Báo cáo khoa học: Refined solution structure and backbone dynamics of the archaeal MC1 protein ppt

Sulfolobus species Crenarchaea have small monomeric proteins with an SH3-like fold, such as Sac7d and Cren7 [3,4], whereas members of the Thermoplasma genus have the dimeric protein Keyw

archaeal MC1 protein

Franc¸oise Paquet, Karine Loth, Herve´ Meudal, Franc¸oise Culard, Daniel Genest and

Ge´rard Lancelot

Centre de Biophysique Mole´culaire, CNRS UPR 4301, Orle´ans, France

Introduction

DNA-binding proteins play a central role in all aspects

of genetic activity within an organism, such as

tran-scription, packaging, rearrangement, replication, and

repair Archaeons have a variety of abundant,

sequence-independent nucleoid proteins, some of which

are able to compact DNA Among the numerous

chro-matin proteins identified in archaeons, only two –

histones and Alba homologs – are present in almost all

archaeal phyla [1]

Archaeal histones (e.g HMfa and HMfb) are

char-acterized by an a-helical histone fold Their monomers

are not stable, and must form homodimers In the presence of DNA, dimers assemble into tetramers and, sometimes, hexamers [2] These archaeal histone tetra-mers wrap  90 bp in less than one circle, resulting in

a horseshoe-shape assembly Histones are replaced by other chromatin proteins in archaeons that lack them, namely the hyperthermophilic Crenarchaea and euryar-chaeal Thermoplasma Sulfolobus species (Crenarchaea) have small monomeric proteins with an SH3-like fold, such as Sac7d and Cren7 [3,4], whereas members of the Thermoplasma genus have the dimeric protein

Keywords

arm; bulges; DNA-binding protein; molecular

dynamics (MD) simulation; NMR relaxation

Correspondence

F Paquet, Centre de Biophysique

Mole´culaire, CNRS UPR 4301, Rue

Charles-Sadron, F-45071 Orle´ans Cedex 2,

France

Fax: +33 2 38631517

Tel: +33 2 38257692

E-mail: francoise.paquet@cnrs-orleans.fr

Database

Structural data are available in the Protein

Data Bank database under the accession

number 2KHL

(Received 26 July 2010, revised 15

September 2010, accepted 20 October

2010)

doi:10.1111/j.1742-4658.2010.07927.x

The 3D structure of methanogen chromosomal protein 1 (MC1), deter-mined with heteronuclear NMR methods, agrees with its function in terms

of the shape and nature of the binding surface, whereas the 3D structure determined with homonuclear NMR does not The structure features five loops, which show a large distribution in the ensemble of 3D structures Evidence for the fact that this distribution signifies internal mobility on the nanosecond time scale was provided by using15N-relaxation and molecular dynamics simulations Structural variations of the arm (11 residues) induced large shape anisotropy variations on the nanosecond time scale that ruled out the use of the model-free formalism to analyze the relaxation data The backbone dynamics analysis of MC1 was achieved by compari-son with 20 ns molecular dynamics trajectories Two b-bulges showed that hydrogen bond formation correlated with u and w dihedral angle transi-tions These jumps were observed on the nanosecond time scale, in agree-ment with a large decrease in 15N-NOE for Gly17 and Ile89 One water molecule bridging NH(Glu87) and CO(Val57) through hydrogen bonding contributed to these dynamics Nanosecond slow motions observed in loops LP3 (35–42) and LP5 (67–77) reflected the lack of stable hydrogen bonds, whereas the other loops, LP1 (10–14), LP2 (22–24), and LP4 (50–53), were stabilized by several hydrogen bonds Dynamics are often directly related to function Our data strongly suggest that residues belonging to the flexible regions of MC1 could be involved in the interac-tion with DNA

Abbreviations

CSP, chemical shift perturbation; MC1, methanogen chromosomal protein 1; MD, molecular dynamics; RDC, residual dipolar coupling.

