Deletions and Insertions in the P53 Tumor Suppressor Gene in Human Cancers: Confirmation of the DNA Polymerase Slippage/Misalignment Model Marc S

Home , Insertion (genetics)

[CANCER RESEARCH 56. 2130-2136. May 1. 1996] Deletions and Insertions in the p53 Tumor Suppressor Gene in Human Cancers: Confirmation of the DNA Polymerase Slippage/Misalignment Model Marc S. Greenblatt, Arthur P. Grollman, and Curtis C. Harris1

Laboratory of Human Carcinogenesis, National Cancer Institute, National Institutes of Health, Bethesda, Maryland 20892 [M. S. G., C. C. H.J, and Department of Pharmacological Sciences. Slate University of New York at Stony Brook, Stony Brook, New York 11794 [A. P. G.J

ABSTRACT monotonie runs. When a repeat sequence mispairs with a tandem (adjacent) or nontandem (nonadjacent) complementary motif, the loop We analyzed all published deletions and insertions in the p53 gene to formed by the intervening oligonucleotide sequence may be deleted or assess the relevance of mutagenesis models. Almost all deletions and duplicated (10-12). Such repeats have been identified at the sites of insertions can be explained by one or more of the following DNA sequence spontaneous and ionizing radiation-induced deletions (13, 14) and at features: monotonie base runs, adjacent or nonadjacent repeats of short tandem sequences, palindromes, and runs of purines or pyrimidines (ho- the sites of deletions in synthetic DNA sequences with adducted bases mocopolymer runs). Increased length of monotonie runs correlates posi (9). Loops may also form between nonadjacent inverted repeats (pal tively with increased frequency of events. Complex frameshift mutations indromes), leading to deletion of the loop and sometimes to complex can be explained by the formation of quasi-palindromes, with mismatch frameshift mutations involving both deletion and base substitution excision and replication using one strand of the palindrome as a template. (12, 15-17). A review of short (<20-bp) germ line deletions and Deletions and insertions in the p53 tumor suppressor gene may reflect insertions in genes associated with human genetic disease confirmed both spontaneous and carcinogen-induced mutagenesis. that these DNA sequence features are the most important factors determining the specificity of these events (11). Quantitative mutational spectrum analysis in vitro shows that: (a) INTRODUCTION the frequency of frameshift mutations increases as the length of the The somatic mutations that accumulate during multistep carcino- monotonie run increases; and (b) single-base deletions are much more frequent in vitro than are single-base insertions (reviewed in Ref. 5). genesis include major chromosomal events (such as translocations, allelic deletions, and gene rearrangements) and point mutations (such The proposed explanation is superior stability of the hydrogen bond ing of extensive repeats, and G:C-rich regions, and intermediate DNA as base substitutions and small deletions or insertions; Ref. l). The relative frequencies of different classes of mutations vary among structures associated with deletions. DNA polymerase infidelity may cancer-related genes. For example, mutations found in the APC tumor lead to spontaneous frameshift errors; different polymerases vary in suppressor gene are most often short deletions or insertions, whereas the frequency of errors and the ratio of deletions:insertions (5). The activating mutations in the ras proto-oncogene are base substitutions surrounding sequence context is another major factor affecting the likelihood of misalignment mutations (9, 18-20). (1). The largest data base of somatic mutations in human cancers exists for the p53 tumor suppressor gene (2, 3). All classes of muta Limited analysis of smaller p53 mutation data bases has confirmed the presence of some of these high-risk sequence features, including tions are found. Point mutations, often accompanied by deletion of the second alÃele,are the most common event. The majority of p53 point monotonie and tandem repeats, at the sites of p53 somatic mutations mutations are base substitutions, but 10â€”15%of both somatic and (10). Long runs of purines or pyrimidines, known as homocopolymers germ line p53 mutations are small, intragenic deletions or insertions (e.g., AGAGAG, CCTTCC, and AAAGGG) have also been noted at (1,4). This data set is large enough to test hypotheses regarding the in sites ofp53 deletions (21). We updated and reviewed the data base of vivo mechanisms of intragenic deletions and insertions. all reported somatic p53 mutations (2) and analyzed the subset of all In vitro mutagenesis studies have established that DNA sequence deletions and insertions in more detail to assess: (a) the extent to context is the most important factor in determining the specificity of which the mechanisms determined in experimental systems could deletions and insertions (reviewed in Refs. 5 and 6). Almost all small account for somatic p53 deletions and insertions; (b) how patterns of deletions and insertions in model systems occur at monotonie runs of observed p53 deletions and insertions might refine existing models; two or more identical bases, or at repeats of 2-8-bp DNA motifs, and (c) whether relative frequencies of deletions and insertions pro either in tandem or separated by intervening nucleotides. Several vide information to distinguish between spontaneous or carcinogen- mechanisms have been proposed. The most well-studied mechanism, induced events. first proposed by Streisinger et al. (7) in 1966, involves slippage or misalignment of the template DNA strands in iterated bases during MATERIALS AND METHODS replication. This leads to either base deletion (if the nucleotides excluded from pairing are on the template strand) or insertion (if they The p53 mutation data base has been described and is available via E-mail are on the primer strand); replication "fixes" these mutations in the from the European Molecular Biology Library (2). We updated the data base daughter cells (5, 6). Misalignment can occur "spontaneously" in to include all p53 mutations published before September 1994, as identified by monotonie runs or can be initiated by a spontaneous or adduct- searches of MEDLINE and Current Contents. All intron and exon sequences at induced base misincorporation that creates a context for slippage (5, the sites of, adjacent to, and up to 40 bp away from insertions and deletions were examined, in some cases with the aid of the Gene Jockey sequence 8, 9). analysis program (Biosoft, Cambridge, United Kingdom). Regression and x2 Misalignment may also occur in repeat sequences that are not analyses were performed using standard statistical software.

