This document describes a study that investigated the interaction between nick-directed DNA loop repair and mismatch repair in human cells. The study constructed DNA substrates containing combinations of mismatches and loops. It was demonstrated that a nick 3' to a large loop can direct loop repair in human cell extracts in a bidirectional manner. However, when a mismatch was near a loop, the efficiency of nick-directed mismatch repair was reduced. This suggests interference between loop repair and mismatch repair when sites are adjacent to avoid double-stranded DNA breaks.

1. Interaction of Nick-directed DNA Mismatch Repair and Loop Repair
in Human Cells*
Received for publication, February 16, 2004, and in revised form, March 25, 2004
Published, JBC Papers in Press, May 18, 2004, DOI 10.1074/jbc.M401675200
Yao-Ming Huang‡, Shee-Uan Chen§, Steven D. Goodman¶, Shang-Hsin Wu‡, Jau-Tsuen Kao‡ʈ,
Chun-Nan Lee‡ʈ, Wern-Cherng Chengʈ, Keh-Sung Tsaiʈ, and Woei-horng Fang‡ʈ**
From the ‡School of Medical Technology and the §Department of Obstetrics and Gynecology, College of Medicine, National
Taiwan University, and the ʈDepartment of Laboratory Medicine, National Taiwan University Hospital, Taipei, Taiwan
100-63, Republic of China and the ¶Division of Diagnostic Sciences, University of Southern California School of Dentistry,
Los Angeles, California 90089-0641
In human cells, large DNA loop heterologies are re-
paired through a nick-directed pathway independent of
mismatch repair. However, a 3؅-nick generated by bac-
teriophage fd gene II protein heterology is not capable
of stimulating loop repair. To evaluate the possibility
that a mismatch near a loop could induce both repair
types in human cell extracts, we constructed and tested
a set of DNA heteroduplexes, each of which contains a
combination of mismatches and loops. We have demon-
strated that a strand break generated by restriction en-
donucleases 3؅ to a large loop is capable of provoking
and directing loop repair. The repair of 3؅-heterodu-
plexes in human cell extracts is very similar to that of
5؅-heteroduplex repair, being strand-specific and highly
biased to the nicked strand. This observation suggests
that the loop repair pathway possesses bidirectional re-
pair capability similar to that of the bacterial loop re-
pair system. We also found that a nick 5؅ to a coincident
mismatch and loop can apparently stimulate the repair
of both. In contrast, 3؅-nick-directed repair of a G-G
mismatch was reduced when in the vicinity of a loop (33
or 46 bp between two sites). Increasing the distance
separating the G-G mismatch and loop by 325 bp re-
stored the efficiency of repair to the level of a single
base-base mismatch. This observation suggests interfer-
ence between 3؅-nick-directed large loop repair and con-
ventional mismatch repair systems when a mispair is
near a loop. We propose a model in which DNA repair
systems avoid simultaneous repair at adjacent sites to
avoid the creation of double-stranded DNA breaks.
DNA loop heterologies can occur as a consequence of DNA
biosynthetic errors or as a result of recombinational strand
transfer between heterologous sequences (1–4). These DNA
loop structures are mutagenic if they are not corrected. It is
known that the DNA mismatch repair system in both pro-
karyotes and eukaryotes is capable of correcting small DNA
loops (for reviews, see Refs. 5–10). The Escherichia coli mis-
match repair system can efficiently process DNA loops up to 7
unpaired nucleotides (11). Genetic studies suggest that, in
yeast, the mismatch repair system may be capable of correcting
DNA loops up to 13 unpaired nucleotides (12). Functional in
vitro mismatch repair assays have shown that the human
mismatch repair pathway can repair DNA loops with at least 8
unpaired nucleotides (13).
There is accumulating evidence suggesting that DNA loop
repair can be carried out by a pathway distinct from the mis-
match repair system (13–19). Conflicting results have been
obtained regarding the strand specificity and the role of strand
breaks in mammalian loop repair. Although single-stranded
loops of varying sizes are repaired efficiently in vivo, there is a
2:1 bias against loop retention (20). Nicks have a minor effect
on strand discrimination for biased repair of 25–57-base single-
stranded loops in monkey cells (21). However, in vitro func-
tional assays have been used to demonstrate that, in the ab-
sence of nicks, repair of multi-based loops is inefficient. Umar
et al. (14) and Littman et al. (16) have identified a vigorous
5Ј-nick-directed loop repair activity of this type using mismatch
repair-deficient extracts and DNA substrates containing 5–16
(14) or Ն27 (16) nucleotides in the loop. Although Umar et al.
(14) found that a 3Ј-nick can direct repair of 2–5-nucleotide
loops, they could not rule out that at least some of this activity
could be attributed to the mismatch repair system. In contrast,
the latter study observed no significant repair of this type for
27- or 216-nucleotide loops (16). Recently, McCulloch et al. (22)
demonstrated that heteroduplexes containing 2–12-nucleotide
loops can be processed bidirectionally by mismatch repair-de-
pendent and -independent pathways in human cells. A study of
the repair mechanism revealed that the processing of the het-
erology in substrates with a 3Ј-strand break is consistent with
the involvement of endonucleases as yet unidentified (22). Mc-
Culloch et al. (23) also found that large loops can be processed
through both a nick-directed and loop-directed process.
Recent studies of E. coli loop repair in our laboratory showed
that a strand break located either 3Ј or 5Ј to the loop is suffi-
cient to direct heterology-dependent repair to nicked strands in
mismatch repair-deficient cell extracts (19). Consequently, it
appears that bacterial nick-directed loop repair has a broader
substrate specificity than that observed in the human pathway
(16). However, the design of 3Ј-nicked substrates in our study is
different from previous eukaryotic loop repair experiments
(19). Thus, a potential activating role of 3Ј-strand break in
human loop repair still needs to be clarified.
Although transformation experiments with E. coli suggest
that loops can be indirectly co-repaired by the dam-directed
mismatch repair system (24), analogous experiments in human
cells have not been reported. Taghian et al. (25) did, however,
evaluate an in vivo extrachromosomal recombination assay in
* This work was supported by Grants NSC91-2314-B-002-243 (to
W.-h. F.) and NSC88-2314-B-002-092-M40 (to J.-T. K.) from the Na-
tional Science Council (Taipei, Taiwan, Republic of China). The costs of
publication of this article were defrayed in part by the payment of page
charges. This article must therefore be hereby marked “advertisement”
in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
** To whom correspondence should be addressed: School of Medical
Technology, College of Medicine, National Taiwan University, 7 Chung-
Shan South Rd., Taipei, Taiwan 100-63, Republic of China. Tel.: 886-
2-2312-3456 (ext. 6926); Fax: 886-2-2371-1574; E-mail: whfang@
ha.mc.ntu.edu.tw.
THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 279, No. 29, Issue of July 16, pp. 30228–30235, 2004
© 2004 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S.A.
