Copyright 0 1997 by the Genetics Society of America

Enhanced Deletion Formation by Aberrant DNA Replication in

Catherine J. Saveson and Susan T. Lovett

Department of Biology and Rosenstiel Basic Medical Sciences Research Center, Brandeis University, Waltham, Massachusetts 02254-9110 Manuscript received September 26, 1996 Accepted for publication February 21, 1997

ABSTRACT Repeated genes and sequences are prone to genetic rearrangements including deletions. We have investigated deletion formation in Eschm'chia coli strains mutantfor various replication functions. Deletion was selected between 787 base pair tandem repeats carried either on a ColElderived plasmid or on the E. coli . Only mutations in functions associated withDNA I11 elevated deletion rates in our assays. Especially large increases were observed in strains mutant in dnaQ the E editing subunit of Pol 111, and dnuB, the replication fork . Mutations in several other functions also altered deletion formation: the a polymerase (dna,the y clamp loader complex (holC, dnaX), and the p clamp (dnaN) subunits of Pol I11 and the primosomal proteins, dnaCandpui. Aberrant replication stimulated deletions through several pathways. Whereas the elevation in dnaB strains was mostly recA- and Zeddependent, that in dnaQstrainswas mostly recA- and Zed-independent. Deletion productanalysis suggested that slipped mispairing, producing monomeric products,may be preferentially in- creased in a dnaQ mutant and sister-strand exchange, producing dimeric replicon products, may be elevated in dnaE mutants. We conclude that aberrant Polymerase I11 replication can stimulate deletion events through several mechanismsof deletion andvia both recA-dependent andindependent pathways.

ANDEMLY repeated DNA sequences, common in The RecA-independent mechanism can mediate dele- T the genomes of many organisms, are vulnerable tion between homologies of several nucleotides, al- to deletions and amplifications. Such rearrangements though the efficiency of deletion is improved dramati- may occur between repeated genes, kilobases in length, cally with increased homology (ALBERTINI et al. 1982; or between short segments, of as little as a few nucleo- DIANOVet al. 1991; WINet al. 1991; PIERCEet al. 1991; tides in length. These tandem repeat rearrangements BI and LIU 1994). are common sources of genetic mutation and human DNA replication may play a role in bothmechanisms genetic disease (MEUTH 1989; KRAWCZAK and COOPER of deletionformation. The events that may trigger 1991; Hu and WORTON1992; NELSON1993). RecA-dependent homologous recombination between Genetic analysis of repeatedsequence rearrange- tandem repeats arenot known butcould include a ments in Escherichia coli has revealed that multiple mech- blocked replication fork. Our genetic studies impli- anisms contribute to theprocess. In particular, we have cated the RecF, RecO and RecR homologous recombi- previously investigated the genetic dependenceof spon- nation proteins of E. coli in the RecAdependent dele- taneousdeletion events between 787 base pair (bp) tion pathway LOVE^ et d. 1993; M. Bzymek and S. T. repeated sequences within the tetA gene of E. coli (Fig- Lovett, unpublished results). These functions are also ure 1) (LOVETTet al. 1993). The majority of recovered required for recombinational repair ofDNA lesions deletions occur by a homologous recombination path- that block replication in the cell (TSENGet al. 1994). way requiring the RecA strand transfer protein. How- Therefore, some tandem rearrangements could result ever, a substantial proportion of these events (30%) from recombinational gap-filling reactions in a blocked occur independently of RecA. replication fork as depicted in Figure 2. Unequal cross- Both RecA-independent and-dependent deletion ing-over between sister or recombination formation occurs between homologoustarget se- between the two repeats on the same chromosome dur- quences. However, there is a minimal homology re- ing the course of this repair may lead to deletion of quirement for the Red-dependent pathway and no one of the repeated segments. Alternatively, as shown such requirement for the Red-independent pathway. in Figure 3, a stalled replication fork may be broken; The RecA-dependent mechanism does not appear to recombination of the broken chromosome with its sis- function on homologies less than -200 bp in length ter to restore thefork could result in deletion of tandem (DIANOVet al. 1991; WINet al. 1991; BI and LIU 1994). repeats (ASAI et al. 1994; KUZMINOV1995a,b). DNA replication is considered to be central to the Corresponding author: Susan T. Lovett, Rosenstiel Basic Medical Sci- ences Research Center MSO29, Brandeis University, Waltham, MA RecA-independent mechanisms that lead to genetic re- 0225491 10. E-mail: [email protected] arrangements. Experimental evidence supports a repli-