Received 9/11/95; accepted 3/4/96. The cosls of publication of this article were defrayed in part by the payment of page RESULTS charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1To whom requests for reprints should be addressed, at Laboratory of Human Since the publication of our original p53 data base (2), the preva Carcinogenesis. National Cancer Institute. National Institutes of Health. Building 37. lence of deletions and insertions among all p53 mutations has de Room 2C05, Bethesda, MD 20892-4255. Phone: (301) 496-2048; Fax: (301) 496-0497. creased from 13 to 11%. This reduction probably reflects a detection 2130

Table I DNA sequence features associated with deletions and insertions of pS3 in human cancers

(%)Class Sequence features associated with events

high-risksequence eventAllAllof feature9797951009794961009796Monotonieruns72807775868356867471Tandemrepeats57585667484467436683Tandemrepeatsonly1112801083308gHomocopolymerruns565531426665521004738Nontandemrepeats192351700404217Palindromes55575004141213

deletionsAll insertionsComplex frameshifts1 -bpdeletions1-bp insertions2-bp deletions2-bp insertions3 + -bpdeletions3 + -bp insertionsn34925978121184827711324Any bias, because most recent reports have examined only exons 5-8, in the sites of 77% (269 of 349) of all events; only 46% of all p53 bp are which missense mutations are more common, and have neglected within these runs (odds ratio for an event occurring at a monotonie exons 2-4 and 9-11, in which the majority of mutations are deletions run, 3.95; 95% confidence interval, 3.00-5.20; P < 0.0001). Re and insertions (1). The data base contains 370 deletions and insertions peated tandem sequences, either adjacent to or distant from each in thep5J gene. Of these, 21 detected by sequencing of cDNA involve other, encompass or flank 30% of deletions or insertions, and palin entire exons or large deletions at splice sites, indicating aberrant dromes flank another 5%. Homocopolymer runs of 4 or more bp occur mRNA splicing. Because splicing mutations often result from either at the sites of 56% of events, but in only 3% are they the only point mutations in splice sequences at the intron-exon junction or associated feature. Homocopolymer runs of 4 or more bp encompass dysregulation of the splicing mechanism and not from loss or gain of 42% of all bp in p53-coding exons (odds ratio for an event occurring DNA, these mutations were not classified as DNA deletions or inser at a homocopolymer run, 1.75; 95% confidence interval, 1.38-2.23; tions. The patterns and DNA sequence features of the remaining 349 P < 0.0001). deletions and insertions (or "events") were analyzed. Short deletions and insertions ( 1 bp) are associated with a different Sizes and Relative Frequencies ofp53 Deletions and Insertions. pattern of sequence features than are longer events (2 or more bp; Simple deletions constituted 74% (259 of 349) of all events (71% of Table 1). For 1-bp changes (n = 118 deletions; n = 48 insertions), one 1 bp, 76% of two bp, and 82% of &3 bp). Simple insertions monotonie runs and direct tandem repeat sequences account for al constituted 22% (78 of 349) of the events; the remaining 3% (12 of most all events. Monotonie runs of 2-5 bp occur at 86 and 83% of 349) were complex frameshift mutations that combined deletion, 1-bp deletions and insertions, respectively. Where no monotonie insertion, and base substitution. In 49% (166 of 349, 118 deletions and repeat exists (n = 24; 14% of 1-bp events), tandem repeat sequences 48 insertions) only 1 bp was affected; 10% (34 of 349, 27 deletions are present at 15 (94%) of 16 1-bp deletions and 7 (88%) of 8 1-bp and 7 insertions) involved 2 bp; and 39% (137 of 349, 113 deletions insertions. Overall, tandem repeats accompany 48% of 1-bp deletions and 24 insertions) altered 3 or more bp. Only six intragenic deletions and 44% of 1-bp insertions; they are the only high-risk DNA se in p53 larger than 30 bp and unassociated with splice sites were quences present at 10% of 1-bp deletions and 8% of 1-bp insertions. reported; the largest was 137 bp. Runs of 4 or more homocopolymer bp occur at 64% of 1-bp events, DNA Sequence Features at Sites of Deletions and Insertions. but in all but one instance (1%), they accompany other high-risk We calculated the proportions of all DNA deletions and insertions sequence features. attributable to each of the following mechanisms (Table 1 and Fig. 1): Deletions of 2 bp (n = 27) are less likely to be associated with monotonie base runs, repeats of short tandem sequences, palindromes monotonie runs (56%) and more likely to be associated with tandem (inverted repeats of dyad symmetry), and runs of four or more purines repeats (67%). In 33% of these 2-bp deletions, tandem repeats are the or pyrimidines (homocopolymer runs). only deletion-promoting DNA sequence features present. However, Almost all [341 (98%) of 349] deletions and insertions can be six of seven insertions of 2 bp are at monotonie runs, and only three explained by one or more of these DNA sequence features. The most are at tandem repeats. Homocopolymer runs accompany 52% of common sequence motifs seen at the site of deletions or insertions are deletions and all seven insertions of 2 bp. Of deletions of 3 or more monotonie runs (two to five consecutive identical bases), present at bp (n = 134), 68% were at monotonie runs, 58% were at direct

Fig. 1. Sequence features associated with p53 deletions and insertions. MONO RUN. monotonie base run; TAN RPT. tandem repeat sequence; HCP RUN, homocopolymer run of four or more bases; NONTAN RPT. nonlandem repeated sequence (nonadjacent repeat sequence flanks event); PAL, palindrome.

