THE M.ECORV DNA-(ADENINE N6)-METHYLTRANSFERASE

USES DNA BENDING FOR RECOGNITION OF AN EXPANDED

ECODAM RECOGNITION SITE Tomasz P. Jurkowski1, Nils Anspach2, Liliya Kulishova3,4, Wolfgang Nellen2 and

Albert Jeltsch1,*

From 1) Biochemistry Lab, School of Engineering and Science, Jacobs University Bremen,

Campus Ring 1, 28759 Bremen, Germany, 2) Abt. Genetik, CINSaT, Universität Kassel,

Heinrich-Plett-Str. 40, 34132 Kassel, Germany, 3) BioRec program, Jacobs University Bremen,

School of Engineering and Science, Campus Ring 1, 28759 Bremen, Germany 4) present address:

Institut für Molekulare Enzymtechnologie der Heinrich-Heine Universität Düsseldorf, FZ-Jülich,

52426 Jülich, Germany

Address correspondence to Albert Jeltsch, School of Engineering and Science, Jacobs University

Bremen, Campus Ring 1, 28759 Bremen, Germany, Phone: +49 421 200 3247, Fax: +49 421 200

3249, a.jeltsch@jacobs-university.de, http://www.jacobs-university.de/ses/ajeltsch/

The M.EcoRV DNA methyltransferase

recognizes GATATC sites. It is related to

EcoDam, which methylates GATC sites.

The DNA binding domain of M.EcoRV is

similar to that of EcoDam suggesting a

similar mechanism of DNA recognition.

We show that amino acid residue K11 of

M.EcoRV is involved in recognition of

Gua1 and R128 contacts the Gua in base

pair six. These residues correspond to K9

and R124 in EcoDam, which recognize

the Gua residues in both strands of the

Dam recognition sequence, indicating that

M.EcoRV and EcoDam make similar

contacts to outermost base pairs of their

recognition sequences and M.EcoRV

recognizes its target site as an expanded

GATC site. In contrast to EcoDam,

M.EcoRV considerably bends the DNA

(59±4°) suggesting indirect readout of the

AT rich inner sequence. Recognition of an

expanded target site by DNA bending is a

new principle for changing DNA

recognition specificity of proteins during

molecular evolution. R128A is inefficient

in DNA bending and binding whereas

K11A bends DNA with relaxed sequence

specificity. These results suggest a

temporal order of the formation of

protein-DNA contacts in which the Gua6-

R128 contact forms early followed by

DNA bending and, finally, the formation

of the K11-Gua1 contact.

1

http://www.jbc.org/cgi/doi/10.1074/jbc.M706933200The latest version is at JBC Papers in Press. Published on October 31, 2007 as Manuscript M706933200

Copyright 2007 by The American Society for Biochemistry and Molecular Biology, Inc.

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from

Sequence specific recognition of DNA by

proteins is essential for life, because the

biological effect of each gene is regulated by

transcription factors, which bind to the

promoter elements of target genes and

stimulate or repress the expression of the

corresponding gene (review: 1). Also many

nucleic acid interacting enzymes, like DNA

methyltransferases (reviews: 2,3,4),

aminoacyl-tRNA synthetases (review: 5),

DNA repair enzymes (review: 6) and

restriction endonucleases (review: 7) interact

with nucleic acids in a highly sequence

specific manner. In general, DNA

recognition follows two paradigms: direct

and indirect readout (review: 1). For direct

readout, proteins form contacts (including

hydrogen bonds and van der Waals contacts)

in the major (and to a lesser degree also the

minor) groove of the DNA to the edges of

the base pairs to probe the DNA sequence

(8). For indirect readout, proteins form

contacts to the DNA backbone. Since

structural preferences and dynamics of the

DNA are sequence dependent, the

interaction energy associated with a

particular DNA conformation can be used to

read the DNA sequence (9).

DNA methyltransferases are an important

example of enzymes that recognize specific

DNA sequences (reviews: 2,3,4). DNA-

(adenine N6)-MTases, as studied here,

transfer a methyl group to the N6-position of

adenine residues embedded in a specific

recognition sequence. In E.coli, the EcoDam

enzyme which methylates DNA at GATC

sites is involved in the coordination of DNA

replication and cell cycle, post-replicative

mismatch repair and regulation of the

expression of several genes (review: 4).

Recently, the structure of the EcoDam

enzyme in complex with specific DNA was

solved illustrating the mechanism of

sequence specific DNA interaction of this

enzyme (10). The protein comprises two

domains, a catalytic domain that is

conserved among DNA MTases and a DNA

binding domain that folds into a five-helix

bundle. The DNA interactions are made by a

ß-hairpin (residues 118-139) which is part of

the DNA binding domain and an N-terminal

extension of the catalytic domain (Fig. 1A).

Key DNA recognition contacts between

EcoDam and the GATC site are a hydrogen

bond between K9 and Gua1, contacts of

L122 and P134 to base pair 3 and an

interaction of R124 to the Gua in the 4th

base pair (see Fig. 1B). Exchange of any of

these residues led to major changes in the

DNA recognition specificity of EcoDam

(10,11). The enzyme forms contacts also

with the backbone of the DNA but there is

no apparent deformation of the DNA (apart

from flipping out the target base for

methylation and its partner base) which

suggests that indirect readout is secondary

for DNA recognition. The EcoDam enzyme

2

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from

is related to the phage T4 encoded T4Dam

enzyme, which also recognizes GATC sites

(Fig. 1A). The T4Dam has a very similar

structure and mode of DNA recognition as

EcoDam (11,12), with the only exception

that recognition of Gua1 is not based on an

interaction with the N-terminal loop of the

enzyme (T4Dam carries an alanine at the

position corresponding to EcoDam K9, see

Fig. 1A), but on a bidentate hydrogen bond

between R130 located at the basis of the ß-

hairpin and the guanine.

The M.EcoRV enzyme, which is the subject

of the present work, is part of a

restriction/modification system that protects

the bacterial host against bacteriophage

infection (review: 4). Although, restriction

enzymes and methyltransferases can

recognize the same sequence, they have

evolved independently and use different

mechanisms for DNA recognition. The

M.EcoRV enzyme is related to the Dam

family (Fig. 1A), but it recognizes GATATC

sites and modifies the first adenine within

the sequence (13). The DNA binding

domain of M.EcoRV is similar to that of

EcoDam, which suggests a similar

mechanism of DNA recognition (14,15). In

EcoDam and T4Dam, the target base is

flipped out of the DNA helix (10,11), which

most likely is also happening in the

M.EcoRV-DNA complex (16). For

methylation, the target base is inserted into a

pocket formed by a DPPY tetrapeptide that

is the most highly conserved part of N-

MTases (review: 2). The M.EcoRV DNA

interaction has been shown to invoke strong

bending of the DNA of approx. 61° (17),

and to induce large structural changes of the

DNA (18). Furthermore, binding of

sinefungin, an AdoMet analogue, increased

DNA binding by M.EcoRV 20 fold (19)

which indicates that there is communication

between the coenzyme and DNA binding

site.

