Báo cáo khoa học: A single mismatch in the DNA induces enhanced aggregation of MutS Hydrodynamic analyses of the protein-DNA complexes pot

A single mismatch in the DNA induces enhanced
aggregation of MutS
Hydrodynamic analyses of the protein-DNA complexes
Nabanita Nag
1
, G. Krishnamoorthy
1
and Basuthkar J. Rao
2
1 Department of Chemical Sciences, Tata Institute of Fundamental Research, Mumbai, India
2 Department of Biological Sciences, Tata Institute of Fundamental Research, Mumbai, India
The DNA mismatch repair (MMR) system, an evolu-
tionarily conserved biochemical pathway, plays an
important role in regulating the genome by correcting
base mismatches arising either from replication errors
(error rate 10
)8
) or from homologous recombination
preventing recombination between DNA molecules
that have high sequence divergence (mismatches) [1–3].
Inactivation of MMR genes results in a significant
increase in the spontaneous mutation rate, thereby
leading to microsatellite repeat instability, where cells
become hyper-recombinogenic, which account for
% 40–50% of hereditary nonpolypopsis colorectal can-
cers in humans [1,4–6].
The most extensively studied adenine methyl directed
MMR pathway of Escherichia coli implicates the parti-
cipation of several gene products, including MutS,
MutL, MutH, DNA helicase II, single-stranded DNA
binding protein, exonuclease I, VII or RecJ exonuclease,

DNA polymerase III holoenzyme and DNA ligase [1,7].
In E. coli, repair is initiated by dimeric MutS protein
that recognizes a mismatch ⁄ insertion–deletion-loop with
an affinity that is only several-fold higher than that of
its binding to homoduplex [8,9]. After mismatch recog-
nition, MutS with the assistance of MutL initiates the
mismatch repair by activating MutH that nicks the
newly synthesized, unmethylated ‘GATC’ sequence
strand [1,10], following which a concerted action of
helicase ⁄ exonuclease ⁄ polymerase and ligase functions
ensue, thereby restoring the correct complementary
sequence in the DNA strand [1,2,7,11–13].
Currently, most efforts in mismatch repair studies
are focused on trying to reveal the finer mechanistic
Keywords
ATP; hydrodynamic radius; mismatch;
MMR; MutS
Correspondence
B.J. Rao, Department of Biological
Sciences, Tata Institute of Fundamental
Research, Homi Bhabha Road, Mumbai
400005, India
Fax: +91 22 2280 4610; 2280 4611
Tel: +91 22 2278 2606
E-mail:
(Received 20 June 2005, revised 19 August
2005, accepted 28 September 2005)
doi:10.1111/j.1742-4658.2005.04997.x
Changes in the oligomeric status of MutS protein was probed in solution
by dynamic light scattering (DLS), and corroborated by sedimentation ana-

lyses. In the absence of any nucleotide cofactor, free MutS protein [hydro-
dynamic radius (R
h
) of 10–12 nm] shows a small increment in size (R
h
14 nm) following the addition of homoduplex DNA (121 bp), whereas the
same increases to about 18–20 nm with heteroduplex DNA containing a
mismatch. MutS forms large aggregates (R
h
>500 nm) with ATP, but not
in the presence of a poorly hydrolysable analogue of ATP (ATPcS). Addi-
tion of either homo- or heteroduplex DNA attenuates the same, due to
protein recruitment to DNA. However, the same protein ⁄ DNA complexes,
at high concentration of ATP (10 mm), manifest an interesting property
where the presence of a single mismatch provokes a much larger oligomeri-
zation of MutS on DNA (R
h
>500 nm in the presence of MutL) as com-
pared to the normal homoduplex (R
h
% 100–200 nm) and such mismatch
induced MutS aggregation is entirely sustained by the ongoing hydrolysis
of ATP in the reaction. We speculate that the surprising property of a sin-
gle mismatch, in nucleating a massive aggregation of MutS encompassing
the bound DNA might play an important role in mismatch repair system.
Abbreviations
AFM, atomic force microscopy; DLS, dynamic light scattering; MMR, mismatch repair system; R
h
, hydrodynamic radius.
6228 FEBS Journal 272 (2005) 6228–6243 ª 2005 FEBS

details of the pathway by using the E. coli system as a
paradigm and by applying the same for the newly dis-
covered eukaryotic MMRs. Several studies have tried
to address two most important issues related to MutS:
(a) how does the system achieve the specific recogni-
tion of mismatches and insertion–deletion-loops? (b)
Following recognition of a mismatch, how does it
communicate the signal of such a mismatch to a dis-
tant landmark, the GATC-tract such that the down-
stream components, namely MutH-UvrD proteins, act
in highly mismatch-specific context? The former issue
has been elegantly addressed in the studies that des-
cribed the high-resolution structures of MutS bound to
a mismatch [14,15] as well as atomic force microscopy
(AFM) images of such complexes on mica surface in
air [16]. From these studies, one infers that MutS
makes specific contacts with the DNA helix in the
vicinity of a mismatch and generates a kink in the
DNA that seems to play a crucial role in mismatch
recognition process [14–16].
The latter issue of how MutS cross talks with
GATC tract has remained largely elusive. Three mod-
els have been proposed to address the same: according
to the first model, mismatch bound MutS undergoes a
conformational change following ATP binding and
hydrolysis that facilitates the recruitment of MutL, fol-
lowed by bidirectional translocation along the DNA
to encounter the downstream components, namely
MutH-UvrD proteins, at GATC tracts [17,18]. In the
second model, the MutS–ADP binary complex under-

goes an ADP to ATP exchange upon binding to mis-
match and forms an ATP hydrolysis independent, but
MutL dependent, sliding clamp along DNA that
encounters downstream MMR components during its
sliding action [9,19]. In the third model MutS remains
at the site of the mismatch following mismatch recog-
nition and interacts with the MutH through space via
MutL mediated crosstalk with MutH, thereby leading
to a loop formation of the intervening DNA [20,21].
Interestingly, additional studies from the proponents
of this model hint at ATP binding in the absence of its
hydrolysis as sufficient to trigger formation of a MutS
sliding clamp [22] of the sort described in the second
model [9,19].
Using nuclease footprinting, gel-shift analyses, and
surface plasmon resonance spectroscopy, it has been
demonstrated that MutS, in an ATP hydrolysis
dependent manner, establishes a near complete cover-
age of mismatch containing DNA, presumably through
a putative ‘treadmilling action’ of protein [23,24].
Essentially this model is a variation of the first one,
where the action of protein translocation on both sides
away from a mismatch, fuelled by the energy of ATP
hydrolysis, obviates the need of looping of intervening
DNA. The in vivo data supporting the MutS–MutL
foci formation suggests the possibility of extensive
recruitment of protein molecules at the sites of mis-
match repair, thereby achieving high enough local con-
centration of protein [25]. Importantly, such models
that implicate high local protein densities rely on the

property of protein aggregation that is presumably
coupled to its action of ATP hydrolysis. The solution
assays used so far to address this aspect of protein
dynamics did not enable one to monitor the same in
real time at its equilibrium conditions. In order to
achieve this, we have used an assay system that
allowed us to monitor the size of the protein complex
through its hydrodynamic properties, as a function of
not only ATP hydrolysis but also its binding to a mis-
match in the duplex DNA. We have observed that
MutS, which remains in dimer–tetramer equilibrium in
physiological conditions [26], has the propensity to
aggregate into dramatically large particles that show
hydrodynamic radii of more than several hundred
nanometers. Interestingly, such a protein aggregation
ensues specifically in the presence of an ongoing ATP
hydrolysis, since it is effectively ‘poisoned’ by the addi-
tion of a poorly hydrolysable analogue of ATP
(ATPcS). Moreover, additions of homo ⁄ heteroduplex
templates suppress the same by the squelching action
of DNA following protein binding. However, interest-
ingly, the protein regains a unique mode of aggrega-
tion even when bound to DNA following an
enhancement in the concentration of ATP. It is here
that the MutS ⁄ MutL system acquires a special prop-
erty of aggregation that is specific to the presence of
a single mismatch, thereby generating large protein ⁄
DNA complexes encompassing the mismatch.
Results
MutS interaction with MutL and stoichiometry

analyses
In this study, we have investigated the changes associ-
ated with the molecular aggregates of mismatch repair
proteins MutS, MutL in relation to their interaction
with mismatch containing DNA and the ongoing ATP
hydrolysis. Here we have mainly used dynamic light
scattering (DLS) to monitor the hydrodynamic radii
(R
h
) of the molecular complexes as a function of reac-
tion time and corroborated the essential findings by
protein fluorescence and other biochemical assays. The
principal players in the system namely, MutS and
MutL proteins showed a reasonably narrow distribu-
tion of R
h
values with a peak at 10 nm and 4 nm,
N. Nag et al. Hydrodynamic analyses of MutS aggregates
FEBS Journal 272 (2005) 6228–6243 ª 2005 FEBS 6229
respectively (Fig. 1A, Table 1). At the concentration
chosen (0.15 lm), the protein preparation exhibited
hardly any large particulate aggregates. Interestingly,
when the two proteins were mixed at 1 : 1 molar ratio
(0.15 lm each), we observed a distinct shift in the dis-
tribution of R
h
values towards a larger size with a
peak at 25 nm (Fig. 1A, Table 1). Such a shift towards
a size larger than that of the individual proteins is
consistent with the model where the two proteins

