Copyright 0 1996 by the Genetics Society of America

Destabilization of Simple Repetitive DNA Sequences by Transcription in Yeast

Monika Wierdl,*
Christopher N. Greene,t’:
Abhijit
Sue Jinks-Robertsont
and Thomas D. Petes*
*Department of Biology and Curriculum in Genetics and Molecular Biology, University of North Carolina, Chapel Hill, North
Carolina 27599-3280 and tDepartment of Biology, fGraduate Program in Genetics and Molecular Biology, and sGraduate Program in
Biochemistry and Molecular Biology, Emory University, Atlanta, Georgia 30322
Manuscript received January 17, 1995
Accepted for publication March 11, 1996

ABSTRACT
Simple repetitive DNA sequences in the eukaryotic genome frequently alter in length. In wild-type
strains, we find that transcription through a repetitive poly GT tractdestabilizesthe tract four- to
ninefold. In mismatch repairdeficient yeast strains, simple repeats are very unstable. High levels of
transcription in such strains destabilize repetitive tracts an additional two- to threefold.

S IMPLE repetitive sequences (microsatellites) are

tandem repeats of a single base or a small number
of bases. Such sequences are found in all eukaryotic
Mostof the available evidence indicates that alter-
ations of repetitive tracts by small numbers of repeats
represent DNA polymerase slippage rather thanrecom-
genomes thus far examined (TAUTZand RENZ 1984; bination events. First, mutations that reduce recombi-
TAUTZet al. 1996). The most common simple repeats nation frequency do not affect the stability of simple
are 20-50-bp tracts of poly GT, poly A and poly GA repeats in Escherichia coli (LEVINSON and GUTMAN 1987)
(HAMADA et al. 1982; TAUTZ and ~ N 1984).Z Although or in yeast (HENDERSON and PETES1992). Second, the
the functions of these sequences (if any) are unclear, meiotic rate of repeat instability is about the same as
the high frequency with whichrepeats alter their length the mitotic rate, although the frequency of meiotic re-
has a number of important consequences. First, the combination is -1000-fold greater than the frequency
polymorphic nature of these repeats in human popula- of mitotic exchange (STRAND et al. 1993). Third, muta-
tions makes them valuable for genetic mapping (WEBER tions in the mismatch repair genes elevate simple repeat
1990). Second, length expansions within trinucleotide instability in E. coli (LEVINSON and GUTMAN 1987), yeast
repeats cause several human genetic diseases (CASKEY (STRAND et al. 1993, 1995), and mammalian cells (re-
et al. 1992; RICHARDS and SUTHERLAND 1994). Third, viewed by KARRANand BIGNAMI1994). This effect is
global microsatellite instability is associated with several consistent with the DNA polymerase slippage model,
types of cancer (AALTONEN et al. 1993; MERLO et al.
since the unpaired repeats would be expected to be a
1994; MIRONOVet al. 1994; ORTHet al. 1994).
substrate for the DNA mismatch repair system. Failure
Two models have been proposed to explain the insta-
to correct the unpaired repeats in mismatch repair-
bility of simple repeats: unequal recombination and
deficient mutants would result in elevated levels oftract
DNA polymerase slippage. Unequal recombination in-
instability.
volves the pairing of repetitive tracts on sister chroma-
tids (or homologous chromosomes) in a misaligned Based on the polymerase slippage model, the rate of
configuration. Crossovers between such misaligned re- simple repeat instability is a function of the rate of DNA
peats generate one larger repetitive tract and one tract polymerase slippage and the rate ofDNA mismatch
with the reciprocal deletion (SMITH1973). In the DNA repair. Two parameters that influence tract stability are
polymerase slippage model ( STREISINGER et al. 1966), a the length and the “purity” of the tract (whether the
transient dissociation of theprimer and template tract contains variant repeats). InE. coli (LEVINSON and
strands during DNA replication is followed by a misa- GUTMAN 1987;FREUND et al. 1989) and mammalian cells
ligned reassociation of the strands leading to unpaired (WEBER 1990), long repetitive tracts are less stable than
repeats (Figure 1).If replication resumes without repair short tracts, presumably because the opportunity for a
of these intermediates, this mechanism will lead to an stable misaligned configuration is greater for the longer
addition of repeats (if the unpaired repeat is in the tracts. In addition, variant bases within repetitive tracts
primer strand) or deletion of repeats (if the unpaired lead to increased stability in mammalian cells (CHONG
repeat is in the template strand) in the newly synthe- et al. 1995).
sized strand. In addition to the length and purity of repetitive
tracts, there may be other factors that affect microsatel-
Corresponding author: Thomas D. Petes, Department of Biology, Uni-
versity of North Carolina, Chapel Hill, NC 27599-3280. lite stability. Transcription has been associatedwith
E-mail: tompetes@email.unc.edu both DNA replication and nucleotide excision repair
Genetics 1 4 3 713-721 (June, 1996)
714 M. Wierdl rt nl.
G~GTG?L,
PC- GTGTGTGTGTGT
T ~"CACACACACACACACACACACA-

FIGL'RE1.-Alterations in the length of a poly GT tract as a consequence of DNA polymerase slippage. During replication,
there is a transient dissociation between the primer (P) andtemplate (T) strands. The reassociation may result in DNA molecules
in which a GT repeat is unpaired in either the primer (left side of figure) or template (right side of figure) strands. If this
mismatch is not remired. continued redication will lead to an addition of one repeat (displaced repeat in primer strand) or a
deletion of one repeat (displaced repeat in template strand).