HTa, a member of the HU family [5,6] Sac7d and

Cren7 have a small b-barrel with or without an

amphi-philic C-terminal a-helix, and HUs have a largely

a-helical body capped by b-sheets that extend into two

b-ribbon arms

Alba homologs, which were first identified in

differ-ent species of Sulfolobus (e.g Sac10b and Sso10b),

bind to both DNA and RNA The Alba dimer has

two extended b-hairpins flanking a central body,

sug-gesting three main points of contact with the DNA [7]

They are present in all archaeons except Halobacteria

and Methanomicrobia (Euryarchaea) For instance,

Methanosarcina sequenced genomes contain one gene

coding for a true archaeal histone, HMm, as well as

genes coding for structural proteins of the methanogen

chromosomal protein 1 (MC1) family [8]

In laboratory growth conditions, MC1 is the most

abundant structural protein present in Methanosarcina

thermophila CHTI55 [9] A large number of charged

residues (24 basic and 12 acidic amino acids) are

dis-tributed all along the protein sequence This small

pro-tein of 93 residues is able to bind and to bend any

dsDNA as a monomer Related to its capacity to

introduce strong DNA conformational changes, MC1

is able to discriminate between different deformations

of the DNA double helix Thus, MC1 recognizes and

strongly binds to four-way junctions [10] and to

mini-circles [11,12] In addition, MC1 is easily able to

recog-nize flexible DNA sequences [13] Visualization of the

linear DNA molecules by electron microscopy reveals

that the binding of MC1 induces sharp kinks with an

angle value of 116 [14] We have previously solved the

three-dimensional structure of this architectural protein

extracted from the M thermophila strain CHTI55 by

using 1H-NMR spectroscopy only [15] The overall

fold of MC1, characterized by its b–b–a–b–b–b

link-ing, is different from those of other known

DNA-bind-ing proteins Site-directed mutagenesis showed that

two residues belonging to the loop b4–b5 (Trp74 and

Met75) are involved in DNA binding [16]

Further-more, hydroxyl radical footprinting, together with a

dystamycin competition experiment, suggested that the

monomeric MC1 binds to DNA through the minor

groove, and that the binding site, covering at least

15 bp, is composed of two areas of contact separated

by nearly 10 bp [13] The static structure, previously

described, could not explain this particular behavior

[15] We therefore decided to continue the structural

study of MC1 with a recombinant13C,15N-labeled

pro-tein expressed in Escherichia coli In this article, we

report heteronuclear NMR experiments that have

enabled us to assign all side chains and to introduce

dihedral angle restraints (u and w angles) Residual

dipolar couplings (RDCs) were also measured in a par-tially aligned sample with radially compressed poly-acrylamide gel to add constraints, particularly in the arm In addition to the structural study, a qualitative analysis of the NMR relaxation and molecular dynam-ics (MD) simulation data was carried out

Results and Discussion

Refined NMR structures of MC1 Chemical shift assignments for MC1 were obtained for 97% of N, HN, Ha, Ca, Cb and C¢ nuclei (Table S1) The refined structures of MC1 were determined by using NOE distances, dihedral angles, hydrogen bonds, and 1DNH RDC restraints (Table 1) The global fold consists of a pseudobarrel with an extension of the b-sheet (b4–b5) forming an arm (Fig 1A,B) The sec-ondary structure elements, namely an a-helix, a1 (25– 32), and five b-strands, b1 (4–9), b2 (15–21), b3 (43– 48), b4 (55–65), and b5 (79–90), are all antiparallel and packed with each other as previously described [15]

An antiparallel b-bulge (B1), composed of Leu8, His16, and Gly17, is present in seven structures, and another antiparallel b-bulge (B2), composed of Val57, Glu87, and Arg88, is observed for all the structures The secondary structure elements are connected

by loops LP1 (10–14), LP2 (22–24), LP3 (35–42), LP4 (50–53), and LP5 (67–77), referred to as ‘arm’ in the text The latter now appears to be remote from the protein core, whereas it was previously described as pulled down on the a-helix Superimposition of the 15 best structures of MC1 clearly shows that the regions with the largest degree of structural variations include the N-terminus, C-terminus, and loops LP1, LP3, LP4, and, especially, LP5 (Fig 1A) Its rmsd value is large (11 A˚), in agreement with the extensive conformational space swept by its residues (Table 1), whereas the rmsd values of the other loops fall between 2 and 2.7 A˚ MC1 can no longer be considered as a spherical pro-tein, but rather as an anisotropic structure defined by the ratio of the principal components of the inertia tensor This ratio differs within the 15 models of MC1, 1.00 : (0.85–0.94) : (0.34–0.43), according to the posi-tion of the arm