interact with each other, which we confirmed using
fluorescence assay (see below). These measurements
suggested that the proteins are amenable for studies
by DLS.
MutS protein was surface labelled with minimal
amount of fluorescamine (see Experimental proce-
dures), a primary amine reactive fluorescent probe,
such that the protein retained its biochemical activity
and exhibited sufficiently high steady-state fluorescence
emission at 477 nm, following excitation at 380 nm.
Fluorescamine labelled MutS was as active as unla-
belled protein in gel shifting ) specifically the mismatch
containing duplex rather than normal duplex ) thereby
revealing that dye binding has not affected the activity
of the protein measurably (data not shown). A fixed
amount of MutS protein (0.25 lm dimer) was titrated
with increasing concentrations of MutL protein and
steady-state intensity of fluorescence emission was
measured at each addition. MutL addition led to a
measurable drop in fluorescence intensity, based on
which we could construct a binding isotherm for MutL
interaction with MutS (Fig. 1B). Interestingly, such an
analyses revealed that the two proteins interact with
each other at an almost 1 : 1 molecular ratio with an
approximate K
d
of 70 ± 20 nm. Since the titration
(Fig. 1B) is close to a case of stoichiometric binding,
the estimated value of K
d

should be taken as the upper
limit. If one assumes that MutS exists largely as a
stable dimer, this result suggests that MutS–L com-
plex comprises of a dimer of each, which is entirely
consistent with the data in the literature [20]. This
Fig. 1. Analyses of the R
h
distribution of MutS as a function of its
interaction with MutL. (A) Analyses of the R
h
distribution of MutS
as a function of its interaction with MutL 0.15 l
M MutL (I), 0.15 lM
MutS (II) and a mixture of MutS and MutL (0.15 lM each) (III). The
samples were incubated in buffer A for 10 min at 22 °C, followed
by DLS analyses as specified. (B) MutS.MutL binding isotherm.
Fluorescamine-labelled MutS (0.25 l
M) was taken in buffer C and
titrated with MutL. The steady-state fluorescence measurements
were carried out with the excitation wavelength set at 380 nm
monitoring the change in fluorescence intensity at 477 nm (maxi-
mum k
em
). The smooth line represents the theoretical fit with
dissociation constant of 70 n
M.
Table 1. Hydrodynamic radii (R
h
in nm) of MutS and MutL in the
presence of Homo- or Hetero duplex DNA of different lengths (All

in minus ATP conditions, see text).
Species R
h
(nm)
MutS 10 (± 1)
MutL 4 (± 1)
MutS-MutL 25 (± 2)
MutS-Homoduplex (121 bp) 14 (± 1)
MutS-Heteroduplex (121 bp) 20 (± 2)
MutS-Homoduplex (61 bp) 14 (± 1)
MutS-Heteroduplex (61 bp) 18 (± 2)
MutS-Homoduplex (16 bp) 10 (± 1)
MutS-Heteroduplex (16 bp) 10 (± 1)
MutS-MutL-Homoduplex (121 bp) 30 (± 3)
MutS-MutL-Heteroduplex (121 bp) 35 (± 3)
Hydrodynamic analyses of MutS aggregates N. Nag et al.
6230 FEBS Journal 272 (2005) 6228–6243 ª 2005 FEBS
experiment not only corroborated the qualitative
conclusion drawn from DLS analyses, but provided
an equilibrium analysis of MutS complexation with
MutL.
Analysis of protein binding to DNA
It is of note that the DNA duplex itself does not show
sufficient scattering intensity in this concentration
range, thus precluding the estimation of its R
h
. Hence
all of the hydrodynamic radii in the following meas-
urements are directly ascribable to the protein species
in solution. Addition of either a single mismatch (het-

ero) or no mismatch (homo) containing duplex DNA
(0.15 lm of molecules) to MutS protein (0.15 lm)
resulted in interesting changes, where the distribution
of R
h
(of MutS peak at 10 nm) shifted towards a lar-
ger size. The particles in the presence of homoduplex
showed a peak at 14 nm whereas that with hetero-
duplex DNA showed a peak at 20 nm (Table 1). As
the duplex length in homo- vs. heteroduplex is identi-
cal, this result is consistent with the model in which
heteroduplex bound MutS appears to be a larger oligo-
mer than that of the homoduplex bound form (see
Discussion). Interestingly, the larger oligomeric state
of MutS, as reflected by higher R
h
, held true when the
duplex target size was reduced to 61 bp from that of
121 bp, but not so at much shorter duplex size of
16 bp (Table 1). In fact, MutS R
h
values obtained with
16 bp duplex (10 nm) were identical to that of free
MutS itself, thereby suggesting that protein failed to
stably bind the short duplex. The trend of the higher
oligomeric protein form associated with heteroduplex
DNA was observed with the MutS–MutL sample as
well, where addition of homo- and heteroduplex DNA
led to a shift of R
h

from 25 nm to 30 nm and 35 nm,
respectively (Table 1). We studied the changes in DLS
associated with DNA binding as a function of time.
Analysis of R
h
distribution pattern as a function of
time revealed that within about 5 min of DNA addi-
tion, MutS protein with heteroduplex DNA yielded
particles distinctly larger than that with homoduplex
DNA (data not shown). The observed difference in R
h
(% 20 nm and 14 nm with hetero and homoduplex,
respectively) remained constant throughout the time
course, suggesting the formation of stable and distinct
particles of bound MutS on these two DNA templates.
It is also important to note that the difference in R
h
between MutS bound to hetero vs. homoduplex was
evident even at a 1 : 1 molar ratio of protein to DNA
(0.15 lm each).
After establishing the basic system of MutS and
MutL and their interaction with DNA, as reflected by
the appropriate changes in R
h
as summarized in the
Table 1, we studied the changes in protein aggregation
in the presence of ATP hydrolysis.
ATP induced oligomerization of MutS
Addition of ATP to MutS led to a time-dependent
increase in the R

h
values of MutS particles. Moreover
the extent of MutS aggregation was clearly ATP con-
centration dependent. At the lowest concentration of
ATP (0.3 mm) tested, the R
h
values increased to % 17–
18 nm from that of 10 nm and the increase ensued
within 2–3 min of ATP addition (Fig. 2A). At the next
higher concentration of ATP (0.6 mm) the rise in R
h
was much more dramatic resulting in 200 nm particles
within the first 2 min and slowly increasing further
beyond 400–500 nm, the limit of detection by DLS, as
a function of time. The width of distribution of R
h
was in the range of % 50 nm in these samples. Next,
two higher concentrations of ATP (1 mm and 10 mm)
brought about rapid aggregation of MutS, generating
particles > 600 nm (Fig. 2A). In fact, it appears that
at these higher ATP concentrations, MutS aggregation
continues to increase even after several minutes of
ATP addition. This experiment demonstrated the ATP
concentration-dependent enhancement in MutS aggre-
gation results in very large (perhaps sedimentable, see
the next portion of the manuscript) particles whose R
h
value exceeded 500–600 nm. We tested whether ADP
also exhibits a similar effect on MutS aggregation by
analysing changes in R

h
as a function of time at two
different concentrations of ADP (1 mm and 10 mm).
The observed changes in R
h
with ADP were signifi-
cantly lower: at 1 mm and 10 mm ADP the R
h
increased to and stabilized at % 30 nm and % 150 nm,
respectively (data not shown). The aggregation was
also not due to the pyrophosphate anion (PP
i
) effect in
ATP as shown by the lack of increase in R
h
when PP
i
(1 mm and 10 mm) was added to MutS protein (data
not shown). This experiment revealed that the
observed effects of MutS aggregation were specific to
ATP rather than to ADP or PP
i
conditions (see
Discussion).
ATPcS addition ‘poisons’ ATP mediated
oligomerization of MutS
We tested the role of ATPcS in ATP induced MutS
aggregations by two different protocols. In the first
protocol increasing amounts of ATPcS were premixed
with 1 mm ATP, and then R

h
changes in MutS were
noted as a function of reaction time. We observed that
the presence of 0.5 mm ATPcS had only a marginal
N. Nag et al. Hydrodynamic analyses of MutS aggregates
FEBS Journal 272 (2005) 6228–6243 ª 2005 FEBS 6231
effect on the changes in R
h
induced by 1 mm ATP
(Fig. 2B) where the R
h
values sharply increased to
more than 400 nm by about 10 min. In contrast when
the concentration of ATPcS that was premixed with
ATP increased to 1 mm, the inhibitory effect on the
increase in R
h
was distinct and dramatic where the
particle size dropped to about 150 nm even after pro-
longed incubation. This experiment suggested that the
presence of ATPcS effectively poisoned the ATP medi-
ated aggregation of MutS. In another protocol we tes-
ted whether the suppression of MutS aggregation by
ATPcS could be reversed by the addition of ATP. As
expected, the control reaction where MutS was incuba-
ted with 1 mm ATPcS alone exhibited no MutS aggre-
gation throughout the incubation period of 30 min
where a particle with an R
h
of 10 nm was observed.

