. 1 caption). Theenzyme used in this experiment was Pfu Turbo DNA polymerase. Numbers on the lanes represent the cooling rate employed in the reaction.M and C, molecular size marker and fragments amplified in conventional PCR with the same template mixture, respectively.


homologous amplicons should arise during PASA-PCRwithout any further modification.

Figure 4 shows evidence for recombination duringPASA-PCR. We used two different templates, KSL1and KSL2 (Fig. 4A), which had an overlapping se-quence (490 bp in length). KSL1 and KSL2 representedthe templates carrying the insert with the C- and N-boundaries, respectively, and the overlapping sequencewas representative of the virtual functional region.

The fragment generated by recombination waspoorly amplified in a single-round PASA-PCR. How-

ever, the recombinant was detected as a major productafter another run of PASA-PCR using the diluted re-action mixture from the previous PASA-PCR as thetemplate solution (Fig. 4B). In addition to its molecularsize (about 580 bp, which corresponded to the predictedsize of the recombinant), the major product of the dou-ble-round PASA-PCR was also confirmed to be therecombination product by DNA sequencing (data notshown).

Figure 4C shows the effect of cooling rate on theefficiency of the recombinant formation/amplification

FIG. 4. In vitro recombination during PASA-PCR. (A) Schematic of in vitro recombination and two constructs used in this experiment. Twooverlapped DNA fragments derived from the firefly luciferase gene were directionally subcloned into pBluescript II KS�. The constructedplasmids, designated as pKSL 1 and 2, carried inserts from positions �186 to �1144 and �651 to �1651 of the luciferase gene, respectively(A of the initiation codon is designated as �1, with positive and negative integers proceeding 3� and 5�, respectively). The names of ampliconsfrom the plasmids are termed as KSL1 and KSL2, respectively. The overlapping region (represented by closed boxes) between the twocorresponded to positions 651–1144, which was 493 bp in length. (B) In vitro recombination between the two amplicons during PASA-PCR.PASA-PCR using Pfu Turbo DNA polymerase with the two plasmid templates was carried out. Lanes M, molecular size marker; lane 1, singleround PASA-PCR with the two templates; lane 2, second round PASA-PCR using the diluted products of the single-round PASA-PCR (lane1) as a template; lane C1, PASA-PCR with pKSL1 alone; lane C2, PASA-PCR with pKSL2 alone. The arrow indicates the recombinantproduct. Numbers on the left of the gel indicate fragment sizes (bp). (C) Effect of cooling rate on recombinant formation/amplification.PASA-PCR employing various cooling rates was carried out using pKSL1 and pKSL2 as templates. The numbers on the lanes represent thecooling rates (°C/s). Lanes C1 and C2 were the same as those in Fig. 3 (left), except for the cooling rate (0.2°C/s). The arrow indicates therecombinant fragment. The open arrowhead represents the heteroduplex products consisting of KSL2 and the recombinant. (D) Effect oftemplate concentration on the recombinant formation/amplification in PASA-PCR with a decrease in cooling rate. The numbers on the lanesindicate concentrations of template mixture (ng each templates in a 50-�l reaction mixture). The reaction condition was the same as that usedin the experiment c, lane 0.2. The arrow indicates the recombinant fragment. The open arrowhead represents the hybrid fragment formedwith KSL2 and the recombinant.


in a single-round PASA-PCR. The decrease in the cool-ing rate to 0.2°C/s caused a marked improvement inthe recombinant formation/amplification, and theproduct was detected as one of the major bands. Inaddition, further stimulation of the recombination wasobserved in the reactions with increased template con-centration (Fig. 4D).

Determination of Spc19-Binding Domain of Spc34

We next applied PASA-PCR to protein–protein inter-action analysis using the yeast two-hybrid system.Much attention is being paid to the construction of agenome-wide protein interaction map, because physi-cal interactions between proteins explain their func-tional relationship in a biological context, which canlead to the systematic annotation of functionally un-known proteins (8, 9).

