Stability of intrastrand hairpin structures formed by the CAG/CTG class of DNA
triplet repeats associated with neurological diseases
Stability of intrastrand hairpin structures formed by the CAG/CTG class of DNA triplet repeats associated with neurological diseases
John
Petruska
,
Norman
Arnheim
and
Myron F.
Goodman*
Department of Biological Sciences, Hedco Molecular Biology Laboratories,
Molecular Biology Program, University of Southern California,
Los Angeles
, CA 90089-1340,
USA
Received March 26, 1996;
Revised and Accepted April 19, 1996
ABSTRACT
Expansions of trinucleotide repeats in DNA, a novel source of mutations
associated with human disease, may arise by DNA replication slippage initiated
by hairpin folding of primer or template strands containing such repeats. To
evaluate the stability of single-strand folding by repeating triplets of DNA bases, thermal melting
profiles of (CAG)
10
, (CTG)
10
, (GAC)
10
and (GTC)
10
strands are determined at low and physiological salt concentrations, and
measurements of melting temperature and enthalpy change are made in each case.
Comparisons are made to strands with three times as many repeats, (CAG)
30
and (CTG)
30
. Evidence is presented for stable intrastrand folding by the CAG/CTG class of
triplet repeats. Relative to the GAC/GTC class not associated with disease, the
order of folding stability is found to be CTG > GAC
[approx] CAG > GTC for 10 repeats. Surprisingly, the folds formed by 30 repeats of CTG
or CAG have no higher melting temperature and are only 40% more stable in free
energy than those formed by 10 repeats. This finding suggests that triplet
expansions with higher repeat number may result from the formation of more
folded structures with similar stability rather than fewer but longer folds of
greater stability.
INTRODUCTION
At least nine human diseases, most of which are neurodegenerative, are known to
result from expansions of trinucleotide repeats in genomic DNA (
1
-
5
). Of all possible classes of repeating triplets in DNA, three have been
associated with disease; namely, the CAG/CTG class, the CGG/CCG class and, most
recently, the GAA/TTC class (
6
). Expansions of the CAG/CTG class of triplet repeats, including permutations
AGC/GCT and GCA/TGC, are associated with Huntington's (HD) and six other
neurological diseases; those of the CGG/CCG class, with Fragile X syndromes,
FRAXA and FRAXE.
Genetic instability in the number of tandemly repeating triplets is a major
characteristic of these disorders (reviewed in
1
-
5
). The number of triplet repeats in disease-associated genes tends to increase in germline transmission from parent to
offspring. Usually, the larger the repeat number the higher the probability of
expansion and the earlier and more severe the disease.
In the present paper, we examine the single-strand folding properties of the CAG/CTG of triplet repeats associated
with neurological diseases. For comparison, we also examine structures formed
by repeating triplets of the GAC/GTC class, not associated with any known
disease. A favored molecular explanation for genetic expansions of tandemly
repeating triplets is primer/template slippage in DNA replication, an idea
first proposed to explain frameshift mutations in simple sequence repeats (
7
).
The tendency for primer to slip on template during replication of triplet repeat
regions is attributed to the ability of primer or template strands to fold into
hairpin structures stabilized by intrastrand basepairing (8-13). By characterising heat denaturation profiles of disease-associated (CAG/CTG) and unassociated (GAC/GTC) triplet repeat
sequences, we are able to evaluate thermal stabilities of triplet repeat
folding that may be important for understanding the slippage process at the
molecular level.
MATERIALS AND METHODS
Strand preparations
Synthetic 30mer DNA strands with 10 triplet repeats-(CAG)
10
, (CTG)
10
, (GAC)
10
and (GTC)
10
-were made using an Applied Biosystems 392 DNA/RNA synthesizer with [beta]-cyanoethyl phosphoramidites. Each strand was purified by
electrophoresis on 12% polyacrylamide gel in 8 M urea, cutting out the desired
band and extracting DNA from the band. Extracted samples at high strand
concentration (absorbance
A
260
= 5-10/cm)
were dialyzed extensively against low ionic strength buffer (5 mM NaH
2
PO
4
, 5 mM Na
2
HPO
4
, 1 mM Na
4
EDTA, pH 7.0) and stored frozen at -70oC.
Synthetic 90mer DNA strands-(CTG)
30
, (CAG)
30
and random (C,A,G)
30
-were purchased from Operon Technologies, Inc. The purchased samples, at
high strand concentration (absorbance
A
260
[approx] 10/cm), were dialyzed against low ionic strength buffer and stored frozen
as above.
