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

In the absence of any nucleotide cofactor, free MutS protein [hydro-dynamic radius Rh of 10–12 nm] shows a small increment in size Rh 14 nm following the addition of homoduplex DNA 121 b

aggregation of MutS

Hydrodynamic analyses of the protein-DNA complexes

Nabanita Nag1, G Krishnamoorthy1and Basuthkar J Rao2

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: bjrao@tifr.res.in

(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 (Rh) of 10–12 nm] shows a small increment in size (Rh

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 (Rh>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 (Rh>500 nm in the presence of MutL) as com-pared to the normal homoduplex (Rh
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.

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 inserrecogni-tion–delerecogni-tion-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

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 (Rh) 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 Rh values with a peak at 10 nm and 4 nm,

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 Rh 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 Kd of 70 ± 20 nm Since the titration (Fig 1B) is close to a case of stoichiometric binding, the estimated value of Kdshould 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 Rhdistribution of MutS

as a function of its interaction with MutL 0.15 l M MutL (I), 0.15 l M

MutS (II) and a mixture of MutS and MutL (0.15 l M 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 (Rhin nm) of MutS and MutL in the presence of Homo- or Hetero duplex DNA of different lengths (All

in minus ATP conditions, see text).

MutS-Heteroduplex (121 bp) 20 (± 2)

MutS-Heteroduplex (61 bp) 18 (± 2)

MutS-Heteroduplex (16 bp) 10 (± 1) MutS-MutL-Homoduplex (121 bp) 30 (± 3) MutS-MutL-Heteroduplex (121 bp) 35 (± 3)

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 Rh 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 Rh (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 Rh, 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 Rhvalues 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 Rh 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 Rh 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 Rh

(
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 Rh

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 Rh 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 Rh values of MutS particles Moreover the extent of MutS aggregation was clearly ATP con-centration dependent At the lowest concon-centration of ATP (0.3 mm) tested, the Rhvalues 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 Rh 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 Rh 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 Rh value exceeded 500–600 nm We tested whether ADP also exhibits a similar effect on MutS aggregation by analysing changes in Rh as a function of time at two different concentrations of ADP (1 mm and 10 mm) The observed changes in Rh with ADP were signifi-cantly lower: at 1 mm and 10 mm ADP the Rh increased to and stabilized at
30 nm and
150 nm, respectively (data not shown) The aggregation was also not due to the pyrophosphate anion (PPi) effect in ATP as shown by the lack of increase in Rhwhen PPi (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 PPi 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 Rh 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

effect on the changes in Rh induced by 1 mm ATP (Fig 2B) where the Rh 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 Rh 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 exhibiincuba-ted no MutS aggre-gation throughout the incubation period of 30 min where a particle with an Rh 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 Rh 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 m M (n), 0.6 m M (h), 1 m M (s), 10 m M (,) 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 m M ATPc S +1 m M ATP (premixed) (,)

or 1 m M ATPc S +1 m M ATP (premixed) (s) or 1 m M 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 l M (h), 0.15 l M (n), 0.3 l M (e), 0.45 l M (s)], followed by DLS analyses as

a function of incubation time.

the particles exhibited an Rh 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 Rhof 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 Mg2+

depend-ence of ATP induced MutS aggregation In four

differ-ent samples that contained varying levels of Mg2+, Rh

value was monitored as a function of time following

ATP addition The sample that contained no Mg2+

showed the least protein aggregation reaching an Rhof

about 100 nm By the addition of 1 mm or more of

Mg2+, ATP induced MutS aggregation was

substan-tially increased generating particles of Rh that were

larger than 400–500 nm (data not shown) The

experi-ment suggested that the ATP induced aggregation was

highly Mg2+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 Rh 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 Rh 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 (Rh
140 nm) as com-pared with that with homoduplex DNA (Rh¼ 50–60 nm) (Fig 3B) The high ATP (10 mm) experi-ment was carried out under conditions in which the

Mg2+level (5 mm) appeared to be limiting To verify that the observed high ATP effect arose from the phys-iologically relevant Mg2+ bound form of ATP, and not from free ATP, we repeated the same experiment

at excess Mg2+ (15 mm) as well The DLS result at high Mg2+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 Rh

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 Rh 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 Rh (het-ero minus homo) was plotted as a fraction of maxi-mum difference observed in Rh 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 Rhdifference (i.e Rhheteroduplex–

Rh 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

Rh in hetero- vs homoduplex was lost when ATP hydrolysis was poisoned by either EDTA or ATPcS This was evident when Rh associated with hetero-duplex set decreased to background level close to that

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 l M each) for 10 min at 22
C in buffer containing 50 m M Hepes

pH 7.5, 50 m M KCl, 5 m M MgCl2, followed by adding ATP at various final concentrations [1 m M (A), 10 m M (B)] and analysing the complexes

by DLS as a function of incubation time High Mg2+control of the same was done by forming MutS-DNA complexes with 0.15 l M of MutS and eitherheteroduplex (n) or homoduplex (s) DNA (0.15 l M each) for 10 min at 22
C in buffer containing 50 m M Hepes pH 7.5, 50 m M

KCl, 15 m M MgCl2, followed by adding 10 m M 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 Rhmeasurement as a function of incubation time The Rh 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.

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 Rhas a function of time Addition of

1 mm ATP caused marginal increase in the Rhvalue 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 Rh 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 Rhreaching

a limit of < 200 nm, whereas that of heteroduplex

DNA revealed rapid growth in protein aggregation

that appeared to go beyond an Rh value of 500 nm

within 15 min (Fig 4B) Again, the effect was clearly

not due to Mg2+ limiting (5 mm) conditions, as a

repeat experiment at high Mg2+ (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 Rhvalues 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 m M

Hepes pH 7.5, 50 m M KCl, 5 m M MgCl2, followed by adding ATP at

various final concentrations [1 m M (A), 10 m M (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
C in

buffer containing 50 m M Hepes pH 7.5, 50 m M KCl, 15 m M MgCl2,

followed by adding 10 m M ATP (C) and analysing the complexes by

DLS as a function of incubation time.

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 Rh 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 Rh 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 Rh

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

ATP

Conc of ATP/ATP γS in mM

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) or ATPcS (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).

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¢-32P-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

D C

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) homoHetero-duplex 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].