This paper is available on line at http://www.jbc.org30228
atNationalTaiwanUniversity,onJanuary17,2011www.jbc.orgDownloadedfrom

2. Chinese hamster ovary cells. They identified a loop repair
activity that processes predicted loop heterologies consistent
with the single-strand annealing model of recombination inter-
mediates. Moreover, this in vivo loop repair activity is inde-
pendent of the general (enzyme complexes of MutS homologs/
enzyme complexes of MutL homologs) mismatch repair (25, 26).
However, in the presence of single-base mismatches, the loop
repair is inhibited (25). Interestingly, in mismatch repair-defi-
cient cells, single-base mismatches are repaired more effi-
ciently in the presence of a loop, consistent with a co-repair
initiated at the loop (25).
We therefore constructed a set of heteroduplexes containing
a combination of loops and mispairs to study human large loop
repair and its interaction with mismatch repair pathways. We
have demonstrated here that nuclear extracts derived from
HeLa cell lines support efficient 3Ј-nick-directed large loop
repair in vitro. We also found a negative interference between
3Ј-nick-directed large loop repair and conventional mismatch
repair systems when a mispair is in the vicinity of a loop. This
may have broad implications for heterologous DNA repair sys-
tems functioning in close proximity to one another.
EXPERIMENTAL PROCEDURES
Materials—E. coli strains NM522 and RS5033, bacteriophage
f1MR1, and the nucleotide numbering at restriction endonuclease
cleavage sites have been described (11, 27). HeLa S3 and LoVo cells
were grown at 37 °C in 5% CO2 atmosphere as described (16). Nuclear
extracts were prepared as described (28). ATP-dependent DNase, E. coli
DNA ligase, T4 polynucleotide kinase, calf intestinal alkaline phospha-
tase, and restriction endonucleases were obtained from commercial
sources.
Construction of f1PM Mutants—A 200-bp HindIII-NheI fragment
(coordinates 29–229) from pBR322 was inserted into f1MR1 at the NheI
and HindIII cleavage sites (coordinate 5621 and 5627 of f1MR1). Sub-
sequently, a 24-bp fragment from phage M13mp18 XbaI and HindIII
digestion was inserted into f1MR1 at the HindIII and XbaI cleavage
sites (coordinate 5821 and 5832 of f1MR1). The phage in this construc-
tion, which retains the HindIII site, is designated f1PM. The phage was
further mutagenized by insertion of a synthetic 21-bp oligonucleotide
linker into the HindIII cleavage site, a 24-bp oligonucleotide linker into
the NheI cleavage site, or a synthetic 32-bp linker into the XbaI cleav-
age site (see Fig. 1 and Table I). As shown in Table I, a randomized site
was used in each oligonucleotide linker so that all four nucleotide
permutations could be obtained from each insertion. Each of the vari-
ants contained a new unique restriction endonuclease recognition se-
quence (see Table I). Phages f1hG-N and f1hC-N-xG were constructed
by insertion of a 292-bp NspI restriction fragment derived from plasmid
pBR322 (coordinates 1816–2108) into SphI-cleaved f1hG and f1hCxG,
respectively. Mutant f1PM phages were identified by restriction anal-
ysis of replicative form minipreparations, and mutant sequences were
confirmed by DNA sequencing.
Construction of Heteroduplex DNA—Heteroduplex DNA substrates
were constructed essentially as described by Lu et al. (29). Phage f1PM
derivative replicative form DNA (1 mg) was linearized with AlwNI,
BanII, or EcoRV and mixed with a 4-fold molar excess of viral DNA,
followed by alkaline denaturation and annealing. After isolation by
hydroxylapatite chromatography, double-stranded DNA was dialyzed,
and linear homoduplex DNA was removed by digestion with ATP-de-
pendent DNase (28). The open circular heteroduplex was purified by
Sephadex G-200 chromatography (Amersham Biosciences) and benzo-
ylated naphthylated DEAE-cellulose chromatography (Sigma) as de-
scribed previously (30). The covalently closed circular substrates were
prepared by E. coli DNA ligase treatment of nicked substrates in the
presence of ethidium bromide (96 mmol of dye/mol of nucleotide) and
then isolated by equilibrium centrifugation in CsCl/ethidium bromide
(29). Substrates containing a nick at the replication origin of the viral
strand were prepared by cleaving covalently closed circular heterodu-
plexes with bacteriophage fd gene II protein (gpII)1
as described (31).
The substrates used in this study are summarized in Table II. By
pairing different f1PM derivatives, a heteroduplex containing a combi-
nation of base-base mismatches and/or a loop can be constructed. The
location of the strand break can be determined by the restriction endo-
nuclease employed to cleave the replicative form DNA used in the
construction (i.e. the following sites at the following locations: AlwNI,
position 2192; EcoRV, position 5667; and BanII, position 5887). Strand
breaks were placed in complementary strands either 3Ј or 5Ј to the
mismatch (see Fig. 1).
Mismatch Repair and Endonuclease Assays—Loop repair in human
nuclear extracts was determined in a manner similar to that described
for human mismatch repair (28). The repair reaction using concen-
trated nuclear extracts was carried out in 40 ␮l containing 0.02 M
Tris-HCl (pH 7.6), 5 mM MgCl2, 0.11 M KCl (derived by titrating extracts
with KCl to determine the optimal concentration), 50 ␮g/ml bovine
serum albumin, 1 mM ATP, 0.1 mM dATP, 0.1 mM dGTP, 0.1 mM dTTP,
0.1 mM dCTP, and 0.4 ␮g (92 fmol) of heteroduplex DNA. The addition
of cell-free human nuclear extracts was optimized to 7.5 mg/ml protein.
Incubation was at 37 °C for 15 min. The reactions were quenched by the
addition of 40 ␮l of 40 mM EDTA (pH 8.0) and 1% SDS, followed by
incubation with proteinase K at 37 °C for 30 min. The DNA was then
purified by phenol extraction and ethanol precipitation, divided into
four aliquots, and analyzed by restriction endonuclease digestion and
agarose gel electrophoresis. The ethidium complexes of DNA products
were quantified using a gel documentation CCD camera (UVP Inc.) (11).
RESULTS
Substrate Specificity of Mismatch Repair in HeLa Cell Nu-
clear Extracts—Using circular f1MR heteroduplex substrates,
previous studies have shown that nuclear extracts of HeLa
cells can at least repair all eight possible base-base mismatches
in a strand-specific manner targeted by a single-strand break
located 808 or 125 bp 5Ј to the mismatches in the complemen-
tary DNA strand (28, 30). Due to the limitation of unique
restriction endonuclease sites in the f1MR DNA, it is not pos-
sible to place a strand break 3Ј to the mismatch in the comple-
mentary strand. A covalently closed circular heteroduplex
treated with gpII protein and MutH can generate strand
breaks 181 and 1024 bp 3Ј to the mismatch in the viral strand
(31). However, these substrates are difficult to make, and the
product yield is very low; thus, only G-T heteroduplexes were
made and tested (31). We therefore designed a new set of
circular heteroduplex molecules in which strand breaks 3Ј to
the mismatch can be conveniently placed in the complementary
strand.
