Proc. NatL Acad. Sci. USA Vol. 79, pp. 1588-1592, March 1982

Gap misrepair : Efficient site-directed induction of transition, , and frameshift in vitro (site-specific mutagenesis/base substitutions/DNA polymerase/fidelity of DNA synthesis/thiophosphate ) DAVID SHORTLE*, PAULA GRISAFI*, STEPHEN J. BENKOVICt, AND DAVID BOTSTEIN* *Department of Biology, Massachusetts Institute ofTechnology, Cambridge, Massachusetts 02139; and tDepartment of Chemistry, The Pennsylvania State University, University Park, Pennsylvania 16802 Contributed by David Botstein, November 13, 1981

ABSTRACT Short single-stranded gaps can be constructed by tagenesis to within afewbase pairs, the degree ofsite-specificity limited exonuclease action at single-stranded breaks (nicks) placed is determined by which ofseveral endonucleolytic reactions are at predetermined sites on closed circular DNA molecules. As ef- used to generate the nick (5-11). ficient primer-templates for DNA polymerase, single-stranded gaps can be repaired in vitro to regenerate an intact DNA duplex. MATERIALS AND METHODS In this report two in vitro reaction schemes are described that produce a high frequency oferrors during repair ("misrepair") of Strains. The Escherichia coli strains HB101 (thr- leu- pro- gaps and thereby allow the efficient recovery ofmutations limited recA- thi- hsdR- hsdMf supE-), BD-1528 (F' met- hsdR- to the sequence at or near the original gap. In the first hsdM+ supE supF ungi-1 nadB7; obtained from B. Duncan), and of these misrepair schemes, nucleotide misincorporations are DB4906 (thr- leu- thi- supE lacY tonA ara- hsdR- hsdMf stimulated by omission of one of the four deoxynucleoside tri- uvrD::Tn5) were used as recipients for transformation with the phosphates; the misincorporations are trapped by the presence of small plasmid pBR322. excess DNA ligase in the reaction mixture. The second misrepair Materials. Micrococcus luteus DNA polymerase I was ob- scheme involves the misincorporation of an excision-resistant a- tained from Miles; restriction endonuclease Cla I was obtained thiophosphate nucleotide, followed by gap filling in the presence from Boehringer Mannheim; all other enzymes were obtained ofall four conventional deoxynucleoside triphosphates. When ap- from New England BioLabs. Units ofenzyme activity are those plied to short gaps constructed at one ofseveral unique restriction ofthe manufacturer. Deoxynucleoside triphosphates were pur- sites on the small plasmid 'pBR322, both gap misrepair methods chased from Schwartz/Mann. The Sp diastereomers of thymi- yielded mutations within the targeted restriction site at high fre- dine 5'-O-(l-thiotriphosphate) (dTTP[aS]) and 2'-deoxyaden- quency (6-42%). A majority of the sequence changes identified osine were prepared were base substitutions; and transitions are approx- 5'-O-(1-thiotriphosphate) (dATP[aS]) imately equally represented. The remaining sequence changes according to Bryant and Benkovic (12). were an of a single base pair and deletions ofone to four Restriction Enzyme Nicking Reactions. Covalently closed base pairs. circular pBR322 DNA (13) was nicked with restriction endo- nucleases HindIII, Cla I, or BamHI by incubating 10 pkg of In the past few years, several methods of site-specific muta- plasmid DNA in a 100-Al solution of20 mM Tris'HCl, pH 7.8/ genesis have been developed to facilitate recovery ofmutations 7 mM MgClJ7 mM 2-mercaptoethanol/gelatin (100 Ag/ml) in genes carried on small circular DNA molecules (1-3). Per- and -a concentration of ethidium bromide (150 ,ug/ml, 75 pig/ haps the most direct of these methods involves the use of syn- ml, or 100 Ag/ml, respectively) determined by titration to give thetic oligonucleotides as primers for DNA synthesis after an- an optimal level of nicking. An amount of restriction endonu- nealing to a single-stranded' circle of wild-type DNA (2, 4). clease was added sufficient to convert ==50-90% of the input Because the nucleotide sequence of the synthetic oligonucleo- DNA to an open circular form on incubation at room temper- tide replaces the homologous wild-type sequence, in principle ature for 2-4 hr. The nicking reaction with the EcoRI enzyme any -base substitution, insertion, or -can consisted of 100 mM Tris'HCl, pH 7.6/50 mM- NaCV5 mM be induced at any site via synthesis of the appropriate oligo- MgCl2/gelatin (100 ug/ml)/ethidium bromide (150 ,ug/ml). nucleotide. However, when the exact