The DNA strands used in our melting studies each had an OH group on the 5'-end as well as on the 3'-end.
Melting studies
Thermal denaturation of DNA samples is first examined in the low ionic strength
buffer (19 mM Na
+
) used to store frozen samples. At this low counterion concentration, melting of
secondary structure occurs <80oC even for correctly paired duplexes such as (CAG)
30
@(CTG)
30
. We examine melting profiles obtained by measuring
A
260
versus temperature T, while T is raised from 20 to 80oC at a constant rate, starting at 2oC/min. A slower rate of heating (0.5oC/min) is used to establish equilibrium conditions needed to
evaluate standard enthalpy change ([Delta]Ho) from melting curve slope by a van't Hoff expression (
14
). When an ordinary cuvette is heated to 2oC/min, a teflon cover tightly wrapped with parafilm is sufficient to
prevent significant evaporation; at slower rates of heating, a layer of mineral
oil, >= 5 mm, on the sample is also needed to prevent excessive evaporation. To
obtain corresponding curves at physiological ionic strength, 1/33 vol 5 M NaCl
is added, raising [Na
+
] to 167 mM. Unless stated otherwise, the total concentration of DNA bases in
each case is ~0.1 mM, with
A
260
[approx] 1 for 1 cm pathlength.
To reveal differences in secondary structure, we present melting curves obtained
at 2oC/min, showing the
A
260
ratio,
A
260
(ToC)/
A
260
(80oC), with 80oC representing the melted state. The observed
A
260
value at 80oC in each case is given in the figure legends. The curve of
A
260
ratio versus T is presented for each strand and also for the correctly paired
duplex formed by annealing equimolar amounts of the two complementary strands.
To establish equilibrium, the following melting profiles were carried out at a
slower heating rate (0.5oC/min): (CTG)
10
, (CAG)
10
, (CAG)
30
, (GAC)
10
, (GTC)
10
@(GAC)
10
and (CTG)
10
@(CAG)
10
in low salt (19 mM Na
+
); (CTG)
10
was also examined at slow rate in high salt (167 mM Na
+
). The slower heating caused essentially no change in the slopes for any of the
curves
>37oC obtained at 2oC/min, indicating that stably folded structures are evaluated close to
equilibrium at both rates, and only these structures are included in Table
1
.
Evaluation of T
m
and
[Delta]Ho
from melting curves
A sigmoidal (S-shaped) melting profile represents a transition from ordered to disordered
states. The temperature at the sigmoidal midpoint, where the curve becomes
linear with steepest slope, is identified as T
m
, the melting temperature. At T = T
m
ordered and disordered structures are considered equally probable, so [Delta]G = 0 and [Delta]S
=
[Delta]H/T
m
are the corresponding free energy and entropy changes for the melting
transition centered on T
m
. When temperature is raised slowly enough to assure equilibrium and the melting
curve is properly normalized, the slope of the linear region centered on T
m
can be used to evaluate the standard enthalpy change [Delta]Ho in the transition, according to a `van't Hoff' formula derived by
Marky and Breslauer (
14
). The normalized slope ([sigma]) around T
m
is measured after drawing linear baselines at low and high temperatures (
14
) and normalizing absorbance values between the baselines so that they range
from 0 to 1, with 0.5 being the value at T
m
. The `van't Hoff' value of [Delta]Ho is obtained by the formula (14), [Delta]Ho
= (2 + 2n)RT
m
2
[sigma], where [sigma] is the normalized slope or first derivative (with respect to T)
at T = T
m
, with R being the gas constant (1.987 cal/mol-K) and n the
molecularity
or number of strands in the ordered state (n = 1 for single-stranded state, 2 for double-stranded). To measure [sigma] as the first derivative, we use a linear least-squares fit to at least 10 data points in the linear
region centered on T
m.
End-labeling and electrophoretic analysis
To distinguish between single-strand structures and dimeric or multimeric associations, samples of each
strand are labeled at the 5'-end with
32
P, using [[gamma]-
32
P]ATP and T
4
polynucleotide kinase, and subjected to 12% polyacrylamide gel electrophoresis
(PAGE) in 0.5* TBE (45 mM Tris-borate pH 8.0, 1 mM EDTA). After frozen samples are thawed, end-labeling is carried out at low temperature (<20oC) to retain unstable associations promoted by
freezing. Each labeled sample is divided into two parts, one kept on ice until
loading on PAGE apparatus, the other kept on ice until ~30 min before loading, when it is preheated to 37oC for 30 min to deliberately denature unstable components. During
PAGE, conducted in cold room, temperature is maintained at 4oC and current is kept low enough (5 mA at 25 V/cm) to avoid heat
denaturation during electrophoresis.