in a variety of experiments. The human transcription tively, at the 5' ends. pNKY48 has a HIS4-URA3 cassette that
was removed as a 1148-bp BnmHI fragment and inserted into
factor CTF is identical to NF-1, an initiation factor for the BnmHI site ofpSR339 to generate pSR348. Since yeast
adenovirus DNA replication (JONES et al. 1987). Several strains with a mutant uru3 gene at the normal URA3 locus
components of the nucleotide excision pathway are as- and an integrated copy of pSR348 grow in medium lacking
sociated with transcription factor complexes in both uracil, this construction results in a functional URA3 fusion
protein. pSR231 contains a pGALl0-LYS2 fusion constructed
prokaryotes and eukaryotes (reviewed by DRAPKIN et al. by ligating a 685-bp BumHI-EroRI GALI-GALlOpromoterfrag-
1994). Fox ~t al. (1994) showed that transcription inter- ment from pBMl3O (JOHNSTON and DAVIS1984) withPsfI
fered with the function of the E. coli MutY mismatch linkers into pSR229. pSR229 is a derivative ofpDPG (FLEIC
repair gene. In addition, a number of researchers (WV rt al. 1986) containing a promoter-less LYS2 gene and was
constructed by digestion with Hind11 and EcoRV followed by
~t al. 1988; RAHMOUNI and W~1.1-s 1989) foundthat creating flush ends with Klenow fragment and ligation of PstI
transcription alters DNA superhelicity, which could in- linkers.
fluence either DNA replication or DNA repair.
DATTAandJINKS-ROBERTSON (1995) recently showed
that transcription increases the level of reversion of a
frameshift mutation in yeast -30-fold. Since such rever-
sion events often represent changesin short runsof the
same base (C. GREENE and S.JINKS-ROBERTSON, unpub-
lished data), this result suggests the possibility that tran-
scription might also alter the stability of microsatellite
sequences. Below, we show that poly GT tracts of 31 or
35 bp aredestabilized four- to ninefold by high levels of
transcription and that this destabilization results partly
from an interference with DNA mismatch repair. CUILI-IO

MATERIALSAND METHODS lacZ

Plasmid constructions: The plasmid pSR348 (Figure 2)
B g l II
contains a pGALlO-LYS2-HIS4-URA3translational fusion de-
rived from pSR339 and pNKY48 ( A L A N 1 and KLECICWER 1987).
pSRSS9 has a 1.2-kb XhoI-BnmHI pGALl0-LYS2fragment, gen- FIGURE2.-Restriction map of plasmid pSR348. This plas-
erated by PCR amplification of pSR231, cloned into theLEU2 mid contains a fusion gene (indicated by shading), in which
integrating vector pRS30.5 (SIKORSKI and HIETER1989). The synthesis of a fusion protein (with segments derived from the
pGALl0-LYS2 fragment was amplified using the primers yeast LYS2, HIS4 and URA3 genes) is controlled by the GALl-
pgallox (5' C G G C T C G A G C G ~ G A ~ ~ ~ A C C C CandAG) 10promoter. This fusion protein results in a Ura+ phenotype.
lys2r (5' CGCGGATCCTGGATGGATCGCTTAGCGCA).The Poly GT tracts were inserted into the unique BgnI site within
primers pgallOx and lys2r have XhoI and BnmHI sites, respec- the fusion gene.
 Transcription-Associated Mutations 715