nucleotide sequence Reactions were stopped by addition of excess EDTA followed change necessary to create a mutant allele with a particular phe- by phenol extraction and ethanol precipitation. notype is unknown, synthesis of the many different oligonu- Gapping Reaction and Purification of Gapped-Circular cleotides required to cover all possibilities becomes inefficient DNA. To convert restriction enzyme-induced nicks into short and laborious. In such situations, the optimal strategy for mu- single-stranded gaps, 5 ug of nicked pBR322 was added to 25 tagenesis is one that induces a variety of point mutations scat- 1.d of70 mM Tris-HCl, pH 8.0/7 mM MgCl2/1 mM 2-mercap- tered over a specificed region of the gene. toethanol containing 0.5 unit of M. luteus DNA polymerase I Methods have been developed (5-7) that allow the direction per Ag of DNA (10). After incubation at room temperature for of mutagenesis to unique nucleotide positions or to sites dis- 60 min, the reaction was stopped with excess EDTA, phenol tributed within a defined segment of DNA. These methods are extraction, and ethanol precipitation. From this mixture of based on the introduction ofa single-stranded break (nick) into DNA forms, circular DNA with single-stranded gaps was se- a DNA molecule, followed by enlargement to a small gap that lectively purified by first ligating linear and nicked-circular can serve as a specific target for an in vitro mutagenic reaction. molecules to a relaxed, covalently closed form in a 50-.ul re- Because the position of the nick determines the region of mu- action consisting of 66 mM Tris HCl, pH 7.6/6.6 mM M'gClJ 5 mM dithiothreitol/gelatin (100 ug/ml)/0.5 mM ATP/150 The publication costs ofthis article were defrayed in part by page charge units of T4 DNA ligase; the reaction was incubated overnight payment. This article must therefore be hereby marked "advertise- at 00C. After phenol extraction and ethanol precipitation, this ment" in accordance with 18 U. S. C. §1734 solely to indicate this fact. DNA mixture was dissolved in 0.2 ml of 10 mM sodium citrate, 1588 Downloaded by guest on September 28, 2021 Genetics: Shortle et al. Proc. Natl. Acad. Sci. USA 79 (1982) 1589

pH 6.0/0.5 mM EDTA/0.2 M NaCVethidium bromide (2 ,ug/ 1.Endonuclease ml) and passed over a 0.2-ml column of acridine yellow ED 2. Exonuclease beads (Boehringer Mannheim). After washing the column-first 5'-- A-T-C-G-A-T -- 3' with the loading solution and then with 10 mM sodium citrate, 3'-- T-A-G-C-T-A --5' pH 6.0/0.5 mM EDTA/0.2 M NaCl without ethidium bro- A-T P04 mide-the open-circular (i.e., gapped) DNA fraction was eluted T-A-G-C-T-A with the same buffer containing 0.5 M NaCl and was concen- Misincorporation Excision trated by ethanol precipitation in the presence of carrier tRNA. of dNTP t 1Ecso Gap Misrepair: Method 1. A typical reaction consisted of 100-200 ng ofgapped pBR322 DNA in 20 ,ul of60 mM Tris HCl, _ A-TN Po= pH 8.0/20 mM 2-mercaptoethanol/1 mM MgAc2/2 mM T-A-G -C-T-A MnCl2/gelatin (100 tWg/ml)/0.5 mM ATP and three deoxynu- Gap filling+ cleoside triphosphates at a concentration of 125 ,uM each. To ligation this mixture was added 150 units ofT4 DNA ligase plus 0.2 unit A-TNG -A-T ofM. luteus DNA polymerase I, and incubation was carried out T-AG-C -T-A at 26°C for 16-18 hr. The percentage ofmolecules that had been Replication /A Ah converted to a covalently closed form was assayed by electro- phoresis ofan aliquot on a 1.2% agarose gel containing ethidium -A-T-N-G-A-T- - A-T-C-G-A-T- bromide at 0.5 ,ug/ml in both the gel and the running buffer. -T- A-N'- C-T-A - T-A-G-C-T-A- runs as a On such gels, circular DNA closed by ligation faintly MUTANT WILD - TYPE fluorescent band with a mobility slightly greater than that of ( ClaI resistant) (ClaI sensitive) negatively supercoiled pBR322. Gap Misrepair: Method 2. In the first reaction, an a-thio- FIG. 1. Outline of steps used to induce point mutations within a was onto the unique restriction site (Cla I is used as an example) by gap misrepair. phosphate deoxyribonucleotide misincorporated In method 1, misincorporation of a dNTP and gap filling plus ligation 3' terminus of a single-stranded gap in a 15-p1 reaction con- all occur in a single reaction mixture; in method 2, a single a-thio- sisting of 130 mM Hepes, pH 7.5/0.2 mM MnCl2/2 mM 2- phosphate nucleotide is misincorporated (essentially irreversibly) in mercaptoethanol/gelatin (100 ,ug/ml)/100 