The spindle pole body (SPB), which is the functionalequivalent of the centrosome in S. cerevisiae, is a com-plex (comprising at least 43 different proteins (10))dynamic organelle that plays an important role in mi-tosis and meiosis. Among the components of SPB, we

are interested in the interaction between SPC34 andSPC19 because a complex interaction network in SPBwas found around their interaction (8). In particular, inaddition to the binding with Spc19, Spc34 also inter-acts with Duo1 or Ydr016c (8), which suggests thatSpc34 likely plays a pivotal role in yeast SPB. How-ever, structural information on Spc34 such as its dis-tinct motives or domain required for the interactionwith other SPB components is yet unavailable. There-fore, we attempted to determine the Spc19-bindingdomain of Spc34 by PASA-PCR.

We constructed a randomly digested Spc34 ORF li-brary on pGAD424 (pGAD-SPC34frag library) and thentransformed S. cerevisiae strain HF7c bearing pGBT-SPC19. A plasmid mixture was prepared from a pool ofapproximately 800 yeast clones that proliferated onselection plates, and then it was used as templatemixture for PASA-PCR. The best amplification profilewas observed when we tested PASA-PCR using PfuTurbo DNA polymerase with the low cooling rate(0.2°C/s). The resulting amplicons (the major band con-verged on 500 bp, Fig. 5A) were reinserted into

FIG. 5. Determination of Spc19-binding domain of Spc34 using yeast two-hybrid assay and PASA-PCR. Clones carrying short fragmentsthat encoded Spc19-binding affinity were amplified by PASA-PCR (A), and the boundary sequences of the amplified products weredetermined by DNA sequencing after reassessment of their interaction with Spc19 (B). (A) Lanes M, P, and C, the size marker, the productamplified by PASA-PCR, and the product amplified by conventional PCR, respectively. The N- or C-boundary of the Spc19-binding regionanalyzed by DNA sequencing of the selected clones was represented by closed arrows with vertical bars on the amino acid sequence of S.cerevisiae Spc34. The comparison of amino acid sequence of S. cerevisiae Spc34 with that of S. exiguus is also shown. A white arrow from D171represents that binding assay between Spc19 and Spc34�N170, which is a Spc34 deletion mutant lacking its N-terminal 170 amino acidresidues. This mutant showed a lower growth rate on the selection plate and a weak �-galactosidase activity.


pGAD424, followed by retransformation of the HF7cthat carries pGBT-SPC19. The insert size of the result-ing clones was the same as that of the one whichconverged by PASA-PCR (data not shown). The bound-ary sequences of several interaction-positive cloneswere determined by DNA sequencing (Fig. 5B).

Interestingly, the termini of the Spc34 fragmentsresponsible for the N-boundaries of the Spc19 bindingregion were located around the sequence which exhib-ited lower similarity with the S. exiguus Spc34 ho-molog. The predicted secondary structure of the Spc34protein showed that the boundaries were locatedaround the putative linker sequence (I148-GSS-VISNSSS-G159, represented by underlined letters inFig. 5). Moreover, strain HF7c (pGBT-SPC19) trans-formed by pGAD-SPC34�N170 (lacking N-terminus1–170 amino acid residues), exhibited a poor growthrate on the selection plate and had an undetectable�-galactosidase activity (represented by a white arrowin Fig. 5), suggesting the decreased binding efficiencyby further deletion. These data suggested that theSpc19-binding domain of Spc34 was assigned to beS155-Q291.

DISCUSSION

We showed the feasibility of the preferential ampli-fication of the shortest amplicon by the PCR-basedmethod and its applicability to the rapid mapping ofprotein-binding region. It was demonstrated thatPASA-PCR efficiently amplifies the shortest ampliconwith high reproducibility, if the template mixture iscomposed of amplicons that are more than 400–500 bp(Figs. 2 and 3). We believe that the low size limitationis sufficient for mapping most of the protein domains.Since the amplicons shown in Fig. 2 carried additivesequences derived from multicloning sites and primerannealing sites, the actual size of insert in the ampli-con was about 240 bp (see Materials and Methods). Itcorresponds to a polypeptide of 80 amino acid residues,which can span a large numbers of protein domains(�100–300 amino acids).