Although this new fold is completely different from those of other known proteins, it has similarities to the small architectural proteins Sac7d and Cren7 belonging

to the Sulfolobus strains of the Crenarchaeota subdo-main (Fig 2) [3,4] All possess a triple-stranded b-sheet (b3–b4–b5) Sac7d and Cren7 cause a single-step sharp kink in DNA ( 60 and  53, respectively) through the intercalation of hydrophobic side chains Despite

their similarity in overall structure, these two SH3-like

proteins differ in the DNA-binding surface Cren7

shows a substantially larger binding site ( 8 bp) than

Sac7d (4 bp), as it possesses a long loop of seven

resi-dues between b3 and b4 in the DNA binding surface

[17] Loop b3–b4 of Cren7 undergoes a significant

con-formational change upon binding of the protein to

DNA, suggesting its critical role in the stabilization of

the proteinÆDNA complex The arm of MC1 can also

be compared with the DNA-binding b-hairpin arms of

HUs, which showed high mobility relative to the core

(Fig 2) The b-ribbon arms wrap around the minor

groove of the DNA and, at the tip of each arm, the

conserved Pro intercalates between base pairs, creating

and⁄ or stabilizing two kinks in the DNA (global

cur-vature between 105 and 140) [18] This variability is

reflected by extensive DNA contacts between 9 bp of

DNA and the b-ribbon arms, and variable contacts between additional DNA and the body of the protein [19] In the case of MC1ÆDNA complexes, we know that the protein covers at least 15 bp and that the binding site is composed of two areas of contact sepa-rated by nearly 10 bp [13] The arm (loop LP5) seems

to be essential to cover such a long sequence In fact, the arm of MC1 has many hydrophobic residues (Pro68, Pro72, Trp74, Met75, and Pro76), which are conserved in different species of Methanosarcina and Halobacteria Site-directed mutagenesis showed that two residues belonging to the loop (Trp74 and Met75) are involved in DNA binding [16] It is clear that the arm of MC1 is essential for DNA binding and bend-ing The interaction mode of MC1 is probably com-pletely different from those of Sac7d and Cren7, which bind and bend DNA by placing their triple-stranded b-sheet (b3–b4–b5) across the DNA minor groove Indeed, the electrostatic potential surface of MC1 reveals that one side of the protein has a considerable number of positively charged residues: Arg4, Lys22, Arg25, Lys53, Lys54, His56, Lys69, Arg71, Lys81, Lys85, Lys86, and Lys91 (Fig 1C) This side, the reverse of the one used by Sac7d and Cren7 of Sulfolo-bus, is a good candidate to interact with the phosphate group of nucleotides

15N-NMR Relaxation for MC1 The 15N-HSQC spectrum of MC1 recorded at

600 MHz showed good dispersion of the crosspeaks (Fig S1) Relaxation data were obtained for 84 back-bone N–H pairs (93 residues minus Pro24, Pro42, Pro68, Pro72, Pro76, Pro82, Gly51, and the two N-ter-minal residues Ser1 and Asn2) at 600 MHz (R1, R2,

15N-NOE) and 800 MHz (R1, R2), and, owing to spec-tral overlap, for 79 residues at 500 MHz (15N-NOE) The experimental relaxation data at 600 and 800 MHz are shown in Figs 3 and S2 respectively The patterns seen for the individual relaxation rate constants at the different field strengths are similar The average value

of R1 is 1.6 s)1 at 600 MHz and 1.1 s)1 at 800 MHz

R2values showed large deviations up to 60% from the mean value (11.5 s)1 at 600 MHz and 13.8 s)1 at

800 MHz) Such variations in R2 values can result from relatively large-amplitude motions, efficient exchange processes, or shape anisotropy effects In our experimental conditions, no significant increase in R2

values was observed for MC1 between 600 and

800 MHz, indicating the absence of efficient exchange processes We observed that R2 values decreased sub-stantially for Gly17, Asp66, Lys69, Asn70, Arg71, and Ile89, whereas R1 values increased, reflecting local

Table 1 NMR constraints and structural statistics.