ANY HCP NONTAN PAL RUN RPT Sequence Feature

2131

DELETION FREQUENCY

0.6

0.5-

0.4 -J

El 1 BP DELETIONS 0.3 â€¢¿ALLDELETIONS

0.2

0.1

0.0

Fig. 2. Relationship between length of monotonie repeat and frequency of deletions and insertions in p53. LENGTH OF REPEAT A, deletions increase in a linear, positive relationship with length of repeat [R = 0.930 for 1 bp (Q), 0.864 for all (*)]. B, insertions increase in frequency only at mono- B INSERTION FREQUENCY tonic runs of five bases. 0.5

0.4

0.3

s ALL INSERTIONS â€¢¿1BP INSERTIONS 0.2

0.1

0.0

LENGTH OF REPEAT

repeats, and only 43% were accompanied by homocopolymer se mutations) have been reported at this G:C-rich sequence with multiple quences. Notably, 43% were flanked by nontandem repeat sequences, runs and direct repeats. and 10% were flanked by palindromes, a sequence feature that rarely Increased Length of Monotonie Runs Correlates with Increased accompanies shorter events. The same general patterns are seen for Frequency of Events. To test the prediction of the DNA polymerase insertions of 3 or more bp (n = 27; Table 1). slippage model that deletions and insertions should be more common Krawczak and Cooper (11) identified a consensus sequence of 6 bp at longer repeats (5), we calculated a regression curve of the frequency (TG A/G A/G G/T A/C), which occurs frequently at sites of germ line of events per site. We divided the number of deletions and insertions deletions in a variety of genetic diseases. We identified this sequence occurring at monotonie runs of 2, 3, 4, and 5 bp by the number of runs six times in the coding exons of the p53 gene; 14 somatic p53 of each length in p53-coding exons times the number of bases per run. deletions or insertions are associated with it, including 11 at one site The resulting number describes the frequency of events per bp at in exon 5 (codons 173-175, TG AGG C). With only one exception, at monotonie runs. One run of 6 bp occurs in exon 11, but because this least one other deletion-associated sequence feature (usually tandem region is infrequently sequenced (only one deletion here has been repeat or palindrome) also is associated with these events. The region reported), it was not included in this analysis. Fig. 2 shows that for of p53 that most frequently contains deletions or insertions is codons deletions of 1 bp, the relationship is positive and linear; deletions are 151-159: CGC CCG CGC CGC ACC CGC GTC CGC GCC. Thirty proportionally more frequent at longer runs (R = 0.930). This rela mutations (9% of all deletions and insertions, and 1% of all p53 tionship is present but weaker when deletions of all lengths are 2132

Table 2 Features of deletions and insertions flanked by palindromes and tandems sequence context, with both stable palindromes and tandems absent from the site. nDeletionsInsertionsLength Exogenous versus Endogenous Mutation Analysis by Tumor Type. The mutational spectrum may be an index of the relative frequencies of endogenous versus exogenous mutations in tumors and (%)2 of paired sequence bp3bp4bp5 may help detect patterns characteristic of exogenous mutagens. Pre viously, we analyzed base substitution patterns among tumor types

bp>5 and noted that substitutions that are thought to be due to exogenous bpLength carcinogens are more frequent in tumors for which epidemiological (%)1-3of deletion studies indicate a link with carcinogen exposure (1, 4). Endogenous bp4-9 bp10-20bp>20 deamination of 5-methylcytosine at CpG sites results in G:C to A:T transitions at CpG dinucleotides; CpG site transitions therefore have bpLength been referred to as "spontaneous" mutations, which can occur in the (%)1-3of insertion bp4-9 absence of exogenous mutagens. The prevalence of CpG site transi bp10-20 tions varies widely among tumor types, generally correlating nega bp>20bpSegment tively with carcinogen association (1). (%)Bothdeleted Because DNA polymerase infidelity alone can produce frameshift primersIntervening mutations, deletions and insertions also have been suggested as a onlyOne loop loopPartialprimer + marker of spontaneous mutation (10). However, chemical carcinogens loopOtherSegmentprimer + and their adducts can induce frameshift mutations in vitro (6, 18) and, therefore, may contribute to frameshift mutations in human cancers. (%)Bothinserted primersIntervening We compared the prevalence of deletions and insertions among tu onlyOne loop mors to determine whether they correlated either positively or nega loopPartialprimer + loopOtherPalindromes1814401739222271443362500751421-290-757-710SO250250Tandems6560523129201533843130602020710-1257-6517-23000602020primer + tively with carcinogen exposure and could thus be used as an indicator of endogenous or exogenous mutations. Deletions and insertions are most common in tumors of the thyroid (17%), head and neck (15%), breast (15%), and sarcomas (15%); they are least common in melanoma (5%) and other skin cancers (8%) and considered, because these events are often associated with other colon (8%), hepatocellular (8%), and cervical carcinomas (8%). Both sequence features. Insertions of 1 or more bp are uncommon at runs of these lists contain tumors that have been frequently (thyroid, head of two, three, or four bases; this portion of the curve is flat. However, and neck, skin, hepatocellular, and cervical) and infrequently (sarco insertions increase sharply in frequency at 5-bp runs (Fig. 2). When mas and colon) associated with environmental mutagens, and both only mutations and sequences in the commonly studied exons 5-8 contain tumors that frequently (thyroid and colon) and infrequently were considered, similar results were obtained (deletions at 4-bp runs (head and neck and hepatocellular) contain CpG site transitions, are slightly underrepresented; data not shown). Therefore, selection markers of mutations unassociated with exogenous mutagens (1). bias in detection of deletions is unlikely to affect these results. The prevalence of deletions and insertions varies in tobacco-asso Nonadjacent Tandem Repeat Sequences and Palindromes. Six ciated cancers (e.g., cervix, 8%; lung, 10%; bladder, 12%; and head ty-five mutations occur at nonadjacent tandem repeats, and 18 occur and neck, 15%). The hot spot region from codons 151-159 is affected at inverted repeats (palindromes). In all but one of these events, the in 16% of 36 lung cancer deletions and insertions, including 3 around length of the repeated motif is 3 or more bp (Table 2). More than codon 157. Codon 157 is a hot spot for base substitutions in lung three-fourths of the sequences deleted or inserted between tandem cancer (1); this sequence may be the target of an important promuta- repeats are between 4 and 20 bp long; 40% are between 4 and 9 bp, genic, tobacco-associated adduci. and 43% are between 10 and 20 bp, but only 12% are more than 20 bp. Of sequences deleted or inserted between palindromes, however, more than three-fourths are 10 bp or greater; 40% are greater than 20 DISCUSSION bp, and few are between 4 and 9 bp. Deletions or insertions flanked by This analysis confirms many of the prior observations of deletions tandem sequences usually include one full copy of the repeat and the and insertions in human disease (10, 11, 21), highlights the relevance complete sequence between the repeats. Those between inverted re of many of the mechanistic models for deletion and insertion derived peats usually include a portion, but not all, of the palindrome; in about from in vitro mutagenesis studies (5, 6, 9), and suggests refinement of one-fourth of cases, only the sequence between the repeat is affected some models. (Table 2). The regression curve of the frequency of deletions and insertions Complex Frameshift Mutations. Complex frameshifts are muta versus the length of a monotonie base run shows that as the length of tions which involve combinations of deletion, insertion, and base a run increases, the probability of deletions (especially 1-bp deletions) substitution. In a model derived from in vitro studies (6, 17), these increases in a linear relationship. However, a different relationship is mutations are explained by formation of quasi-palindromic loops, noted for insertions. A run of repeated bases must reach a threshold of which are treated as double-stranded DNA. Deletions, insertions, and 5 bp before insertions are likely to occur in the p53 gene. This substitutions are generated by excision of mismatches and template- threshold pattern has been demonstrated for other systems in vitro, directed replacement of bases complementary to one side of the loop. such as deletions in DNA polymerase a replication of free DNA (5). Twelve complex frameshift mutations appear in the data base. Nine Sequence context surrounding runs of identical length can contribute of these can be explained by the formation of palindromes with to differences in deletion and insertion frequency (5, 19, 20); there template-directed mutation as described by the above model (exam fore, a simple length-frequency relationship is difficult to interpret. ples in Fig. 3; Ref. 17). Two others occur at the sites of tandem repeats However, the same sequences in p53 are at risk for both deletions and of 5-6 bp. Only one complex frameshift is not explained by DNA insertions; the fact that their incidence-versHj-length curves differ 2133