The limit of the effective range in PASA-PCR wasaffected by the elongation rate of DNA polymerase(Fig. 2C) and the cooling rate between the denatur-ation and annealing/extension steps (Fig. 3). Thesedata suggest that the “step-by-step primer extension”across cycles caused by the abbreviated annealing/ex-tension (�5 s) plays an essential role in preferentialamplification. This amplification technique increasesthe difference in the number of cycles required for eachamplicon doubling, which results in highly biased am-plification during reiterative thermocycling.

In addition, the data shown in Fig. 3 suggest that theprematurely extended primers generated by the abbre-viated extension also affect the efficiency of preferen-

tial amplification. In this experiment, the decreasedcooling rate that enhanced antecedent annealing of theprematurely extended primers clearly showed im-proved preferential amplification. Since the prema-turely extended primers have higher annealing tem-perature than the unextended ones, they tend tooccupy their templates during the annealing/extensionsteps. This would cause a decrease in effective tem-plate concentration and a decrease in priming fre-quency of the unextended primers. Then the prema-ture product extending on the shortest ampliconswould likely anneal to the opposite extending strandearlier than those extending on longer templates,which supplies an available template for the unex-tended primers. As a result, the effective concentrationof the template that is available for the subsequentpriming increases, which likely leads to biased ampli-fication.

In vitro recombination between overlapping se-quences is another essential factor in achieving thesystematic determination of protein functional regions.Figure 4 indicates that the prematurely extendedprimers likely participate in recombinant generationas well as in efficient preferential amplification. This isbecause the amount of available templates in the sub-sequent annealing/extension steps is different for eachextending primer. For example, in Fig. 4A, the T7pprimer extended on KSL1 is not easily annealed toKSL2 in the subsequent cycles, which indicates a de-crease in the template concentration for the extendedprimer. On the other hand, for the one extended onKSL2, either KSL2 or KSL1 is available in the subse-quent cycles, which is an advantage over the T7pprimer extending on KSL1. As a result, generation ofprecisely “trimmed” amplicons was facilitated by invitro recombination during PASA-PCR.

As far as we tested, the positive effect of the lowercooling rate (�0.5°C/s) on the preferential amplifica-tion was significantly observed in PASA-PCR using PfuTurbo DNA polymerase rather than that using Ex TaqDNA polymerase. In fact, as shown in Fig. 5A, PASA-PCR using Pfu Turbo DNA polymerase with the lowcooling rate facilitated the convergence of amplicons.Since the lower cooling rate in PASA-PCR using ExTaq DNA polymerase resulted in little or no facilitationin the preferential amplification/recombinant genera-tion in many cases (data not shown), the effect of thelower cooling rate was likely related with the elonga-tion rate of DNA polymerase.

The efficiency of recombinant generation betweensimilar sequences should be also discussed. Somegenes contain repetitive sequences in their protein-coding region. When PASA-PCR is applied to the am-plification of such genes, it might yield a short, artifac-tual PCR product(s) that lacks the region responsiblefor their function, due to unfavorable recombination.


This could be a critical drawback in functional domainmapping using PASA-PCR. The frequency of such un-favorable recombination is, however, quite low in prac-tice, for the following reason. To generate recombinantamplicon, the DNA sequences of such repetitive unitsmust be almost identical and must have specific lengthto allow the stable annealing of prematurely extendedprimers. It should be noted that the remaining part ofprematurely extended primers that is not involved inthe annealing to other repetitive units likely preventssuch unfavorable recombination by anchoring itselfonto the original annealing site. In fact, recombinantformation between tandem tethered genes (encodinggreen fluorescent protein and EYFP (Clontech); 72% ofsequence similarity) was not detected under the typicalconditions of PASA-PCR (Ikeuchi et al., unpublisheddata). This data strongly suggest that unfavorable re-combination between similar sequences negligibly oc-curs during PASA-PCR.

We successfully determined the Spc19-binding re-gion of Spc34 using the yeast two-hybrid system andPASA-PCR (Fig. 5). The Spc19-binding region of Spc34was determined to be the latter half of its protein-coding region. The N-boundary of the Spc19-bindingregion was mapped around the putative linker se-quence, supporting that this empirically determinedboundary was plausible. The weaker binding activity ofthe N-deleted clone, pGAD-SPC34�N170, stronglysuggests the importance of the putative linker se-quence in the intrinsic conformation of the Spc19-bind-ing domain.