uM dTTP[aS] (or one reaction, followed by gap filling and ligation in a second reaction. dATP[aS])/50-100 ng of DNA. In some reactions, either dTTP or dGTP was also present at a concentration of 20 ,M. After ditions used, an average offive or six nucleotides were removed addition of0.3 unit ofKlenow fragment polymerase, incubation from the nicked strand, predominantly in the 5' -- 3' direction. was carried out at 26°C for 14-16 hr. The reaction was stopped Circular DNA gapped at a particular restriction site was purified by making the mixture 10 mM in EDTA, followed by phenol and used as substrate for an in vitro "gap misrepair" reaction extraction and ethanol precipitation. The precipitate was dis- in which the short stretch ofDNA missing at the single-stranded solved in a few microliters of 2 mM Tris HCl, pH 8.0/0.2 mM gap was synthesized by DNA polymerase under conditions that EDTA. To fill in the remainder of the single-stranded gap, the promote a high frequency of nucleotide misincorporation. same M. luteus DNA polymerase I reaction used in method Because the bacterial DNA polymerases used for repair syn- 1-as described above except for the addition of all four deoxy- thesis possess a 3' -- 5' "editing" exonuclease, the occasional nucleoside triphosphates-was carried out. DNA recovered misincorporated nucleotide will be efficiently excised. To elim- from this reaction was used directly for transformation. inate this undesired competing reaction, two different ap- Recovery and Analysis of Mutant Plasmids. Several nano- proaches were taken. In method 1 (gap misrepair via nucleotide grams of mutagenized pBR322 were used to transform E. coli omission), misincorporation is promoted by omitting one ofthe strain HB101 by the protocol ofDagert and Ehrlich (14). Trans- four deoxynucleoside triphosphates from the reaction mixture formants were selected on LB agar plates containing ampicillin (17) and by adding manganese ion (18) plus T4 DNA ligase and at 250 ,ug/ml. To allow for segregation ofmutant plasmids from ATP. Under these conditions, only noncomplementary nucleo- wild-type, 24-48 independent colonies were picked and re- tides are available for incorporation at those positions where the streaked twice. A small amount of plasmid DNA was prepared omitted nucleotide would otherwise be correctly incorporated. from each isolate by a modification ofthe method ofHolmes and Infrequently, DNA synthesis proceeds past such a point by Quigley (15) and was digested with the cognate restriction en- misincorporation; if the remainder of the single-stranded gap donuclease for the site that had been mutagenized. In most is filled in and the strand immediately closed by DNA ligase, experiments, plasmid DNA was also prepared from a pool of the misincorporated nucleotide will be trapped in a nonexcis- 1000 or more transformed cells. The percentage of this DNA able state. In method 2 (gap misrepair via nonexcisable nu- pool that had lost the restriction site was estimated by com- cleotides), an "excision-resistant" a-thiophosphate analogue of parison of the circular pBR322 band intensity before and after a deoxynucleoside triphosphate is misincorporated onto the 3' exhaustive restriction enzyme digestion. For several mutant terminus of a single-stranded gap as a first step. Because these plasmids identified among the independent isolates, a large nucleotide analogues are efficient substrates for DNA poly- preparation of DNA was made and the nucleotide sequence merases (19, 20) but are resistant to 3' -> 5' exonuclease activity surrounding the lost restriction site was determined (16). once incorporated into a DNA strand (ref. 21; unpublished re- sults), their misincorporation onto a 3' terminus becomes ef- RESULTS fectively irreversible. As a consequence, misincorporation The basic steps of the gap misrepair mutagenesis methods de- events accumulate over time. In a second reaction, the re- scribed above are illustrated in Fig. 1. A short single-stranded mainder ofthe single-stranded gap is filled in by DNA synthesis gap was generated at a unique restriction site on the small plas- from this stable, mismatched 3' terminus, and the strand is mid pBR322 in two steps. First, the DNA was nicked with the sealed by DNA ligase. restriction endonuclease in the presence of ethidium bromide With both methods, the product of a completed reaction is (6, 8) and then a limited gap was formed by exonucleolytic