CODONS 155-161 CODONS 235 - 241

CTC AcxcG GT CG T1 AA TC 1A GÂ¡/Ac/ CC A CC CAâ€” GT AC T T AC CC*â€¢ Aâ€”TSs$Â£?GC Cde/3 A fcC (r) del 4 C.GCC \G/ ins 1 T "" CGr-A7"Gâ€” yAtoT^AAC fc ^ A-T Gâ€”C C Câ€”GT Câ€”GTÂ©G Câ€”G T GCTCAC Aâ€” T C â€”¿GAâ€” TTG

CODONS 199-213 INTRON 6 - EXON 7, CODON 227

Fig. 3. Complex frameshift mutations fit a model of quasi-palindrome formation with DNA excision TAG TA and replication using one side of the palindrome as CA C a template. Deleted bases are circled; bold italics. C C â€”¿G C new bases. G.A^v\cT\\TT AÂ°T AGâ€”GCQ TT AC _AAdel GC T AT CTT GT G ^^jG CT CC G â€”¿ins6 to A T 10 "~ TTGCGTGTGAGTATTTGGCG Cmult fc C G^ subs, C â€”¿ CTG 1 T Ã±delie â€”¿/G\Tâ€” :C C VT/ fA\Tâ€” GG AT C â€”¿ TC â€”¿TT [AG C/l TC VA/GG TC GT TA GT GGGgc AT^AA GA

*QT^lTTGGGGA AC CA TTO GAG TU'GGATTCQTCTATA xG_L/'>Â»delie

suggests that the intermediate events involved in their generation Several models involving DNA damage-induced template mis differ. One potential explanation is that repair of short heteroduplex alignment or slippage during DNA replication have been proposed as DNA structures predisposing to deletions is less efficient than that of mechanisms of frameshift mutagenesis (5, 6, 8, 9, 17-19, 26-29). The those leading to insertion (see below). following sequence of reactions may lead to base deletions: DNA The proportion of p53 deletions and insertions that represent spon polymerase encounters a damaged or adducted base at the replication taneous events due to DNA replication infidelity versus mutagen- fork, and DNA synthesis is blocked, often transiently, by helix de induced changes remains unclear. Frameshift mutations that seem to formation or adduci bulk. If chain extension is delayed, the nucleotide(s) at the 3' primer terminus may pair with a complementary form spontaneously in genomic DNA may be generated by DNA polymerase-induced errors in replication or may arise as a result of base(s) 5' to the lesion. This misaligned template generates a repli endogenous or exogenous damage to template DNA. Endogenous cation product with deleted bases. The complementary base pairing at DNA damage is generated by reactive oxygen species (22), products the primer-template junction facilitates extension of the misaligned chain. Base substitutions occur when insertion of an "incorrect" base of lipid peroxidation (23, 24), or other metabolic reactions that alter DNA structure (25), thereby complicating the assignment of these opposite the lesion is followed rapidly by chain extension. Adduct- mutations to specific environmental agents. induced base misincorporation may also increase the stability of a We compared the prevalence of frameshift mutations in tumor misaligned intermediate. types with epidemiological and mutational spectrum evidence sup Frameshift mutations are induced in prokaryotic models by some porting carcinogen-induced and endogenous mutagenesis (1); associ bulky chemical mutagens, including aflatoxin B, and arylamines. The ation with cancers linked to carcinogens would strengthen the infer ability of a damaged nucleotide to generate frameshift mutations is ence that frameshifts were induced by exogenous damage. Deletions determined by the nature of the base inserted opposite the adduci, the and insertions did not correlate well with these indices. This suggests surrounding sequence context (adduct-induced frameshift mutations that differences among tumor types in their prevalence may reflect occur more frequently in monotonie or tandem base runs), and the rate both endogenous and carcinogen-induced mutagenesis, such as organ- of translesional DNA synthesis estimated from kinetic analysis (9, 19, specific carcinogens or tissue-specific properties that regulate repli 20, 30-32). Studies with chemically defined DNA adducts have cation fidelity. The presence in lung cancers of a hot spot for frame- established the mutational specificity of some carcinogen-induced shift mutations in the G:C-rich sequence at codons 151-159, near a lesions in vitro (31, 33). Although the frequency and sites of muta base substitution hot spot (1). suggests that carcinogen-mediated tions are not always predictable, the relative propensity for some DNA frameshifts may occur in vivo. lesions to promote frameshift deletions in vitro can be estimated (9). 2134