Starting from phage f1MR1 (27), we prepared a set of four
f1h derivatives that contain single-base differences (Fig. 1 and
Table I). This set of f1h derivatives permits construction of
heteroduplexes representing all eight possible base pair mis-
matches. In each of these heteroduplexes, the heterology is
located at the same position (coordinate 5833). Moreover, as
shown in Table II, each mismatch is located within overlapping
restriction endonuclease recognition sites; and in every case,
the rest of the sequences are identical. The generation of over-
lapping restriction sites was accomplished while maintaining
the overall sequence environment by perfect alignment of four
restriction endonuclease recognition sequences with only a sin-
gle-base difference. This set of heteroduplexes allows correction
on either DNA strand to be directly determined under condi-
tions in which the effects of sequence environment on repair
efficiency are minimized. Digestion of the heteroduplex DNA
with AlwNI (Fig. 1) and the indicator restriction endonuclease,
whose recognition site was inactive because of the heterology,
yielded a 6.6-kb fragment only (Fig. 2A). Similar digestion of
DNA in which the recognition sequence had been restored by
repair reaction yielded 3.0- and 3.6-kb fragments (Fig. 2C).
As shown in Table III, all eight base-base mismatch hetero-
duplexes are subject to strand-specific correction directed by a
strand break 165 bp 3Ј or 75 bp 5Ј to the mismatch in the
complementary strand. As judged by repair levels, all 5Ј-sub-
strates were effectively repaired, with efficiency comparable
with that observed in previous studies (30). All the 3Ј-hetero-
duplexes were also effectively repaired with efficiency similar
to the repair of 5Ј-substrates (Table III). It is evident that the1
The abbreviation used is: gpII, gene II protein.
Interaction of Human Mismatch Repair and Loop Repair 30229
atNationalTaiwanUniversity,onJanuary17,2011www.jbc.orgDownloadedfrom

3. repair specificity is dependent on the mismatch recognition and
fails to distinguish the polarity of the strand break.
Strand-specific Loop Repair Directed by a Nick 3Ј to the
Heterologies—A previous study has shown that mismatch-in-
dependent repair of a DNA loop that is 27 nucleotides in length
or larger in human cells can be directed only by a 5Ј-nick, but
not a 3Ј-nick (16). To further characterize this observation, we
prepared f1PM-based looped substrates containing a nick gen-
erated by restriction endonucleases in the complementary
strand 3Ј to a 32-nucleotide loop (Fig. 1B and Table II, xC32
and xV32). In contrast to the previous results, the looped het-
eroduplexes with 3Ј-nicks were effectively processed by human
cell extracts. As shown in Fig. 2D and Table IV, our results
suggest that 3Ј-nicks can efficiently direct loop repair in a
FIG. 1. Heteroduplex substrates. The map of circular 6647-bp f1PM phage shows restriction sites relevant to this study. The coordinates and
locations of the single-strand breaks tested are indicated by lines that contact one DNA strand only. Combinations of different insertion derivatives
allowed us to prepare the variety of heteroduplexes shown. A–G are types of heteroduplexes prepared for this study. C, complementary strand; V,
viral strand.
TABLE I
Construction of f1PM derivatives for heteroduplex preparation
Bacteriophage f1PM derivatives were constructed by insertion into the following synthetic duplexes at specific sites. With different substitutions
at position N, a set of f1 phages was constructed for this study.
At the HindIII site,
V 5Ј-AGCTACGGTCCNTAAGGTGGT
C TGCCAGGNATTCCACCATCGA-5Ј
At the NheI site,
V 5Ј-CTAGTCGGCCNACGTGAGGATGGA
C AGCCGGNTGCACTCCTACCTGATC-5Ј
And at the XbaI site
V 5Ј-CTAGTGTCTTGATGCCTGTCCGGNACCTCCTA-3Ј
C ACAGAACTACGGACAGGCCNTGGAGGATGATC-5Ј
f1 mutant
Insertion site and substitution at position N
HindIII Marker XbaI Marker NheI Marker
f1hA A PflMI
flhC C Bsu36I
f1hG G CpoI
f1hT T AflII
flxG G PpuMI
f1hAxG A PflMI G PpuMI
f1hCxG C Bsu36I G PpuMI
f1hGxG G CpoI G PpuMI
f1hTxG T AflII G PpuMI
f1hGxC G CpoI C BglI
f1nGhCxC C Bsu36I C BglI G EagI
f1hG-N G CpoI
f1hC-N-xG C Bsu36I G PpuMI
Interaction of Human Mismatch Repair and Loop Repair30230
atNationalTaiwanUniversity,onJanuary17,2011www.jbc.orgDownloadedfrom

4. strand-specific way. The repair levels were comparable with
corresponding 5Ј-heteroduplexes in Table IV and previously
reported 5Ј-heteroduplexes (16). Repair of these DNAs dis-
played a substantial bias (2.5–10-fold) toward the nicked DNA
strand.
To ascertain that 3Ј-nick-directed loop repair is a general
pathway, we tested a subset of looped M13LR heteroduplex
substrates (19). The results obtained with M13LR heterodu-
plexes further confirmed that human cell extracts can process
3Ј-nicked loops (Table IV). All of the M13LR 3Ј-heteroduplexes
containing loops of 16–45 nucleotides were processed with
levels and strand bias similar to those of f1PM substrates.
Significant repair was also observed for the 3Ј-heteroduplexes
containing a 429-nucleotide unpaired region, but this substrate
was processed with less strand bias than those containing the
smaller non-homologous segment. We attribute the higher loop
removal activity in the continuous viral strand of V429 as the
loop-directed reaction described by McCulloch et al. (23).
Whereas all of the 3Ј-heteroduplexes generated by restriction
endonuclease digestion contained the strand break in the com-
plementary strand, tested substrates included several with the
loop present in either the complementary or viral strand (Table
IV). Consequently, the strand-specific asymmetry observed for
repair of 3Ј-heteroduplexes cannot be attributed to the pres-
ence of a simple loop in a particular strand. Rather, this effect
must be due to the 3Ј-strand break. Furthermore, the sequence
TABLE II
Heteroduplex substrates
Circular heteroduplexes containing a set of base pair mismatches
and/or a loop were prepared using the phage DNAs shown. The types of
heteroduplex are as shown in Fig 1. Complementary and viral strands
are designated C and V, respectively.