The plasmid pSR348 has a single BglII site within the LYS2 Strains SJR453-31 and SJR454-35 are gal80 derivatives of
sequence where annealed oligonucleotides containingthe MBW1-31 and MBW1-35, respectively. These strains were con-
poly GT repeat were inserted. pMBWl was derived from structed by transformation of theparental strains with a
pSR348 by insertion of a 33-bp poly GT repeat at the unique BamHI fragment containing a gal80 gene disrupted by inser-
BglII site. Two complementary oligonucleotides with cohesive tion of a hisGURA3-hisC cassette, derived from plasmid
BglII ends and the following sequences were synthesized: 5' pSR372 (provided by J. FRIDOVICH-KEIL, Emory University).
GATCGTCGACA(TG)lGTACTCGAG 3' and 5' GATCCTC- The resulting Ura+ transformants were then plated on me-
GAGTA(CA)IGTGTCGAC3'. These oligonucleotides were an- dium containing5FOA to select for loss of the URA3 insertion
nealed by the method described by HENDERSON and PETES within the gal80 gene. The gal80 mutation was confirmed by
(1992),and ligated to BglII-digestedpSR348. The ligated PCR and by testing the isolates for sensitivity to Zdeoxygalac-
products were transformed into the E. coli strain DH5a. A tose (PIATT 1984).
positive clone was identified by restriction analysis and the Measurements of simplesequencestability: The starting
DNA sequence confirmed by themethod of KRAFT et al. strains were phenotypically Ura- as a consequence of the out-
(1988). In pMBWl, the poly GT tract was oriented such that of-frame insertion of the poly GT tract (either 35 or 31 bp).
the GT repeats were in the transcribed strand. The plasmid Consequently, we monitored alterations that restored the cor-
pMBW2 was constructed as described above, except theoligo- rect reading frame by measuring the frequency of Ura+ colo-
nucleotides had a 31-bp poly GT repeat in the opposite orien- nies. Yeast strains were pregrown under conditions that lead
tation within pSR348 compared with pMBWl. to high or low transcription levels of the fusion gene using
The plasmid pSR376, used to monitor thelevel of induction two different protocols (as described below). For -20 inde-
of transcription from the PGALI-GALlOpromoter, is a U R A 3 pendent colonies derived from each strain, we measured the
marked integratingplasmid containing a pGAL-LacZfusion. It frequency of Ura+ cells in each colony by plating cells to
was derived from pRY131 (YOCUM et al. 1984) by deleting medium (SGuracil) that contained galactose (to induce ex-
a 1.9-kb SpeI fragment containing the 2-pm plasmid DNA pression of the fusion gene) and lacked uracil. Ura+ colonies
replication origin. Linearizing pRS3'76 by StuI digestion tar- were counted after 2 days incubation at 30". The rates of Ura+
gets integration at URA3. events per generation were calculated using the method of
Mediaand growth conditions: Standard SD-complete and the median (LEA and COULSON 1949).
YEPD media were used (SHERMAN 1991). YEPGal plates were Two methods of elevating transcription levels were used.
identical toYEPD, with galactose replacing glucose as a carbon GAL80 strains were pregrown on plates containing SD com-
source. Carbon sources for the various media were 2% glucose plete medium (glucose as carbon source, low rate of transcrip-
(YEPD and SD-complete); 2% galactose (YEPGal); 2% galac- tion of fusion gene) or on plates with SG complete medium
tose, 1%raffinose (SG-complete); and 2% glycerol, 1%ethanol (galactose as carbon source, high transcription rate of fusion
(YEPGE). Cells were grown at 30". For one protocol, cells were gene). In other experiments, isogenic GAL80and gal8Ostrains
pregrown on plates containing SDcomplete or SG-complete were pregrown on rich medium containing ethanol andglyc-
media for 3 days. In the other protocol, cells were pregrown erol (YEP-GE), resulting in low (GAL80 strains) and high
for 5 days on plates containing YEP-GE medium. (gal80strains) levels of expression of the fusion gene. Individ-
Yeast strain constructions: All yeast strainsused in this ual colonies derived from the pregrown cells were examined
study were derived from AMY125 (MATa a d d his7-2 leu2- for the frequency of Ura+ cells as described above.
3,112 trpl-289 ura3-52; STRANDet al. 1993) by transformation Analysis of thelength of repetitivetracts by hot PCR
with various plasmids. The strain MBW1-33was created by Lengths of poly GT tracts in Ura+ colonies were measured by
transformation (SHERMAN et al. 1983) of AMY125 with EcoRI- the method described by FARRER et al. (1994). PCR primers
digested pMBWl DNA. The insertion of pMWl at the LEU2 complementary to the LYS2 coding sequences flanking the
locus was confirmed by Southern blot analysis (SAMBROOK et poly GT tract (5' CCAACGTGGTCATTTAATGAGC and 5'
al. 1989). Since the poly GT insertion was in-frame with the GCTTGAACTCGTCTAATTTG) were used to amplify an
fusion protein, MBW1-33 was Ura+. -200-bp product. The 20 pI PCR reactions contained: 200-
MBW1-31 and MBW1-35 strains were obtained as spontane- 500 ng yeast DNA, 12.5 pM of the primers, 1 unit Taq poly-
ous 5-fluoroorotic acid-resistant (5FOA') derivatives of merase (Perkin-Elmer), and 2.5 pCi "'P-dATP in a solution
MBW1-33 and hadpoly GT tracts of 31 and 35 bp, respectively. of 10 mM Tris (pH 8.3), 50 mM KCl, 0.001% gelatin, 3 mM
Single colonies ofMBW1-33 were spread on solid medium MgC12,and 0.25 mM dNTPs. PCR products were analyzed on
containing 5FOA (which selects for Ura- cells; BOEKEet al. 6% sequencing gelswith control DNA samples containing
1984) and 5FOA' papillae were identified. The lengths of the poly GT tracts of 31, 33 and 35 bp.
poly GT tracts in the Ura- isolates were determined by the Assay of transcription induction by the GAL10 promoter:
hot PCR method described below. Strain MBW2-31 was iso- The plasmid pSR376 was transformed into MBW1-31 and
genic to MBW1-31 except that the poly GT repeat had the SJR453-31 to create the strains SJR474-31 and SJR475-31. Sin-
opposite orientation. This strain was constructed by transfor- gle copy insertion at the U r n 3 locus was confirmed by South-
mation of AMY125 with EcoRI-treated pMBW2. ern analysis. These strains were grown in 5-ml cultures con-
The strain MS121 (provided byM. STRAND, Univ. of North taining either 2% glycerol and 2% ethanol (YEP-GE) or
Carolina) is isogenic to AMY125 except for a disruption of cultures containing 2% galactose (YEPGal). Early log phase
the MSH2 gene by a URA3 insertion. A 5FOA' derivative of cultures were harvested by centrifugation. The cultures were
MS121 was transformed with pMBWl to construct the strain diluted and permeabilized cells were assayed for P-galactosi-
MBW3-33. Southern analysis confirmed that the Leu+ Ura+ dase activity as described by AUSUBEI. et al. (1987).
transformant contained a single integrated copy of the plas- statistical analyses: For each strain, the numbers of Urd+
mid. The strain MBW3-35 was a 5FOA' derivative of MBW3- cells were determined for -20 independent colonies per ex-
33. By the PCR method described below, we showed that periment, andeach experiment was done atleast twice. From
MBW3-35 contained a 35-bp poly GT tract (instead of the 33- the measured numbers, rates were calculated by the method
bp tract of MBW3-33). Strain SJR482-35 is a pmsl derivative of the median (LEAand CCXILSON 1949). Therates obtained
of MBW1-35 constructed by two-step transplacement using for each experiment were then averaged for each strain or
the BstXI-treated plasmid pSR211; this plasmid (provided by growth condition.
D. MALONEY,Univ. of California, Berkeley) contains a dele- To determine whether the rates of instability were signifi-
tion mutant allele of PMSl with flanking sequences inserted cantly different for one strain compared with another (or one
in YIp5. The deletion in SJR482-35 was confirmed by PCR. growth condition compared to another), we converted the
716 M. Wierdl et al.