In addition to the effects of specific mutagens, DNA repair capacity related genes, including APC in colon cancers (50) and pl6'"k4 is an important factor affecting mutagenesis. The rates and types of (CDKN2 and MTS-1) in a variety of tumors (51-54). Interestingly, spontaneous and induced mutations in Salmonella typhimurium can 1-bp deletions in APC are less often associated with monotonie runs vary dramatically depending on the DNA repair capacity of the strain but are often associated with tandem repeats, perhaps reflecting the (17, 34). In eukaryotes, mismatch repair involves a complex of several influence of impaired mismatch repair due to abnormalities in the repair enzymes. One of these, MSH2, binds to palindromes with high hMLHl and HMSH2 genes in colon tissue. Of 24 deletions and five affinity, and to large and small insertions and G/T mismatches with insertions described in pl6'nt4, only 38% (11 of 29) affect 1 bp, and moderate affinity (35). MSH2 mutants are defective in the repair of 48% (14 of 29) affect 3 or more bp. DNA sequence features in this mismatches and small insertion/deletion mispairs (36); thus, this sys preliminary data base are similar to those seen in p53. Monotonie runs tem is likely responsible for repair of the misalignments leading to (86%) and tandem sequences (72%) are the most frequent motifs frameshifts. Germ line mutations in hMSH2 (37, 38) and other mis present at the sites of deletions. match repair enzymes have been linked to hereditary nonpolyposis Deletions and insertions in p53 are likely underreported, because colon cancer (39-41). Individuals who inherit these mismatch repair many investigators concentrate on exons 5-8, the highly conserved abnormalities demonstrate genome-wide insertions and deletions in DNA-binding region in which most missense mutations occur. More repetitive DNA sequences (replication errors). This "mutator pheno- complete analysis of all p5J-coding exons would give a more thor type" may contribute to intragenic deletions in cancer-related genes ough picture of mutational patterns. Missense mutations constitute (42). However, replication errors and the mismatch repair status in 77% of mutations in exons 5-8 but less than half of the mutations cases of colon or other tumors containing p53 mutations have not yet outside of these regions (1). This is probably due to less frequent been reported. selection of missense mutants in exons 2-4 and 9-11. Single-base These mismatch repair enzymes also are involved in homologous substitutions in the DNA-binding domain often alter p5J-induced recombination, as are other proteins, including RecA and its eukary- transactivation (1). However, single-amino acid substitutions in the otic homologues, which bind to homologous DNA regions and pro transactivation domain (amino acids 20-42) do not affect transacti mote pairing and strand exchange (43-46). The heteroduplex DNA vation; some double substitutions do (55). Frameshifts occurring sequences formed during this process are generally longer than the within or 5' to the DNA-binding domain eliminate the p53 DNA- intermediate structures proposed for the mispairs that lead to frame- binding function and thus would alter p53 function and cell cycle shifts (47, 48), and it is uncertain how these recognition and repair kinetics every time they occur. The carboxyl terminus (exons 9-11) of systems are involved in mutagenesis. p53 also is important in the oligomerization and nuclear localization Palindromes and nonadjacent repeat sequences often flank dele of the p53 protein (reviewed in Ref. 56) and regulates its sequence- tions and insertions larger than 3 bp in vitro (5, 6, 16) and in this specific DNA binding (57-60). report. The patterns we observed suggest that the intermediate path ways differ for events initiated by these two sequence motifs. Palin ACKNOWLEDGMENTS dromes generally flank larger deletions, suggesting that short loops between palindromic dyad elements are less likely to be deleted than We are grateful to Dorothea Dudek for excellent editorial assistance. Peter are longer loops. The presence of short loops in the dyads associated Shields for assistance with statistical analysis, and Thomas Kunkel for critical with complex frameshifts (Fig. 3) supports this explanation. The data review of the manuscript. also indicate that, although direct repeats of 2-bp motifs can predis pose to short deletions, more than 2 bp homology are required for both REFERENCES