Type Heteroduplex V strand C strand
1A T-G f1hT f1hC
1A T-T f1hT f1hA
1A A-G f1hA f1hC
1A A-A f1hA f1hT
1A G-A f1hG f1hT
1A G-G f1hG f1hC
1A A-C f1hA f1hG
1A C-C f1hC f1hG
1B xC32 f1PM f1xG
1B xV32 f1xG f1PM
1C TGxC32 f1hT f1hCxG
1C TGxC32 f1hT f1hCxG
1C TTxC32 f1hC f1hAxG
1C CTxC32 f1hC f1hAxG
1C AAxC32 f1hG flhTxG
1C GAxC32 f1hG f1hTxG
1C GGxC32 f1hA f1hGxG
1C ACxC32 f1hA f1hGxG
1C CCxC32 f1hC f1hGxG
1D GGxV32 flhGxG f1hC
1E hC21xGG flxG f1hGxC
1F hGG-N-xC32 f1hG-N f1hC-N-xG
1G nC24-3Ј-GGЈ f1nGhCxC f1hGxC
FIG. 2. Repair of heteroduplexes containing both mismatches
and loops in HeLa nuclear extracts. Repair reactions with HeLa
nuclear extracts were performed as described under “Experimental
Procedures.” DNA products were digested with AlwNI (Fig. 1) and the
appropriate restriction endonuclease (Tables I and II) and were then
subjected to agarose gel electrophoresis to score mismatch repair occur-
ring on each DNA strand. A, heteroduplex 3Ј-AAxC32 treated with
heat-inactivated extracts and then hydrolyzed with diagnostic restric-
tion endonuclease; B, repair reaction of 3Ј-AAxC32; C, repair reaction of
3Ј-A-A; D, repair reaction of 3Ј-xC32. Bars pointing to the 6.6-kb frag-
ment represent unrepaired substrates, and bars pointing to 3.6- and
3.0-kb fragments indicate corrected products. o, repair occurring on the
open (nicked) complementary strand; c, repair occurring on the closed
viral strand.
TABLE III
Efficiency of mismatch repair in HeLa nuclear extracts
Repair was measured in nuclear extracts (7.5 mg/ml protein) as
described under “Experimental Procedures.” 3Ј-Substrates are hetero-
duplexes with a strand break 165 bp 3Ј pump to the mismatch; 5Ј-
substrates have a strand break 75 bp 5Ј to the mismatch. o, repair
occurring on the open (nicked) strand; c, repair occurring on the closed
strand. Repair results are the averages of two experiments. Complete
repair would correspond to 23 fmol.
Heteroduplex
Repair
3Ј-Substrate 5Ј-Substrate
o c o c
fmol
T-G 11.4 Ϯ 0.9 2.8 Ϯ 1.6 12.0 Ϯ 2.0 2.3 Ϯ 0.3
T-T 12.9 Ϯ 0.3 2.1 Ϯ 1.1 12.0 Ϯ 2.2 2.4 Ϯ 0.8
C-T 13.2 Ϯ 1.5 2.2 Ϯ 1.9 13.8 Ϯ 0.8 1.8 Ϯ 1.7
A-A 12.4 Ϯ 0.6 2.1 Ϯ 1.3 11.9 Ϯ 1.9 1.8 Ϯ 1.3
G-A 13.0 Ϯ 1.1 1.9 Ϯ 0.6 12.2 Ϯ 1.2 1.7 Ϯ 1.0
G-G 12.7 Ϯ 1.7 3.9 Ϯ 0.6 12.1 Ϯ 2.2 2.8 Ϯ 1.3
A-C 12.3 Ϯ 1.0 1.6 Ϯ 1.1 11.9 Ϯ 0.7 2.2 Ϯ 0.3
C-C 8.3 Ϯ 3.2 1.9 Ϯ 0.2 6.7 Ϯ 3.8 1.2 Ϯ 1.1
TABLE IV
Efficiency of 3Ј- and 5Ј-nick-directed loop repair in HeLa
nuclear extracts
Repair was measured in nuclear extracts (7.5 mg/ml protein) as
described under “Experimental Procedures.” f1PM-based 3Ј-substrates
are heteroduplexes with a strand break 165 bp (3Ј-hC21) or 198 bp
(3Ј-xC32 and 3Ј-xV32) 3Ј to the loop; M13mp 18-based 3Ј-substrates
have a strand break 49 bp (3Ј-V22, 3Ј-V429, and 3Ј-C27) or 70 bp
(3Ј-C16) 3Ј to the loop. o, repair occurring on the open (nicked) strand;
c, repair occurring on the closed strand. Repair results are the mean Ϯ
S.E. determined from at least three separate experiments. Complete
repair would correspond to 23 fmol for f1PM-derived substrates and 21
fmol for M13mp18-derived heteroduplexes.
Heteroduplex
Repair
3Ј-Substrate 5Ј-Substrate
o c o c
fmol
f1PM
xC32 5.9 Ϯ 0.9 0.5 Ϯ 0.5 4.7 Ϯ 0.5 1.5 Ϯ 0.4
xV32 7.2 Ϯ 0.3 2.4 Ϯ 0.3
hC21 4.7 Ϯ 0.2 2.0 Ϯ 0.5
M13 mp18
C16 8.8 Ϯ 0.3 0.9 Ϯ 0.2 4.5 Ϯ 0.3 0.4 Ϯ 0.4
V22 3.8 Ϯ 0.3 1.7 Ϯ 0.02 6.3 Ϯ 0.1 0.8 Ϯ 0.2
C27 8.6 Ϯ 0.5 1.6 Ϯ 0.09 3.8 Ϯ 0.6 1.7 Ϯ 0.7
V45 3.0 Ϯ 0.3 0.8 Ϯ 0.11 4.8 Ϯ 0.4 0.9 Ϯ 0.3
V429 3.2 Ϯ 0.1 2.5 Ϯ 0.3 5.1 Ϯ 0.3 0.5 Ϯ 0.2
Interaction of Human Mismatch Repair and Loop Repair 30231
atNationalTaiwanUniversity,onJanuary17,2011www.jbc.orgDownloadedfrom

5. contents in the region of heterologies and the nicks are very
different between the f1PM and M13LR (19) looped substrates.
It therefore seems highly unlikely that the repair events are a
consequence of nonspecific activities.
In previous studies, heteroduplexes with a 3Ј-nick were gen-
erated by gpII incision in the viral strand (16). Although these
substrates are suitable for the human mismatch repair system
(31), the same nick may not be conducive for the loop repair
pathway. To clarify this issue, a subset of f1PM-based looped
substrates containing 3Ј-nicks generated by gpII protein were
tested in human extracts. As shown in Fig. 3, both gpII-incised
substrates and covalently closed circular heteroduplexes (ccc)
were only marginally repaired, and the repair levels were much
less compared with substrates containing nicks generated by
restriction endonucleases (3Ј-198). A gpII-generated nick in
M13LR-based C22 (19) located 5Ј to the heterology was also not
effective in stimulating loop repair (data not shown). This
observation indicates that gpII-incised substrate is refractory
to human loop repair because of the nature of the gpII nick
rather than the orientation of the strand breaks. The differ-
ences between nicks generated by restriction endonuclease di-
gestion and gpII incision are considered further under
“Discussion.”