NMR constraints

Distance restraints

Total dihedral angles

Structural statistics for the ensemble of the 15 lowest-energy

structures

Average violations per structure

Hydrogen bonds ‡ 0.5 A˚ 0

RDC constraints rmsd (Hz) 0.75

Average pairwise rmsd (A ˚ ) Backbone atoms Heavy atoms

a1, b1–5 (50 residues) 1.22 ± 0.26 1.85 ± 0.26

Average rmsd (A ˚ ) after fitting the secondary structure elements

(a1, b1–5) as in Fig 1

Ramachandran analysis

Most favored region (%) 79.4

Allowed region (%) 19.6

Generously allowed (%) 0.7

motions Large variations of 15N-NOE were observed

along the sequence at 600 MHz, particularly for Thr3,

Arg4, Gly17, Asp66–Ile79, Ile89, and Glu93, for which

15N-NOE < 0.65; these residues clearly possess

con-siderable internal motions on the nanosecond time

scale It is interesting to locate these residues in the

structure: they belong to bulges B1 (Gly17) and B2

(Ile89), loop LP5 (Asp66, Ala67, Lys69, Asn70, Arg71,

Ala73, Trp74, Met75, Glu77, Lys78, and Ile79), and the termini (Thr3, Arg4, and Glu93)

Although the structure of the MC1ÆDNA complex has not yet been solved, relaxation measurements on the complex have been conducted (Fig S3) Besides six Pro residues, resonance overlap precluded the interpre-tation of relaxation data for seven residues (Phe19, Arg25, Gly51, Asp66, Lys86, Ile89, and Glu90)

C

LP4

LP5

N LP2

β3

β5

180°

β4

180°

A

B

C

Fig 1 (A) Superimposition of the 15 low-est-energy structures fitted on the second-ary structure elements (B) Ribbon diagram

of the lowest-energy solution structure of MC1 (C) Solvent-accessible surface area of MC1 color-coded by surface charge (blue and red correspond to basic and acidic regions, respectively).

C

C

N

N

N

N

C

Fig 2 Structures of some archaeal chroma-tin proteins other than the histones and Alba homologs (A) Sac7d (Protein Data Bank ID code: 1AZP) and (B) Cren7 (Protein Data Bank ID code: 3LWI) specific to Sulfolobus (Crenarchaea) (C) MC1 (Protein Data Bank

ID code: 2KHL) specific to Methanosarcina (Euryarchaea) and (D) HU monomer (Protein Data Bank ID code: 1P71) specific to Thermoplasma (Euryarchaea).

Several residues belonging to arm LP5 (Ala67, Lys69,

Asn70, Ala73, Met75, and Glu77) exhibit an increase

in15N-NOE If we compare the sites that exhibit

back-bone chemical shift perturbations (CSPs) upon DNA

binding with those that exhibit an increase in NOE

upon DNA binding, we can conclude that the arm

becomes much less mobile after binding with DNA

(Fig 4) This is reminiscent of the structure and

dynamics of the highly mobile b-arms in the free

pro-tein HU, which become much less mobile after binding

with DNA In the model proposed by Tanaka, the

DNA-binding arms can move as rigid arms, creating

sufficient room for accepting DNA [20] The tips of

the arms are highly flexible, and once the DNA has

moved inwards, the arms close and the tips of the

arms wrap around the DNA

The amplitudes and time scales of the intramolecular

motions experienced by the protein backbone are

com-monly determined from the15N-NMR relaxation data,

by using the model-free approach suggested by Lipari

and Szabo [21,22] and extended by Clore et al [23]