number of Ura+ cells to a rate measurement for each individ- DNA sequence analysis showed that all such derivatives
ual colony by using the number of Ura+ cells for each colony
had alteredpoly GT tractsthat restored the correctread-
as the median value and then employing the method of the
median described by LEA and COULSON(1949). These rates ing frame for the fusion URA3 gene.
were ranked for the two strains (or conditions) being com- Destabilization of simplerepeats bytranscription:
pared in order from highest to lowest using the same number The fusion URA3 gene is transcribed from the galactose-
of colonies for each strain (or growth condition). We then inducible GALI-10 promoter, which is positively regu-
determined by chi-square analysis whether one strain had sig-
nificantly more colonies ranked in the top half of the rate lated by Gal4p and negatively regulated by Gal80p
values than the other strain. Chi-square values >3.85 (P < (JOHNSTON and CARLSON 1992). Ingal80 strains grown
0.05) were judged to be significant. For example, for strain in medium containing glycerol and ethanol as carbon
MBW2-31, we measured 37 rates under conditions of low tran- sources or in GAL80 strains grown in medium con-
scription and 37 under conditions of high transcription. When
these 74 rates were ranked in a single list, 33 of the highest taining galactose, transcription occurs at a high rate
37 rates were derived from the high-transcription experiments. from this promoter. In GAL80 strains grown in medium
If there is no significant difference in the rate of instability containing glucose or glycerol and ethanol, transcrip-
under the high- and low-transcription conditions, one would tion occurs at a low rate. Consequently, to determine
expect equal numbers (18.5) of the 37 topranked rates to be
derived from cells under the two conditions. By chi-square the effect of transcription on simple repeat instability,
analysis, we found that our observed result was a significant we used two different growth protocols. In the first pro-
departure (x' = 21, P < 0.001) from that expectation. tocol, we compared the rateof the appearance of Ura'
Confidence limits (95%) were calculated forthe rates colonies in a GAL80 strain that was pregrown on me-
shown in Tables 1 and 3. Using Table A-25a of DIXON and
MASSEY (1969), we calculated 95% confidence limits on the
dium containing glucose (low transcription) with the
median number of Ura' cells for each experiment. For exam- rate measuredin the same strain pregrown on galactose
ple, for one of the three experiments in which we examined (high transcription). In the second protocol, we com-
the rate of instability in MBW1-35 under low transcription pared the rate of Ura+ colonies in a GAL80 strain that
conditions, we measured the numberof Ura+ cells in 15inde- was pregrown on medium containing ethanoland glyc-
pendent colonies and these numbers were ranked. The me-
dian number of Ura+ cells was 71. From Table A-25a, if the erol (low transcription), with the rate in an isogenic
number of independent observations is 15, one can conclude gal80 strain pregrown on the same typeof medium
that the 95% confidence intervals are between the 4th and (high transcription).
11th ranked values. For this experiment, these values were 58 Our Ura' rate measurements are summarized in Ta-
and 88 Ura+ cells. Using these numbers and theaverage num-
ber of cells per colony, we calculated rates using Table 3 of ble l. Similar results were obtained with both methods
LEAand COULSON(1949), resulting in arate estimate for this of inducing high transcription levels. For the GAL80
experiment of 1.1 X 10-5/cell division with 95% confidence strain with the 35-bp tract of poly GT (MBW1-35) pre-
limits between 0.9 and 1.2 X (Table1). grown on medium containing glucose (low transcrip-
-
tion),we found a rateof Ura' derivativesof 1 X
RESULTS mitotic division. Since we find (as described below) that
all of these Ura' derivatives had alterations in length
Experimental system: To analyze the stability of sim- of the polyGT tract, this rate represents the rate of
ple repetitive sequences, we used a frameshift assay, simi- tract instability. When MBW1-35 wasgrown in galactose-
lar to those described previously (HENDERSON and PETES containing medium (high transcription), the rate in-
1992). We inserted an in-frame 33-bp tract of poly GT creased about sixfold, a statistically significant differ-
into the coding sequence of a URA3 fusion gene in ence (x' = 19.9, P < 0.001). The rate of Ura' colonies
the plasmidpSR348 (Figure 2) to create the plasmid was ninefold higher for theguZ80 strain SJR454-35 (high
pMBW1. A yeast strain with a uru? mutation (AMYl25) transcription) compared with the GAL80 strain MBW1-
was transformed withthisplasmid and Uraf trans- 35 (low transcription) when both strains were grown in
formants were isolated. Insertion ofpMBWl into the medium containing glycerol and ethanol (x*= 26.3,
chromosome resulted in a strain (MBW1-33) in which P < 0.001).
the fusion gene is located between duplicated copies of In similar comparisons with strains containing poly
LEU2. When we selected Ura- derivatives of MBWl-33 GT insertions of 31 bp (strains MBW1-31 and SJR453-
by using 5FOA, we found that some of these derivatives 31), the high transcription levels elevated the rate of
contained alterations of the poly GT tract leading to tract instability by about three- to fivefold (Table 1).
out-of-frame insertions; two such derivatives had 31-bp Statistically significant differences were found compar-
(MBW1-31) and 35-bp(MBW1-35)polyGT tracts. In ing rates for MW1-31 pregrown on glucose- (low tran-
addition, we found derivatives of MBW1-33 in which the scription) or galactose-containing (high transcription)
fusion gene was lost as a consequence of recombination medium (x' = 8.7, P < 0.01), and comparing rates
between the flanking LEU2 genes. Consequently, rather for MW1-31 (low transcription) and SJR453-31 (high
than assessing tract instability by measuring the fre- transcription) pregrown on YEPG medium ( x 2 = 24.5,
quency of 5FOAR cells derived from MBW1-33, we moni- P < 0.001).
tored the frequency of Ura' cells derived from the Ura- For both protocols, the induction of instability by
strains MBW1-31 and MBW1-35. As described below, high transcription levels was about twofold less for the
 Transcription-Associated Mutations 717

TABLE I
Increased destabilizationof simple repetitive sequencesby high levels of transcription

Average rate of Ura'Fold increase by

Strain" GAL80 genotype/media*
Transcription Rate of Ura+ derivatives
derivatives X transcription
MBW1-35 GAL80/SD-complete Low 1.1 (0.9-1.2), 1.o 1 .o
0.9 (0.8-1.1)
MBW1-35 GAL80/SGcomplete High 5.3 (3.4-13.8), 6.2 6.2
7.1 (4.3-16.1)
MBW1-35 Low 0.5 (0.4-0.6), 0.5 1 .o
0.2 (0.2-0.3),
0.4 (0.2-0.9),
0.8 (0.6-1.1)
SJR454-35 High 4.3 (3.7-4.9), 4.4 8.8
4.5 (4.0-7.2)
MBW1-31 Low 1.1 (1.0-1.3), 1.3 1 .o
1.5 (1.1-2.5)
MBWl-31 High 5.0 (3.1-7.3), 4.6 3.5
4.1 (3.1-5.5)
MBWl-31 GAL80/YEP-GE Low 1.4 (1.0-2.0), 1.3 1 .o
1.2 (0.9-1.7),
1.4 (1.1-2.8)
SJR45331 High 7.8 (5.6-9.3), 6.2 4.8
5.9 (5.1-8.5),
4.9 (3.4-7.0)
MBW2-31 Low 1.9 (1.1-2.3), 2.2 1.o
2.5 (1.8-3.1)
MBW2-31 High 8.7 (6.8-13.0), 6.6 3.0
4.5 (3.6-5.8)
"The numbers 31 and 35 after the strain name indicates the length of the GT tract in base pairs.
SD-complete medium contains dextrose, SG-complete medium contains galactose, and YEP-GE medium contains glycerol
and ethanol.
'Numbers in parentheses represent 95% confidence values for the rates measured in individual experiments.