Interaction of Loop Repair and Mismatch Repair in the Same
Heteroduplexes with a Nick 3Ј to the Heterologies—Since pre-
vious studies suggested that a nick 3Ј relative to a large loop is
unable to provoke specific repair in human cells, we wished to
test whether a mismatch near a loop could provoke the repair
of both as shown in E. coli (24). According to previous excision
tract mapping of the human mismatch repair pathway, proc-
essing a 3Ј-heteroduplex would excise the nucleotides from the
strand break to 90–160 nucleotides beyond the mismatch (32).
If a loop is co-repaired during mismatch correction, the dis-
tance between the two sites would need to be within this range.
Based on this assumption, we designed a set of heteroduplexes
to test this idea. f1hx phage were constructed by insertion of
the 32-mer oligonucleotide at the XbaI site of f1h phage (Fig. 1).
Combinations of different f1h and f1hx phage permits construc-
tion of heteroduplexes containing the eight possible base pair
mismatches and a loop of two possible configurations (Table II).
We constructed a set of 3Ј-nick-mismatch-loop heteroduplexes
(Fig. 1, C and D; and Table II) that contained the 32-base loop
in either the complementary or viral strand separated by 33
nucleotides from several possible base-base mismatches. Table
V compares the efficiency of correction of all heteroduplexes as
scored by restriction endonuclease digestion, and Fig. 2B illus-
trates the behavior of one repair example in the restriction
assay. The results obtained with Types 1C and 1D in Table V
show that both mismatches and loops were repaired, but the
levels were lower than those of the repair of the mismatch
(Table III) or loop alone (Table IV). For example, compared
with the results in Tables III and IV, the repair efficiency of
3Ј-GGxC32 was 8.5 fmol for G-G mispair (61% of 3ЈG-G in Table
III) and 5.1 fmol for C32 (85% of 3Ј-xC32 in Table IV). Although
the rates of these reactions may demonstrate an even greater
difference, these measurements are statistically significant.
Unlike the E. coli system (24), the two human repair pathways
appear to be subject to negative interference when a mismatch
is in proximity to a loop.
We also constructed and tested a 3Ј-nick-loop-mismatch het-
eroduplex substrate (Fig. 1E) to determine whether the relative
location of the mismatch and loop to the nick would behave
differently. As shown with Type 1E in Table V, similar to the
results with Types 1C and 1D, the repair levels of both the
mismatch and loop in 3Ј-hC21xGG were lower than those of
otherwise separate heteroduplex repairs. The repair of a G-G
mispair was further decreased to the level of the loop repair. To
understand the details of how adjacent mismatches and loops
are processed in human extracts, repair time courses were
measured for 3Ј-hC21xGG as well as a single mismatch (3Ј-
G-G) and a single loop (3Ј-xC32). As shown in Fig. 4, repair of
a 3Ј-G-G single-base mispair occurred at a higher rate than
that of loop repair of 3Ј-xC32, as expected. However, time
course experiments for 3Ј-hC21xGG revealed that repair of a
G-G mismatch in this heteroduplex was reduced to a level
similar to that of loop repair alone. Moreover, the rate of
processing of the G-G mismatch was almost identical to the
rate of repair of the C21 substrate. The time course analysis of
3Ј-hC21xGG was particularly informative since the repair of
both the loop and mismatch could be evaluated simultaneously.
The apparent reduction in the reaction rate of mismatch repair
to the level of the adjacent loop repair suggests a single path-
way monopolizing the reaction of both lesions. Based on simi-
larities in the reaction rates of loop repair and reduced co-
repair of mismatches and loops, it seems reasonable to infer
that the corrections of both heterologies in 3Ј-hC21xGG are
mediated by the loop repair pathway alone.
The Repair Interference between the Mismatch and Loop in a
3Ј-Heteroduplex Is Due to the Proximity of Two Sites—The
simplest interpretation of our observation of repair interfer-
ence between neighboring mismatches and loops is that two
distinct repair machineries require appropriate spacing for
processing the heterologies. Thus, there is a competition for the
space near the heterologies. To test this idea, we increased the
distance between a G-G mismatch and a C32 loop to 325 bp by
inserting a 292-bp NspI-digested DNA fragment from pBR322
(Fig. 1F). As shown in Table V, the repair efficiency of a G-G
mismatch in the 3Ј-hGG-N-xC32 heteroduplex had reverted
almost to the level of a single G-G mismatch repair (Table III).
Hence, distancing the loop from the mismatch restores signif-
icant strand-specific repair to approximately 65% of single-loop
repair.
In addition to assessing both heterologies on the same side of
the nick, the mismatch and loop were also placed on different
sides of the nick (Fig. 1G). The repair efficiency of a G-G
mismatch remained essentially unchanged compared with the
level of single G-G mismatch repair (93% of 3Ј-G-G repair)
(Table III). However, the correction efficiency of the 24-nucle-
otide loop significantly increased (almost 150% of better re-
paired 3Ј-xV32 in Table IV).
FIG. 3. Loop repair of covalently closed circular heterodu-
plexes and gpII-incised substrates. Repair reactions were per-
formed as described under “Experimental Procedures.” Substrates con-
taining 32 nucleotides of unpaired loop in the complementary strand
(xC32) and in the viral strand (xV32) were assayed. Heteroduplexes of
3Ј-198 contained a BanII-generated strand break in the complementary
strand 198 bp 3Ј to the heterologies. ccc, covalently closed circular
substrate; gpII (3Ј-170), substrates with a gpII-incised strand break in
the viral strand 170 bp 3Ј to the loops. White bars, repair occurring
on the complementary (C) strand; black bars, repair occurring on the
viral (V) strand. The error bars represent one S.D. from three
determinations.
Interaction of Human Mismatch Repair and Loop Repair30232
atNationalTaiwanUniversity,onJanuary17,2011www.jbc.orgDownloadedfrom

6. To further confirm that the interference between the mis-
match and loop in the 3Ј-nicked substrates is due to the prox-
imity of the two sites, heteroduplexes containing single mis-
pairs and single loops were examined by a competition assay in
which 3Ј-A-A was incubated with nuclear lysate in the presence
of an equimolar concentration of 3Ј-V27 substrate. As shown in
Fig. 5, the repair levels of combined substrates were increased
for 3Ј-A-A and unchanged for 3Ј-V27 in comparison with the
results of standard reactions. In this case, it is apparent that
3Ј-A-A and 3Ј-V27 did not compete with the repair machineries
of human cells.
A 5Ј-Nick Stimulates Co-repair of Loops and Mismatches—
Since the results described above suggested that there was an
interference in 3Ј-heteroduplex repair of an adjacent mismatch
and loop in mismatch repair-proficient HeLa extracts, we
wished to test whether a nick placed 5Ј to or between the
heterologies would affect the repair of either or both. We chose
the GGxC32 lesion (Fig. 1C and Table II) for these manipula-
tions and used 3Ј- and 5Ј-substrates of a single G-G mispair
(Fig. 1A) and a single xC32 loop (Fig. 1B) for comparison. To
assess the dependence of the reaction on the location of the
strand break, we prepared modified GGxC32 substrates in
which a single-strand break was placed at a BanII site (Fig. 6).