Genetics 146: 457-470 (June, 1997) 458 C. J. Saveson and S. T. Lovett fetA+ :.:.:.:.:.:.;.:.:..:.:.:.:.:.:.:.:.: :.:.:.:.:.:.:.:..:+:+:.:.:.:

tetAdu EmRV Nrul :.:.:.:.:.:.:.: ...... :::::::::I

FIGLIRE1.-Diagram of the ktA locus and the 787-bp dupli- cation used in these experiments. Duplication of the segment from the EcoRV to the NmI sites of the gene disrupts the tdA gene. Deletion of one copy of the tandem repeat restores the IdA' gene structure and can be selected by tetracycline- resistance. cative mechanism for the RecA-independent deletion of both short (TRINH and SINDEN1991) and long (LO- VETT and FESCHENKO1996) tandem repeats. Exposure C. Branch migration J/ of singlestrand DNA during the replication process may provide an opportunity for tandem sequence ho- mologies to interact. A simple slipmispairing model (Figure 4A) proposes that slipped realignment between the nascent strand and its template leads to deletions or amplifications (STREISINCERet al. 1967; ALBERTINI et al. 1982). Additional models propose that deletionsmay arise by strand misalignment within a replication fork D. Resynthesis J. *rn??%?*?A?Y*?* ..,. .,,,,,!Rx*~:?~w$$~??x~ :.:.. concomitant with sisterchromosome exchange (Figure 3 :;I := K. ~:II~~~~~:*~.<; := ...... :s., ::::., .:+ n..6:: ...... 4B) (LOVEITet al. 1993; BI and LIU 1996). The location ...... :.:.>...... x& ...... >:.:+>:.:.:.:.:.:.>>:.:.>:.:.:. of these slipped mispairing events within closely juxta- ...... <...... +?? posed and singlestranded regions of the replication ...... &. .:.:...... :.:...:.:., .*.:..<*+- fork would explain why a strand transfer protein such >.*.Mm$$:>y* w k.& g< .:$;*??x%K&Y -< fi:::..<+wa &% -> * * e s:;*; ss:*... as Red is not needed for such reactions. ... Because DNA replication is implicated in both RecA- dependent and -independentdeletion mechanisms, we E. Junctionresolution c have investigated the genetic relationship between DNA replication components and deletion formation. There are three DNA in E. coli: Polymerase I11 bears the major responsibility for chromosomal replica- tion, Polymerases I and I1 are both believed to be in- volved in repair pathways. Polymerase I is required for some plasmid replication initiation and may also play FIGURE2.-Recombinational gapfilling (post-replication) repair. (A) DNA synthesis cannot proceed past a replication- arole inOkazaki fragment joining (KORNBERGand blocking lesion, giving rise to a daughterstrand gap. (B) The BAKER1992). gap can be filled by recombination with the sister chromo- Polymerase I11 itself is a complex enzyme with multi- some. (C) Branch migration of the Holliday junction can ple subunits encoded by different genes (BAKER and bypass the lesion and cause the incomplete nascent strand to WICKNER1992; MARIANS 1992; KELMAN and O'DONNELL switch templates. (D) Replication is completed on the sister template. (E) The Holliday junction is resolved. This particu- 1995; MARIANS 1996). The polymerase activity is pro- lar resolution produces a cross-over between sister chromo- vided by the a subunit, encoded by dnaE. Its proofread- somes. ing e subunit, possessing activity, is en- coded by dnaQ. The function of the final subunit of O'DONNELL1991). The subunit,encoded by dnaN, the core polymerase, 8, encoded by hoE, is apparently is a factor that forms a clamp around the nonessential (STUDWELL-VAUGHANandO'DONNELL DNA to anchorthe polymerase (STUKENBERGet al. 1993; SLATERet al. 1994). It is believed that both leading 1991). This clamp is loaded by a polymerase-associated and lagging strandsare coordinately synthesized by complex (O'DONNELL1987), which includes the y an asymmetric polymerase dimer (JOHANSON and (also encoded by dnaX), 6 (holA), 6' (holB), x (holCj, MCHENRY1984). The T subunit, encodedby dnuX, facil- and t,b (hoD) proteins. itates the interaction between the two core polymerase The , encoded by dnaC, must be recruited to molecules (MCHENRY1982; STUDWELL-VAUGHANand the replication fork for synthesis of RNA primers. This Hyperdeletion in Replication Mutants 459

A. deletion products from selected strains. We present evi- dence that aberrantor incomplete replication via DNA polymerase I11 stimulates deletion at tandem repeats via several mechanisms.