31-bp tract than for the 35-bp tract. The difference ap- or growth condition. The GAL80 strain SJR474-31 had
pears to reflect a lower rate of instability for the 35-bp 360-fold more ,&galactosidaseactivity when grown in
tract under conditions of low transcription, rather than mediumcontaining galactose (YEPGal) than when
a higher level of instability of the 35-bp tract under grown in medium with glycerol and ethanol (YEP-GE).
conditions of high transcription (Table 1). This issue The gal80 strain SJK475-31, grown in YEP-GE, had 460-
will be discussed further below. fold more /%galactosidase activitythan the GAL80strain
We also examined the effect of the orientation of the SJR474-31, grown under the same conditions.
poly GT tract on transcription-stimulated tract instabil- Sequence analysis of altered tracts: In our previous
ity. In MBW1-31, the transcript of the fusion URA3 gene studies of alterations in poly GT tracts, we found that
will contain a poly CA tract in the mRNA. The strain most changes involve addition or deletion of one or
MBW2-31 is identical toMBW1-31, except for the orien- two repeats with -10% of the alterations representing
tation of the poly GT tract. High levels of transcription larger deletions or insertions (HENDERSON and PETES
resulted in a 3.0-fold induction (statistically significant 1992; STRAND et al. 1993, 1995). To confirm that the
with x2 = 21, P < 0.001) of tract instability in MBW2- Ura+ cells derived from the strains used in this study
31 (Table l ) ,similar to the 3.5-fold induction observed also resulted from tract alterations and to determine
by the same protocol for MBWl-31. the types of alterations, we examined tract lengths by
Level of transcription induction: To confirm levels analyzing PCK products on DNA sequencing gels.
of high and low transcription from the GAL10 pro- These results are summarized in Table 2.
moter, we inserted a single copy of a PGALlOlacZ re- All Ura' derivatives examined hadpoly GTtract alter-
porter gene intostrains MBWl-31(GAL80) and SJR453- ations that restored the correct reading frame of the
31 (ga180) to create thestrains SJR474-31 and SJR475-31, fusion protein. In strains with35-bp tracts, the most
respectively. Measurements of &galactosidase activity of common alteration was a deletion of one repeat, al-
four independent cultures were done for each strain though a few tracts with additions of two repeats were
718 M. Wierdl et al.

TABLE 2
Tract length alterations in high and low transcription strains

No. of tracts with additions or

deletions of base pairs
Strain" GAL80 genotype/mediah Transcription -4 -2+4 +2 Total
MBW1-35 GAL80/SD-complete Low 0 19 0 4 23
MBW1-35 GAL80/SGcomplete High 0 21 0 2 23
SJR454-35 gd80/YEP-GE High 0 15 0 1 16
MBWl-31 GAL80/SD-complete Low 0 0 20 0 20
MBW1-31 GALKO/SGcomplete High 1 0 18 0 19
SJR453-31 g&O/YEP-GE High 0 0 15 0 15
MBW2-31 GAL80/SD-complete Low 0 0 15 0 15
MBW2-31 GAL8U/SGcomplete High 0 0 15 0 15
" The numbers 31 and 35 after the strain name indicates the length of the GT tract in base pairs.
" SD-complete medium contains dextrose, SGcomplete medium contains galactose, and YEP-GE medium contains glycerol
and ethanol.