As shown in Table V, the repair levels of a coincident G-G
mismatch and C32 loop with a nick at an EcoRV site 3Ј to the
heterologies were lower than the repair levels of an individual
mismatch (Table III) or loop (Table IV). In Fig. 6, the results
are arranged for comparison (sets I and III). When the nick was
moved to the BanII site, which is 48 bp 5Ј to the loop and 81 bp
5Ј to the mismatch, it activated the repair of both. As shown in
Fig. 6 (set IV), the repair levels of the G-G mismatch increased
to 120% and those of the C32 loop increased to 180% compared
with the repair levels of single heterology of the same nick
position (Fig. 6, set II). Therefore, a nick 5Ј to an adjacent
mispair and loop apparently can stimulate the repair of both in
mismatch-proficient human cells.
A 3Ј-Nick Stimulates Co-repair of Loops and Mismatches in
Mismatch-deficient Cell Extracts—The results described above
suggest that, in mismatch repair-proficient human extracts,
the loop repair pathway dominates mismatch repair activity for
processing an adjacent loop and mismatch in a 3Ј-heteroduplex.
Here, we decided to test whether the loop repair system alone
is sufficient to repair base mismatches in the same manner as
with co-repair. We first prepared mismatch repair-deficient
LoVo cells for analysis (14, 16). Since it was difficult to grow a
high density of this attached cell line, we managed to make
only a limited amount of lysate with a concomitant lower pro-
tein concentration. Although this led to lower loop repair levels
compared with the HeLa activity described above (65–90% of
HeLa extracts) (Table VI, Type 1B), in several selected 3Ј-
heteroduplexes containing an adjacent loop and mispair, the
LoVo extracts were capable of processing mispairs to levels
similar to those of the correction of loops (Table VI, Types 1C,
1D, and 1E). Thus, 3Ј-nick-directed loop repair in the mutant
cells correlates with the repair of mismatches, a likely conse-
quence of co-repair.
TABLE V
Interaction of mismatch repair and loop repair in HeLa nuclear extracts
Repair was measured in nuclear extracts (7.5 mg/ml protein) as described under “Experimental Procedures.” The type of substrates is as
illustrated in Fig. 1. Repair results are the mean Ϯ S.E. determined from at least three separate experiments. Complete repair would correspond
to 23 fmol.
Type Heteroduplex
Repair
Mismatch Loop
o c o c
fmol
1C 3Ј-GGxC32 8.5 Ϯ 1.2 1.8 Ϯ 0.3 5.1 Ϯ 1.0 1.3 Ϯ 0.3
1C 3Ј-TGxC32 6.8 Ϯ 0.2 2.1 Ϯ 0.6 4.3 Ϯ 0.4 1.2 Ϯ 0.4
1C 3Ј-TTxC32 8.2 Ϯ 0.5 0.6 Ϯ 0.5 4.2 Ϯ 0.6 2.0 Ϯ 0.5
1C 3Ј-AGxC32 8.2 Ϯ 0.9 2.0 Ϯ 0.2 4.8 Ϯ 0.7 1.4 Ϯ 0.4
1C 3Ј-GAxC32 9.7 Ϯ 1.2 2.5 Ϯ 0.5 4.1 Ϯ 0.7 1.8 Ϯ 0.1
1C 3Ј-ACxC32 7.5 Ϯ 0.4 2.1 Ϯ 0.3 4.2 Ϯ 0.4 2.0 Ϯ 0.8
1C 3Ј-AAxC32 7.0 Ϯ 0.5 2.3 Ϯ 0.4 3.7 Ϯ 0.8 1.3 Ϯ 0.1
1C 3Ј-CCxC32 7.3 Ϯ 0.2 1.6 Ϯ 0.1 3.3 Ϯ 0.4 1.2 Ϯ 0.5
1D 3Ј-GGxV32 7.4 Ϯ 1.1 2.4 Ϯ 0.2 4.3 Ϯ 0.5 1.3 Ϯ 0.4
1E 3Ј-hC21xGG 5.2 Ϯ 0.5 2.4 Ϯ 0.3 5.6 Ϯ 0.3 1.3 Ϯ 0.4
1F 3Ј-GG-N-xC32 13.8 Ϯ 0.2 3.9 Ϯ 0.4 3.8 Ϯ 0.9 1.4 Ϯ 0.4
1G nC24–3Ј-GG 13.0 Ϯ 0.2 3.12 Ϯ 0.4 11.1 Ϯ 0.1 3.2 Ϯ 0.4
FIG. 4. Time course of nick-directed correction for mismatches
and loops. Repair reactions were performed as described under “Ex-
perimental Procedures.” At the times indicated, each reaction was
quenched with EDTA and SDS and stored on ice. All samples were
subsequently processed in parallel. The closed circles reflect repair of
the 3Ј-G-G substrate, and the closed squares are the results of the assay
for the 3Ј-xC32 DNA. The open symbols are the data from the 3Ј-
hC21xGG heteroduplex: circles are the data for the G-G mismatch, and
squares are the correction of the C21 loop. These results are the aver-
ages from two (3Ј-G-G) or three (3Ј-xC32 and 3Ј-hC21xGG) independent
measurements.
FIG. 5. Mixed substrate repair of mismatches and loops. Repair
reactions were performed as described under “Experimental Proce-
dures.” Heteroduplex correction of 3Ј-A-A and 3Ј-V27 was performed
separately (Standard), or the reaction was scaled up 2-fold with 23 fmol
of each substrate mixed together (Combined). The error bars represent
one S.D. from three determinations.
Interaction of Human Mismatch Repair and Loop Repair 30233
atNationalTaiwanUniversity,onJanuary17,2011www.jbc.orgDownloadedfrom

7. DISCUSSION
In this study, we clearly show that human cell extracts can
efficiently process looped heteroduplexes containing 12–429
unpaired nucleotides with a strand break 3Ј to the heterology.
The repair reaction is strand-specific and highly biased to the
nicked strand. Our results appear to be in conflict with those of
Littman et al. (16) and McCulloch et al. (23), who found that
there was limited processing of heteroduplex loops adjacent to
a single-strand break 3Ј to the heterology. In contrast, we
observed a significant degree of 3Ј-nick-dependent processing
of large loops in human cell extracts. To reconcile the differ-
ences between the previous results and this study, we propose
that the substrate design of our 3Ј-heteroduplexes is suitable
for loop repair, whereas the 3Ј-heteroduplexes of previous stud-
ies are not. The heteroduplexes prepared by gpII treatment
proved to be good substrates for human mismatch repair (31).
Previous studies used gpII to make an incision 3Ј to the heter-
ologies in viral DNA strands. However, the incision made by
gpII may not be suitable for the loop repair pathway. The
cleaving-and-rejoining activity of gpII for replication initiation
requires magnesium (33). Although other divalent cations such
as calcium, barium, and manganese could be substituted for
magnesium, the enzyme was found covalently linked to the
nicking site within the origin or strand joining the viral and
complementary strands to form a hairpin structure (33, 34).