diges- a covalently closed circular DNA with one or more mismatched tion with DNA polymerase I from M. luteus (8). Under the con- base pairs at the site of the original single-stranded gap. On Downloaded by guest on September 28, 2021 1590 Genetics: Shortle et al. Proc. Natl. Acad. Sci. USA 79 (1982) transformation of such DNA into an appropriate host cell, the The nucleotide sequence surrounding the lost restriction site misincorporated nucleotide(s) on one strand should segregate was determined by the method of Maxam and Gilbert (16) for from the wild-type sequence on the other strand at the first several mutants from each experiment. As shown in Fig. 2, most round ofreplication (or by random mismatch repair), resulting of the mutants induced by misrepair at the HindIII, Cla I, and in a fixed base substitution mutation in halfofthe progeny mol- BamHI sites have each acquired a single base substitution at ecules. For method 1, the expected result is that base substi- the position expected based on the nucleotide omission-i.e., tutions should only occur at positions in the single-stranded gap the first nucleotide position beyond the restriction enzyme-in- where the omitted nucleotide would otherwise have been cor- duced nick in one of the two strands. One exception is H103, rectly incorporated, the actual base change being determined where the mutation is shifted one position away from a nick. by the nucleotide misincorporated in its place. For method 2, However, four ofthe mutant plasmids contain from one to three base substitutions are expected at positions where the a-thio- additional base substitutions in the nucleotide sequences flank- phosphate nucleotide has been misincorporated-i.e., the po- ing the mutant restriction site. Every base substitution can be sition of the 3' terminus when misincorporation occurred. In explained as resulting from a misincorporation event at positions this case, although the mutational site may be somewhat in- where the omitted nucleotide could have been correctly incor- determinate due to movement of the 3' terminus by the 3' porated during 5' -* 3' DNA synthesis from a restriction en- -*5' exonucleolytic activity of DNA polymerase, the mutation zyme-induced nick. For this reason, we presume that the mu- should be a base substitution determined by the a-thiophos- tations located outside the restriction site arose by limited nick phate nucleotide used. translation occurring prior to strand closure by DNA ligase or Method 1 (Gap Misrepair via Nucleotide Omission). In a test by misrepair of a gap longer than six nucleotides. experiment, purified pBR322 containing a single-stranded gap Unlike the situation at the HindIII, Cla I, or BamHI sites, at the HindIII site was used as substrate for repair synthesis in DNA polymerase proceeding 5' -* 3' from a nick in the EcoRI the presence of excess DNA ligase (as described in Materials site encounters two adjacent positions where the omitted nu- and Methods) with either all four nucleotide triphosphates cleotide dATP would otherwise be incorporated. For DNA syn- added or with dATP alone omitted. For both reactions, the thesis to advance beyond this point in the absence ofdATP re- extent ofgap filling as assayed by conversion ofthe input DNA quires two consecutive misincorporations. Such double aberrant to aclosed circular form was 90% or greater. This DNAwas used events would be expected to occur at significantly lower fre- to transform E. coli strain HB101 to ampicillin resistance. Of quencies than single misincorporations. The frequency with 40 independent transformants with DNA repaired in the re- which EcoRI site mutants were recovered from DNA misre- action with all four dNTPs present, none yielded plasmid DNA paired in the absence of dATP was lower (6%) than that at the resistant to cleavage by the HindIII endonuclease. However, HindIII site (14%) where only a single misincorporation is re- 5 of 36 transformants (14%) with DNA repaired in the absence quired. The nucleotide sequence changes in three EcoRI site of dATP contained mutant plasmids that had lost the HindIII mutants shown in Fig. 2 occur only at the second of the two A site. positions. To reach this position starting from a 3' terminus at Similar experiments were carried out at three other unique a nick, DNA polymerase must have incorporated an A residue restriction sites (Cla I, BamHI, and EcoRI) on the plasmid at the first