In the two-hybrid system, yeast clones expressingtwo fused proteins with a weaker binding affinity oftenproliferate at lower growth rate on the selection plate.Since the plasmid mixture of the binding positive di-gests of Spc34 was prepared from the pool of yeastcolonies proliferated on the selection plate, we probablymissed clones carrying Spc34 fragments of muchweaker Spc19-binding affinity. In fact, although indis-tinguishable �-galactosidase activity was detected, theclone harboring pGAD-SPC34�N170 grew on the se-lection plate, suggesting weak interaction betweenSPC19 and SPC34�N170. Therefore, the classificationof the growth rate of clones before the plasmid mixturepreparation is possibly useful for the intensified deter-mination of the region responsible for such weak inter-action.

Nonetheless, the results shown in Fig. 5 indicatethat this method is applicable to the systematic anal-

ysis of protein-binding regions. In fact, the proceduredeveloped in this study requires no preliminary infor-mation about the conserved motives or three-dimen-sional structures of the target proteins. In addition,this determination of binding-domains is basically car-ried out in a formatted procedure, which allows high-throughput determination of binding-domains in ge-nome-wide scale. Moreover, it should be emphasizedthat, aside from the rapid mapping of protein-bindingdomain, PASA-PCR can be applied for other types offunctional (e.g., enzymatic activities) domain mapping,if an efficient system for functional screening of a frag-ment library is available.

REFERENCES

1. Fromont-Racine, M., Rain, J. C., and Legrain, P. (1997) Towarda functional analysis of the yeast genome through exhaustivetwo-hybrid screens. Nature Genet. 16, 277–282.

2. Hua, S-b., Luo, Y., Qiu, M., Chan, E., Zhou, H., and Zhu, L.(1998) Construction of a modular yeast two-hybrid cDNA libraryfrom human EST clones for the human protein linkage map.Gene 215, 143–152.

3. Pawson, A. J. (Ed.) (1998) Protein Modules in Signal Transduc-tion, Springer Verlag, Berlin.

4. Mills, D. R., Peterson, R. L., and Spiegelman, S. (1967) Anextracellular Darwinian experiment with a self-duplicating nu-cleic acid molecule. Proc. Natl. Acad. Sci. USA 58, 217–224.

5. Meyerhans, A., Vartanian, J-P., and Wain-Hobson S. (1990)DNA recombination during PCR. Nucleic Acids Res. 18, 1687–1691.

6. Judo, M. S. B., Wedel, A. B., and Wilson, C. (1998) Stimulationand suppression of PCR-mediated recombination. Nucleic AcidsRes. 26, 1819–1825.

7. Zhao, H., Giver, L., Shao, Z., Affholter, J. A., and Arnold, F. H.(1998) Molecular evolution by staggered extension process(StEP) in vitro recombination. Nature Biotechnol. 16, 258–261.

8. Ito, T., Tashiro, K., Muta, S., Ozawa, R., Chiba, T., Nishizawa,M., Yamamoto, K., Kuhara, S., and Sakaki, S. (2000) Towardprotein–protein interaction map of the budding yeast: A compre-hensive system to examine two-hybrid interactions in all possi-ble combinations between the yeast proteins. Proc. Natl. Acad.Sci. USA 97, 1143–1147

9. Uetz, P., Giot, L., Cagney, G., Mansfield, T. A., Judson, R. S.,Knight, J. R., Lockshon D., Narayan, V., Srinivasan, M., Pochart,P., Qureshi-Emili, A., Li, Y., Godwin, B., Conover, D., Kalb-fleisch, T., Vijayadamodar, G., Yang, M., Johnston, M., Fields, S.,and Rothberg, J. M. (2000) A comprehensive analysis of protein-protein interactions in Saccharomyces cerevisiae. Nature 403,623–627.

10. Wigge, P. A., Jensen, O. N., Holmes, S., Soues, S., Mann, M., andKilmartin, J. V. (1997) Analysis of the Saccharomyces spindlepole by matrix-assisted laser desorption/ionization (MALDI)mass spectrometry. J. Cell. Biol. 141, 967–977.


a method for functional mapping of protein–protein binding domain by preferential amplification of...

Documents