The M.EcoRV/EcoDam system has been

studied as a paradigm of how proteins

change their DNA recognition during

evolution. One central prerequisite to answer

this question is to determine which structural

elements of M.EcoRV correspond to

EcoDam, meaning how the six base pair

recognition sequence of EcoRV should be

aligned to the EcoDam sequence (Fig. 1C).

Given the fact, that both enzymes methylate

the second adenine in a GAT sequence, an

alignment of the GAT parts of both

sequences appeared likely (20,21). Then, the

difference between M.EcoRV and EcoDam

DNA recognition sequence would lie in the

ATC part of the EcoRV site and M.EcoRV

would require some structural elements for

its recognition. The availability of the

T4Dam and EcoDam-DNA complex

structure shed new light on this question,

because the part of the ß-hairpin which is

3

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from

responsible for the recognition of the TC

part of the target site is fully conserved

between EcoRV and EcoDam as well as

other members of the family (10,11). This

observation would suggest that the last bases

of the target sites correspond to each other

and the recognition of the first part has been

altered between M.EcoRV and EcoDam.

This model is supported by the observation

that it is the recognition of Gua1 that is

altered between EcoDam and T4Dam (10)

suggesting that this is an evolutionary

flexible contact. Here, we show that both

these models are incorrect, because

recognition of the first and last base pairs of

the target site is conserved between

M.EcoRV and EcoDam. Hence, M.EcoRV

recognizes GATATC sites as an expanded

GATC site, with DNA bending introduced

to accommodate the increased distance

between the first at last bases of the target

site. Furthermore, we show that R128A is

inefficient in DNA binding and bending

whereas K11A shows relaxed sequence

specificity in DNA binding and bending.

These results suggest that the DNA contact

of R128 is an early contact between the

enzyme and the DNA whose formation is

followed by bending of the DNA and

making of the contact between K11 and

Gua1.

Experimental procedures

Oligonucleotides - Oligonucleotides were

purchased from Thermo Hybaid (Ulm,

Germany). We used one canonical substrate

LD20 and 13 near cognate substrates in

which one base pair of the recognition

sequence had been altered. Hemimethylated

substrates were used to specify which strand

of the near cognate substrates became

methylated. Double-stranded substrates were

prepared by mixing equimolar (20 µM)

amounts of each strand, heating to 95°C for

5 min, and slowly cooling to room

temperature. Following oligonucleotides

were used (all written in 5’ to 3’ orientation,

Bt stands for Biotinylation, mA for 6-

methyladenine):

LD20: d (GAT CGT AGA TAT CTC ATC

GA) / Bt-d (TCG ATG CGmA TAT CTA

CGA TC)

LD20 G1T: d (GAT CGT ATA TAT CTC

ATC GA) / Bt-d (TCG ATG CGmA TAT

ATA CGA TC)

LD20 G1C: d (GAT CGT ACA TAT CTC

ATC GA) / Bt-d (TCG ATG CGmA TAT

GTA CGA TC)

LD20 G1A: d (GAT CGT AAA TCT CTC

ATC GA) / Bt-d (TCG ATG CGmA TAT

TTA CGA TC)

LD20 T3A: d (GAT CGT AGA AAT CTC

ATC GA) / Bt-d (TCG ATG CGmA TTT

CTA CGA TC)

4

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from

LD20 T3C: d (GAT CGT AGA CAT CTC

ATC GA) / Bt-d (TCG ATG CGmA TGT

ATA CGA TC)

LD20 T3G: d (GAT CGT ATA GAT CTC

ATC GA) / Bt-d (TCG ATG CGmA TCT

ATA CGA TC)

LD20 A4C: d (GAT CGT AGA TCT CTC

ATC GA) / Bt-d (TCG ATG CGmA GAT

CTA CGA TC)

LD20 A4T: d (GAT CGT AGA TTT CTC

ATC GA) / Bt-d (TCG ATG CGmA AAT

CTA CGA TC)

LD20 A4G: d (GAT CGT AGA TGT CTC

ATC GA) / Bt-d (TCG ATG CGmA CAT

CTA CGA TC)

LD20 T5C: (GAT CGT AGA CAC CTC

ATC GA) / Bt-d (TCG ATG CGG TGT

CTA CGA TC)

LD20 T5A: d (GAT CGT AGA TCA CTC

ATC GA) / Bt-d (TCG ATG CGT GAT

CTA CGA TC)

LD20 T5G: d (GAT CGT AGA TAG CTC

ATC GA) / Bt-d (TCG ATG CGC TAT

CTA CGA TC)

LD20 C6T: d (GAT CGT AGA TAT TTC

ATC GA) / Bt-d (TCG ATG CAmA TAT

CTA CGA TC)

Site-directed mutagenesis, protein

expression and protein purification - The

M.EcoRV mutants were generated by the

megaprimer site-directed mutagenesis

method as described before (22). The

pGEXMRV plasmid (23), which encodes

M.EcoRV N-terminally fused to GST, was

used as a template. The mutagenic primer

contained the designated single substitution

and the screening restriction site to confirm

the mutation. Mutant plasmids were selected

by restriction analysis of marker sites

introduced together with the mutation. The

absence of non-desired mutations was

confirmed by complete sequencing of all

variant genes. Protein expression was

performed as described before (20), except

that cells were harvested 3h after the

induction of protein expression. Purification

of the GST-fusion proteins was done by

chromatography over GSH-Sepharose

(Pharmacia Biotech) basically as described

before (20), except that the sonication buffer

contained 500mM NaCl. Usually purified

proteins were >95 % pure (see Suppl. Fig. 1

for some representative examples of the

protein purifications used). Protein

concentrations were determined by

densitometric analysis of Coomassie stained

SDS PAGE gels.

Kinetic analysis using in vitro methylation

assay - Quantitative determination of the

catalytic activity of the wild type M.EcoRV

and its variants was carried out in vitro with

a biotin/avidin methylation assay as

described (24). All the methylation reactions

were performed under single turnover

conditions, because multiple turnover

experiments are not suited to probe the

5

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from

transition state of the DNA methylation

reaction by M.EcoRV, since the turnover of

M.EcoRV is limited by product release (25).