This approach is applicable for the case of statistically

independent overall tumbling and internal motions In

the case of MC1, large-amplitude internal motions on

the same scale as global rotation are detected for at least 16 N–H vectors, ruling out use of the model-free formalism

MD analysis Consistent with experimental observations, the protein core was stable at 300 K during the MD simulation The average backbone rmsd calculated with the sec-ondary structure atoms of the 2000 snapshots was 2.2 ± 0.1 A˚ Such deviations are characteristic of pro-tein simulations carried out in the presence of solvent [24,25] The backbone rmsd calculated with all of the residue atoms starts at 5.8 A˚ and increases up to 15 A˚ during the 20 ns trajectory time, showing large motions of the loops and the arm (Fig S4)

Rotational diffusion Knowing the rotational diffusion tensor is essential for a detailed analysis of intramolecular motions in nonspherical proteins When the shape of a molecule changes over time, its associated rotational diffusion tensor varies The eigenvalues of the diffusion tensor

2.5

A

B

C

0.5 1.0 1.5 2.0

R1

R2

0.0

10 15 20

0 5

1

0 0.2 0.4 0.6 0.8

3 8 13 18 23 28 33 38 43 48 53 58 63 68 73 78 83 88 93

Sequence

Fig 3 Backbone 15 N-relaxation data for

1.6 m M free MC1 at 600 MHz (A)

Longitudi-nal relaxation rate (B) Transverse relaxation

rate (C) Heteronuclear NOE.

show variations of 25% on the nanosecond time scale

during the 20 ns trajectory (Fig 5) These variations,

resulting from the position of the arm, are correlated

with internal motions of the overall protein, as can be

observed for the Ca–Ca or HN–HN distances between

Arg71 and Leu92 (Fig 6) The length of the arm

Ala67–Glu77 showed high variation (10.4–16.8 A˚), as

indicated by the distance Ca–Cabetween Val65 at the

end of strand b4 and Arg71 at the extremity of

arm LP5 This stretch was made up of complex motions

in the arm, as shown by the variations in Ca–Ca dis-tances between Ala67–Arg71, Ala67–Lys78, Arg71– Val65, and Arg71–Ile79 The motion of the arm is cen-tered on a hinge composed of Ala67 and Glu77 More-over, loops LP1 and LP3 exhibited substantial conformational changes during the trajectory, as shown by variations in the Ca–Ca distances between Glu11–Asp43, Gly13–Leu92, and Gly35–Lys62 Dur-ing the trajectory, the location of loop LP1 changed in relation to strands b3 and b5, as indicated by varia-tions in the Glu11–Asp43 and Gly13–Leu92 distances

Internal correlation functions The internal autocorrelation functions are calculated within the molecular reference frame of the superposed structures Figure 7A shows the time-correlation func-tions for three representative residues in different parts

of MC1 The upper N–H vector (Gln26 in the helix) shows a rapid (< 10 ps) decay of C(t) from 1.0 to

 0.9, arising from vibrational motion This correla-tion funccorrela-tion is typical for residues in relatively rigid parts of MC1, such as the a-helix and the b-strands, excluding bulges The correlation functions for two residues, Val18 in a bulge and Asn70 in the arm, are also shown The fast decay of the Val18 N–H vector (Sf2= 0.85) is followed by a slow motion on a nano-second time scale with an order parameter, S2, of 0.55 The third C(t) of Asn70 is composed of three decays The fast decay (< 10 ps) is followed by an intermedi-ate decay (100–500 ps) that primarily arises from libra-tional motion This intermediate motion is common to residues belonging to loops LP1, LP3, LP4, and LP5,

180°

180°

A

B

Fig 4 (A) Residues that exhibit a significant increase in 15 N-NOE upon DNA binding are

in blue, and those with an intermediate increase are in marine (B) Residues that exhibit significant CSP upon DNA binding are in red, and those with intermediate changes are in orange.