RESULTS
The thermal stability of single-strand folding by triplet repeat sequences is examined at low (19 mM) and
physiological (167 mM) salt concentrations, using melting curves obtained by
measuring
A
260
versus T at a constant heating rate. At a rate of 2oC/min, we observe in some cases low-temperaure (<37oC) sigmoidal features of unstable associations between strands
promoted by freezing as well as higher temperature sigmoidal profiles of stably
folded structure. At a slower rate (0.5oC/min) the low-temperature features become much less conspicuous, because sigmoidal
slope is reduced as expected for non-equilibrium associations between two or more strands, but the higher-temperature profiles show very little change as expected for single
strand folding close to equilibrium. Since slower heating, which requires an
oil layer to retard evaporation, does not significantly alter the melting
profiles >37oC, only melting curves obtained at 2oC/min (without oil) are presented.
The disease-associated triplet sequence, 5'-CAG-3', and its complement 5'-CTG-3', are each examined in
two strand lengths, 30mer strands with 10 triplet repeats and 90mer strands
with 30 repeats. Also, for comparison, the `sister' sequences having the same
bases in reverse order, 5'-GAC-3' and 5'-GTC-3', which are not associated
with any known disease, are each examined in 30mer strands with 10 repeats. The
denaturation profiles of single-strand folded structures are also compared with those of their correctly
paired double-helical structures formed by slowly melting and annealing equimolar
mixtures of the two complementary strands.
Melting curves for CAG and CTG repeats
At low ionic strength (19 mM Na
+
), where even normal DNA duplexes melt <80oC, sigmoidal melting curves are obtained (Fig.
1
a) for each of the 30mer strands, (CAG)
10
and (CTG)
10
, as well as for their correctly paired duplex, (CAG)
10
@(CTG)
10
. The corresponding curves for the 90mer counterparts, (CAG)
30,
(CTG)
30
and (CAG)
30
@(CTG)
30
, are shown in Figure
1
b. Included for comparison is the approximately linear curve (Fig.
1
b, light dashed line) obtained for (C,A,G)
30,
a `randomized' version of (CAG)
30
.
Melting curves for GAC and GTC repeats
Sigmoidal melting features are also observed for GAC and GTC repeats (Fig.
2
a), which have the same bases as CAG and CTG respectively, but in reverse order.
The curves shown for (GAC)
10
and (GTC)
10
in Figure
2
a are obtained with strand concentrations similar to those in Figure
1
a, with nearly equivalent absorbance in the denatured (80oC) state (see legend to Fig.
2
a). The (GAC)
10
@(GTC)
10
duplex formed by annealing the two strands has a melting curve (Fig.
2
a, dashed line) almost identical to that observed for (CAG)
10
@(CTG)
10
in Figure
1
a. However, the single strands have very different profiles. The strand with GTC
repeats shows only one sigmoidal transition (T
m
= 39oC), whereas that with GAC repeats shows two (34 and 47oC).
Analysis by end-labeling and electrophoresis
By labeling strand 5'-ends with
32
P and using electrophoresis (12% PAGE) to separate structures by size, we
confirm that the unstable structures promoted by freezing are mainly dimeric
associations between like strands. However, because strands are diluted by end-labeling and by loading and subsequent electrophoresis on the gel, the
amount of inter-strand association seen in the gel is likely to be less than in melting
curves. Figure
3
shows PAGE results obtained with and without preheating (to 37oC for 30 min) to melt the unstable components of (CTG)
10
and (GAC)
10
having T
m
= 29
and 34oC respectively, seen in Figures
1
a and
2
a. The preheated samples are in lanes on the left (Fig.
3
); the unheated samples are on the right, in the same order. By comparing each
`preheated' lane on the left with the corresponding `unheated' lane on the
right, and referring to the size of duplex DNA markers (on the extreme right),
we can gauge the approximate size of structure removed by preheating.
Stabilization of folded structures by added salt
By adding NaCl to raise counterion concentration from low (19 mM) to
physiological (167 mM) levels, we find that single-strand folding of triplet repeats is greatly stabilized, much like normal
duplex DNA (
15
-
18
). For example, in the case of single strands with 10 or 30 repeats, increases
in T
m
of 13-15oC are observed (Table
1
), compared with 16-17oC
for their correctly paired duplexes.