palindromes and nontandem repeat elements to form stable structures. 1. Greenblatt. M. S., Bennett, W. P., Hollstein. M., and Harris. C. C. Mutations in the Longer dyad sequences are required for palindromes than for tandems. p53 tumor suppressor gene: clues to cancer etiology and molecular pathogenesis. The deletions associated with tandem sequences usually exactly in Cancer Res., 54: 4855-4878. 1994. 2. Hollstein. M.. Rice, K.. Greenblatt, M. S.. Soussi, T.. Fuchs, R.. Sorlie. T.. Hovig. E., corporate the intervening loop and one copy of the repeat, confirming Smith-Sorensen, B., Montesano, R., and Harris. C. C. Database of p53 gene somatic the existing model (11, 13, 14, 49). However, the p53 deletions mutations in human tumors and cell lines. Nucleic Acids Res., 22; 3547-3551. 1994. associated with palindromes usually include portions of both flanking 3. Hollstein. M., Shomer, B.. Greenblatt. M.. Soussi, T., Hovig, E., Montesano, R., and Harris, C. C. Somatic point mutations in the p53 gene of human tumors and cell lines: motifs; the exact deletion generally cannot be predicted. This suggests updated compilation. Nucleic Acids Res., 24: 141-146. 1996. that different deletion mechanisms occur, perhaps involving variable 4. Hollstein, M., Sidransky, D., Vogelstein, B., and Harris. C. C. p53 mutations in human cancers. Science (Washington DC), 253: 49-53. 1991. actions of DNA repair proteins (see above). The specific system or 5. Kunkel. T. A. Misalignment-mediated DNA synthesis errors. Biochemistry. 29. mutagen also may influence repair. For example, the spectrum of 8003-8011. 1990. 6. Ripley, L. S. Frameshift mutation: determinants of specificity. Annu. Rev. Genet., 24: spontaneous and induced mutations in 5. typhimurium usually changes 189-213, 1990. when a gene for an error-prone DNA repair enzyme is inserted, but no 7. Streisinger, G., Okada. Y., Emrich. 1, Newton. J.. Tsugita. A.. Terzaghi, E., and change is seen in mutations induced by the heterocyclic amine Glu- Inouye. M. Frameshift mutations and the genetic code. This paper is dedicated to P-l (17, 26). Professor Theodosius Dobzhansky on the occasion of his 66th birthday. Cold Spring Harbor Symp. Quant. Biol.. 31: 77-84. 1966. Analysis of complex frameshift mutations confirms that DNA 8. Benebek. K.. and Kunkel. T. A. Frameshift errors initialed by nucleotide misincor- replication and repair mechanisms may be involved in their genesis. poration. Proc. Nati. Acad. Sci. USA. 87: 4946-4950. 1995. 9. Shibulani. S.. and Grollman, A. P. On the mechanism of frameshift (deletion) Prior studies of complex frameshift mutations induced in S. typhi mutagenesis in vitro. !. Biol. Chem., 268: 11703-11710, 1993. murium generated a model of adduct-mediated mispairing, quasi- 10. Jego, N., Thomas, G., and Hamelin. R. Short direct repeats flanking deletions, and palindromes, and DNA replication in palindromes (17). In this system, duplicating insertions in p53 gene in human cancers. Oncogene, 8: 209-213. 1993. 11. Krawczak. M.. and Cooper. D. N. Gene deletions causing human genetic disease: the presence of an error-prone DNA repair gene enhances the forma mechanisms of mutagenesis and the role of the local DNA sequence environment. tion of complex frameshifts. We show that complex mutations in p53 Hum. Genet., 86: 425-441, 1991. are consistent with DNA excision repair and synthesis in palindromes, 12. Albertini, A. M.. Hofer, M., Calos, M. P., and Miller, J. H. On the formation of spontaneous deletions: the importance of short sequence homologies in the generation with one strand of the palindrome used as a template (Fig. 3). This of large deletions. Cell, 29: 319-328, 1982. may be triggered by an adduct-induced substitution mutation that 13. Lutze, L. H., Cleaver, J. E., Morgan, W. F., and Winegar, R. A. Mechanisms involved in rejoining DNA double-strand breaks induced by ionizing radiation and restriction increases the stability of a palindrome (17). enzymes. MutÃ¢t.Res., 299: 225-232, 1993. Germ line and somatic mutation have been reported in other cancer- 14. Lutze, L. H., Winegar. R. A., Jostes, R., Cross, F. T., and Cleaver, J. E. Radon-