Interestingly when 1 mm ATP was added to this con-
trol at the midpoint of incubation, we observed the
induction of MutS aggregation and the particle size
gradually increased to about 70 nm, which suggested
that addition of ATP tends to partially reverse the poi-
soning effect of ATPcS. These controls taken together
suggest that MutS aggregation critically depends on
the level of ATP hydrolysis rather than ATP binding.
We studied this issue further in the following experi-
ments.
ATP induced aggregation of MutS is protein
concentration dependent
ATP induced aggregation of MutS was measured at
different concentrations of protein as a function of
time after adding ATP. The R
h
values obtained from
this study revealed that protein aggregation was least
at the lowest concentration of MutS (0.05 lm) where
Fig. 2. ATP hydrolysis induced aggregation of MutS. (A) Time
course of MutS aggregation as a function of ATP concentration.
Different concentrations of ATP were added to MutS protein
(0.15 l
M) in buffer A, followed by DLS analyses as a function of
time.[0 m
M (¯), 0.3 mM (n), 0.6 mM (h), 1 mM (s), 10 mM (,)of
ATP]. (B) ATP induced aggregation of MutS is inhibited by ATPcS.
In four independent reactions, 0.15 l
M of MutS was incubated with
either 1 m

M ATP (h) or 0.5 mM ATPc S+1mM ATP (premixed) (,)
or 1 m
M ATPc S+1mM ATP (premixed) (s)or1mM ATPcS(n),
followed by DLS analyses as a function of incubation time. In a
separate experiment, 1 m
M ATP was added to an ongoing reaction
containing 1 m
M ATPcS at its 15th min of incubation (d), followed
by DLS analyses. (C) Rate of ATP induced aggregation of MutS
depends upon the protein concentration. ATP (1 m
M) was added to
MutS taken at various concentrations [0.05 l
M (,), 0.1 lM (h),
0.15 l
M (n), 0.3 lM (e), 0.45 lM (s)], followed by DLS analyses as
a function of incubation time.
Hydrodynamic analyses of MutS aggregates N. Nag et al.
6232 FEBS Journal 272 (2005) 6228–6243 ª 2005 FEBS
the particles exhibited an R
h
of 100 nm that stayed
constant throughout the time course (Fig. 2C). At the
next highest concentration of MutS (0.1 lm), there was
an increase in the rate of MutS aggregation where the
particles reached an R
h
of 500 nm in about 30 min. It
appears that in this reaction protein aggregation con-
tinued to occur even after 30 min of incubation. In the
other samples where the protein concentrations were

> 0.1 lm (0.15, 0.3 and 0.45 lm), aggregation was
much more rapid resulting in particles of about
500 nm size within first 5–10 min and then the particle
size appeared to increase further with time (Fig. 2C).
In the next experiment we analysed the Mg
2+
depend-
ence of ATP induced MutS aggregation. In four differ-
ent samples that contained varying levels of Mg
2+
,R
h
value was monitored as a function of time following
ATP addition. The sample that contained no Mg
2+
showed the least protein aggregation reaching an R
h
of
about 100 nm. By the addition of 1 mm or more of
Mg
2+
, ATP induced MutS aggregation was substan-
tially increased generating particles of R
h
that were
larger than 400–500 nm (data not shown). The experi-
ment suggested that the ATP induced aggregation was
highly Mg
2+
dependent.

After establishing the basic conditions that influence
MutS aggregation, we studied the same in the presence
of duplex DNA targets that contained or did not
contain a mismatch (heteroduplex or homoduplex,
respectively).
MutS aggregation in the presence of duplex DNA
senses a single mismatched base pair
The role of ATP hydrolysis
MutS–DNA complexes were formed at 1 : 1 ratio,
ATP (1 mm) was added and then R
h
was analysed as
a function of time. As shown earlier (Table 1), before
the addition of ATP we recovered MutS–homoduplex
and MutS–heteroduplex complexes of about 14 and
20 nm in size, respectively. Following ATP addition
there was only a marginal increase in R
h
of both the
complexes where the former reached a size of
24–25 nm and the latter 20–22 nm (Fig. 3A). It is
important to note that MutS had shown extensive
aggregation reaching a particle size of about 500–
600 nm in the same conditions that contained no
duplex DNA (Fig. 2A). In contrast, the current experi-
ment, in which DNA was present, MutS aggregation
was significantly reduced suggesting that the protein
was sequestered on DNA such that free protein aggre-
gation induced by 1 mm ATP was dramatically
reduced. Moreover, reduction in MutS aggregation

was observed even with DNA targets (such as short
16-mer homo ⁄ heteroduplexes, or ssDNA 121-mer
oligonucleotides) that are poor binders of the protein
(data not given), implying that protein disaggregation
must have been brought about by relatively weak pro-
tein–DNA contacts. To test whether MutS aggregation
in the presence of DNA is affected by high concentra-
tion of ATP, we repeated the same experiment at
10 mm ATP and observed a surprising effect of MutS
aggregation that was significantly higher in the pres-
ence of heteroduplex DNA (R
h
% 140 nm) as com-
pared with that with homoduplex DNA (R
h
¼
50–60 nm) (Fig. 3B). The high ATP (10 mm) experi-
ment was carried out under conditions in which the
Mg
2+
level (5 mm) appeared to be limiting. To verify
that the observed high ATP effect arose from the phys-
iologically relevant Mg
2+
bound form of ATP, and
not from free ATP, we repeated the same experiment
at excess Mg
2+
(15 mm) as well. The DLS result at
high Mg

2+
essentially reproduced (Fig. 3C) the results
obtained earlier (Fig. 3B), confirming that the effect of
high ATP concentration was genuine where ) specific-
ally ) the presence of a mismatch induced a higher
level of protein aggregation (see Discussion). To test
whether mismatch specific enhanced aggregation of
MutS requires the sustained presence of ongoing ATP
hydrolysis, the following control experiments were car-
ried out. MutS–DNA (hetero ⁄ homo) reactions were
initiated at 10 mm ATP, followed by poisoning of
ATP hydrolysis by either EDTA or ATPcS (10 mm)at
early (3 min) or late (20 min) time-points of DLS-
time-course and analysing further the changes in R
h
.
We surmised that effective poisoning of ongoing ATP
hydrolysis by EDTA or ATPcS might unravel its role
in the maintenance of mismatch induced MutS aggre-
gation, if any. The R
h
analyses as a function of time
revealed that addition of EDTA or ATPcS had signifi-
cantly lowered MutS aggregation specifically in a mis-
match containing reaction. The specificity of such an
effect was evident when the relative change in R
h
(het-
ero minus homo) was plotted as a fraction of maxi-
mum difference observed in R

h
between hetero and
homoduplex sample at the final time-point (40 min) of
the reaction (Fig. 3D). As expected, in the normal con-
trol experiment where neither EDTA nor ATPcS was
added, the relative R
h
difference (i.e. R
h
heteroduplex–
R
h
homoduplex) kept on increasing as a function of
reaction time, thereby corroborating the specificity of
mismatch induced MutS aggregation described earlier
(Fig. 3C). Interestingly such a differential increase in
R
h
in hetero- vs. homoduplex was lost when ATP
hydrolysis was poisoned by either EDTA or ATPcS.
This was evident when R
h
associated with hetero-
duplex set decreased to background level close to that
N. Nag et al. Hydrodynamic analyses of MutS aggregates
FEBS Journal 272 (2005) 6228–6243 ª 2005 FEBS 6233
of homoduplex reaction (Fig. 3D), thereby revealing
the critical requirement of ongoing ATP hydrolysis
for sustained maintenance of mismatch specific MutS
aggregation (see Discussion).