In order to determine comparable catalytic

activities of wild type and mutants, equal

concentrations of the enzymes (1 μM) were

used in the individual reactions. All the

reactions were performed in 40 µL reaction

mixture containing 1 µM oligonucleotide, 1

µM enzyme, 2.6 µM [3H]AdoMed, 500

ng/mL BSA, 1 mM EDTA, 50 mM Tris-HCl

(pH 7.5) and 100 mM NaCl. All kinetics

were done at least in triplicate and the

results averaged.

Kinetic data analysis - Linear reaction

progress curves were analyzed by linear

regression to determine the initial slope

(kmet). Hyperbolic reaction curves were

analyzed by a single exponential fit followed

by a linear phase:

CPMtheo(t) = F (1-e-kst t) + kmt t + BL (eq. 1)

kst single turnover rate constant for DNA

methylation, [1/min]; kmt multiple turnover

rate constant for DNA methylation

[CPM/min]; F size of the exponential phase

in [CPM]; BL background.

Since the single turnover rate (kst) of

M.EcoRV is much faster than the multiple

turnover rate (kcat) (25), under our conditions

the single turnover rate dominates the results

with kmt contributing less than 5% to kmet.

To determine the initial rate of enzyme turnover (kmet) the equation 1 is differentiated at t→0 to obtain:

kmet = F kst + kmt (eq 2)

The initial rate depends on the saturation

level of substrate with enzyme, which

depends on the DNA binding constant.

Since wild type M.EcoRV and all mutants

bind to the L20 substrate with an affinity of

about Kd=1x10-7 M or weaker (15), the

initial slopes determined here reflect kst/Kd

which is a well established measure of

enzyme specificity (26).

To compare the relative activity of wild type

and variant enzyme for methylation at a

modified target site (X), specificity factors

were calculated, where denotes the

methylation rate of one particular substrate

by one M.EcoRV variant or the wild type:

substratevariantk

substrateX typewild

GATATC typewild

substrateXvariant

GATATCvariant

/kk/kkS −

−

=

An increase in specificity is indicated by

S>1, loss of specificity by S<1.

Electrophoretic mobility shift experiments -

In order to investigate the DNA binding and

bending by M.EcoRV, two different 150 bp

DNA fragments called E (for “end”) and M

(for “middle”) were used. Both substrates

had a single EcoRV target site, E at the end

of the fragment and M in the middle. Both

6

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from

substrates were constructed by PCR

amplification of pET28a(+) vector using two

sets of oligonucleotides: 5’-ATC CCA CTA

CCG AGA TAT CCG CACC-3’ / 5’-TGT

CCG GTT TTC AAC AAA CCA TGC-3’

for E and 5’-AAT CCT GTT TGA TGG

TGG TTA ACG3’ / 5’-TGG TTG CCA

ACG ATC AGA TGG CG-3’ for M. Both

substrates were radioactively labeled with γ-32P-ATP using T4 polynucleotide kinase

(NEB).

The band shift assay was carried out as

previously described (17). In 20 µl, the

reaction mixtures contained 50 mM HEPES-

NaOH, pH 7.0, 1 mM EDTA, 100 mM NaCl,

5 mM dithiothretoitol, and 1 µg of

acetylated bovine serum albumin, 0.2 mM

sinefungin (Sigma), approx. 100 pM of each

amplified DNA and 100 nM of M.EcoRV

wild type protein (200 – 800 nM for the

M.EcoRV variants). The reactions were

incubated at room temperature for 15-30

min and loaded on a 8% non-denaturating

polyacrylamide gel run in 0.5 x TBE buffer,

followed by 6 hours electrophoresis at room

temperature (300V, 250 mA). The separated

complexes were visualized by

autoradiography using the phosphoimager

system (Fuji).

Scanning-force microscopy experiments -

A 406 bp DNA substrate was obtained by

PCR amplification of the pET28a(+) using

the primers: 5’-GCG TAT TGG GCG CCA

GGG 3’ and 5’- GGC TGG CAT AAA TAT

CTC ACT CGC-3’. It contains one EcoRV

site located in the center of the sequence.

DNA-protein complexes were established in

a total volume of 10 µl by combining 30 nM

DNA with either 250 nM of M.EcoRV (wt),

300 nM of K11A or 800 nM of R128A

protein in 50 mM HEPES (pH 7,5), 250 mM

NaCl, 1 mM EDTA and 100 µM of

sinefungin. The protein concentration had to

be increased in case of the mutants, to

observe a sufficient numbers of complexes.

Samples were incubated for 15 min at room

temperature. Then, 1 µl of the complex

solution was mixed with 9 µl of 5 mM

MgCl2 solution and deposited on freshly

cleaved mica (Plano, Wetzlar), allowed to

adhere for 40 seconds and washed with 1 ml

of bi-distilled water afterwards. The sample

was dried using compressed air.

Protein-DNA complexes were observed by

tapping mode in air using a Multimode

AFM with a Nanoscope III controller

(Digital instruments, Santa Barbara, CA).

We used NST-NCHF silicon cantilevers

(Nascatec, Stuttgart) with a nominal spring

constant of 50 N/m and a resonance

frequency of ~350 kHz. All images were

obtained with a scanning speed of 0.5-1 Hz

and a resolution of 512×512 pixels.

Experiments were carried out blinded, i.e.

the operator of the microscope did not know

7

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from

the previous data for each variant. To

remove background slope, raw images were

flattened using the Nanoscope software.

Particles on the DNA with a diameter of >8

nm and exceeding the height of the DNA

molecule by >50% were considered to be

bound protein.

Since the orientation of the DNA could not

be determined in the images, the distance of

a bound protein to the closest end of the

molecule was measured. When two proteins

were bound to one DNA molecule, distances

were measured with respect to the closest

end of each single protein. Distances were

evaluated after subdividing the 406 bp DNA

substrate into 14 sections of approx. 5 nm

each. The specific binding site is located in

segment 14. Segments 2, 3, 9, and 10

contain GAT sites which was important for

the analysis of the K11A variant. Bending

angles were measured by drawing straight

lines through the DNA arms adjacent to a

binding protein using the section tool of the

Nanosope software (Version 6.11r1). The

angle is given as deviation from the straight

line across the binding protein.