2135

induced deletions in human cells: role of nonhomologous strand rejoining. Cancer H. T., Trent. J. M., de la Chapelle. A., Kinzler. K. W.. and Vogelstein, B. Mutations Res., 52: 5126-5129, 1992. of a mutS homolog in hereditary nonpolyposis colorectal cancer. Cell, 75: 1215- 15. Nalbantoglu. J.. Hartley. D., Phear, G., Tear, G.. and Meuth. M. Spontaneous deletion 1235, 1993. formation a( the aprt locus of hamster cells: the presence of short sequence homol- 39. Peltomaki, P.. Aaltonen, L. A., Sistonen. P.. Pylkkanen, L., Mecklin, J. P.. Jarvinen, ogies and dyad symmetries at deletion termini. EMBO J., 5: 1199-1204. 1986. H.. Green, J. S., Jass, J. R., Weber, J. L., Leach, F. S., Peterson, G. M.. Hamilton. 16. Glickman, B. W.. and Ripley, L. S. Structural intermediates of deletion mutagenesis: S. R.. de la Chapelle, A., and Vogelstein, B. Genetic mapping of a locus predisposing a role for palindromic DNA. Proc. Nati. Acad. Sci. USA, 81: 512-516, 1984. to human colorectal cancer. Science (Washington DC), 260: 810-812, 1993. 17. Levine, J. G.. Schaaper. R. M.. and DeMarini, D. M. Complex frameshift mutations 40. Aaltonen, L. A.. Peltomaki, P., Leach, F. S.. Sistonen. P.. Pylkkanen, L., Mecklin, mediated by plasmid pKM 101: mutational mechanisms deduced from 4-aminobiphe- J. P., Jarvinen. H., Powell. S. M.. Jen, J.. Hamilton, S. R.. Peterson, G. M., Kinzler, nyl-induced mutation spectra in Salmonella. Genetics, 136: 731-746, 1994. K. W., Vogelstein, B., and de la Chapelle, A. Clues to the pathogenesis of familial 18. Lambert. I. B., Napolitano. R. L., and Fuchs. R. P. Carcinogen-induced frameshift colorectal cancer. Science (Washington DC). 260: 812-815, 1993. mutagenesis in repetitive sequences. Proc. Nati. Acad. Sci. USA, 89: 1310-1314. 41. Thibodeau, S. N.. Bren. G., and Schaid. D. Microsatellite instability in cancer of the 1992. proximal colon. Science (Washington DC), 260: 816-819, 1993. 19. Belguise-Valladier, P., and Fuchs. R. P. Strong sequence-dependent polymorphism in 42. Markowitz. S., Wang. J., Myeroff. L.. Parsons. R.. Sun, L.. Lutterbaugh, J., Fan, R. S., adduct-induced DNA structure: analysis of single /V-2-acetylaminofluorene residues Zborowska, E., Kinzler, K. W., Vogelstein, B., Brattain, M.. and Willson, J. K. bound within the Narl mutation hot spot. Biochemistry. 30: 10091-10100, 1991. Inactivation of the type II TGF-ÃŸreceptor in colon cancer cells with microsatellite 20. Parris, C. N., Levy, D. D., Jessee, J.. and Seidman, M. M. Proximal and distal effects instability. Science (Washington DC). 26S: 1336-1338, 1995. of sequence context on ultraviolet mutational hotspots in a shuttle vector replicated in 43. Radman, M.. and Wagner, R. Mismatch recognition in chromosomal interactions and xeroderma cells. J. Mol. Biol.. 236: 491-502. 1994. speciation. Chromosoma (Beri.), 102: 369-373, 1993. 21. Redston. M. S.. Caldas, C., Seymour, A. B.. Hruban, R. H., da Costa, L., Yeo. C. J.. 44. Hollingsworth, N. M.. Ponte, L.. and Halsey, C. MSH5. a novel MutS homolog, and Kem, S. E. p53 mutations in pancreatic carcinoma and evidence of common facilitates meiotic reciprocal recombination between homologs in Saccharom\ces involvement of homocopolymer tracts in DNA microdeletions. Cancer Res.. 54: cerevisiae but not mismatch repair. Genes & Dev., 9: 1728-1739. 1995. 3025-3033, 1994. 45. Baker, S. M., Bronner, C. E., Zhang, L., Plug, A. W., Robatzek, M., Warren, G., 22. Breimer, L. H. Molecular mechanisms of oxygen radical carcinogenesis and mu Elliott, E. A., Yu, J., Ashley, T., and Amheim. N. Male mice defective in the DNA tagenesis: the role of DNA base damage. Mol. Carcinog., 3: 188-197. 1990. mismatch repair gene PMS2 exhibit abnormal chromosome synapsis in meiosis. Cell, 23. Nath, R. G.. and Chung, F. L. Detection of exocyclic 1JV2-propanodeoxyguanosine 82: 309-319, 1995. adducts as common DNA lesions in rodents and humans. Proc. Nati. Acad. Sci. USA, 46. de Wind, N.. Dekker, M.. Berns, A., Radman, M.. and te Riele. H. Inactivation of the 91: 7491-7495, 1994. mouse Afs/i2 gene results in mismatch repair deficiency, methylation tolerance, 24. Chaudhary, A. K., Nokubo, M.. Reddy, G. R.. Yeola, S. N., Morrow, J. D.. Blair. hypcrrecombination, and predisposition to cancer. Cell. 82: 321-330, 1995. I. A., and MarneÂ»,L. J. Detection of endogenous malondialdehyde-deoxyguanosine 47. Conley, E. C.. and West, S. C. Underwinding of DNA associated with duplex-duplex adducts in human liver. Science (Washington DC). 265: 1580-1582, 1994. pairing by RecA protein. J. Biol. Chem., 265: 10156-10163, 1990. 25. Lindahl. T. Instability and decay of the primary structure of DNA. Nature (Lond.). 48. Hsieh, P., Camerini-Otero. C. S.. and Camerini-Otero, R. D. The synapsis event in the 362: 709-715, 1993. homologous pairing of DNAs: RecA recognizes and pairs less than one helical repeat 26. Levine, J. G.. Knasmuller, S., Shelton. M. L., and DeMarini, D. M. Mutation spectra of DNA. Proc. Nati. Acad. Sci. USA, 89: 6492-6496. 1992. of Glu-P-1 in Salmonella: induction of hotspot frameshifts and site-specific base substitutions. Environ. Mol. Mutagen., 24: 11-22, 1994. 49. Studier, F. W., Rosenberg, A. H., Simon. M. N., and Dunn. J. J. Genetic and physical mapping in the early region of bacteriophage T7 DNA. J. Mol. Biol.. 135: 917-937, 27. Gupta, P. K.. Johnson, D. L., Reid, T. M., Lee, M. S., Romano. L. J., and King, C. M. Mutagenesis by single site-specific arylamine-DNA adducts. Induction of mutations 1979. 