The role of MutL
We tested further whether such high ATP induced mis-
match specific aggregation of MutS ensues even in the
Fig. 3. ATP induced aggregation of MutS in presence of hetero ⁄ homo- duplex DNA. MutS-DNA complexes were formed by incubating
0.15 l
M of MutS with either heteroduplex (n) or homoduplex (s) DNA (0.15 lM each) for 10 min at 22 °C in buffer containing 50 mM Hepes
pH 7.5, 50 m
M KCl, 5 mM MgCl
2
, followed by adding ATP at various final concentrations [1 mM (A), 10 mM (B)] and analysing the complexes
by DLS as a function of incubation time. High Mg
2+
control of the same was done by forming MutS-DNA complexes with 0.15 lM of MutS
and eitherheteroduplex (n ) or homoduplex (s) DNA (0.15 l
M each) for 10 min at 22 °C in buffer containing 50 mM Hepes pH 7.5, 50 mM
KCl, 15 mM MgCl
2
, followed by adding 10 mM ATP (C) and analysing the complexes by DLS as a function of incubation time. (D) MutS-DNA
complexes (homo- or heteroduplex containing) were formed as described (Fig. 3B) to which either ATPcS or EDTA (10 m
M each) was added
at the third or 20th minute of the reaction time-course (arrows), followed by R
h
measurement as a function of incubation time. The R
h
differ-
ences between hetero and homoduplex-containing reactions reached a maximum at the 40th min with respect to which those at other time-
points [(nR
h
at x
th

min) ⁄ (nR
h
at 40th min); nR
h
¼ R
h(het)
–R
h(homo)
] are expressed as a function of time. Decrease in nR
h
observed following
the addition of ATPcS (open triangles) or EDTA (open circles) was similarly expressed as a function of time.
Hydrodynamic analyses of MutS aggregates N. Nag et al.
6234 FEBS Journal 272 (2005) 6228–6243 ª 2005 FEBS
presence of MutL. DNA (0.15 lm) was added to
MutS–MutL complex (0.15 lm each) to facilitate a
1 : 1 complex, followed by the addition of ATP and
measurement of R
h
as a function of time. Addition of
1mm ATP caused marginal increase in the R
h
value of
complexes where heteroduplex and homoduplex DNA
samples showed a plateau at about 45 nm and 35 nm
particles, respectively (Fig. 4A). The marginal increase
in the R
h
value of protein–DNA complexes in the
presence of MutL at 1 mm ATP was qualitatively sim-

ilar to that of minus MutL set (Fig. 3A). The same
experiment in the presence of MutL at high ATP
(10 mm) revealed a dramatic enhancement in the
aggregation of protein that was highly specific to the
presence of a mismatch. The reaction containing
homoduplex DNA exhibited a slow rise in R
h
reaching
a limit of < 200 nm, whereas that of heteroduplex
DNA revealed rapid growth in protein aggregation
that appeared to go beyond an R
h
value of 500 nm
within 15 min (Fig. 4B). Again, the effect was clearly
not due to Mg
2+
limiting (5 mm) conditions, as a
repeat experiment at high Mg
2+
(15 mm) resulted in
the same effect (Fig. 4C), where a single mismatch pro-
voked higher aggregation of MutS in the presence of
high ATP. These experiments suggested a surprising
property of MutS where large protein aggregates form
in a mismatch specific manner, selectively under high
ATP (10 mm) conditions. It should be stressed that the
observation of particles with such large R
h
values and
the dramatic discrimination in the size of complexes in

hetero vs. homoduplex DNA in the presence of high
(% 10 mm) concentrations of ATP was very robust
and reproducibly seen in a large number of repeat
experiments.
Effect of ADP and salt
It is to be noted that the discrimination rendered by
the presence of a single mismatch in the DNA on the
Fig. 4. ATP induced aggregation of MutS-MutL in presence of
homo ⁄ heteroduplex DNA. MutS-MutL-DNA complexes were formed
by incubating of MutS-MutL (preincubated for 5 min by mixing both
at 0.15 l
M each) with either heteroduplex (n) or homoduplex (s)
DNA (0.15 l
M each) for 10 min at 22 °C in buffer containing 50 mM
Hepes pH 7.5, 50 mM KCl, 5 mM MgCl
2
, followed by adding ATP at
various final concentrations [1 m
M (A), 10 mM (B)] and analysing the
complexes by DLS as a function of incubation time.MutS-MutL-DNA
complexes were formed by incubating of MutS-MutL (preincubated
for 5 min by mixing both at 0.15 l
M each) with either heteroduplex
(n) or homoduplex (s) DNA (0.15 l
M each) for 10 min at 22 °Cin
buffer containing 50 m
M Hepes pH 7.5, 50 mM KCl, 15 mM MgCl
2
,
followed by adding 10 m

M ATP (C) and analysing the complexes by
DLS as a function of incubation time.
N. Nag et al. Hydrodynamic analyses of MutS aggregates
FEBS Journal 272 (2005) 6228–6243 ª 2005 FEBS 6235
level of MutS aggregation was lost when we substi-
tuted high ATP with high ADP (10 mm) (data not
shown). In fact at high ADP, the changes in R
h
as a
function of time in homo- vs. heteroduplex DNA
reached about 100 nm, with essentially no difference
between the two sets, again reiterating the specific role
of ATP and its hydrolysis in MutS aggregations (see
Discussion).
We tested the effect of salt (150 mm KCl) on the
formation as well as stability of mismatch induced
MutS aggregation. Normal MutS–DNA reaction con-
tains 50 mm KCl (see Experimental procedures) to
which an additional 100 mm KCl was added either at
the start or at the 20-min time-point of the reaction.
Interestingly, addition of salt at the start of the reac-
tion essentially abrogated mismatch induced dis-
crimination of MutS aggregation, where hetero- as
well as homoduplex reactions showed similar level
of increase in R
h
as a function of time (data not
shown). On the other hand, the same level of salt
added following mismatch induced aggregate forma-
tion (at 20 min) had barely any effect: Higher R

h
attained by hetero- as compared to the homoduplex
reaction was stable even in the presence of high salt
(data not shown). This experiment suggests that the
molecular interaction properties between MutS–MutS
and MutS–DNA that govern the formation vs. the sus-
tenance of mismatch induced MutS aggregation are
significantly different.
AFM imaging of the same samples suggested that
DLS results was not due to aggregation of just a small
subpopulation of MutS protein in the sample, but
rather reflected the entire protein population gener-
ating large particles of about 200–300 nm size in the
presence of mismatch as compared to smaller sized
particles (of about 100 nm) with normal homoduplex
DNA (data not shown) (DLS being more sensitive to
larger particles can mask the presence of smaller parti-
cles even in situations where, in mixtures of both large
and small particles, the major population is smaller in
size). AFM imaging of free protein, in the absence of
ATP, revealed particle distribution consistent with
dimeric ⁄ tetrameric forms [26], while the same in ATP
(1 mm) resulted in massive particles with a concomit-
ant loss of dimeric ⁄ tetrameric forms (data not shown),
reiterating that DLS results stemmed from uniformly
large-scale aggregation of MutS. Moreover, due to
large-scale aggregation of protein and the relatively
short duplex (121 bp) used in the system, it was not
possible to relate the status of aggregation in terms of
the position of mismatch in the duplex. AFM image

analyses followed by computation of attendant volume
changes in MutS particles as a function of ATP
concentration and DNA length is a separate study that
is currently underway.
Mismatch-dependent MutS aggregation as
revealed by centrifugation assays
In the following sedimentation assays, we monitored
MutS aggregation states in a variety of conditions, des-
cribed earlier, and tried to establish the general validity
of DLS results. MutS protein incubated with increas-
ing concentrations of either ATP or ATPcS was centri-
fuged followed by assaying the protein concentrations
in the supernatant as well as the pellet. In the set con-
taining ATPcS, the entire protein sample was recov-
ered in the supernatant (Fig. 5A) and no protein was
detected in the pellet fractions (data not shown). In
the same conditions the ATP set exhibited nucleotide
cofactor concentration dependent aggregation of MutS
where at about 3 mm ATP a significant fraction of
MutS was recovered in the pellet fraction with a
A
1.2
1.0
0.8
0.6
0.4
0.2
0.0
02468
10 12