Results and Discussion

We applied double mutant cycles for the

identification of enzyme-DNA contacts

relevant for DNA recognition. To this end,

we prepared variants of M.EcoRV in which

candidate amino acid side chains were

exchanged against alanine and studied their

methylation activity using the cognate

substrate GATATC and several near-

cognate substrates, in which one base pair

had been altered. The underlying idea of

these experiments is illustrated in Fig. 2A. If

an enzyme-DNA contact is present between

amino acid D and base pair Z, the wild type

enzyme will methylate near-cognate

substrates altered at base pair Z at reduced

rate because of the loss of the contact. Since

a variant mutated at D lacks the contact it

often will show reduced activity at the

cognate site. However, it will also show

reduced ability to discriminate between

cognate and near cognate substrates having

altered bases at Z. Therefore, the near

cognate substrate Z will show a less

pronounced drop in activity (when

compared with the activity at the cognate

site) with mutant D than with the wild type

enzyme. It is this absence of decline in

activity that is indicative of a contact

between D and Z. In contrast, removal of

one contact between the enzyme and the

DNA might make recognition of other bases

become more stringent, because a complex

of the protein mutated at D with a near-

cognate substrate that carries a base pair

exchange at a site other than Z will contain

two defects, one introduced by the mutation

in the protein the other by the base pair

exchange. Hence, if an enzyme variant

shows an increased accuracy of recognition

8

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from

of a near-cognate substrate, one can

conclude that the mutated amino acid side

chain is not involved in recognition of that

particular base pair, but it may be involved

in recognition of other base pairs or in

indirect readout of the DNA sequence.

Selection of M.EcoRV variants - On the

basis of the EcoDam crystal structure and

the alignment of M.EcoRV with members of

the Dam family (Fig. 1), we selected several

residues for a mutational study, namely:

K11, which corresponds to EcoDam K9,

Q13 and K18 which are on the N-terminal

loop and hence should be close to the DNA.

Structural modeling suggested that these

residues might be candidates for the

recognition of the inner AT bases of the

M.EcoRV sequence (data not shown). The

R128 and P138 residues correspond to

EcoDam R124 and P134. R145 presumably

is located at the basis of the ß-hairpin which

forms many of the specific contacts to the

DNA in EcoDam and T4Dam (10,11). These

M.EcoRV variants had been studied already

(20), but in the former project all

experiments were designed based on the

assumption that the GAT parts of the

M.EcoRV and EcoDam recognition sites

should be aligned (which is not correct),

such that the influence of the exchanges had

only been studied on the 3’ part of the

recognition sequence. Also, in that paper

only one near-cognate substrate (A4C in the

nomenclature used here) had been used with

the mutant enzymes. Here, we investigated

methylation of 13 near-cognate substrates by

all variants, an approach that was successful

in the identification of enzyme-DNA

contacts of the EcoDam enzyme (10,11).

Catalytic Activity of the M.EcoRV and its

variants at cognate substrates - To

investigate the specific recognition of the

DNA by the M.EcoRV, we exchanged 6

amino acid residues to alanine, purified the

proteins and determined their DNA

methylation activity using the L20 substrate

(which contains a cognate GATATC site) in

vitro. The single turnover methylation rate

constant of wild type M.EcoRV was 0.18

(±0.03) min-1, which is similar to values

determined earlier (20). The relative

turnover rates of M.EcoRV and its variants

averaged over 3-5 experiments are given in

Fig. 2B, representative examples of

individual methylation experiments are

shown in Suppl. Fig. 2. The K11A, R128A

and R145A showed strong reduction in

activity, which indicates that these residues

are functionally important for the enzyme.

The activities of Q13A, K18A and P138A

were only slightly reduced or not reduced at

all. However, this does not rule out a role of

these residues in DNA recognition, because

the EcoDam P134 is a key residue for the

recognition of the third base pair, although

9

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from

the P134A variant also did not exhibit

reduced activity at the cognate substrate (11).

Specificity of DNA Methylation Activity by

M.EcoRV - To determine the stringency of

recognition of the individual base pairs of

the M.EcoRV site, we have performed

methylation kinetics using a set of 13 near

cognate substrates that differ from the

cognate GATATC site by single base pair

exchanges. The rates of methylation were

compared with the rate of methylation of the

L20 reference substrate to obtain a

specificity profile for M.EcoRV (Fig. 2C,

Suppl. Fig. 2). All experiments were

performed using hemimethylated substrates

to direct the methyl group transfer to one

target strand. As shown previously (20,21),

M.EcoRV displayed considerable activity

towards cognate and near-cognate

substrates; relative to GATATC near-

cognate substrates were methylated with the

10-1000 reduced rates. In general, the

specificity of M.EcoRV is lower than that of

EcoDam as indicated by the observation that

the relative activity of EcoDam was below

1% with all near-cognate substrates and 4

out of 8 substrates were methylated with

relative rates of 1/1000 or below (11). In

contrast, M.EcoRV methylated 9 out the 13

near-cognate substrates with relative rates

that are higher than 1% and all of them were

methylated with a relative rate higher than

1/1000.

The activity of M.EcoRV towards the A4C

substrate was reduced by about one order of

magnitude, which is in agreement with an

earlier result (20). It is interesting to note

that the T3C and T5C exchanges were most

inhibitory to catalytic activity. This result

distinguishes M.EcoRV from EcoDam,

which tolerated transitions better than

transversions (11). Like in the EcoDam

enzyme (10), at position Gua1, the Gua to

Ade exchange was tolerated best by

M.EcoRV. In EcoDam this result is in

agreement with the observation that the

enzyme contacts the Gua1 by a hydrogen

bond to the N7 atom, which is available in

guanine and adenine residues.

To identify amino acid residues, involved in

specific contacts to the target sequence, we

performed methylation kinetics using all

near-cognate substrates with the M.EcoRV

variants as well. The specificity profiles of

the six M.EcoRV variants studied here are

shown in Fig. 2D. Due to its low overall

activity the detection limit with the R128A

variant was at a level of 1% of the activity

observed with L20; with all other variants

relative activities of 0.1% could be analyzed

quantitatively. In order to compare these

specificities with the wild type enzyme, a

specificity factor S was calculated which

compares the relative specificity of

M.EcoRV variant and wild type.

10

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from

The accuracy of recognition (A) of each

base pair can be described as the relative

loss of activity at near cognate sites

modified at this base pair (A= kGATATC/kX-

substrate). Changes in the DNA recognition of

enzyme variants are identified by calculating

the S factor which is defined as the ratio of

the accuracies of the mutant and wild type

(S=Avariant/Awild type). S>1 indicates that

recognition is more stringent in the variant

than in wild type (=higher specificity), an S

factor <1 indicates that the recognition is

reduced in the variant to a certain degree

(=lower specificity). As described above,

increased specificity of a variant indicates

that the amino acid residue exchanged does

not play a role in the recognition of that

particular base pair. Reduced specificity

indicates the potential presence of an

enzyme-DNA contact.