This is consistent with the fact that preparation of gpII-incised
substrates in the presence of calcium or barium requires pro-
teinase K treatment to resolve an otherwise inseparable pro-
tein-DNA complex (16, 31). Thus, the incision reaction of gpII
might contain a sufficiently aberrant structure so that the
mismatch repair pathway is unaffected (31), but the loop repair
activity is severely compromised (Fig. 4) (16, 23). In contrast,
for the construction of 3Ј-heteroduplexes, we employed a pro-
tocol similar to one that had been used to create 5Ј-heterodu-
plexes. Thus, other than the orientation between the nicks and
heterologies, the nature of 3Ј-nicks in our substrates is exactly
the same as the 5Ј-nicks. The substantial strand-specific repair
of 3Ј-heteroduplexes in our study clearly demonstrates that the
loop repair pathway in human cells is capable of repairing large
loops with a 3Ј-strand break. These findings indicate that large
loop repair activity in human cells possesses bidirectional proc-
essing capability, consistent with parallel results in E. coli (19).
Based on similar results in prokaryotes and eukaryotes, large
loop repair may be a conserved activity in living organisms. It
will be of great interest to identify components of each system
to look for orthologous genes.
Previous studies of 3Ј-nick-directed human mismatch repair
tested only a limited set of base-base mispairs (31, 32). We have
constructed a new set of 5Ј- and 3Ј-nick heteroduplexes con-
taining each of eight possible mismatches in the same sequence
context. The results from analysis of the substrate specificity of
5Ј-nicked heteroduplexes in HeLa cell extracts are consistent
with earlier findings that 5Ј-G-G and 5Ј-G-T are better repaired
than 5Ј-C-C (30). Here, we further expanded this to include the
repair specificity of heteroduplexes containing a strand break
3Ј to the mismatch. Our results show that human mismatch
repair specificity for 3Ј-nicked heteroduplexes is similar in
extent to the repair specificity for 5Ј-nicked substrates (Table
III). It is clear that the repair specificity is dependent on
mismatch recognition and ignores the polarity of the strand
break.
Transformation experiments with E. coli suggested that
loops of unmethylated DNA strands can be corrected by the
dam-directed mismatch repair system if a repairable mismatch
is in the vicinity (24). A similar result was also obtained from
an in vivo recombination event when a predicted loop was
adjacent to a single-base mismatch (25). To clarify whether this
human mismatch repair pathway possesses a similar capabil-
ity, a set of heteroduplexes containing a mismatch, a loop, and
a nick in different orientations were tested in HeLa nuclear
extracts for repair. We used the repair specificity of 3Ј-nicked
mispaired substrates (Table III) as a basis of comparison; we
found interference in 3Ј-heteroduplex repair of an adjacent
mismatch and loop. The simplest interpretation of our findings
is that two independent repair machineries compete at the site
of the lesions and that steric exclusion occurs in favor of the
loop repair system of the mismatch. Previous studies of human
mismatch-dependent excision tract mapping showed that the
reactions span a broad range from the strand break to a point
ϳ200 nucleotides beyond the heterology (31). When we moved
the loop 325 bp away from a G-G mismatch, the repair level of
G-G was restored to the basal level. This observation appears
to support the steric exclusion idea.
FIG. 6. Repair efficiency and nick positions. Repair reactions
were performed as described under “Experimental Procedures.” Sub-
strates with Individual G-G or C32 heterology and Co-existing G-G and
C32 heterologies were assayed. All heteroduplexes tested contained a
restriction endonuclease-generated strand break; the nick positions
relative to the heterologies are illustrated at the top. White bars, repair
occurring on the nicked complementary strand; black bars, repair oc-
curring on the closed viral strand. The error bars represent one S.D.
from three determinations.
TABLE VI
Co-repair of loops and mispairs in mismatch repair-deficient human
nuclear extracts
Repair was measured in LoVo nuclear extracts (7.5 mg/ml protein) as
described under “Experimental Procedures.” The type of substrates is
as illustrated in Fig. 1. Repair results are the average determined from
at least two separate experiments. Complete repair would correspond to
23 fmol. ND, not determined.
Type Heteroduplex
Repair
Mismatch Loop
o c o c
fmol
1B 3Ј-xC32 5.5 0.6
1B 3Ј-xV32 4.6 0.5
1C 3Ј-GGxC32 5.2 1.1 4.0 1.1
1C 3Ј-AAxC32 4.9 0.9 4.2 1.4
1D 3Ј-GGxV32 4.0 1.2 3.7 0.9
1E 3Ј-hC21xGG 3.9 ND 4.5 0.8
Interaction of Human Mismatch Repair and Loop Repair30234
atNationalTaiwanUniversity,onJanuary17,2011www.jbc.orgDownloadedfrom

8. In two previously used mammalian transformation systems,
it was demonstrated that loops separated by Ͼ200 bp are
rarely co-repaired, but complete co-repair is seen when these
markers are Ͻ60 bp apart (20, 35). In our biochemical analysis,
we observed that the 3Ј-nick-directed repair of a single-base
mismatch was affected by a loop 33–45 bp away, whereas
increasing the distance separating the two sites to 325 bp
diminished the influence of the loop. This observation is con-
sistent with the above in vivo observation, indicating that loop
repair tracts are relatively small compared with the mismatch
repair pathway (31, 36, 37).
When mismatches and loops were placed at sites distant
from a nick (Fig. 1G), the repair efficiency of a 24-nucleotide
loop significantly increased. As judged by the repair levels and
topological orientation of these two heterologies, it is unlikely
that that repair of both sites was via the same repair machin-
ery, e.g. it is possible that different ends of the strand break
were independently utilized by the mismatch and loop repair
systems. Thus, the strand break would persist in the reaction
for much longer than their lifetime during the normal course of
single-heterology repair. Therefore, the higher level loop repair
of this substrate might be due to better accessibility of the
strand break.
Although 3Ј-nick-directed repair of a G-G mismatch was
reduced when in the vicinity of a loop, the repair of a 5Ј-
heteroduplex containing adjacent heterologies of a mismatch
and a loop showed a synergistic effect by being better that the
repair of either individual heterology (Fig. 6). Genetic data
indicate that single-base mismatches may dominate loop-spe-
cific repair, perhaps by recruiting nick-directed repair, but the
directionality of the nicks was not discussed (25). In our bio-
chemical analysis, the enhanced co-repair of 5Ј-heteroduplexes
is consistent with this observation; however, the repair of 3Ј-
heteroduplexes appears to be the exact opposite; loop repair
dominates mismatch repair. In mismatch repair-deficient cells,
in vivo loop-specific repair tends to increase repair of single-
base mismatches, which is likely due to co-repair initiated by
loop repair (25). The results from our in vitro assay of mismatch
repair-deficient cell extracts clearly support this idea. The dif-
ferences in processing of 3Ј- and 5Ј-heteroduplexes may be due
to the nature of their polarity in DNA. Previous studies had
shown that processing mismatch repair involves a repair patch
of up to hundreds of nucleotides (31). Comparatively, loop re-
pair seems to be restricted to the region near the heterologies
(16). Repair of 3Ј-mismatches involves differential access to the
DNA lesion for DNA excision (3Ј 3 5Ј) and resynthesis (5Ј 3
3Ј). Therefore, it is conceivable that the cause of mismatch
repair interference may be a time/position discordance depend-
ing on the steps of repair, whereas in contrast, 3Ј-nick-directed
loop repair is much more localized.