1
2
3
4
56 7
ATP
Conc. of ATP/ATPγS in mM
Fraction of Muts in the supernatant
B
Fig. 5. Effect of nucleotide cofactor (ATP or ATPcS) concentration
on MutS aggregation as assessed by Centrifugation assay. MutS
(0.5 l
M) protein was incubated in buffer A at 25 °C for 10 min in
the presence of varying concentrations of ATP (s)orATPcS(n),
followed by centrifugation assay (see Experimental procedures) to
analyse MutS concentration in the supernatant fractions by Brad-
ford Dye binding. The fraction of total MutS recovered in the
supernatant fractions is plotted as a function of ATP ⁄ ATPcS
concentrations (A). Analyses of all the pellet fractions for MutS on
10% SDS ⁄ PAGE (B) (lane 1 corresponds to the pellet-equivalent
recovered from minus ATP control without centrifugation step;
lanes 2–7 correspond to pellet fractions of 0, 1, 3, 5, 7, 10 m
M ATP
containing samples, respectively, following centrifugation assay).
Hydrodynamic analyses of MutS aggregates N. Nag et al.
6236 FEBS Journal 272 (2005) 6228–6243 ª 2005 FEBS
concomitant reduction of the same in the supernatant
(Fig. 5A and B). At higher ATP concentration, MutS
aggregation was so severe that essentially all the pro-
tein was converted into a sedimentable fraction. This

experiment demonstrated that MutS aggregation is
highly ATP concentration dependent and corroborated
the DLS results described earlier in this study
(Fig. 2A). In order to verify whether ATP induced
MutS aggregation encompasses the bound DNA in the
complexes, we repeated the centrifugation assay on
MutS-labelled DNA duplex samples. In this experi-
ment, we included MutL along with MutS (0.4 lm
each) and incubated with an equimolar concentration
of 5¢-
32
P-labelled 121-mer hetero ⁄ homoduplex DNA at
increasing concentrations of ATP, followed by a cen-
trifugation assay. The pellet samples recovered in this
assay were treated with EDTA-SDS followed by analy-
sis in a native gel and the recovered labelled DNA was
imaged on a PhosphorImager. The result showed that
hetero and homoduplex DNA was rendered sedimenta-
ble by MutS aggregation in an ATP dependent
manner. In this assay the samples without or a low
amount of ATP showed hardly any sedimentable
DNA while at a concentration of ATP higher than
A
B
DC
Fig. 6. Effect of ATP concentration on aggregation of MutS-MutL-DNA complexes as assessed by centrifugation assay. MutS-MutL-DNA
complexes were formed by incubating of MutS-MutL (preincubated for 5 min by mixing both at 0.4 l
M each) with either heteroduplex or
homoduplex DNA (0.4 l
M each) for 10 min at 25 ° C in buffer A, followed by adding ATP at various final concentrations, incubating for

another 10 min and analysing the complexes by Centrifugation assay. One set of the experiment contained radiolabelled duplex DNA where
the common CLL strand (see Experimental procedures, Table 2) carried
32
P at its 5
¢
-end and the other set the same DNA in unlabelled form.
Pellet fractions from the first set were denatured with 20 m
M EDTA, 1% SDS, analysed on 8% native PAGE, followed by PhosphorImager
scanning of the dried gel. [(A) Heteroduplex DNA (B) Homoduplex DNA; lane 1, labeled CLL strand; lane 2, input duplex label; lane 3, pellet-
equivalent recovered from minus ATP control without centrifugation step; lanes 4–9, pellet fractions of 0, 1, 3, 5, 7, 10 m
M ATP containing
samples, respectively, following centrifugation assay]. Pellet fractions from the second set (containing unlabeled duplex DNA) were heat
denatured with SDS loading buffer, analysed by 10% SDS ⁄ PAGE, followed by silver staining to visualize both proteins and DNA. [(C) Hetero-
duplex DNA (D) homoduplex DNA; lane 1, pellet-equivalent recovered from minus ATP control without centrifugation step; lanes 2–7, pellet
fractions of 0, 1, 3, 5, 7, 10 m
M ATP containing samples, respectively, following centrifugation assay].
N. Nag et al. Hydrodynamic analyses of MutS aggregates
FEBS Journal 272 (2005) 6228–6243 ª 2005 FEBS 6237
5mm essentially all the DNA became sedimentable
(Fig. 6A,B) (PhosphorImager quantitative data not
shown). This result mirrored the sedimentation prop-
erty of the free protein observed in the earlier assay
suggesting that the DNA sedimentation accompanied
that of protein. In order to verify whether all of the
components of the complex ) namely MutS, MutL
and DNA ) are rendered sedimentable by high ATP,
we repeated the same experiment and analysed protein
as well as DNA simultaneously in the same gel using
the silver staining protocol. As shown earlier
(Fig. 6A,B), the silver stained gel in this experiment

revealed sedimentable DNA at a concentration of ATP
higher than 5 mm (Fig. 6C,D). Both hetero- and
homoduplex DNA were almost equally sedimentable.
In the same assay one could observe cosedimentation
of both MutS and MutL protein at high ATP. Silver
staining, being more sensitive than Coomassie blue
stainng, revealed some background retention of MutS-
L-DNA on the tubes even in the absence of centrifuga-
tion (lane 1, Fig. 6C,D). Samples containing high ATP
showed a signal for all these three components that
were significantly higher than the background. Taken
together, all of these centrifugation experiments dem-
onstrated high ATP induced aggregation of MutS–
DNA complexes that are highly sedimentable. It is
important to point out that MutS–DNA complexes
obtained with homo- vs. heteroduplex DNA targets
exhibited similar sedimentation properties in the cen-
trifugation assay, although DLS analyses revealed
larger complexes with heteroduplex DNA (Fig. 4B),
suggesting that centrifugation assay fails to discrimin-
ate the size differences associated with MutS–DNA
complexes, but quantitatively scores essentially all
complexes.
Discussion
This study involves the analysis of the changes associ-
ated with MutS aggregation in response to ATP bind-
ing ⁄ hydrolysis and its mismatch recognition in duplex
DNA. The study aims primarily to understand large
aggregational changes associated with MutS to help
model how the protein might transduce the informa-

tion of a single mismatch across a long physical dis-
tance in the duplex DNA. In contrast with many
biochemical techniques such as foot-printing and ana-
lytical centrifugation used by others, the DLS analyses
presented here offers an equilibrium study of the com-
plex changes brought about by MutS ATP hydrolysis
and mismatch binding. Therefore the current study
addresses the dynamic changes of the system more
comprehensively.
MutS and MutL interact with each other
In the first part of this report we have shown that the
system is highly amenable for studies by DLS where
the distribution of R
h
(hydrodynamic radius in nm), as
a function of added components, revealed signatures
of bonafide protein–protein interactions. MutS protein
showed hydrodynamic radius (R
h
) of 10 nm, which is
comparable to the MutS-dimer described in the crystal
structure 125 · 90 · 70 A
˚
3
dimension [14,15]. Addition
of MutL (R
h
4 nm) to MutS (R
h
10 nm) at an equimo-

lar ratio yielded a particle with a significantly higher
R
h
value (25 nm) (Fig. 1A). Indeed the complexation
of MutL with MutS did lead to 1 : 1 stoichiometric
complexes under these conditions, and was established
by an independent experiment involving fluorescence
titration (Fig. 1B). It appears that the MutS–MutL
complex with an R
h
value of 25 nm does reflect a
particle that is somewhat larger than a simple 1 : 1
complex of protein dimers [27,28].
Binding of MutS–MutL to DNA: sensing
of mismatch
Similarly MutS binding to homoduplex DNA led to an
increase in R
h
(DR
h
4 nm) that was smaller than the
increase observed with the heteroduplex DNA of the
same size (DR
h
10 nm) (Table 1). Such an enhanced
increase in the R
h
with heteroduplex DNA is highly
consistent with the conversion of dimeric MutS to that
of tetramer either during or following mismatch recog-

nition [26]. Interestingly, time course analysis of R
h
changes following DNA addition seems to suggest
that MutS interaction with heteroduplex DNA rapidly
generates a particle size similar to that of homoduplex
DNA (14 nm) following which the particle size increa-
ses further. This again is consistent with the model
where initial binding of MutS to the heteroduplex will
be equivalent to that of homoduplex, following which
mismatch recognition leads to tetramerization of the
protein in the heteroduplex reaction. Moreover the dif-
ference in R
h
following hetero vs. homoduplex binding
by MutS was observed even in the presence of MutL
where the R
h
value suggested the involvement of
MutS–MutL complex in the recognition of the mis-
match, which is consistent with a large body of
published literature on MutS–MutL system [24,27,28].
Amplification of single mismatch by MutS–MutL
The most important finding of the study relates to the
massive aggregation of MutS in the presence of ATP
where protein molecules with an initial R
h
of about
Hydrodynamic analyses of MutS aggregates N. Nag et al.
6238 FEBS Journal 272 (2005) 6228–6243 ª 2005 FEBS
10 nm are converted into large particles (with an R

h
of
several hundred nanometers) within less than 5 min
following ATP addition (Fig. 2A). Surprisingly, the
extent of aggregation was highly ATP concentration
dependent in a range far above the micromolar bind-
ing affinity reported for ATP binding with MutS
[29,30], possibly reflecting the role of additional puta-
tive low affinity ATP binding sites in this system.
Expectedly, the ATP mediated protein aggregation was
inhibited by the poisoning action of ATPcS, implying
the role of ATP hydrolysis in MutS aggregation
(Fig. 2B). Moreover, protein aggregation induced by
ADP was significantly lower than that of ATP, reveal-
ing the specificity of the same with ATP. This observa-
tion is highly consistent with earlier report where
gel-filtration analysis revealed that higher order oligo-
merization of MutS was favoured specifically by ATP
hydrolysis [31]. Presence of homo- as well as hetero-
duplex DNA significantly reduced ATP induced aggre-
gation of protein, suggesting the possibility that the
binding of protein to DNA somehow interferes with
the polymerization of free protein (Fig. 3A,B). How-
ever, most intriguingly, protein aggregation reappeared
even in the presence of DNA at a high concentration
of ATP (Fig. 3B). In this high ATP regime, protein
aggregation was not related to limiting Mg
2+
, as the
same was observed even at high concentration of