The S factors of all variants are displayed in

Fig. 3. On the basis of the accuracy of the

individual methylation rates as shown in Fig.

2, S factors >10 or <0.1 are highly

significant and will be discussed in the

following paragraphs. There were two

outstanding results, namely the loss of

specificity of K11A and R128A for

recognition of the first and last base pair,

respectively. No other variant than R128A

showed loss of recognition of the last base

pair. For the first base pair it was only K11A

that displayed a consistent loss of

recognition with all three substrates tested.

R128A showed a minor loss of recognition

with G1A, but this effect was small (about

3-fold) and the same variant showed an

increase in recognition of G1T. In addition,

we observed loss of recognition with one (or

several) of the near cognate substrates

substituted at the third position with K11A,

Q13A, R128A, P138A and R145A. Because

of its lack of specificity, this effect was not

attributed to direct base recognition.

Recognition of the outer base pairs of the

M.EcoRV site – The K11A variant

displayed the largest changes in specificity,

because it showed a pronounced increase in

specificity in the interaction with the A4G

and C6T substrates. At the same time, it

displayed a considerable loss of specificity

for the G1T substrate. In EcoDam K9, the

residue homologous to K11 in M.EcoRV,

forms a hydrogen bond to the N7-position of

Gua1 (10). In the EcoDam-DNA complex

K9 (which is K11A is the only variant of

M.EcoRV which showed a reduced

recognition of the first base pair, suggesting

that it forms a similar hydrogen bond. The

K11A exchange almost did not affect the

interaction with G1A, while G1C and

particularly G1T showed loss of recognition.

The same pattern of residual activities had

been observed with EcoDam. It supports the

notion that K11 contacts the N7 of Gua1,

11

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from

because such contact cannot contribute to

the discrimination of guanine and adenine.

In extrapolation these results suggest a

comparable role of the N-terminal loop in

M.EcoRV and EcoDam, which explains the

finding that cofactor binding influences

DNA recognition by M.EcoRV (19),

because we have recently shown that slight

structural rearrangements of this loop in

EcoDam couple coenzyme binding,

recognition of the first base pair and

insertion of the flipped target base into the

base binding pocket of the enzyme (27).

The R128A variant showed a 20 fold

reduced activity at GATATC sites and it

displayed no detectable activity at most near

cognate sites. However, the GATATT site

was modified with the same rate as

GATATC indicating a clear loss of

recognition of the C6T substrate, which is

accompanied by increased selectivity for

G1A and exchanges at the 4th base pair.

These results can be compared with the

EcoDam R124A variant which showed a

strongly reduced catalytic activity at the

canonical and no activity at most near

cognate sites except GATG and GATT

which were methylated at the same rate as

the GATC site (11). Therefore, the

properties of the M.EcoRV R128A and the

EcoDam R124A variant are very similar to

each other. Combined with the position of

M.EcoRV R128 in the alignment of

M.EcoRV and EcoDam and the structural

data demonstrating that EcoDam R124

contacts the Gua of the fourth base pair in

the EcoDam-DNA complex, these data

strongly suggest that M.EcoRV R128 forms

a contact to the Gua in the sixth base pair of

the GATATC site in M.EcoRV.

Recognition of the inner part of the

M.EcoRV recognition sequence - The

Q13A variant showed an increased

recognition of all near cognate substrates

except T3C. Therefore, it is unlikely that

Q13 is involved in any base specific contact.

K18A and P138A did not show large

changes in the specificity profile when

compared with the wild type enzyme. Hence,

the P138 residue seems not to be involved in

the recognition of the Thy5, although in

EcoDam P134 has a van der Waals contact

to the TA base pair that plays an important

role in sequence readout (11). This

difference in the results obtained with

M.EcoRV and EcoDam suggests that the

structure of the M.EcoRV-DNA complexe

differs from the EcoDam-DNA complex at

Thy5. We also could not identify any other

residue that appears to be responsible for the

recognition of the central TAT part of the

M.EcoRV site.

DNA bending by M.EcoRV - The lack of

direct contacts to the central part of the

recognition sequence observed here for

12

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from

M.EcoRV is very similar to what is seen

with the R.EcoRV restriction enzyme, were

crystal structure analyses did not provide

any evidence of direct readout of the inner

AT base pair (28). However, R.EcoRV

bends the DNA during specific complex

formation (28,29). Since the stacking energy

of TA and AT dinucleotide steps is lowest,

the unstacking that accompanies bending

enables the enzyme to interact specifically

with GATATC sites. M.EcoRV had been

reported to bend its target DNA upon

specific binding (17) like many other DNA

MTases do as well (30-32). To investigate

the influence of the mutants investigated

here on DNA bending, we have set up an

electrophoretic mobility shift assay to detect

DNA binding. It uses two DNA fragments

of equal length that carry the EcoRV site in

the middle (M) or towards the end (E) of the

fragment. DNA bending would result in a

reduced gel mobility of the M fragment as

compared to E (33), which is what we

observed reproducibly (Fig. 4).

DNA bending by EcoRV variants - Given

the observation that K11 and R128

recognize the inner and outer base pair of

the EcoRV sequence, we wanted to

determine the effect of an exchange of one

of these residues to alanine on DNA binding

specificity and DNA bending. Therefore,

DNA binding by K11A, Q13A, R128A and

R145A was studied by gel retardation

experiments. All four proteins showed

reduced DNA binding such that higher

protein concentrations had to be used. Under

these conditions, non-specific binding was

observed as well (Fig. 4) and the specifically

shifted band was only detectable with K11A,

Q13A and R128A. However, the difference

in the relative mobility of the M and E

fragments was much smaller with the

variants (Fig. 4). This result could mean that

either specificity of DNA binding is reduced

and many complexes bind and bend at non-

specific sites or that specific binding no

longer led to DNA bending. In order to

distinguish between these alternatives, and

to derive DNA bending angles Scanning

force microscopy (SFM) experiments were

carried out.

SFM analysis of DNA binding specificity

and DNA bending - SFM experiments were

carried out in tapping mode using a 406 bp

DNA substrate containing one GATATC

site located in the center of the molecule.

Representative examples of SFM images

obtained are shown in Fig. 5; the data are

summarized in Supplemental Table 1. As

shown in Fig. 5A, the wild type enzyme

bound specifically to segment 14 which

contains the specific binding site (about 30%

of all bound proteins). Since there are 400

non-specific target sites and one specific site,

specific binding is preferred about 120 fold.

As observed previously for other nucleic

13

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from

acid binding proteins, there is an additional

binding preference to segment 1 that

represents the open ends (34-36). The K11A

(Fig. 5B) and even more R128A (Fig. 5C)

variants showed reduced DNA binding.