1.9

1.3

1.5

1.7

Dx

7 s

Dy

7 s

Dz

7 s

Time (ns)

1.6

1.8

1.2

1.4

2.8

3.0

3.2

2.4

2.6

0 2 4 6 8 10 12 14 16 18 20

Time (ns)

0 2 4 6 8 10 12 14 16 18 20

Time (ns)

0 2 4 6 8 10 12 14 16 18 20

Fig 5 Fluctuations in the anisotropic rotational diffusion

eigen-values Dx, Dyand Dzalong the 20 ns trajectory.

in the last case with larger amplitude Finally, a slower

decay reaches a plateau value S2 of 0.10 after 8.3 ns

This time value is close to the average harmonic mean

correlation time of 8.6 ± 0.3 ns calculated using

hydronmrduring the trajectory

The MD-derived order parameters S2 values for the

b-strands and the a-helix are consistent with the

exper-imental relaxation data (Fig 7B) Slow motions were

detected for Ser1, Asn2, Leu92, Glu93 (terminal

resi-dues), Gly17 to Phe19 (bulge B1), Arg34 to Gly37

(loop LP3), Gly51 and Thr52 (loop LP4), Ala67 to

Lys78 (loop LP5), and Glu87 to Ile89 (bulge B2) The

largest amplitudes were observed for Ser36 (S2= 0.13)

and from Ala67 to Asn70 (0.1 < S2< 0.16) The

resi-dues involved in the two bulges have S2 values around

0.6, which is consistent with the15N-NOE values The

calculated S2 values in the loops are lower than

expected, particularly in loops LP3 and LP4 A recent

study provides evidence for a specific link between

force field deficiencies and disagreement between

experimental and MD order parameters [26] MD

sim-ulations using three MD force fields (comprising

amberff03) overestimate the flexibility of backbone

N)H vectors at the borders of secondary structure and

in loops Specific inaccuracies in the treatment of hydrogen bonding could be responsible for increased flexibility in silico In the case of MC1, the conforma-tional changes observed during the trajectory are consistent with the crosspeaks observed on the NOESY spectra Low values of S2 computed with the correlation functions indicated slow motions with large amplitude However, these values can only be obtained with large uncertainties, as a trajectory of 20 ns allowed us to calculate a correlation function only over

10 ns Achieving reliable correlation functions requires several repetitions of occurrences on the time scale of the trajectory An isolated occurrence generates waves

on the correlation function that have little significance,

as seen for Val18

Correlated motions on the nanosecond time scale

The trajectories of some dihedral angles and distances were examined in the two b-bulges and in the loops For bulge B1, a hydrogen bond was alternately present

12 14

16

13–92

14

6 8 10 12

8

10

35–62

10 12 14

9

11

71–67

4 6 8

12

5 7

12

6 8 10

9

11

67–78

8 10 12

48

52

71–92

3 5 7

40 44

Time (ns) Time (ns)

92 13 11 43

35 62 7978

65 67 71

92

13

11 43

35 62 79 78

65 67 71

A

B

Fig 6 (A) Some C a –C a distances (A ˚ )

along the 20 ns simulation trajectory.

(B) Snapshots of MC1 after the equilibrium

period at 0 and 17.5 ns Black lines indicate

distances between specific residues that

show shape variations during the trajectory.

between NH(Leu8) and CO(His16) for 9 ns, and

between NH(Leu8) and CO(Val17) for 9.5 ns

(Fig 8A,C) Thus, at least one of the two hydrogen

bonds was present for 18.5 ns on the 20 ns trajectory

time The transitions occurred at 1, 2, 10, 14, 15 and

19 ns, and are correlated with the motions of the

dihe-dral angles w(His16), u(Gly17), and w(Val18) This

flip-flop leads to slow internal motions with large

amplitude, as seen for C(t) of NH(Gly17) and

NH(Val18) (S2= 0.55) For bulge B2, a unique

hydrogen bond between NH(Val57) and CO(Arg88)

was present for 13 ns in periods of 2–4 ns (Fig 8B,C)