In Table
1
, along with T
m
are presented [Delta]Ho, the standard enthalpy change upon melting as deduced from melting
curve slope, and the corresponding free energy change [Delta]Go evaluated at 37oC. To obtain [Delta]Go = [Delta]Ho - T[Delta]So, we use the entropy
expression, [Delta]So = [Delta]Ho/T
m
, with T
m
and T in degrees K, and [Delta]Ho obtained from the normalized slope at T
m
. The result is [Delta]Go = [Delta]Ho(T
m
- T)/T
m
whose value at T = 37oC is shown in each case (Table
1
). Since [Delta]Ho is positive (heat absorbed) and T
m
in degrees K is positive, the calculated [Delta]Go
has the same sign as T
m
- 37oC and indicates the stability of folded structure at 37oC
.
Comparison of melting temperature T
m
, standard enthalpy change [Delta]H
o
, and corresponding free energy change [Delta]Go at 37oC required to melt secondary structure formed by DNA triplet
repeats at low and physiological ionic strengths
a
Repeat structure
Ionic strength, [Na
+
]
0.02 M
0.17 M
T
m
[Delta]Ho
[Delta]Go
(37oC)
T
m
[Delta]Ho
[Delta]Go
(37oC)
oC
kcal/mol
kcal/mol
oC
kcal/mol
kcal/mol
Hairpin folding
(CTG)
30
51
72
3.1
66
76
6.5
(CAG)
30
46
48
1.4
60
50
3.5
(CTG)
10
51
52
2.2
66
55
4.7
(CAG)
10
47
36
1.1
60
38
2.6
(GAC)
10
47
39
1.2
62
40
3.0
(GTC)
10
39
48
0.3
53
50
2.4
Base-paired duplex
(CAG)
10
@(CTG)
10
67
190
16.7
83
220
28.3
(67)
(188)
(16.6)
(85)
(226)
(30.3)
(GAC)
10
@(GTC)
10
66
200
16.9
83
230
29.2
(74)
(200)
(21.3)
(87)
(230)
(31.9)
a
The standard free energy change for melting, [Delta]Go = [Delta]Ho(T
m
- T)/T
m
, is evaluated at T = 37oC = 310 K, using T
m
and [Delta]Ho obtained from sigmoidal midpoint and normalized slope of melting
curve as described in Materials and Methods. Also shown for each duplex are
expected values (in brackets), based on nearest-neighbor doublet evaluations in normal double-helical DNA in 0.02 M and higher Na
+
concentrations (
15
).
As shown in Table
1
, the addition of 0.15 M NaCl, to raise [Na
+
] from 0.02 to 0.17 M, greatly stabilizes the single-strand folding of triplet repeats, raising both T
m
and [Delta]Go
by substantial amounts. We note that [Delta]Go, being equivalent to the product ([Delta]Ho/T
m
)(T
m
- T), tends to change in proportion to T
m
- T for strands of equal length, since [Delta]Ho/T
m
stays relatively unchanged. The folding of 10 or 30 CTG repeats has T
m
elevated by 15oC (51-66oC), the somewhat less stable folding of CAG repeats, by 13-14oC. Comparable elevations are also seen for 10 GAC
and GTC repeats. These elevations in T
m
are almost as large as we find for the correctly paired duplexes, (CAG)
10
@(CTG)
10
and (GAC)
10
@(GTC)
10
, namely ~16-17oC, which agree with known salt-induced T
m
increases in normal DNA duplexes (
15
-
18
).
In contrast, the interstrand associations promoted by freezing show much less
stabilization by salt. The dimeric association of (CTG)
10
shows only a 7oC
increase in T
m
(29-36oC), while that of (GAC)
10
shows a 3oC
decrease
(34-31oC). Since their T
m
values remain <37oC, such associations are unstable under physiological conditions and
therefore are not included in Table
1
.
DISCUSSION
Single-strand folding by the CAG/CTG class of triplet repeats has now been
examined with chemical probes, NMR, electrophoresis, thermal melting analysis
and theoretical calculations (
10
-
13
,
19
,
20
). All these studies, including our own, are consistent with the idea that such
repeats form stable hairpin folds that may be obstacles causing slippage in DNA
replication. Our results suggest, however, that thinking of these structures as
`classical' hairpins with stem lengths proportional to repeat number (
21
,
22
) may be an oversimplification. The melting profiles of strands with 30 repeats
provide an important new insight when compared with their 10 repeat
counterparts. The T
m
values for (CTG)
30
and (CAG)
30
(Fig.