also observed. In strains with 31-bp tracts, almost all of DISCUSSION

the altered tracts had gained a single repeat; only one
tract with a deletion of 4 bp was observed. Of 146 tract We find that high levelsof transcription stimulate
alterations examined, none contained an addition or instability of simple repetitive sequences, and that this
deletion of more thantwo repeats. The types ofchanges effect is partly dependent on a functional DNA mis-
observed with high transcription levels were not differ- match repair system. Below,we discuss the implications
ent from those observed with low transcription levels. of these results.
In addition, the orientation of the poly GT tract did As pointed out in RESULTS, the induction of tract
not affect the types of tract alterations. instability by high levels of transcription is larger for
Effects of transcription on simple repeat stabilityin strains with the 35-bp repeat (six- to ninefold) than for
mutants deficient for DNA mismatch repair:Mutations strains with the 31-bp repeat (three- to fivefold). This
that reduceDNA mismatch repair dramatically increase difference reflects a lower rate of instability in strains
simple repeat instability in prokaryotes (LEVINSON and with the 35-bp repeat compared with strains with the
GUTMAN 1987) and in eukaryotes (STRAND et al. 1993; 31-bp repeat under conditionsof low transcription (Ta-
HEMMINKI et al. 1994;WINDet al. 1995). In yeast, simple ble l ) . Previously, we observed that insertions of single
repeats are destabilized -200-fold by mutations in the GT repeats were more frequent than deletions by two-
PMSI, MSH2, or MLHl genes (STRAND et al. 1993) and to threefold in wild-type strains under conditions of low
-40-fold by mutations in MSH? (STRAND et al. 1995). transcription (HENDERSON and PETES1992; STRAND et
In strain MBW1-35, there is a six- to ninefold stimula- al. 1993, 1995). For the 35-bp repeat, deletion of 2 bp
tion of repeat instability by transcription (Table 1). is required to restore the reading frame, whereas an
Strains MBW3-35 and SJR482-35 are isogenic with addition of 2 bpis required to restore the reading frame
MBW1-35, except for mutations in the msh2 and pmsl forthe 31-bp repeat. The difference in the rates of
genes, respectively. In both mismatch repair defective instability for strains with 35 and 31 bp repeats under
strains, high levels oftranscription destabilized the poly conditions of low transcription may thus reflect a bias
GT repeats an additional two- to threefold (Table 3). imposed by the selection system. Since strains with 35
The destabilizing effect of transcription,although and 31 bp repeats have similar rates of instabilityunder
smaller in the mismatch repair deficient strains than in conditions of high transcription, transcription-induced
wild-type strains, is still statistically significant: a chi- instability results in similar frequencies of additions and
square value of 19 ( P < 0.001) for the comparison of deletions.
MBW3-35 under pregrowth conditions leading to high In previous experiments, -10% of theframeshift
and low transcription and a chi-square value of20 ( P events detected in wild-type cells with a 33-bp poly GT
< 0.001) for thecomparison of SJR482-35 under condi- tract were deletions or additions involving more than
tions leading to high and low transcription. two repeats (STRAND et al. 1993, 1995). It is not clear
We also examined the types of tract alterations in the whether these larger alterations also represent DNA
m h 2 strain MW3-35 and the pmsl strain SJR482 (Table polymerase slippage or a different mechanism (for ex-
4). Under conditions of high or low transcription, we ample, recombination).In > 100 tracts analyzed in our
found exclusively deletions of one repeat or additions experiments, no large deletions or insertions were de-
of two repeats. The ratio between these two classes of tected. Since the experiments reported herewere done
events was not significantly affected by transcription. in the same genetic background as those of STRAND et
 Mutations Transcription-Associated 719

TABLE3
Transcription destabilizes simple repeat tractsin the absence of mismatch repair

Averagerate of Ura+ Relativerate of

Strain”
Relevant genotype Transcription Rate of Ura+ derivativesb
derivatives destabilization“
MBW1-35
Low type Wild 1.1 X (0.9-1.2) 1.0 X 10-~ 1 .o
0.9 X (0.6-1.2)
MBW1-35 type Wild High 5.3 X (3.4-13.8) 6.2 X 10-~ 6.2
7.1 X (4.3-16.1)
MBW3-35 mh2 Low 1.1 X lo-’1.3
(0.7-1.7) X lo-’ 130 (1.0)
1.2 X lo-’ (0.8-1.6)
1.6 X lo-’ (1.1-2.3)
MBW3-35 msh2 High 2.6 X lo-’3.0
(2.0-4.2) 300 X 10-3 (2.3)
2.8 X lo-’ (1.2-7.5)
3.7 X lo-’ (2.4-6.9)
SJR482-35 Pmsl Low 0.8 X lo-’0.9
(0.5-0.9) X lo-’ 90 (1.0)
1.0 X 10-3 (0.5-1.2)
SJR482-35 Pml High 2.6 X lo-’2.3
(1.9-2.9) X 10-~ 230 (2.6)
1.9 X lo-’ (1.5-2.8)
The numbers 31 and 35 after the strain name indicates the length of the GT tract in base pairs.
Numbers in parentheses represent 95% confidence values for the rates measured in individual experiments.
‘Rates are relative to MBW1-35 grown under low transcription conditions. Numbers in parentheses represent relative rates
for the mismatch repair deficient strains grown under low and high transcription conditions.

al. (1993, 1995), the lack of large alterations does not sion. Thus, the effect examined by DATTAand JINKS-
reflect strain differences. It could be a consequence of ROBERTSON would probably not be detectable in our
the different fusion genes used to assay tract instability experiments.
or a consequence of the location of the assayed gene The observed frequency of tract instability is likely to
(located within the chromosome in our experiments be a function of the error rate of DNA polymerase and
and on a plasmid in the experiments of STRANDet al. the efficiency of DNA mismatch repair. The error rate
(1993, 1995). of DNA polymerase reflects both the DNA polymerase
The six- to ninefold induction of microsatellite insta- slippage rate and the rate with whicherrors are corrected
bility by high levels oftranscription is lessthan the-30- by the associated proofreading exonuclease activity.
fold induction of reversion observed for a lys2 Thus, if the frequency with whichDNA polymerase gen-
frameshift mutation in studies done with the same pro- erates mismatches is a, and the frequency of mismatches
moter (DAITA and JINKS-ROBERTSON 1995). It is likely escaping the mismatch repair systemis 6, then the o b
that the induction of alterations by transcription in served frequency of simple repeat instability is ab, assum-
these two experiments has different causes. DATTAand ing that the frequencies of both processes are indepen-
JINKS-ROBERTSON (1995) suggested that high levels of dent. Transcription could elevate simple repeat
transcription may lead to increased levels ofDNA dam- instability in a variety of ways: by decreasing the rate of
age that is repaired (in part) by the error-prone REV3 mismatch repair, by increasing the error rate of DNA
encoded DNA polymerase. This effect increases the mu- polymerase, by affecting both DNA polymerase and mis-
tation rate from 10-g/division to 3 X 10-8/division. In match repair, or by activating a new pathway (for exam-
our experiments, high levelsof transcription elevate ple, recombination) leading to simple repeat instability.
tract alterations from -10-5/division to 6 x 10-5/divi- Our results limit the possible explanations for the

TABLE 4
Tract alterations observed in mh2 (MT3W3-35) and pml (SJR482-35) mismatch repair mutants

No. of tracts with additions or

deletions of base pairs
Strain“ GAL80 genotype/mediab +4
Transcription -2 Total
MBW3-35 GAL80/SD-complete Low 28 6 34
MBW3-35 GAL80/SGcomplete High 28 9 37
SJR482-35 GAL80/SD-complete Low 13 2 15
SJR482-35 GAL80/SGcomplete High 13 0 13
The numbers 31 and 35 after the strain name indicates the length of the GT tract in base pairs.
SD-complete medium contains dextrose, SG-complete medium contains galactose.
720 M. Wierdl et al.