At the same time and for 7.5 ns, a

water-mediated hydrogen bond between NH(Val57) and

CO(Glu87) was formed when the distance was

 6.4 A˚ The two strands were thus completely

sepa-rated for 7 ns of the trajectory time, which could

explain the greater flexibility of this bulge and the slow

internal motions of Glu87, Arg88 and Ile89 with very

large amplitude These motions were correlated with

the dihedral angle transitions of w(Glu87), u(Arg88),

and u(Ile89) The presence of these hydrogen bonds

was consistent with the homonuclear NOEs found in

this region [15]

Loop LP1 was stabilized with two hydrogen bonds,

NH(Asp10)–CO(Asn14) and NH(Gly13)–CO(Asp10),

throughout the trajectory time Similarly, the hydrogen bond NH(Gln26)–CO(Gln23) stabilized the short loop LP2 for 19 ns

The lack of stable hydrogen bonds in loop LP3 (35– 42) corresponds with large motions of the N)H vectors for Gly35, Ser36, and Gly37 However, this nonstructured loop is probably not important in the DNA binding, because the number of residues between Gly35 and Ile45 (MC1-CHTI55 numbering) varies from 3 to 14 in different species of Halobacteria and Methanomicrobia[15]

Loop LP4 (50–53) was stabilized by two hydrogen bonds, NH(Thr52)–CO(Glu49) and NH(Leu92)– CO(Lys53), binding the loop to strand b5 for 11 ns Supplementary hydrogen bonds involving the side chains NH2(Arg48)–CO(Thr52) and OH(Thr52)– CO(Glu49) contribute to the stiffness of the structure for a short time Moreover, two NOE crosspeaks, OH(Thr52)–NH(Thr52) and OH(Thr52)–NH(Glu49), were observed on the free protein MC1 NOESY spec-tra, owing to a slower exchange process with water This could be explained by hydrogen bonds involving the hydroxyl proton of Thr52

In arm LP5 (67–77), Ala67, Lys69 and Asn70 have global S2 values of  0.1, whereas the other resi-dues have S2 values in the range 0.26–0.46 This

A

1.0

Q26

0.4

0.6

0.8

V18

0.0

0.2

Time (ns)

N70

0.6

0.8

1

0

0.2

0.4

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91

Sequence

Fig 7 (A) Three representative internal cor-relation functions computed on a trajectory

of 20 ns for Gln26 in the a-helix, Val18 in a bulge, and Asn70 in the arm (B) Residue profile of the MD-derived S2.

corresponds to a combination of slow motions of

large amplitude In accordance with this, only four

hydrogen bonds were observed: NH(Lys71)–CO

(Pro68), NH2(Lys71)–CO(Pro76), NH(Met75)–CO

(Pro72), and NH(Trp74)–CO(Pro72) for 6.5, 11, 16

and 0.3 ns respectively

Summary

In summary, the structure of MC1, consisting of a

pseudobarrel with an extension of the b-sheet (b4–b5)

forming an arm of 11 residues, has been refined The

global fold is now compatible with the biochemical

data and a DNA-binding site covering at least 15 bp

The structure features five loops that show a large

distribution in the ensemble of 3D structures Evidence for the fact that this distribution signifies internal mobility on the nanosecond time scale is provided by using 15N-relaxation and MD simulations These local conformational changes in MC1 could facilitate DNA binding, with two areas of contact separated by nearly

10 bp Moreover, the flexibility of MC1 builds up con-formations with large positively charged areas that are highly favorable for binding with the phosphate groups of nucleotides Some residues belonging to arm LP5 (Ala67, Lys69, Asn70, Ala73, and Met75) and to bulge B2 (Ile89) are involved in motions on the nanosecond time scale, and could be related to the interaction with DNA A study of a DNAÆMC1 com-plex is currently underway

200 300 400

H16

50 150

E87

B A

0 100 200 300

G17

50 150

250 R88

100 200 300

V18

50 100 150

200 I89

7.0

1.0 3.0 5.0

(Å) 5.0

2.0 4.0 6.0

5.0

NH57 - CO88

1.0 3.0

1.0 3.0

C

Fig 8 Dihedral angle and distance

transi-tions as a function of time of the two

bulges (A) Bulge Leu8, His16 and Gly17 (B)

Bulge Val57, Glu87 and Arg88 (C) Schemes

of the bulges The dotted lines and the time

characterized the presence of hydrogen

bonding during the trajectory.