1
b) are no higher than those of (CTG)
10
and (CAG)
10
(Fig.
1
a) and their [Delta]Ho and resultant [Delta]Go for melting are only 40% higher (Table
1
). Since T
m
stays the same, and [Delta]Ho (as indicated by sigmoidal slope) only increases by 40%, instead of
by 200% as predicted in proportion to repeat number (
10
,
12
), it seems likely that 30 repeats tend to form more complex hairpin folds with
stems not much longer than those formed by 10 repeats.
To our knowledge, this is the first time [Delta]Ho
has been experimentally evaluated in addition to T
m
for single-strand folding of triplet repeats. The combination of [Delta]Ho and T
m
embodied in [Delta]Go
provides a more reliable measure of folding stability than T
m
alone. The T
m
and [Delta]Go
values for 10 repeats (Table
1
) both indicate that folding stability in 19 mM Na
+
counterion decreases in the order, (CTG)
10
> (GAC)
10
[approx] (CAG)
10
> (GTC)
10
. A previous study of 15 repeats, using much less Na
+
(1 mM) and T
m
evaluated by electrophoretic mobility melting profiles (
19
,
20
), indicates (GTC)
15
[approx] (GAC)
15
> (CAG)
15
[approx] (GTC)
15
. The main difference is that we find T
m
= 47oC for (CAG)
10
, the same as for (CAG)
30
(Table
1
), whereas the value reported for (CAG)
15
is only 38oC (
20
). Since T
m
and [Delta]Go are strongly salt-dependent, as shown in Table
1
, some disagreement may be expected when different salt concentrations are used
by different methods.
We note that for CNG repeats (CNGCNG...), single-strand folding yields a hairpin stem stabilized by base pairing between GC
doublets, known to have attractive base stacking interactions even in low salt
(
15
,
16
). However, for GNC repeats (GNCGNC...), hairpin folding results in base pairing
between CG doublets, which have less favorable stacking in low salt (
15
,
16
). As salt is added, CG stacking becomes more attractive while GC stacking stays
the same, so that CG and GC doublets form equally stable stacked base pairs in ~1 M salt (
15
). At this time little is known about how the mispairs flanking base-paired doublets in hairpin stems affect doublet stacking interactions.
Hairpin stems of CTG or GTC repeats have pyrimidine mispairs (T opposite T)
whereas those of CAG or GAC repeats have purine mispairs (A opposite A). As
seen in Table
1
, the purine mispairs result in lower enthalpy changes for melting. At each salt
concentration, both (CAG)
10
and (GAC)
10
show lower [Delta]Ho
than (CTG)
10
and (GTC)
10
. This result is consistent with expectation, since purines are larger than
pyrimidines and therefore more likely to interfere with normal stacking of GC
and CG doublets. However, [Delta]Ho
is not a good index of stability at physiological temperature. As pointed out in
the Results section, the proper stability index ([Delta]Go at 37oC) correlates much better with T
m
than with [Delta]Ho. The reason is that stability is determined by the product of two
terms one of which (T
m
- 37oC) changes much more than the other ([Delta]Ho/T
m
), for equal numbers of repeats.
We see that (CTG)
10
folding is ~2 kcal/mol more stable than (GTC)
10
at low to physiological salt concentrations (Table
1
). This is the biggest difference in stability we find for single-strands with 10 repeats. The difference is consistent with the idea that
GC doublets form stronger stacked base pairs than CG doublets even when flanked
by T mispairs. However, when flanked by A mispairs, the difference in stability
between GC and CG doublets is no longer apparent. We find (CAG)
10
and (GAC)
10
have similar folding stability, intermediate between (CTG)
10
and (GTC)
10
. As seen in Table
1
, hairpin folds of triplet repeat sequences are much less stable than their
correctly paired duplexes, their [Delta]Go
values being about an order of magnitude smaller.
Upon increasing repeat number, we find (CTG)
30
and (CAG)
30
single strands have [Delta]Ho and [Delta]Go values only 1.4 times as large as their shorter
counterparts with 10 repeats, and no higher T
m
(Table
1
). This is perhaps the most surprising result, considering that strands with
three times as many repeats might be expected to form a hairpins three times as
long and therefore have [Delta]Ho and [Delta]Go about three times as large.