effects of transcription on tract instability. Since tran- BOEKE, J. D., F. LACROUTE and G. R. FINK,1984 A positive selection
for mutants lacking orotidine-5'-phosphate decarboxylase activ-
scription destabilizes poly GT tracts even in mismatch ity in yeast: 5-fluoroorotic acid resistance. Mol. Gen. Genet. 197:
repairdeficient strains, the effect of transcription in 345-346.
wild-type strains is not solely a consequenceof reducing CASKEY, C. T., A. PIZZUTI, Y.-H. Fu, R. G. FENWICK, JR. and D.L.
NEI.SON,1992 Triplet repeat mutations in human disease. Sci-
the efficiency of the MSH2/PMSl mismatch repair sys- ence 256: 784-789.
tem. If the effect of transcription was solely on theDNA CHONG,S . S., A. E. McCu.r., J. COTA,S. H. SUBRAMONY, H. T. O m
polymerase error rate, then one would expect that the et al., 1995 Gametic and somatic tissue-specific heterogeneity
of the expanded SCAl CAG repeat in spinocerebellar ataxia type
effects of transcription and mismatch repair defects 1. Nat. Genet. 10: 344-350.
would be multiplicative. By this model, the expected DATTA,A., and S . JINKS-ROBERTSON, 1995 Association of increased
level of instability for MBW3-35 (grown under condi- spontaneous mutation rates with high levels of transcription in
yeast. Science 268: 1616-1619.
tions leading to high levels of transcription) should be DIXON,W. J., and F. J. MASSEY, JR., 1969 Introduction to Statistical
-800, instead of the observed value of 300 (Table 3). Analysis. McGraw-Hill Book Company, New York.
The third model that can be ruled out is the activation DRAWN, R., A. SANCAR and D. REINBERG, 1994 Where transcription
meets repair. Cell 77: 9-12.
of a mechanism of tract instability that is independent FARBER, R. A., T. D. PKTES,M. DOMINSKA, S. S. HUDGENS and R. M.
of the DNA mismatch repair system. By this model, the LISKAY,1994 Instability of simple sequence repeats in a mam-
effects of transcription would be additive with the ef- malian cell line. Hum. Mol. Genet. 3: 253-256.
FLEIG,U. N., R. D. PRIDMORE and P. PHII.IPPSEN, 1986 Construction
fects of mutations of eliminating DNA mismatch repair. of LYS2 cartridges for use in genetic manipulations of Saccharo-
Thus, this model predicts a relative rate of tract instabil- myces cereuisiae. Gene 46: 237-245.
ity for MBW3-35 (grown under conditions leading to FOX,M. S., J. P. RADICELLA and K. YAMAMOTO,1994 Some features
of base pair mismatch repair and its role in the formation of
high levels oftranscription) of 136, a value considerably genetic recombinants. Experientia 50: 253-260.
below the relative rate observed (300). FREUND, A,"., M. BICHARA and R. P. P. FUCHS,1989 Z-DNA forming
The conclusion that appearsto fit the databest is that sequences are spontaneous deletion hot spots. Proc. Natl. Acad.
Sci. USA 86: 7465-7469.
transcription has two effects, reducing the efficiency of HAMADA, H., M. G. PETRINOand T. KAKUNAGA,1982 A novel re-
DNA mismatch repair two- to threefold and increasing peated element with Z-DNA-forming potential is widely found
the error rate of DNA polymerase about two- to three- in evolutionarily diverse eukaryotic genomes. Proc. Natl. Acad.
Sci. USA 79: 6465-6469.
fold. The neteffect of these two processes is to increase HEMMINKI, A,, P. PELTOMAKI, J.-K. MECKLIN, H. JARVINEN, R. SAL<)-
tract instability in a wild-type strain about sixfold. One VAARA et al., 1994 Loss of the wild type M L H l gene is a feature
explanation for these effects is that the transcription of hereditarynonpolyposiscolorectal cancer. Nat. Genet. 8:
405-410.
complex may displace DNA polymerase and the mis- HENDERSON, S. T., and T. D. PETES,1992 Instability of simple se-
match repair proteins from DNA (Fox et al. 1994). The quence DNA in Saccharomyces cereuisiae.Mol. Cell. Biol. 12: 2749-
displacement of DNA polymerase could increase the 2757.
JOHNSTON,M., and M. CARISON,1992 Regulation of carbonand
rate ofDNA polymerase slippage or reduce the effi- phosphate utilization, pp. 200-227 in The Molecular and Cellular
ciency of the proofreading exonuclease. Alternatively, Biology of the Yeast Saccharomyces, Ed. 2. edited by E. W. JONES,
the increased transcription could influence supercoil- J. R. PRINGLE and J. R. BROACH. Cold Spring Harbor Laboratory
Press, Cold Spring Harbor, NY.
ing in domains of the chromosome, reducing the effi- JOHNSTON, M., and R.W.DAVIS, 1984 A system of shuttle vectors
ciency of DNA-DNA polymerase interactions and DNA- and yeast host strains designed for efficient manipulation of DNA
mismatch repairproteininteractions. Whatever the in Saccharomyces cerevisiue. Mol. Cell. Biol. 4 1440-1448.
JONES, K. A.,J. T.KADONAGA, P. J. ROSENFELD, T.J. KELLYand R.TJAIN,
mechanism involved in transcription-induced destabili- 1987 A cellular DNA-binding protein that activates eukaryotic
zation of simple repeats, our results indicate that the transcription and DNA replication. Cell 48: 79-89.