position, either as dATP contaminating one of the pBR322. In each experiment, the deoxynucleoside triphosphate other three dNTPs or as rATP (22) which is present in the re- omitted from the misrepair reaction represented the first nu- action as cofactor for T4 DNA ligase. This result suggests that, cleotide that would normally be incorporated at the 3' hydroxyl at sites in DNA where one type of base pair occurs in runs of end at restriction enzyme-induced nicks-that is, for pBR322 two or more, all nucleotide positions might not be readily mut- DNA molecules gapped at the Cla I site, dCTP was omitted. For molecules gapped at the BamHI or EcoRI sites, misrepair Hindl SITE (-dATP) ClgI SITE (-dCTP) was carried out in the absence ofdGTP or dATP, respectively. wt AAGCTTTAA wt G TTTGACAGCTTATCATCGATAAGC than 50% T TCGAAAT T C AAACTGTCGAATAGTAGCTATTCG In each ofthese three misrepair reactions, greater of I- - G- 101 ------the input DNA was converted to a covalently closed form. After H101 ------A-T-- -0- ____-_- - transformation with DNA misrepaired at the Cla I site and sin- H102 - ---C---- C102,103-A- gle colony isolation, 12 of 36 transformants (33%) contained - - --G------T------H103 G- C104 --T------G mutant plasmids lacking the Cla I site. Of36 transformants ob- - tained with DNA misrepaired at the BamHI site, 5 (14%) con- C H104 - C - C105 C---A --- A ------C ------tained BamHI site mutants, whereas only 3 EcoRI site mutants -G- G---T -- - T ------G------were identified among 48 transformants (6%) with DNA mu- H105 T - - - -GG C106 T---C---A------A------A - - -G - - - T ------T ------tagenized at this site. -A - - --CC Control DNA polymerase reactions with all four dNTPs in- cluded were run in parallel with each of the three misrepair BomHI SITE(-dGTP) E Qc RI SITE (-dATP) wt GIGATCC wt GIAATTC reactions, and a pool of transformants (>1000) obtained with CCTAGG C TTAAOG this "correctly" repaired DNA was screened en masse to de- - - B 101 - E 10 ---A termine the frequency ofrestriction-site mutants. In each case, ----T----A - - --T- - no restriction enzyme-resistant DNA was detected, whereas B102,103 - -- -A- E 102 ---C - - G - - similar screening of the corresponding misrepaired DNA re- - - - - T - vealed a percentage of restriction site mutant DNA in close E 103 ---l agreement with the values obtained from screening 36-48 transformants individually. Therefore, most, if not all, of the FIG. 2. Results with method 1 (gap misrepair via nucleotide omis- recovered restriction site mutants were induced as a result of sion): restriction site mutants induced in plasmid pBR322 with M. Iu- teus DNA polymerase I in a reaction from which the nucleotide shown omitting one dNTP from the polymerase reaction. Because the in parentheses was omitted. The positions of restriction endonuclease DNA undergoing mutagenesis at a restriction site represents cleavage are shown by arrows; a dashed line indicates the wild-type molecules that had been nicked by the cognate restriction en- base pair was present in the mutant; a black rectangle indicates dele- zyme, preexisting mutations were, ifanything, selected against. tion of the base pair. Downloaded by guest on September 28, 2021 Genetics: Shortle et al. Proc. NatL Acad. Sci. USA 79 (1982) 1591

able by using gap misrepair via nucleotide omission. HindM SITE Cla I SITE EcoRI SITE The pattern ofbase alterations from the nucleotide omission 1 2 34 5 6 23456 123456 A G C T T ATC G AT GiAATTC gap misrepair experiments can be summarized by noting that T T C G A1A TA GCITA CT TA Af the 17 sequenced mutant plasmids contained 26 single-base 1'2'3'4' 5'6' V' 2'3'4'5'6' 1' 2' 3'4'5' 6' changes, with transversions as frequent as transitions (12 vs. 13); dTTP- ocS H301 T----- C301 *- E301 - - --A- one deletion mutation was found. Of the 10 mutational events (H304-H310) A------T - induced by DNA synthesis in the absence of dATP, G substi- C302 ---A-- E302 - - - - -A H302 -T I--- _ - - - tution occurred six times, T substitution occurred twice, C sub- -HA--i-6 - - - T - - T stitution occurred once, and deletion of single AT base pair H303 1 1 C303 --T - - - E 304 - T - - - - occurred once. The 13 mutation events in the absence ofdCTP - - A -- - - A - - - - C304 -A were all base substitutions-6 Ts, 4 Gs, and 3 As; in the absence -T of dGTP there were 2 substitutions of T and 1 of A. dATP-XS Method 2 (Gap Misrepair via Incorporation ofNonexcisable H201 -- T--- C201 --T --- E202 -T---- Nucleotides). This method consists oftwo sequential reactions: - - A - - - - -A-- - - A-- - - H202,203 -- - G -- C202 --lii E203 ---A-- misincorporation of an a-thiophosphate nucleotide onto the 3' - - - C - - -204 - - - - -T- - terminus of a single-stranded gap, followed by gap filling and H204 - - - A - - E204 - - T--- ligation. These two reactions were carried out with dTTP[aS] - - - T - - - - A- - - by using as substrate pBR322 DNA gapped at the HindIII site, E201 - - -1 -- v the Cla I site, or the EcoRI site. After the second reaction, the A plasmid DNA was used to transform E. coli strain HB101 to FIG. 3. Results with method 2 (gap misrepair via nonexcisable ampicillin-resistance, and 24 independent transformants were nucleotide): restriction site mutants induced in plasmid pBR322 by screened for mutant plasmid DNA with the cognate restriction misincorporation of an a-thiophosphate nucleotide in a two-step re- enzyme. Three HindIll-resistant mutants (13%), ten Cla I-re- action series. A dashed line indicates the wild-type base pair was pres- sistant mutants (42%), and five EcoRI-resistant mutants (20%) ent in the mutant; a black rectangle indicates deletion of the corre- spondingbase pair; a caret indicates insertion of abase pair. *Mutants were identified. H202, H203, and C201 were induced by gap misrepair reactions with The nucleotide sequence changes are shown in the top part dATP[aS] to which dGTP (for H202 and H203) or dTTP (for C201) had ofFig. 3. Ofthe 10 mutants sequenced, 7 contained single-base also been added (see text). Presumably, these mutations resulted from substitutions consistent with the misincorporation ofaT residue misincorporation of the conventional nucleotide. onto the 3' terminus of a gap generated from a restriction en- zyme-induced nick on one of the two strands. In the case of dNTP to the misincorporation reaction. For instance, when mutants H301, C304, and E302, the 3' terminus was presum- pBR322 gapped at the HindII1 site was the substrate for mis- ably moved one or two positions in the 3' -> 5' direction from incorporation mutagenesis with dATP[aS], dGTP was used. In the nick by the editing exonuclease of DNA polymerase prior this case, DNA polymerase can correctly incorporate A and G, to the misincorporation event. The three exceptional mu- advancing the 3' terminus to positions labeled 3 or 4' in Fig. tants-H302, H303, and C301-have acquired deletions ofone 3. Any subsequent misincorporation wouldbe expected to occur or two base pairs. only at the next position after the terminal G-i.e., positions To explain the origin of these deletions at sites where mis- 4 or 3'. To the misincorporation reaction with pBR322 gapped incorporation might have been expected, two types of mecha- at the Cla I site as substrate, dTTP was added to block the 3' nisms can be invoked. Either nucleotide positions in the tem- -* 5' exonuclease activity (27), thus directing misincorporation plate strands are skipped over by an aberrant event during the to positions labeled 3 and 4' in Fig. 3. Finally, dGTP was added in vitro polymerase reactions, or deletions are formed in vivo for pBR322 gapped at the EcoRI site, assuring that the poly- after transformation by an abortive attempt at excision repair, merase could advance to positions 3 and 4' and directing mis- possibly as a consequence of the fact that thiophosphate-con- incorporation to positions 4 and 3'. taining phosphodiester linkages are resistant to a variety ofnu- On screening plasmid DNA from 24 individual HB101 trans- cleases (21, 23-25). With regard to the second mechanism, it formants obtained with mutagenized DNA from each of these seemed significant that the host cell for DNA transformation three misrepair reaction experiments, six HindIlI site mutants, in these experiments was a recA- strain; therefore, DNA lesions four Cla I site mutants, and six EcoRI site mutants were iden- generated by incomplete excision, such as apyrimidinic sites, tified. Four mutants from each class were chosen for nucleotide might not be reparable via recombination with the wild-type sequence determination around the lost restriction site. As sequence present on the other strand or via the inducible error- shown in Fig. 3, 8 of these 12 mutants consist of single-base prone repair pathway (26). substitutions, but only 4 mutations (H201, H204, E203, E204) To test the possible influence of the