Mg
2+
(Fig. 3B vs. Fig. 3C). Interestingly, formation as
well as the sustenance of mismatch induced aggrega-
tion of MutS critically requires the presence of ongo-
ing ATP hydrolysis: midway poisoning of the same by
either EDTA (that chelates Mg
2+
cofactor) or ATPcS
(that competitively blocks hydrolysable form of ATP)
attenuates mismatch specific MutS aggregation
(Fig. 3D), suggesting that the process is likely to be
dynamic. However, under the conditions of the current
in vitro study, the aggregation process itself appears to
be rather slow as revealed by several minutes of incu-
bation required before the high R
h
particles are evident
(Fig. 3C,D). Most likely this is due to the prevalent
reaction conditions in vitro and may not reflect the
physiological setting. The effect of a single mismatch
in the DNA provoking a distinctly higher level of
MutS aggregation was further accentuated in the pres-
ence of MutL where high ATP regime resulted in the
generation of massive particles with an R
h
of several
hundred nanometers (Fig. 4B,C). The effect was highly
specific to ATP as the discrimination of a single mis-
match was lost when ATP was replaced by equal con-

centration of ADP, where similar R
h
values (100 nm)
were recovered for homo- vs. heteroduplex samples
with MutS. It is surmised that the large MutS aggre-
gates that are mismatch specific in ATP should encom-
pass the mismatch DNA itself as an intrinsic part of
the particle. This indeed is so, was borne out by the
sedimentation analyses of the complexes. Centrifuga-
tion experiments showed that these particles were
highly sedimentable only at high ATP (Fig. 6A,B) and
such sedimenting complexes encompass not only the
protein but also the DNA (Fig. 6C,D). Experiments
performed with ADP (1 mm and 10 m m) revealed a
much lower extent of MutS aggregation where the dis-
crimination rendered by the mismatch was lost, as
revealed by similar R
h
values (100 nm) in homo- vs.
heteroduplex DNA samples. We conjecture that the
propensity of MutS to undergo ATP-induced aggrega-
tion even in the absence of DNA might form the basis
of such massive MutS–MutL-heteroduplex DNA com-
plexes. The hydrodynamic size of such particles con-
taining protein–DNA complexes seem to suggest that
they may in fact represent assemblies that are connec-
ted by intermolecular interactions encompassing sev-
eral rather than single DNA molecules. However, the
techniques used do not allow us to distinguish the
same. In the current study we have uncovered a novel

facet of the MutS–DNA interaction system that facili-
tates protein aggregation following mismatch recogni-
tion in the DNA, but surprisingly such an aggregation
requires a high ATP concentration. The effect is
clearly ATP hydrolysis mediated since ATPcS can
effectively abrogate the same. We are currently trying
to understand the importance of high ATP in such a
phenomenon that might implicate the existence of
putative low affinity ATP binding sites in the system.
The phenomenon of MutS aggregation in relation to
DNA binding has been well described in the literature.
For example, a very early study by Su and Modrich
[32] showed that the mismatch bound form of MutS
was highly oligomeric in nature. Subsequent footprint-
ing studies further corroborated extensive coverage of
heteroduplexes by MutS protein in ATP specific condi-
tions [20,24]. More recent AFM analysis by Hall et al.
[33] of Mlh1–Pms1 heterodimers from Saccharomyces
cerevisiae bound to duplex targets showed the high
propensity of the mismatch repair proteins to exten-
sively coat the DNA by cooperative binding, some-
times interacting simultaneously with different DNA
targets thereby generating large intermolecular com-
plexes of the type described in the current study.
Mismatch induced aggregation of MutS
in relation to current models of MMR
All of these independent solution assays point to a
novel property of the system where MutS ⁄ MutL
exhibits a propensity to aggregate into large particles
N. Nag et al. Hydrodynamic analyses of MutS aggregates

FEBS Journal 272 (2005) 6228–6243 ª 2005 FEBS 6239
selectively in the presence of a single mismatch and
high ATP. It is highly likely that such an intrinsic
property of MutS to recognize and amplify mismatch
signals could indeed be used by the cells in the context
of MMR. Current models of MMR are rather sketchy:
it is unclear how the signal of a mismatch transduces
to its adjacent hemimethylated GATC tract over a
long distance. The prevailing competing models based
on a large body of biochemical data from E. coli as
well as eukaryotic MutS ⁄ MutL proteins invoke either
ATP-hydrolysis dependent translocation of MutS ⁄
MutL [17,18] or ATP hydrolysis independent passive
sliding clamp of MutS ⁄ MutL [9,19,22] or alternatively,
MutS ⁄ MutL complex stationed at mismatch cross-talk-
ing through space with the GATC tract sites [20,21]. A
variation of the first model proposed by others and us
invokes a near complete coverage of mismatch con-
taining DNA by MutS through a ‘treadmilling action’
of the protein that is highly ATP hydrolysis dependent
[23,24]. The current study, demonstrating a mismatch
specific highly aggregated state of MutS encompassing
bound heteroduplex DNA, is strongly consistent with
this model, where the presence of a mismatch can be
relayed across large distances thereby cross-talking
with GATC-specific excision steps. In addition, the
massively aggregated MutS-heteroduplex complexes
might reflect the propensity of the system that finally
culminate into MutS foci formation at the sites of
DNA mismatch repair in the cells [25]. We conjecture

that a requirement of high ATP for such massive
aggregation of MutS in vitro might reflect how this
property of MutS might elegantly couple MMR to the
regions of the replication fork, when a high local
concentration of dNTPs force a\high level of base-
mismatch incorporations by DNA polymerase, especi-
ally when the exonucleolytic proof-reading function of
the polymerase is compromised [34,35]. Future studies
encompassing AFM imaging will address the mechan-
istic basis of how high ATP induces MutS aggregation
on DNA that is so specific of a single mismatch on a
long duplex tract, and map the long distance propaga-
tion of mismatch-specific MutS cues, an important
aspect of MMR pathway.
Experimental procedures
Materials
ATP was from Sigma-Aldrich (Munich, Germany). ATPcS
was from Roche Diagnostics (Penzberg, Germany), Brad-
ford reagent was from Bio-Rad (Hercules, CA), Fluoresc-
amine was from Molecular Probes (Eugene, OR).
Oligonucleotides were from DNA technology (Aarhus C,
Denmark). Sep-Pak C-18 cartridge was from Waters
Corporation (Milford, MA, USA), and 0.02 lm13mm
Anodisc filter was from Whatman International Ltd
(Maidstone, Middlesex, UK).
DNA substrate
All oligonucleotides used in this study were purified by elec-
trophoresis on a 10% denaturing polyacrylamide gel con-
taining 8 m urea. The full-length oligonucleotide was excised
from the gel and eluted into autoclaved buffer (10 mm

Tris ⁄ HCl pH 8.0, 1 mm EDTA) by diffusion, followed by
desalting through a Sep-pak C-18 cartridge [36]. Final con-
centration of purified DNA was determined by measuring
the absorbance of an aliquot at 260 nm. The concentrations
expressed pertain to that of molecules.
DNA substrates used in all assays were a single G.T-mis-
matched duplex (121 bp) (Heteroduplex) and its corres-
ponding G.C-matched duplex (Homoduplex), the names
and their corresponding sequences are given in Table 2.
Table 2. Names and lengths of the oligonucleotide sequences used for preparing either heteroduplexs (G.T mismatch at the centre) or the
corresponding homoduplexes (the position of mismatch and the corresponding normal match are highlighted by bold and underline).
Name Size (nt) Sequence
CLL 121 5¢-TCACCATAGGCATCAAGGAATCGCGAATCCGCCTCGTTCCGGCTAAGTAACATGGAGCAG
GTCGCG
ATTTCGACACAATTTATCAGGCGAGCACCAGATTCAGCAATTAAGCTCTAAGCC- 3¢
GTL 121 5¢-GGCTTAGAGCTTAATTGCTGAATCTGGTGCTCGCCTGATAAATTGTGTCGAAATCCGCGA
TCTGCTCC
ATGTTACTTAGCCGGAACGAGGCGGATTCGCGATTCCTTGATGCCTATGGTGA-3¢
GCL 121 5¢-GGCTTAGAGCTTAATTGCTGAATCTGGTGCTCGCCTGATAAATTGTGTCGAAATCCGCGA
CCTGCTCC
ATGTTACTTAGCCGGAACGAGGCGGATTCGCGATTCCTTGATGCCTATGGTGA-3¢
CLE 61 5¢-GCCTCGTTCCGGCTAAGTAACATGGAGCAG
GTCGCGGATTTCGACACAATTTATCAGGCGA-3¢
GTE 61 5¢-TCGCCTGATAAATTGTGTCGAAATCCGCGA
TCTGCTCCATGTTACTTAGCCGGAACGAGGC-3¢
GCE 61 5¢-TCGCCTGATAAATTGTGTCGAAATCCGCGA
CCTGCTCCATGTTACTTAGCCGGAACGAGGC-3¢
CLS 16 5¢-TAGGTACG
GTCCATGC-3¢
GTS 16 5¢-GCATGGA