J, MATERIALSAND METHODS B. Bacterial strains, media and antibiotics: Strains were grown on LB media: 1% Bacto-tryptone, 0.5% yeast extract, 0.5% sodium chloride and 0.005% thymine, and, for plates, 1.5% agar (WIl.LETTS et dl. 1969). Growth temperatures are indi- cated in the text. Minimal complete media consisted of 56/ 2 salts (WILLETTSet al. 1969),0.2% glucose, 1 pg/ml thiamine, JI 50 pg/ml each of required amino acids and, for plates, 2% C. Degradation and recombination agar. LCG media for preparation of P1 lysates and transduc- with sister chromosome dqradatQn tions consisted of LB media supplemented with 1% glucose and 2 mM calcium chloride, and, for plates, 1% agar. A 0.7% agar concentrationwas used for LB and LCG top agar. Media were supplemented with antibiotics: kanamycin (Km) at 60 pg/ml, tetracycline (Tc)at 8-15 pg/ml, chloramphenicol (Cm) at 15 pg/ml and ampicillin (Ap) at 100 pg/ml (for plasmid resistance genes) and 30 pg/ml (for chromosomal genes). J. Isogenic strains used for deletion assays were constructed by P1 transduction and aredescribed in Table 1. These strains include derivatives of AB1 157 for plasmid deletion assays (A) and STL695 for chromosomaldeletion assays (B). P1 uirA phage lysates and transduction were performed as described (MILLER1992). For many strains, a TnlQkan element linked to the mutation of interest was used for selection in strain FIGURE3.-Replication fork collapse and repair(Kuzminov constructions. In thesecases, control strains were constructed 1995b). (A) Inhibition of polymerization stalls the replication carrying only the TnIOkun element to ensure thatthis marker fork. (B) Nuclease attack of single-strand DNA breaks the did not contributeto the deletionphenotype. Deletion assays fork. (C) Thebroken end invades its sister chromosome. Dur- confirmed that deletion rates in these control strains were ing this step unequal recombination can occur at a repeated within 50% of the wild-type value (data not shown). Mutants sequence (dark hatching). (D) Thefork is reestablished with in dnuQ were assayed for mutator phenotype on rifampin a deletion on one chromosome. plates (at 100 pg/ml). Deletion assays: Deletion formation was measured using association is mediated by interactions with the dnaB plasmid pSTL55 (LOVE= et al. 1993). Thisplasmid is a deriva- tive of pBR322, conferring ampicillin resistance, and contains protein, also the replication fork helicase (TOUGUet al. a tandem repeat within the tetA gene (Figure 1). The dupli- 1994). The primase, like the P clamp, is required at cated region spans 787 bp between the EcoRV and NruI sites the start of each Okazaki fragment, placing a potential of tetA. Plasmid DNA was introduced into strains via electro- greater burden on these functions for lagging strand poration (DOWERet ul. 1988).Chromosomal deletion fre- synthesis. Other proteins that are required for primo- quencies were measured in strains carrying an insertion of bla and tetAdup787 from pSTL55 on the E. coli chromosome some assembly and may also be componentsof the pri- at lacZ (LOVE= et al. 1993). Deletion of one tandem repeat mosome include the piA, @‘B, pzC, dnaC and dnaT results in recovery of a functional tetA gene and confers resis- gene products (ARAI and KORNBERG1981). An alterna- tance to tetracycline. tive primosome assembly occurs at the E. coli origin of Strains containing either thechromosomal or plasmid con- replication, using the dnd protein in addition to dnaB struct were grown at their permissive temperatures on LB plates containing ampicillin. Individual colonies were then and dnaC (FUNNELLet al. 1987; MASN et al. 1990). picked whole, and grown in LB+Ap broth for 2 hr. The cul- In this work, we have investigated whether aberrant tures were diluted in 56/2 buffer and the numberof Ap‘ and replication promotes deletions at repeated sequences. Tc‘cells in the population was determined by plating cell We constructed a series of isogenic strains containing dilutions on LB+Ap and LB+Ap+Tc media. Because of poor mutations that impairselected components of the repli- growth in rich medium, polA and pnA mutant strains were grown on minimal-complete agar plates and in minimal-liquid cation machinery. Because there are three DNA poly- media with the appropriate antibiotics. Deletion assays were merases in E. coli, we have tested the genetic depen- performed for 210 independent cultures in parallel with the dence of deletion on each polymerase gene as well as wild-type control strain. Deletion rates were calculated by the other functions associated with DNA replication. E. coli method of the median (Lw and COULSON1949) for the en- mutant strains were surveyed for effects on deletion tire set of assays using the formula: deletion rate = M/N, where M is the calculated number of deletion events and N formation using both plasmid and chromosomal dele- is the final average number of Ap‘ cells in the 1-ml cultures. tion assays. To gain insight into themechanisms of dele- Mis solved by interpolation from experimental determination tion during aberrant replication, we also analyzed the of ro, the median number of Tc‘ cells determined among the 460 C. J. Saveson and S. T. Lovett

A. Simple slippedmispairing B. Sister-strandslipped mispairing

f Nascent strand wk::~s7~::;*;;~x...... ~.:.~.~.:.:.~.:.:~...."...... :.:.:.:. '.' 1II ...... ????>% ...... ".::.:.A...... E?%dR???%?.,.!?...... :*