What is the biological significance of our observed interac-
tion between mismatch and loop repair? Heteroduplex DNA
containing a mismatch and a nearby loop can be formed during
genetic recombination in crosses with parental DNA strands,
each containing one of the heterologous sequences. The repair
of such heterologies is essential for maintaining genetic stabil-
ity. It has been shown in E. coli that adjacent DNA lesions can
cause repair interference. It has been suggested that this may
prevent simultaneous repair of proximal lesions, resulting in
potentially lethal double-stranded breaks. If this is the case, we
need to reconcile why a nearby 5Ј-nick enhances loop repair,
whereas a 3Ј-nick impedes it. Characterization of these activi-
ties should clarify the distinction of native DNA repair activi-
ties observed here. This study provides a means to identify
potential genetic targets of this process in human cells.
REFERENCES
1. Streisinger, G., Okada, Y., Emrich, J., Newton, J., Tsugita, A., Terzaghi, E.,
and Inouye, M. (1966) Cold Spring Harbor Symp. Quant. Biol. 31, 77–84
2. Meselson, M. (1988) in Recombination of Genetic Material (Low, K. B., ed) pp.
91–113, Academic Press, Inc., San Diego, CA
3. Hastings, P. J. (1988) in Genetic Recombination (Kucherlapati, R., and Smith,
G. R., eds) pp. 397–428, American Society for Microbiology, Washington,
D. C.
4. Petes, T. D., Malone, R. E., and Symington, L. S. (1991) in The Molecular and
Cellular Biology of the Yeast Saccharomyces: Genome Dynamics, Protein
Synthesis, and Energetics (Broach, J., Jones, E., and Pringe, J., eds) pp.
407–512, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY
5. Modrich, P., and Lahue, R. S. (1996) Annu. Rev. Biochem. 65, 101–133
6. Umar, A., and Kunkel, T. A. (1996) Euro. J. Biochem. 238, 297–307
7. Jiricny, J. (1998) Mutat. Res. 409, 107–121
8. Kolodner, R. D., and Marsischky, G. T. (1999) Curr. Opin. Genet. Dev. 9, 89–96
9. Buermeyer, A. B., Deschenes, S. M., Baker, S. M., and Liskay, R. M. (1999)
Annu. Rev. Genet. 33, 533–564
10. Hsieh, P. (2001) Mutat. Res. 486, 71–87
11. Fang, W.-h., Wu, J.-Y., and Su, M.-J. (1997) J. Biol. Chem. 272, 22714–22720
12. Wierdl, M., Dominska, M., and Petes, T. D. (1997) Genetics 146, 769–779
13. Genschel, J., Littman, S. J., Drummond, J. T., and Modrich, P. (1998) J. Biol.
Chem. 273, 19895–19901
14. Umar, A., Boyer, J. C., and Kunkel, T. A. (1994) Science 266, 814–816
15. Miret, J. J., Milla, M. G., and Lahue, R. S. (1993) J. Biol. Chem. 268,
3507–3513
16. Littman, S. J., Fang, W.-h., and Modrich, P. (1999) J. Biol. Chem. 274,
7474–7481
17. Corrette-Bennett, S. E., Parker, B. O., Mohlman, N. L., and Lahue, R. S. (1999)
J. Biol. Chem. 274, 17605–17611
18. Corrette-Bennett, S. E., Mohlman, N. L., Rosado, Z., Miret, J. J., Hess, P. M.,
Parker, B. O., and Lahue, R. S. (2001) Nucleic Acids Res. 29, 4134–4143
19. Fang, W.-h., Wang, B.-J., Wang, C.-H., Lee, S.-J., Chang, Y.-T., Chuang, Y.-K.,
and Lee, C.-N. (2003) J. Biol. Chem. 278, 22446–22452
20. Weiss, U., and Wilson, J. H. (1987) Proc. Natl. Acad. Sci. U. S. A. 84,
1619–1623
21. Weiss, U., and Wilson, J. H. (1989) Somatic Cell Mol. Genet. 15, 13–18
22. McCulloch, S. D., Gu, L., and Li, G.-M. (2003) J. Biol. Chem. 278, 3891–3896
23. McCulloch, S. D., Gu, L., and Li, G.-M. (2003) J. Biol. Chem. 278, 50803–50809
24. Carraway, M., and Marinus, M. G. (1993) J. Bacteriol. 175, 3972–3980
25. Taghian, D. G., Hough, H., and Nickoloff, J. A. (1998) Genetics 148, 1257–1268
26. Bill, C. A., Taghian, D. G., Duran, W. A., and Nickoloff, J. A. (2001) Mutat. Res.
485, 255–265
27. Su, S. S., Lahue, R. S., Au, K. G., and Modrich, P. (1988) J. Biol. Chem. 263,
6829–6835
28. Holmes, J., Jr., Clark, S., and Modrich, P. (1990) Proc. Natl. Acad. Sci. U. S. A.
87, 5837–5841
29. Lu, A. L., Clark, S., and Modrich, P. (1983) Proc. Natl. Acad. Sci. U. S. A. 80,
4639–4643
30. Kat, A., Thilly, W. G., Fang, W.-h., Longley, M. J., Li, G.-M., and Modrich, P.
(1993) Proc. Natl. Acad. Sci. U. S. A. 90, 6424–6428
31. Fang, W.-h., and Modrich, P. (1993) J. Biol. Chem. 268, 11838–11844
32. Fang, W.-h., Li, G.-M., Longley, M., Holmes, J., Thilly, W., and Modrich, P.
(1993) Cold Spring Harbor Symp. Quant. Biol. 58, 597–603
33. Meyer, T. F., and Geider, K. (1979) J. Biol. Chem. 254, 12642–12646
34. Greenstein, D., and Horiuchi, K. (1989) J. Biol. Chem. 264, 12627–12632
35. Ayares, D., Ganea, D., Chekuri, L., Campbell, C. R., and Kucherlapati, R.
(1987) Mol. Cell. Biol. 7, 1656–1662
36. Wang, H., and Hays, J. B. (2002) J. Biol. Chem. 277, 26136–26142
37. Wang, H., and Hays, J. B. (2003) J. Biol. Chem. 278, 28686–28693
Interaction of Human Mismatch Repair and Loop Repair 30235
atNationalTaiwanUniversity,onJanuary17,2011www.jbc.orgDownloadedfrom