TCGTACCTA-3¢
GCS 16 5¢-GCATGGA
CCGTACCTA-3¢
Hydrodynamic analyses of MutS aggregates N. Nag et al.
6240 FEBS Journal 272 (2005) 6228–6243 ª 2005 FEBS
Annealing between CLL and GTL strands forms a 121-
mer duplex with G.T mismatch at the 61st base from
5¢ end of CLL and the corresponding Watson–Crick
matched homoduplex is formed by annealing the CLL with
the GCL strand. Similarly, annealing between the and GTE
strands forms a 61-mer duplex with a G.T mismatch at the
31st base from the 5 ¢ end of CLE, whereas a corresponding
homoduplex is formed by annealing CLE with GCE.
Annealing between CLS and GTS strands forms a 16-mer
duplex with G.T mismatch at the ninth base from the
5¢ end of CLS and the corresponding homoduplex is
formed by annealing CLS and GCS.
Protein purification
The MutS clone was from L. Worth, NIEHS. The mutS
gene is in His-tag expression vector pQE30. The protocol
followed to purify MutS is as described [37]. The His-tag
was not cleaved from the protein, as it does not seem
to alter the biochemical properties of MutS [29]. The
MutL clone was obtained from M. Winkler, and the
His-tagged MutL was purified as described [37]. Protein
concentrations were measured using Bradford Reagent.
All protein concentrations expressed pertain to protein
dimers.
DNA labelling and annealing
Labelling of DNA with

32
Pat5¢ end was performed as des-
cribed [38]. Complementary strands (see Table 2) were
annealed at a 1 : 1 ratio (10 lm of each strand in a total
volume of 20 lL) in 20 mm Tris ⁄ HCl pH 7.5 and 10 mm
MgCl
2
by heating the sample for 5 min at 90 °C, followed
by slow cooling to room temperature. Analysis of an ali-
quot of annealed sample on native polyacrylamide gel
revealed that the annealed duplex was well resolved from
the single-stranded controls and annealing was achieved
with > 90% of efficiency.
MutS–DNA complex formation
MutS–DNA complexes are formed by adding duplex DNA
(homo- or heteroduplex) to MutS in buffer (50 mm Hepes
pH 7.5, 50 mm KCl, 5 mm MgCl
2
,1mm dithiothreitol),
followed by incubation of the sample for 10 min at room
temperature (% 22 °C). Similarly, MutS–MutL-DNA com-
plexes are formed by adding duplex DNA to a mixture
containing MutS and MutL. In experiments involving
nucleotide cofactors such as ATP or ATPcS, MutS-DNA
(or MutS–MutL-DNA) complexes were first formed, fol-
lowed by the addition of nucleotide cofactors. In experi-
ments involving only MutL, the protein was analysed in the
same buffer. Concentrations and time ⁄ temperature of incu-
bations are as specified.
Fluorescence labelling of MutS with

fluorescamine
Fluorescamine is nonfluorescent until its coupling to pri-
mary amines [39]. MutS was labelled with fluorescamine in
the molar ratio of 10 : 1 with respect to the protein in buf-
fer C (20 mm Na-phosphate pH 8.0, 50 mm KCl and 5 mm
MgCl
2
), incubated for 15 min in the dark at 4 °C. The
unreacted fluorophore reacts with water giving rise to non-
fluorescent product and thus eliminating the need for
removal [36]. Steady-state fluorescence measurements were
carried out using a SPEX fluorolog FL111 Fluorimeter in
T-format, with the excitation wavelength set at 380 nm
monitoring the changes in fluorescence intensity at 477 nm.
The binding isotherm was fitted to the following equation:
F
obs
¼ 1 þ
½MutL
2

0
ðF
max
À 1Þ
½MutL
2

0
þ K

d
ð1Þ
where F
obs
is fluorescence intensity, F
max
is fluorescence
intensity at the end of the reaction, [MutL
2
]
0
is total con-
centration of MutL dimer (Free and MutS-MutL complex)
present and K
d
is dissociation constant. During the titration
of fluorescamine-labelled MutS with MutL (Fig. 1B), it was
ensured that the observed decrease in fluorescence intensity
is not due to any bleaching effect by minimizing the expo-
sure time and by checking with control samples.
DLS: measurement of hydrodynamic radius
DLS experiments were performed at 22 °C on a DynaPro-
MS800 dynamic light scattering instrument (Protein
Solutions Inc., VA) with an inbuilt Laser at 820 nm, by mon-
itoring the scattered light at 90° with respect to irradiation
direction. Buffer solutions were filtered carefully through
20 nm filters (Whatman Anodisc 13) to remove dust parti-
cles. The particulate matter, if any, in the DNA and protein
samples was removed by centrifugation (13 800 g) in a table-
top Eppendorf Centrifuge at 4 °C for 10 min.

For particles undergoing simple Brownian motion, the
autocorrelation function G(s) associated with the scattered
light intensity is given as:
GðsÞ¼
< IðtÞ:Iðt þ sÞ >
< I >
2
¼ 1 þ expðÀDq
2
sÞð2Þ
where I(t) and I(t+s) are the intensity of scattered light at
any time t and t+s, respectively, with s being the delay
time. The angle bracket represents averaging over various
times t.<I> is the time-averaged scattered intensity. D,
the translational diffusion constant, is given by Stokes-
Einstein relationship:
D ¼
kT
6pgR
h
ð3Þ
where k is Boltzman constant, g is the viscosity and R
h
is the hydrodynamic radius of the diffusing particle at
N. Nag et al. Hydrodynamic analyses of MutS aggregates
FEBS Journal 272 (2005) 6228–6243 ª 2005 FEBS 6241
temperature T. q, the modulus of the wave vector is given
by:
q ¼
4pg

k
sinðh

2Þð4Þ
where g, k and h are the refractive index, wavelength of the
irradiation source and the scattering angle, respectively.
When the system is polydisperse in size, G(s) is given by:
GðsÞ¼1 þ
X
k
j¼1
C
j
expðÀD
j
q
2
sÞð5Þ
where D
j
is the translational diffusion constant of the j
th
species with hydrodynamic radius R
j
h
.C
j
gives the fractional
contribution of the j
th

species to the scattered intensity. The
observed autocorrelation curves (at least 10 collections each
collected for 10 s) were analysed either by ‘Regularization’
software or ‘DynaLS’ software provided by the manufac-
turer of the instrument to generate a distribution of R
h
.
The quality of each individual autocorrelation curve was
scrutinized by monomodel analysis before being subjected
to either ‘Regularization’ or DynaLS analysis of the entire
set (¼ 10) of curves. The goodness of the fit by DynaLS
software was evaluated by visual inspection of residual
distribution (random and featureless residual distribution
represents acceptable fit) and residual value (ResSTD
< 0.001). Synthetic beads of 6 nm diameter (provided by
Protein Solutions Inc., Charlottesville, VA, USA) and BSA
(3 nm) were used as standards. In these cases, the recovered
distributions were single narrow Gaussian distributions
with peak value of R
h
at 6.0 or 3.0 nm, respectively. In
these monodisperse samples, the width associated with dis-
tribution of R
h
values could be taken as limiting widths
controlled mainly by the S ⁄ N of measurement. Widths sig-
nificantly higher than these values could then be taken to
represent the actual polydispersity of the sample.
Furthermore, analysis of G(s) by Eqn (5) assumes the
particles as hard spheres. Hence the recovered R

h
value
could be interpreted as follows: the translational dynamics
of the particle being studied is similar to that of a hard
sphere of radius R
h
.
A typical DLS experiment involves the addition of reac-
tion buffer (50 lL; 50 mm Hepes pH 7.5, 50 mm KCl,
5mm MgCl
2
,1mm dithiothreitol or other buffers as speci-
fied in the respective figure legends)to the quartz cuvette,
followed by ascertaining that the buffer system is free of
particles as reflected by very low R
h
(0.1–0.2 nm) values
associated with it. A small aliquot (1–2 lL) of stock protein
sample (MutS ⁄ MutS–MutL), that is cleared of particles by
prior centrifugation (as described above), is added to the
buffer (the final concentration of protein dimer was
0.15 lm or as specified in the legend), followed by collec-
tion of light scattering autocorrelation curves to obtain dis-
tribution of R
h
. In experiments involving MutS interaction
with DNA, the change in protein R
h
was monitored, in real
time, following the addition of either hetero or homoduplex