Both variants displayed a reduced specificity,

because only about 20% all molecules were

bound to the specific site. Close inspection

of the binding profile of K11A revealed that

additional binding occurred mainly in

segments that contain degenerate NNNATC

sites (segments 2, 3, 9, 10). We could not

determine a rule describing how R128A

selects its alternative binding sites. These

results confirm a role of K11 and R128 in

specific DNA recognition by M.EcoRV.

When bending angles are determined and

plotted for protein-DNA complexes, there

was a peak at small angles below 20° and a

second peak at 59.4° ±4.3° for the wild type

and at 60.9° ±3.7° for the K11A mutant but

not for R128A indicating that R128A has

almost lost its ability to bend the DNA (Fig.

6). The SFM derived bending angle for wild

type M.EcoRV is very similar to an angle

determined previously in detailed gel

retardation and phasing analyses (61±3°)

(17). The small angle peak reflects the

intrinsic flexibility of the DNA, since it was

also observed without the addition of protein

(Fig. 6D). When only complexes showing

the specific bending of 60° were plotted

along the length of the molecule, almost all

of them were found in segment 14 for the

wild type. This result indicates that with

wild type M.EcoRV stable DNA bending

only occurs at specific sites. With K11A

bending was also observed at degenerate

NNNATC sites (segments 2, 3, 9, 10, 14,

see Fig. 7 and Suppl. Fig. 3). Therefore, the

exchange of K11 by Ala leads to relaxation

in DNA recognition and stable bending

occurs at degenerate target sites as well.

This result is in agreement with the general

loss of specificity in the interaction with the

first and third base pair observed in the

kinetic experiments.

The SFM results obtained with R128A and

K11A are in agreement with the gel

retardation analyses which indicated a loss

of the difference in the electrophoretic

mobility of the M and E substrates which

carry the GATATC sites in the middle or at

the end. In the case of R128A, this can be

explained by the general and strong

reduction of specific bending. In the case of

K11A, DNA bending occurs at the specific,

but also at additional sites. Therefore, the

influence of the position of the GATATC

sites on the electrophoretic mobility of the

DNA molecule gets smaller and both

substrates run similarly in a gel shift with

K11A.

Conclusions

14

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from

Our results demonstrate that M.EcoRV

recognizes the outer base pairs of the

GATATC target site with residues

corresponding to those that recognize the

outer base pairs of the GATC target site in

the EcoDam enzyme. Hence, the EcoRV site

is recognized as an expanded EcoDam site,

with DNA bending being introduced to

shorten the distance between the outermost

base pairs of the recognition site. This

allows M.EcoRV to interact with a 6 bps site

using structural elements positioned for an

interaction with bases separated by only four

base pairs in unbend conformation (Fig. 8).

In EcoDam, the distance between Gua1 and

Cyt4 is 14.6 Å, which is already larger than

the distance required for 4 base pairs,

because EcoDam, like T4Dam, intercalates

an aromatic residue into the DNA, thereby

expanding it to a length corresponding

roughly to 5 base pairs (10,11). However,

interaction with 6 base pairs would

correspond to a further elongation of the

helix to about 17 Å in B-DNA. Taking these

numbers, geometric calculations show that

bending the DNA by 57° would be required

to bring the first and sixth base pair in

M.EcoRV in a distance of 14.6 Å. Although

this is a very rough estimate, which does not

consider the helical structure of the DNA, it

fits almost perfectly to the bending angle of

M.EcoRV observed by us and others. The

change in helical positioning of the outmost

GC base pairs by insertion of two additional

bases in the GATC sequence, may be

compensated by the unwinding that often

accompanies DNA bending. For example, in

the case of the R.EcoRV restriction enzyme,

the minor grooves sides of both guanines

point in one direction and face the body of

the enzyme (28). Given the structure of

EcoDam, we propose that the major groove

of the bent DNA will faces the protein in the

M.EcoRV-DNA complex. The interaction of

K11 and R128 with their target bases serves

as a binding clamp that enforces the bending

of the DNA. We speculate that the KPNR

insertion in the M.EcoRV sequence that

occurs at the edge of the ß-hairpin may

function as a wedge to enforce DNA

bending and allow expansion of the

recognition sequence. DNA bending could

be a general mechanism for expansion of the

recognition site of DNA methyltransferase

and restriction enzymes, like M.EcoRI and

R.EcoRV.

We could not identify base specific contacts

to the inner part of the GATATC site, which

suggests that indirect readout is operative.

The EcoRV restriction enzyme follows the

same mechanism for the recognition of the

inner AT part of the GATATC site (28).

Indirect readout of the inner base pairs of the

M.EcoRV site explains why in M.EcoRV

the T3C and T5C exchanges are most

deleterious for catalysis, although transitions

were the most well tolerated exchanges in

15

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from

the EcoDam system (11). Since it is AT rich

DNA that is easily bendable, introduction of

a single CG base pair would change this

property and, thereby, interfere with

catalysis. Bending of the DNA by M.EcoRV

also explains abnormalities at the central

base pairs like strong binding of M.EcoRV

to substrates with the central bases modified

by bulky adducts (37) and the strong change

in 2-aminopurine fluorescence with GAT-2

aminopurine-TC substrates (18).

DNA bending by adenine-N6 MTases has

been also shown for M.EcoRI which bends

DNA by 51° +/-17° (31). Later, Allan and

colleagues showed that the H235D exchange

in M.EcoRI led to loss of DNA bending

which was accompanied by a 10-fold loss of

DNA and cofactor binding (38). These

results are comparable with our data for the

M.EcoRV R128A variant. The EcoRI

H235D variant showed an increase in

selectivity for the target site, which is seen

for the R128A variant here as well, because

near cognate sites were methylated with

reduced rates (with the notable exception of

C6T and partially T3G). However, the exact

molecular role of H235 in M.EcoRI could

not be elucidated.

Our results allow arranging some of steps in

enzyme DNA interaction in their temporal

order (Fig. 8). DNA MTases localize their

target site by linear diffusion (39,40).

T4Dam could be crystallized in complex

with partial recognition sites, which only

contained the TC part of the GATC site (11).