Experimental procedures

Preparation of13C,15N-labeled MC1

The proteins were expressed in BL-21(DE3) cells

trans-formed with the pET24a–mc1 plasmid Protein doubly

iso-tope-enriched Celtone-rich medium (Martek Biosciences,

cells were first grown in LB medium, and then, at a

D600 nm of 0.7, they were collected and resuspended in

expression of the protein was performed for 2 h after

addition of 0.1 mm isopropyl thio-b-d-galactoside

Purifi-cation of the proteins was performed by SP-Sepharose

(GE Healthcare Europe GmbH, Orsay, France)

chroma-tography followed by Ultrogel AcA 54 chromachroma-tography

The concentration of protein was determined by

absorp-tion spectrophotometry, with a molecular absorbance

The NMR protein sample was prepared by concentrating

MC1 to 1.6 mm (100 mm acetate buffer, pH 5.1, 800 mm

obtained with different concentrations of MC1 in the same

buffer conditions Decreasing the protein concentration by

a factor of 20 (1.6 mm to 0.08 mm) showed no significant

The DNA oligonucleotides used for NMR were purchased

from Eurogentec (Lie`ge, Belgium) (OliGold oligonucleotides

quality) The single-stranded 15 bp oligodeoxynucleotides

were characterized by NMR and annealed at a 1 : 1 ratio

The MC1ÆDNA complex was prepared by slowly adding

the 7.5 mm DNA duplex solution (10 mm phosphate buffer,

1.6 mm protein solution (10 mm phosphate buffer, pH 6,

NMR spectroscopy and structure calculations

Two-dimensional and three-dimensional NMR experiments

spec-trometer at 299 K Spectra were processed with nmrpipe

[28], and analyzed with nmrview [29] Backbone and side

chain resonance assignments were obtained from the

stan-dard triple resonance experiments [30]

reference Interproton distances were derived from NOESY

datasets obtained at mixing times of 100, 150 and 200 ms

Backbone dihedral angle restraints were determined with

radially compressed 7% polyacrylamide gel (6.0–4.2 mm)

[32,33]

Structures were calculated with NOE distance, hydrogen bond, u and W angle and RDC constraints, using aria2 (version 2.2) [34] The aria2 protocol (cns 1.1) used simu-lated annealing with torsion angle and Cartesian space dynamics with the default parameters RDC restraints within the aria2 protocol were incorporated at the last

ana-lyzed with the module program [35] Fifteen structures from six independent runs were selected on the basis of total energies and restraint violation statistics, to represent the structure of MC1 in solution The electrostatic potential was calculated by using the pdb2pqr server (version 1.6) [36] and apbs software [37] The figures were prepared with

Determination and analysis of15N-relaxation parameters (R1, R2, and NOE) for MC1 NMR relaxation experiments were measured at 299 K on

a Varian 500 MHz (NOE), Varian INOVA 600 MHz

acquired with 32 scans per t1 point, with a recycle delay

relaxation delays of 10, 20, 30, 50, 70, 90, 150, 210 and

relaxation delays of 10, 20, 30, 50, 70, 90, 110 and 150 ms were used for data collection at 800 MHz The errors in

distri-butions of the measured volume V within the V ± DV range and by repeating the fit with this procedure 1000

600 MHz with a 3 s presaturation period and a 2 s relaxa-tion delay; the reference experiment had an equivalent 5 s

the volumes of a crosspeak in the spectra collected with and without proton saturation Both were acquired with

64 scans All experiments were run twice in the same

measured by using nmrview software [29] Uncertainties

in the volumes were measured from the duplicate spectra After obtaining volumes of crosspeaks and their errors, the above time series were fitted from a single exponential decay function

Relaxation experiments for the MC1ÆDNA complex were

relaxa-tion delays of 10, 100, 200, 300, 500, 800, 1000 and

130 and 150 ms at 299 K

bind-ing were analyzed with samplex [40]