50. Nagase. H., and Nakamura, Y. Mutations of the APC gene. Hum. MutÃ¢t..2: 425-434, at multiple sites. J. Biol. Chem.. 264: 20120-20130, 1989. 1993. 28. Moriya, M., Takeshita. M., Johnson, F.. Peden, K.. Will. S., and Grollman. A. P. 51. Yeudall, W. A., Crawford, R. Y., Ensley, J. F., and Robbins. K. C. MTSI/CDK4I is Targeted mutations induced by a single acetylaminofluorene DNA adduci in mam malian cells and bacteria. Proc. Nati. Acad. Sci. USA, 85: 1586-1589. 1988. altered in cell lines derived from primary and metastatic oral squamous cell carci noma. Carcinogenesis (Lond.). /5: 2683-2686. 1994. 29. Wang, C. I., and Taylor, J. S. In vitro evidence that UV-induced frameshift and substitution mutations at T tracts are the result of misalignment-mediated replication 52. Ohta. M., Nagai, H., Shimizu. M.. Rasio, D., Berd. D., Mastrangelo, M.. Singh, A. D., past a specific thymine dimer. Biochemistry, 31: 3671-3681, 1992. Shields, J. A., Shields, C. L., Croce. C. M., and Huebner, K. Rarity of somatic and germline mutations of the cyclin-dependent kinase 4 inhibitor gene CDK41 in mela 30. Burnouf. D., Koehl, P.. and Fuchs. R. P. Single adduct mutagenesis: strong effect of noma. Cancer Res., 54: 5269-5272, 1994. the position of a single acetylaminofluorene adduct within a mutation hot spot. Proc. Nati. Acad. Sci. USA, 86: 4147-4151, 1989. 53. Spruck, C. H., Gonzalez-Zulueta, M., Shibata, A., Simoneau, A. R., Lin, M. F.. 31. Grollman. A. P.. and Shibutani. S. DNA Adducts: identification and Biological Gonzales, F., Tsai, Y. C., and Jones, P. A. piÃ³ gene in uncultured tumours. Nature (Lond.), 370: 183-184. 1994. Significance. Mutagenic specificity of chemical carcinogens as determined by studies of single DNA adducts. IARC Sci. Pubi.. /25: 385-397, 1994. 54. Okamoto, A., Hussain, S. P., Hagiwara, K., Spillare. E. A.. Rusin. M. R., Demetrick, 32. Mendelman. L. V.. Petruska. J.. and Goodman. M. F. Base mispair extension kinetics. D. J., Serrano, M.. Hannon. G. J.. Shiseki, M., Zariwala, M., Xiong, Y., Beach, D. H.. Comparison of DNA polymerase alpha and reverse transcriptase. J. Biol. Chem., 265: Yokota, J.. and Harris. C. C. Mutations in the pl6INK4/MTSl/CDKN2. pl5INK4B/ 2338-2346, 1990. MTS2 and pl8 genes in primary and metastatic lung cancer. Cancer Res.. 55: 33. Singer, B., and Essigmann, J. M. Site-specific mutagenesis: retrospective and pro 1448-1451. 1995. spective. Carcinogenesis (Lond.), 12: 949-955. 1991. 55. Lin. J., Chen, J., Elenbaas, B.. and Levine, A. J. Several hydrophobic amino acids in 34. Bronstein, S. M.. Cochrane, J. E., Craft, T. R.. Swenberg, J. A., and Skopek. T. R. the p53 amino-terminal domain are required for tran.scriptional activation, binding to Toxicity, mutagenicity, and mutational spectra of /V-ethyl-W-nitrosourea in human cell mdm-2 and the adenovirus 5 E1B 55-kD protein. Genes & Dev., 8: 1235-1246. 1994. lines with different DNA repair phenotypes. Cancer Res., 51: 5188-5197, 1991. 56. Hupp. T. R., and Lane. D. P. Allosteric activation of latent p53 tetramers. Curr. Biol., 35. Alani. E., Chi, N. W., and Kolodner, R. The Sacchartnn\ces cerevisiae Msh2 protein 4: 865-875, 1995. specifically binds to duplex oligonucleotides containing mismatched DNA base pairs 57. Bayle, J. H., Elenbaas, B.. and Levine, A. J. The carboxyl-terminal domain of the p53 and insertions. Genes & Dev.. 9: 234-247, 1995. protein regulates sequence-specific DNA binding through its nonspecific nucleic 36. Alani. E., Reenan. R. A., and Kolodner. R. D. Interaction between mismatch repair acid-binding activity. Proc. Nati. Acad. Sci. USA, 92: 5729-5733. 1995. and genetic recombination in Saccharomyces cerevisiae. Genetics, 137: 19-39, 1994. 58. Foord, O. S.. Bhattacharya. P.. Reich. Z.. and Rotter, V. A DNA binding domain is 37. Fishel, R., Lescoe, M. K., Rao, M. R.. Copeland, N. G., Jenkins, N. A.. Garber, J., contained in the C-terminus of wild type p53 protein. Nucleic Acids Res., 19: Kane. M.. and Kolodner, R. The human mutator gene homolog MSH2 and its 5191-5198, 1991. association with hereditary nonpolyposis colon cancer. Cell, 75: 1027-1038, 1993. 59. Wolkowicz, R., Peled, A., Elkind, N. B., and Rotter, V. Augmented DNA-binding 38. Leach, F. S., Nicolaides, N. C., Papadopoulos, N., Liu. B., Jen, J., Parsons, R., activity of p53 protein encoded by a carboxyl-terminal alternatively spliced mRNA is Peltomaki. P.. Sistonen. P.. Aaltonen, L. A.. Nystrom-Lahti. M., Guan. X. Y.. Zhang. blocked by p53 protein encoded by the regularly spliced form. Proc. Nati. Acad. Sci. J.. Meltzer. P. S., Yu, J. W., Kao. F. T.. Chen, D. J. Cerosaletti, K. M.. Foumier, USA. 92: 6842-6846. 1995. R. E., Todd. S., Lewis, T., Leach. R. J., Naylor. S. L., Weissenbach, J., Mecklin. J. P., 60. Jayaraman, J.. and Prives, C. Activation of p53 sequence-specific DNA binding by Jarvinen, H.. Petersen, G. M.. Hamilton, S. R., Green. J.. Jass, J.. Watson. P., Lynch, short single strands of DNA requires the p53 C-terminus. Cell, 81: 1021-1029. 1995.

2136

Downloaded from cancerres.aacrjournals.org on September 30, 2021. © 1996 American Association for Cancer Research. Deletions and Insertions in the p53 Tumor Suppressor Gene in Human Cancers: Confirmation of the DNA Polymerase Slippage/Misalignment Model

Marc S. Greenblatt, Arthur P. Grollman and Curtis C. Harris

Cancer Res 1996;56:2130-2136.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/56/9/2130

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/56/9/2130. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.