Models for hairpin folding by 10 triplet repeats
Shown in Figure
4
are some hairpin folding models that may explain our results for single strands
with 10 repeats. To account for the high stability of (CTG)
10
folding (Fig.
1
a and Table
1
), hairpin structure I is suggested for this strand (Fig.
4
). This compact fold yields the longest possible duplex stem made with basepairs
between GC doublets (G@C pair followed by C@G pair) and mispairs of T opposite T. Structure II for (CTG)
10
, with one less G@C bp and a short protruding `sticky end' end permitting weak dimerization, is
also suggested (Fig.
4
) to account for the minor component melting at 29oC (Fig.
1
a).
Comparison with theory
Theoretical calculations have been made assuming that larger numbers of CAG or
CTG repeats form proportionally longer and more stable hairpins (
10
,
12
). According to these calculations, enthalpy and resultant free energy changes
for hairpin melting should be about three times as high for 30 repeats as for
10 repeats, whereas we find them to be only 1.4 times as high (Table
1
). Such calculations may not fully take into account the flexibility of structures
containing mispairs and their potential to bend or kink into shorter domains of
folding. The greater degrees of freedom available to structures with repeated
mismatches may allow multiple short hairpins to form preferentially over longer
ones.
The exact structure of mispairs in triplet repeat hairpins is not known, but
there are some interesting theoretical possibilities. For example, in the case
of CAG hairpins, with mispaired As flanked by normal C@G and G@C base pairs, a `triad' structure has been proposed (
11
), in which each A is unstacked and hydrogen-bonded to G on the opposite strand, while allowing G to remain hydrogen-bonded to C in the normal manner. The proposed bonding of A to G, if
it occurs, apparently does not make CAG hairpins more stable than CTG hairpins
which lack such bonding opportunities. On the contrary, CTG hairpins are
significantly more stable, with both higher T
m
and larger [Delta]H
o
for melting than those formed by CAG repeats (Table
1
).
Relation to slippage
The exact cause of a DNA slippage event, the number of repeats added or
subtracted per event and the likelihood of such events occurring as DNA
polymerase traverses a trinucleotide repeat region is unknown. Formation of a
hairpin structure on the template strand might result in a deletion if
polymerase can bypass the hairpin. Alternatively, hairpin formation on the
elongating primer strand might result in addition of repeats. Paternal
transmissions of disease alleles in pedigrees (reviewed in
27
-
30
) and in single sperm and single molecule studies (
27
-
30
), reveal a preponderance of expansions suggesting that deletion caused by
polymerase bypass of template hairpin structures is relatively rare. Data from
single sperm typing studies on Huntington's disease germline mutations support
the idea that blocks to polymerase lead to additions of a random number of
repeats per slippage event (
26
). Alleles with more repeats have a higher germline mutation frequency and add a
higher average number of repeats. Such alleles may form more complex secondary
structures, e.g., several smaller hairpins instead of one long one, consistent
with the physical-biochemical data presented here and lead to the addition of more repeats.
It has been suggested that the potential for slippage is greatest when the
lagging strand is the template (
27
-
30
). The idea proposed is that when an Okazaki fragment is initiated within a
repeated region, slippage of the elongating strand can occur at both its ends.
Each slippage event in a triplet repeat region must involve the melting of an
integral number of three basepair units in the primer/template duplex. The heat
absorption required is >20 kcal/mol of base-paired CAG/CTG triplet at physiological ionic strength (15), in accordance
with our finding of [Delta]Ho = 220 kcal/mol for the duplex with 10 base-paired triplets (Table
1
). This energy requirement can only be partially offset by hairpin folding in
the dissociated strand, since hairpins are much less stable than correct
duplexes. However, the repeat region in the template lagging strand, rendered
single stranded by the movement of the replication fork or by exonuclease
activity in mismatch repair, may form hairpin structures spontaneously. The
presence of stable hairpin folds in the template may contribute to primer
slippage by acting as physical blocks to primer elongation by polymerase. Thus
stalling of the polymerase at template folds might be a primary cause for
slippage of the nascent strand leading to triplet additions.
ACKNOWLEDGEMENT
This work was supported by National Institutes of Health grants AG11398, GM21422
and GM36745.
REFERENCES
1 Ashley, C.T., Jr and Warren, S.T. (1995) Annu. Rev. Genet.29, 703-728.
2 Sutherland, G.R. and Richards, R.J. (1995) Proc. Natl Acad. Sci. USA92, 3636-3641.MEDLINE Abstract