rate of transcription is another parameter influencing KARRAN,P., and M. BINGAMI, 1994 DNA damage tolerance, mis-
match repair and genome instability. BioEssays 16: 833-839.
microsatellite stability. KRAFT,R.J., K. S. TARDIFF, K. S . KRAUTER and L. A. LEINWARD, 1988
Using miniprep plasmid DNA for sequencing double stranded
We thank M. STRAND, D. MALONEY, J. FRIDOVICH-KEIL and A. MOR- templates with Sequenase. Biotechniques 6: 544-545.
RISON for providing yeast strains or plasmids, E. SIA for comments LEA,D. E., and C. A. COWLSON, 1949 The distribution of the number
on the manuscript, and D. GORDENIN and F. STUARTfor advice on of mutants in bacterial populations. J. Genet. 49: 264-285.
the statistical analysis. This research was supported by grants from LEVINSON, G., and G. A. GUTMAN, 1987 High frequencies of short
the National Institutes of Health (GM-50892 to T.D.P. and GM-38464 frameshifts in poly-CA/TG tandem repeats borne by bacterio-
to S.J.-R.) and the National Science Foundation (MCB-9405953 to phage M13 in Escherichia coli K-12. Nucleic Acids Res. 15: 5323-
S.J.-R.). 5338.
MERLO,A,, M. MARRY, E. GABRIELSON, R. VOLLMER, s. B.BAM.IN et
al., 1994 Frequent microsatellite instability in primary small cell
LITERATURE CITED lung cancer. Cancer Res. 5 4 2098-2100.
MIRONOV, N. M., M. A,". AUGELON, G. I. POTAPOVA, Y.OMORI, 0. v.
AALTONEN, L. A,, P. PELTOMAKI,F. S . LEACH, P.SISTONEN, L. F'YLKKA- GORBUNOV et al., 1994 Alterations of (CA)n DNA repeats and
NEN et al., 1993 Clues to the pathogenesis of familial colorectal tumor supressor genes in human gastric cancer. Cancer Res. 5 4
cancer. Science 260: 812-816. 41 -43.
A L A N IE.,
, and N. KLECKNER, 1987 A newtype of fusion analysis ORTH,It, J. HUNG, A. GAZDAR, A. BOWCOCK,J. M. MATHIS et al., 1994
applicable to many organisms: protein fusions to the URA? gene Genetic instability in humanovarian cancer cell lines. Proc. Natl.
of yeast. Genetics 117: 5-12. Acad. Sci. USA 91: 9495-9499.
AUSUBEL, F. M., R. BRENT,R. E. KINGSTON, D. M. MOORE, J. G. SEID- PINT, T.,1984 Toxicity of 2deoxygalactose to Saccharomyces cerevis-
MAN et al., 1994 Cuwent Protocols in Molecular Biology, vol. 2.John iae cells constitutively synthesizing galactose-metabolizing en-
Wiley & Sons, Inc., New York. zymes. Mol. Cell. Biol. 4 994-996.
 Transcription-AssociatedMutations 721
RAHMOUNI, A. R., and R. D. WELLS,1989 Stabilization of 2 DNA in within tracts of simple repetitive DNA in Saccharomyces cereuisiae.
vivo by localized supercoiling. Science 246: 358-363. Proc. Natl. Acad. Sci. USA 92: 10418-10421.
RICHARDS, R. I., and G. R. SUTHERLAND, 1994 Simple repeat DNA STREISINGER, G., Y. OKADA, J. EMRICH,J. NEWTON,A. TSUGITAet aZ.,
is not replicated simply. Nat. Genet. 6 114-116. 1966 Frameshift mutations and the genetic code. Cold Spring
SAMBROOK, J., E. FRITSCHand T. MANIATIS, 1989 Molecular Cloning; Harbor Symp. Quant. Biol. 31: 77-84.
A Labmatoly Manual. Cold Spring HarborLaboratory Press, Cold TAUTZ,D., and M. RENZ,1984 Simple sequences are ubiquitous
Spring Harbor, NY components ofeukaryotic genomes. Nucleic Acids Res. 1 2 4127-
SHERMAN, F., 1991 Getting started with yeast, pp. 12-18 in Guide to 4138.
Yeast Genetics and Molecular Biology, edited by C. GUTHRIEand G. TAUTZ,D., M. TRICKand G. A.DOVER, 1996 Cryptic simplicity in
R. FINK. Academic Press, San Diego. DNA is a major source of genetic variation. Nature 3 2 2 652-656.
SHERMAN, F., G. R. FINKand J. B.HICKS, 1983Methods in yeast WEBER, J. L., 1990 Informativeness of human (dcdA)n-(dGdT)n
genetics. ColdSpring HarborLaboratory Press, Cold Spring Har- polymorphism. Genomics 7: 524-530.
bor, N Y . WIND,N. D., M. DEKKER, A. BERNS,M. R AO w and H.T. RIELE,1995
SIKORSKI, R. S., and P. HIETER,1989 A system of shuttle vectors and Inactivation of the mouse Msh2 gene results in mismatch repair
yeast host strains designed for efficient manipulation of DNA in deficiency, methylation tolerance, hyperrecombination, andpre-
Saccharomycescerevisiae. Genetics 122: 19-27. disposition to cancer. Cell 82: 321-330.
Wu, H.-Y., S. SHW,J. C. W m c and L.F. LIU, 1988 Transcription
SMITH, G. P., 1973 Unequal crossover and the evolution of
generates positively and negatively supercoiled domains in the
multigene families. Cold Spring Harbor Symp. Quant. Biol. 38:
template. Cell 53: 433-440.
507-513.
YOCUM,R. R., S. HANLEY, R. WEST,JR. and M. PTASHNE, 1984 Use
STRAND, M.,T. A. PROLIA,R. M. LISKAY and T. D. PETES,1993 De- of ZacZ fusions to delimit regulatory elements of the inducible
stabilization of tracts of simple repetitive DNA in yeast by muta- divergent GALI-GAL10 promoter in Saccharomyces cerevisiae. Mol.
tions affecting DNA mismatch repair. Nature 365: 274-276. Cell. Biol. 4: 1985-1998.
STRAND,M., M. C. EARLEY,G. F. CROUSEand T. D. PETES,1995
Mutations in the MSH3 gene preferentiallyleadtodeletions Communicating editor: P. L. FOSTER