In these complexes the characteristic

bidentate hydrogen bond of R116 to the Gua

(corresponding to R128 in M.EcoRV) was

already formed. This observation suggests

that formation of this contact is an early

event of site recognition. In M.EcoRV,

removal of R128 prevents DNA bending,

which indicates that the contact precedes the

bending event. Mutation of K11 in

M.EcoRV does not prevent bending, but

rather makes it less specific suggesting that

the contact of K11 to Gua1 is formed only

after bending and that the docking of the

first part of the recognition site to K11 is

used to discriminate bending at target and

near cognate sites. If this docking is less

tight, the overall efficiency of bending at the

specific site is reduced (compare 42% of all

complexes bound at the specific site for

K11A with 54 % for wild type). In addition,

the K11A forms a more stable interaction

with bend DNA at NNNATC sites than wild

type M.EcoRV, where bending at non-

GATATC sites was not detectable.

In summary, our data shed light on the

mechanism of molecular evolution that leads

to changes in the DNA recognition

specificity of DNA interacting enzymes.

Furthermore, they show the temporal order

of molecular event associated with DNA

16

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from

17

recognition by M.EcoRV which likely can

be generalized to other enzymes of the

EcoDam family.

Acknowledgements. This work has been

supported by the DFG (JE252/2 and

JE252/5).

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from

References

1. Garvie, C. W., and Wolberger, C. (2001) Mol Cell 8(5), 937-946 2. Cheng, X. (1995) Annu. Rev. Biophys. Biomol. Struct. 24, 293-318 3. Cheng, X., and Roberts, R. J. (2001) Nucleic Acids Res. 29(18), 3784-3795. 4. Jeltsch, A. (2002) ChemBioChem 3, 274-293 5. Lee, S. W., Cho, B. H., Park, S. G., and Kim, S. (2004) J Cell Sci 117(Pt 17),

3725-3734 6. Singer, B., and Hang, B. (1997) Chem Res Toxicol 10(7), 713-732 7. Pingoud, A., and Jeltsch, A. (2001) Nucleic Acids Res 29(18), 3705-3727 8. Seeman, N. C., Rosenberg, J. M., and Rich, A. (1976) Proc Natl Acad Sci U S A

73(3), 804-808 9. Travers, A. A. (1989) Annu Rev Biochem 58, 427-452 10. Horton, J. R., Liebert, K., Bekes, M., Jeltsch, A., and Cheng, X. (2006) J Mol Biol

358(2), 559-570 11. Horton, J. R., Liebert, K., Hattman, S., Jeltsch, A., and Cheng, X. (2005) Cell

121(3), 349-361 12. Yang, Z., Horton, J. R., Zhou, L., Zhang, X. J., Dong, A., Zhang, X., Schlagman,

S. L., Kossykh, V., Hattman, S., and Cheng, X. (2003) Nat. Struct. Biol. 10(10), 849-855

13. Nwosu, V. U., Connolly, B. A., Halford, S. E., and Garnett, J. (1988) Nucleic Acids Res 16(9), 3705-3720

14. Lauster, R., Kriebardis, A., and Guschlbauer, W. (1987) FEBS Lett 220(1), 167-176

15. Roth, M., Helm-Kruse, S., Friedrich, T., and Jeltsch, A. (1998) J. Biol. Chem. 273, 17333-17342

16. Jeltsch, A., Roth, M., and Friedrich, T. (1999) J. Mol. Biol. 285, 1121-1130 17. Cal, S., and Connolly, B. A. (1996) J Biol Chem 271(2), 1008-1015 18. Gowher, H., and Jeltsch, A. (2000) J. Mol. Biol. 303, 93-110 19. Szczelkun, M. D., and Connolly, B. A. (1995) Biochemistry 34(34), 10724-10733 20. Beck, C., Cranz, S., Solmaz, M., Roth, M., and Jeltsch, A. (2001) Biochemistry

40(37), 10956-10965. 21. Beck, C., and Jeltsch, A. (2002) Biochemistry 41(48), 14103-14110 22. Jeltsch, A., and Lanio, T. (2002) Methods Mol. Biol. 182, 85-94 23. Friedrich, T., Roth, M., Helm-Kruse, S., and Jeltsch, A. (1998) Biol Chem 379(4-

5), 475-480 24. Roth, M., and Jeltsch, A. (2000) Biol. Chem. 381, 269-272 25. Jeltsch, A., Friedrich, T., and Roth, M. (1998) J Mol Biol 275(5), 747-758 26. Fersht, A. (1998) Structure and Mechanism in Protein Science, W.H. Freeman

and Company, New York 27. Liebert, K., Horton, J. R., Chahar, S., Orwick, M., Cheng, X., and Jeltsch, A.

(2007) J Biol Chem, in press 28. Winkler, F. K., Banner, D. W., Oefner, C., Tsernoglou, D., Brown, R. S.,

Heathman, S. P., Bryan, R. K., Martin, P. D., Petratos, K., and Wilson, K. S. (1993) Embo J 12(5), 1781-1795

18

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from

29. Stöver, T., Kohler, E., Fagin, U., Wende, W., Wolfes, H., and Pingoud, A. (1993) J Biol Chem 268(12), 8645-8650

30. Dubey, A. K., and Bhattacharya, S. K. (1997) Nucleic Acids Res 25(10), 2025-2029

31. Garcia, R. A., Bustamante, C. J., and Reich, N. O. (1996) Proc Natl Acad Sci U S A 93(15), 7618-7622

32. Rasko, T., Finta, C., and Kiss, A. (2000) Nucleic Acids Res 28(16), 3083-3091 33. Hagerman, P. J. (1990) Annu Rev Biochem 59, 755-781 34. Bonin, M., Oberstrass, J., Lukacs, N., Ewert, K., Oesterschulze, E., Kassing, R.,

and Nellen, W. (2000) Rna 6(4), 563-570 35. Bonin, M., Oberstrass, J., Vogt, U., Wassenegger, M., and Nellen, W. (2001) Biol

Chem 382(8), 1157-1162 36. Klaue, Y., Kallman, A. M., Bonin, M., Nellen, W., and Ohman, M. (2003) Rna

9(7), 839-846 37. Cal, S., and Connolly, B. A. (1997) J Biol Chem 272(1), 490-496 38. Allan, B. W., Garcia, R., Maegley, K., Mort, J., Wong, D., Lindstrom, W.,

Beechem, J. M., and Reich, N. O. (1999) J Biol Chem 274(27), 19269-19275 39. Surby, M. A., and Reich, N. O. (1996) Biochemistry 35(7), 2201-2208 40. Urig, S., Gowher, H., Hermann, A., Beck, C., Fatemi, M., Humeny, A., and

Jeltsch, A. (2002) J Mol Biol 319(5), 1085-1096

19

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from

Figure legends

Fig. 1: Sequence and structure comparisons of M.EcoRV and EcoDam.

A) Multiple sequence alignment of M.EcoRV (GATATC) with members of the EcoDam family

(all GATC). Important residues in M.EcoRV and EcoDam are labeled.