recA- lesion carried by can be simply explained by misincorporation ofdATP[aS]. For strain HB101, the preparation of mutagenized pBR322 DNA another 3 mutants (H202, H203, C201) the base substitution from which the deletion mutants H302 and H303 were re- occurred at the predicted position and is consistent with mis- covered was used to transform two different recA+ E. coli incorporation of the added conventional nucleotide, either strains-DB1528 and DB4906. On screening a total of 48 in- dGTP or dlTP. The remaining 5 mutations are assumed to have dependent transformants, 7 additional HindIII resistant mu- been generated by strand slippage during DNA synthesis in tants were identified. On sequence analysis, all seven were vitro (E201) or by an in vivo phenomenon occurring in a recA- found to have acquired the same base substitution as mutant background at the site(s) of misincorporated a-thiophosphate H301 with no deletions. Although the numbers are small, this nucleotides, as discussed above. result suggests that the frequency ofdeletion mutations induced When the data from all six experiments with a-thiophosphate by this method can be reduced by use of a recA+ strain for nucleotides were examined, the rate of mutagenesis-as de- transformation. fined by the percentage ofrestriction enzyme-resistant mutants In a series of mutagenesis experiments with the nucleotide among the total transformants screened-varied from 12-40%, dATP[aS], an effort was made to fix the position of the 3' ter- with a value of 50% being the theoretical upper limit. Among minus of the single-stranded gap by adding one conventional those mutants that were sequenced, single-base substitutions Downloaded by guest on September 28, 2021 1592 Genetics: Shortle et al. Proc. Natl. Acad. Sci. USA 79 (1982)

ofthe type expected from misincorporation ofone analogue res- the frequency of these deletions can be controlled by modifi- idue constitute 50-60% of the total. The remainder ofthe mu- cations in the reaction conditions or variation in the bacterial tants consisted predominantly of either very small deletions or hosts used for transformation-or both. Nor is it clear whether unexpected types of base substitutions. all kinds ofpoint mutations can be readily induced. Such a de- termination will depend upon experiments using a-thio dCTP DISCUSSION and a-thio dGTP. Nevertheless, given the results to date plus From the results presented in this report (summarized in Table the relatively straightforward chemistry involved, it seems rea- 1), we conclude that a variety of base substitutions can be in- sonable to expect that a-thiophosphate nucleotide misincor- duced at relatively high rates by using DNA polymerase to cat- poration may provide a simple, efficient method ofinducing all alyzed the misincorporation ofnucleotides at specific sites. The types of single-base substitutions. gap misrepair methods allow site-directed induction of trans- version mutations without chemical synthesis of oligonucleo- We thank Scott Putney, Tom Kunkel, and Paul Schimmel for pro- tides. Like previous versions of in vitro mutagenesis directed viding us with data prior to publication; we are also indebted to Douglas Koshland and Dan Nathans for helpful discussions. This research was to sites by specific nicking (5-7, 9), gap misrepair has the ad- supported by grants to D. B. from the American Cancer Society (MV90) vantages and disadvantages of highly localized mutagenesis and the National Institutes of Health (GM18973 and GM21253) and to without complete specification of the mutant sequence in ad- S.J. B. from the National Institutes of Health (GM13306). D. S. is a fel- vance. Therefore, this approach may be particularly useful for low of the Helen Hay Whitney Foundation. genetic investigations, whereas constructions ofpredetermined variant sequences may still best be accomplished by using 1. Shortle, D., DiMaio, D. & Nathans, D. (1981) Annu. Rev. Genet chemical synthesis (2, 4). With gap misrepair mutagenesis, the 15, 265-294. frequency is high enough to encourage screening of individual 2. Smith, M. & Gillam, S. (1981) in Genetic Engineering, Principles isolates one by one, even by biochemical methods. and Methods, eds. Setlow, J. K. & Hollaender, A. (Plenum, New York), Vol. 3, pp. 1-32. The two methods (misrepair via nucleotide omission and 3. Timmis, K. N. (1981) in Society for General Microbiology Sym- misrepair with