DNA (1–2 lL) with either ATP or ATPcS (1–2 lL) (see
legends for details).
Centrifugation assay
Samples (typically 25 lL) containing either free MutS pro-
tein or MutS-DNA complexes (prepared as described in the
respective Legends) were centrifuged at 13 800 g in a table-
top Eppendorf Centrifuge at 4 °C for 20 min. A fixed
volume of sample retrieved from the supernatant and the
pellet retrieved after complete removal of the supernatant
were denatured in buffers containing SDS, followed by ana-
lyses of protein and DNA on either SDS ⁄ PAGE or native
PAGE (see legends for details).
References
1 Modrich P & Lahue R (1996) Mismatch repair in repli-
cation fidelity, genetic recombination and cancer biol-
ogy. Annu Rev Biochem 65, 101–133.
2 Harfe BD & Jinks-Robertson S (2000) DNA mismatch
repair and genetic instability. Annu Rev Genet 34, 359–399.
3 Minnick DT & Kunkel TA (1996) DNA synthesis error,
mutator and cancer. Cancer Survey 28, 3–20.
4 Kolodner R (1996) Biochemistry and genetics of eukary-
otic mismatch repair. Genes Dev 10, 1433–1442.
5 Umar A & Kunkel TA (1996) DNA replication fidelity,
genetic recombination and cancer cell. Eur Biochem 238,
297–307.
6 Jiricny J & Marra G (2003) DNA repair defects in
colon cancer. Curr Opin Genet Dev 13, 61–69.
7 Schofield MJ & Hsieh P (2003) DNA Mismatch Repair:
Molecular Mechanisms and Biological Function. Annu
Rev Microbiol 57, 579–608.

8 Su SS, Lahuae RS, Au KG & Modrich P (1988)
Mispair specificity of methyl directed DNA mismatch
correction in vitro. J Biol Chem 263, 6829–6835.
9 Gradia S, Acharya S & Fishel R (1997) The human mis-
match recognition complex hMSH2-hMSH6 functions
as a novel molecular switch. Cell 91, 995–1005.
10 Cooper DL, Lahue RS & Modrich P (1993) Methyl
directed mismatch repair is bidirectional. J Biol Chem
268, 11823–11829.
11 Buermeyer AB, Deschenes SM, Baker SM & Liskay
RM (1999) Mammalian DNA mismatch repair. Annu
Rev Genet 33, 533–564.
12 Yang W (2000) Structure and function of mismatch
repair proteins. Mutat Res 460, 245–256.
13 Hopfner KP & Tainer JA (2000) DNA mismatch repair:
the hands of a genome guardian. Struct Folding Design
8, R237–R241.
14 Lamers MH, Perrakis A, Enzlin JH, Winterwerp HHK,
Wind N & Sixma TK (2000) The crystal structure of
DNA mismatch repair protein MutS binding to G.T
mismatch. Nature 407, 711–717.
Hydrodynamic analyses of MutS aggregates N. Nag et al.
6242 FEBS Journal 272 (2005) 6228–6243 ª 2005 FEBS
15 Obmolova G, Ban C, Hsieh P & Yang W (2000) Crystal
structures of mismatch repair protein MutS and its
complex with a substrate DNA. Nature 407, 703–710.
16 Wang H, Yang Y, Schofield MJ, Du C, Fridman Y,
Lee SD, Larson ED, Drummond JT, Alani E, Hsieh P
& Erie D (2003) DNA bending and unbending by MutS
govern mismatch recognition and specificity. Proc Natl

Acad Sci USA 100, 14822–14827.
17 Allen DJ, Makhov A, Grilley M, Taylor J, Thresher R,
Modrich P & Griffith. JD (1997) MutS mediates hetero-
duplex loop formation by a translocation mechanism.
EMBO J 16, 4467–4476.
18 Blackwell LJ, Martik D, Bjornson KP, Bjornson ES &
Modrich P (1998) Nucleotide-promoted release of
hMutSalpha from heteroduplex DNA is consistent with
an ATP-dependent translocation mechanism. J Biol
Chem 273, 32055–32062.
19 Gradia S, Acharya S & Fishel R (2000) The role of mis-
matched nucleotides in activating the hMSH2-hMSH6
molecular switch. J Biol Chem 275, 3922–3930.
20 Schofield MJ, Nayak S, Du Scott TH, C & Hsieh P
(2001) Interaction of Escherichia coli MutS and MutL
at a DNA Mismatch. J Biol Chem 276, 28291–28299.
21 Junop MS, Obmolova G, Rausch K, Hsieh P & Yang
W (2001) Composite active site of an ABC ATPase:
MutS uses ATP to verify mismatch recognition and
authorize DNA repair. Mol Cell 7, 1–12.
22 Selmane T, Schofield MJ, Du Nayak S, C & Hsieh. P
(2003) Formation of a DNA mismatch repair complex
mediated by ATP. J Mol Biol 334, 949–965.
23 Blackwell LJ, Wang S & Modrich P (2001) DNA chain
length dependence of formation and dynamics of hMut-
Salpha.hMutLalpha.heteroduplex complex. J Biol Chem
276, 33233–33240.
24 Joshi A & Rao BJ (2002) ATP hydrolysis induces
expansion of MutS contacts on heteroduplex: a case for
MutS treadmilling? Biochemistry 41, 3654–3666.

25 Smith BT, Grossman AD & Walker GC (2001) Visuali-
zation of mismatch repair in bacterial cells. Mol Cell 8,
1197–1206.
26 Bjornson KP, Blackwell LJ, Harvey S, Batinger C,
Allen D & Modrich P (2003) Assembly and molecular
activities of MutS tetramer. J Biol Chem 278, 34667–
34673.
27 Wu H-T & Marinus MG (1999) Deletion mutation ana-
lysis of the mutS gene in Escherichia coli. J Biol Chem
274, 5948–5952.
28 Grilley M, Welsh KM, Su S-S & Modrich P (1989) Iso-
lation an Characterisation of Escherichia coli mutL.
J Biol Chem 264, 1000–1004.
29 Worth L Jr, Bader T, Yang J & Clark S (1998) Role of
MutS ATPase activity in MutS,1-dependent block of in
vitro strand transfer. J Biol Chem 273, 23176–23182.
30 Antony E & Hingorani MM (2003) Mismatch recogni-
tion-coupled stabilisation of Msh2-Msh6 in an ATP-
bound state at the initiation of DNA Repair. Biochemis-
try 42, 7682–7693.
31 Bjornson KP, Allen DJ & Modrich P (2000) Modula-
tion of MutS ATP Hydrolysis by DNA Cofactors.
Biochemistry 39, 3176–3183.
32 Su S-S & Modrich P (1986) Escherichia coli mutS-
encoded protein binds to mismatched DNA base pairs.
Proc Nati Acad Sci USA 83, 5057–5061.
33 Hall MC, Wang H, Erie DA & Kunkel TA (2001) High
affinity cooperative DNA binding by the yeast Mlh1-
Pms1 heterodimer. J Mol Biol 312, 637–647.
34 Bebenek K, Matsuda T, Masutani C, Hanaoka F &

Kunkel TA (2001) Proofreading of DNA polymerase
g-dependent replication errors. J Biol Chem 276, 2317–
2320.
35 Baker RO & Hall JD (1998) Impaired mismatch exten-
sion by a herpes simplex DNA. Polymerase mutant with
an editing nuclease defect. J Biol Chem 273 , 24075–
24082.
36 Sambrook J, Fritsch EF & Maniatis T (1989) Molecular
Cloning: A Laboratory Manual, Cold Spring Harbor
Laboratory Press, 11.39.
37 Feng G & Winkler M (1995) Single step purification of
Hi6
6
-MutH, Hi6
6
-MutL and Hi6
6
-MutS repair proteins
of Escherichia coli. Biotechniques 19, 956–965.
38 Joshi A, Sen S & Rao BJ (2000) ATP-hydrolysis-depen-
dent conformational switch modulates the stability of
MutS–mismatch complexes. Nucl Acids Res 28, 53–861.
39 Nakaya K, Yabuta M, Iinuma F, Kinoshita T &
Nakumara Y (1975) luorescent labelling of the surface
proteins of Erythrocyte membranes using cyclo-
heptaamylose-fluorescamine complex. Bioc Biop Res
Com 67, 760–766.
FEBS Journal 272 (2005) 6228–6243 ª 2005 FEBS 6243
N. Nag et al. Hydrodynamic analyses of MutS aggregates