a-thiophosphate nucleotides) have properties posium, eds. Glover, S. W. & Hopwood, D. A. (Cambridge that make them differentially advantageous in different circum- Univ. Press, Cambridge, England), Vol. 31, pp. 49-109. stances. The nucleotide omission method appears to allow the 4. Wallace, R. B., Schold, M., Johnson, M. J., Dembek, P. & Itak- misincorporation ofall three remaining dNTPs at roughly equal ura, K. (1981) Nucleic Acids Res. 9, 3647-3656. It this result is in contrast to 5. Muller, W., Weber, H., Meyer, F. & Weissmann, C. (1978)J. frequency. should be noted that Mot Biol 124, 343-358. the failure to observe high levels of -pyrimidine 6. Shortle, D. & Nathans, D. (1978) Proc. Natl Acad. Sci. USA 75, mispairing under other, possibly more physiological, conditions 2170-2174. (28, 29). It may be possible to control the pattern of misincor- 7. Shortle, D., Koshland, D., Weinstock, G. M. & Botstein, D. poration by varying the relative concentrations of the three (1980) Proc. Natl. Acad. Sci. USA 77, 5375-5379. deoxyribonucleotide triphosphates in the misrepair reaction. 8. Parker, R. C., Watson, R. M. & Vinograd, J. (1977) Proc. Natl. However, two properties of this mutagenic DNA polymerase Acad. Sci. USA 74, 851-855. 9. DiMaio, D. & Nathans, D. (1980)J. Mol Biol 140, 129-142. reaction may limit its use in the in vitro construction of point 10. Shortle, D. & Botstein, D. (1981) in Molecular and Cellular mutations. First, when the template for DNA synthesis is longer Mechanisms of Mutagenesis, eds. Lemontt, J. F. & Generoso, than a few nucleotides multiple base substitutions can occur. W. M. (Plenum, New York), in press. More important, sites where the omitted nucleotide would be 11. Greenfield, L., Simpson, L. & Kaplan, D. (1975) Biochim. Bio- correctly incorporated at two or more adjacent positions may phys. Acta 407, 365-375. 12. Bryant, F. R. & Benkovic, S. J. (1979) Biochemistry 18, not always be efficiently mutable by this reaction. Addition of 2825-2828. low concentrations ofthe "omitted" nucleotide may reduce this 13. Clewell, D. B. & Helinski, D. R. (1969) Proc. Natl Acad. Sci. problem. USA 62, 1159-1166. Gap misrepair via misincorporation of an a-thiophosphate 14. Dagert, M. & Ehrlich, S. D. (1979) Gene 6, 23-28. nucleotide can also induce point mutations at high frequency. 15. Holmes, D. S. & Quigley, M. (1981) AnaL Biochem. 114, In this case, there appears to be a greater measure of predict- 193-197. a mutations are 16. Maxam, A. M. & Gilbert, W. (1977) Proc. Nati Acad. Sci. USA ability because majority ofthe recovered single- 74, 560-564. base substitutions of the type expected from the nucleotide 17. Kunkel, T. A. & Loeb, L. A. (1979) J. Biol Chem. 254, used. Amongthe unexpected mutations, the predominant types 5718-5725. are deletions ofone to four base pairs. It is not yet clear whether 18. Loeb, L. A., Weymouth, L. A., Kunkel, T. A., Gopinathan, K. P., Beckman, R. A. & Dube, D. K. (1979) Cold Spring Harbor Table 1. Summary of nucleotide sequence changes induced by Symp. Quant. Biol 43, 921-927. 19. Burgers, P. M. J. & Eckstein, F. (1979) J. Biol Chem. 254, gap misrepair mutagenesis 6889-6893. Type of mutation Method 1 Method 2 20. Vosberg, H. P. & Eckstein, F. (1977) Biochemistry 16, 3633-3640. 21. Kunkel, T. A., Eckstein, F., Mildvan, A. S., Koplitz, R. M. & Transition 13 3 Loeb, L. A. (1981) Proc. Natl Acad. Sci. USA 78, 6734-6738. Transversion 12* 20 22. Van de Sande, J. H., Loewen, P. C. & Khorana, H. G. (1972)J. Deletion 1 6t Biol Chem. 247, 6140-6148. Insertion 0 1 23. Burgers, P. M. J. & Eckstein, F. (1979) Biochemistry 18, 592-596. Total lesions 26 30 24. Putney, S. D., Benkovic, S. J. & Schimmel, P. R. (1981) Proc. Total mutant plasmids 17 29 Natl Acad. Sci. USA 78, 7350-7354. 25. Eckstein, F. (1979) Acc. Chem. Res. 12, 204-210. Method 1 and method 2 are described in the text. Numbers given are 26. Witkin, E. (1976) BacterioL Rev. 40, 869-907. the simple sum of mutations analyzed and cannot (particularly with 27. Englund, P. T. (1972)J. MoL Biol. 66, 209-224. method 2) be construed to have more than a very limited statistical 28. Fersht, A. R. & Knill-Jones, J. W. (1981) Proc. NatL Acad. Sci. significance. USA 78, 4251-4255. * All four possible transversions are represented. 29. Topal, M. D. & Fresco, J. R. (1976) Nature (London) 263, t Of these, four had more than one adjacent base pair deleted. 285-289. Downloaded by guest on September 28, 2021