B) Schematic figure of the DNA contacts of EcoDam (10,11).

C) Possible alignments of the recognition sequences of M.EcoRV and EcoDam.

D) Crystal structure of EcoDam complexed with DNA. The amino acid residues making specific

contacts to recognized bases are represented in a space fill mode and colored in dark grey.

Fig. 2: Relative rates of methylation of various substrates by M.EcoRV and M.EcoRV variants.

A) Principle of the double mutant cycle approach for the identification of protein-DNA contacts

applied here. In this example removal of amino acid D will abolish recognition of the Z base.

This means that the specificity of the variant on near-cognate substrates in which the 4th base pair

is altered is lost. In contrast, recognition of the W, X and Y bases becomes more specific.

B) Catalytic activity of M.EcoRV and its variants on the canonical GATATC substrate. Rates are

given relative to the wild-type enzyme activity.

C, D) DNA methylation specificities of M.EcoRV and its variants. In these panels, all activities

were given in comparison to the methylation activity with the canonical substrate. Rates are given

relative to the GATATC substrate. The y-axis is scaled to reflect the detection limit of activity in

the individual experiments.

All results in B-D are averaged over 3-5 independent experiments, the error bars denote the

standard error of the mean.

Fig. 3: Specificity factors of M.EcoRV and M.EcoRV variants. Specificity factors are given

relative to the wild type enzyme. An increase in specificity is indicated by S>1, loss of specificity

by S<1. For bars marked by * or ↑, no activity was detectable with the variant. In cases marked

by ↑, the lower detection limit of the variants’ activity was used to calculate a lower estimate for

the increase in specificity. In cases marked by the * no bar is shown, because the relative activity

of the wild type with this substrate was below the detection limit of activity for the variant.

20

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from

Fig. 4: DNA binding and bending analysis by gel retardation experiments.

A) Gel retardation analysis of DNA binding and bending by M.EcoRV and M.EcoRV variants.

The experiments were carried out with two DNA substrates of equal length, each bearing a single

M.EcoRV site which was located centrally in the fragment labeled with M (“middle”) and

towards the end of the molecule in the fragment labeled with E (“end”).

B) Quantitative analysis of the relative mobility of the E and M fragments in 3-5 independent gels.

For each experiment, µE and µM were determined and their ratio plotted in the figure as one line.

The thicker line labeled with an asterisk represents the average value.

Fig. 5: Distribution of M.EcoRV wild type and mutant enzymes on the substrate containing the

GATATC binding site in the center (segment 14). Enzymes bound to segments 1 to 13 are bound

to one of the corresponding segments on either arm. In A) the DNA molecule is schematically

shown with the GATATC site at segment 14 being depicted by a grey box, and positions of

degenerate EcoRV sites represented by black balls for NNNATC sites, diamonds for NATATC

sites and triangle for GATC sites.

A) Distribution of wild type enzyme, B) of mutant K11A and C) of mutant R128A. The total

numbers of counted protein molecules are given in for each variant. Representative SFM images

of wild type M.EcoRV, K11A and R128A mutant protein bound to the DNA substrate are shown

on the right-hand site of each plot. In each case the height is 8 nm and the bar is 100 nm.

Fig. 6: DNA bending angle distribution of M.EcoRV and its variants determined at all binding

positions. For the wild type (A) and for mutant K11A (B) a peak at 60° bending was observed in

addition to a peak between 0° and 20°. No 60° peak was found for mutant R128A (C) and for

DNA molecules incubated without enzyme (D). In the control without protein, bending was only

measured in the center of the DNA molecule.

21

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from

Fig. 7: The position of all complexes displaying a bending by 60° is plotted along the length of

the DNA molecule. For the wild type (A), 60° bending is mainly seen in segment 14, while the

mutant K11A (B) binds and bends in additional segments containing close to cognate binding

sites (GAT). The DNA molecule is schematically shown in (A) with the GATATC site at

segment 14 being depicted by a grey box, and positions of degenerate EcoRV sites represented by

black balls for NNNATC sites, diamonds for NATATC sites and triangle for GATC sites.

Fig. 8: Schematic summary of the results.

A) Model to explain the expansion of target site by DNA bending in the evolution from 4 to 6

base pair recognizing enzyme.

B) Reaction mechanism of M.EcoRV. The enzyme searches for its target site by linear diffusion.

It pauses at NNNATC sites, initiates DNA bending and contacts the first part of the recognition

sequence. If the sequence matches the catalytic cycle will continue. The K11A lacks one contact

to the first part of the recognition sequence. Hence DNA bending is less efficient and it becomes

less specific. R128A lacks one contact to the second part of the recognition sequence. Therefore,

the efficiency of recognition of NNNATC sites and initiation of DNA bending is reduced.

22

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from

Figures

Fig. 1:

K11 Q13 K18

R128 P138 R145

ß-hairpinR124 P134 Y138 (T4dam: R130)

K9

A

5’ G A T C 3’3’ C T mA G 5’

R124

K9

P134

L122

B Possible alignments of the recognition sequen-ces of EcoDam and M.EcoRV

GATATCGATCGATATCGATC

GATATCGATC

GATATCGATC

GATATCGA TCGATATCGA TC

CDNA recognitionby EcoDam D

G A T CK9 P134 R124

Crystal structure of EcoDam complexedwith DNA

23

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from

Fig. 2:

Wild type M.EcoRV

0.001

0.01

0.1

1

L20

G1A G1T

G1C T3G

T3A

T3C

A4G A4

TA

4C T5G

T5A

T5C

C6T

Wild type M.EcoRV

0.001

0.01

0.1

1

L20

G1A G1T

G1C T3G

T3A

T3C

A4G A4

TA

4C T5G

T5A

T5C

C6T

Methylation of GATATC substrates

0

0.4

0.8

1.2

wt

K11

A

K18

A

Q13

A

R12

8A

P138

A

R14

5A

Methylation of GATATC substrates

0

0.4

0.8

1.2

wt

K11

A

K18

A

Q13

A

R12

8A

P138

A

R14

5A

0.001

0.01

0.1

1

L20

G1A G1T

G1C T3G

T3A

T3C

A4G A4T

A4C T5G

T5A

T5C

C6T

Q13A

0.001

0.01

0.1

1

L20

G1A G1T

G1C T3G

T3A

T3C

A4G A4T

A4C T5G

T5A

T5C

C6T

Q13A

0.001

0.01

0.1

1

L20

G1A G1T

G1C T3G

T3A

T3C

A4G

A4T

A4C

T5G

T5A

T5C

C6T

K11A

0.001

0.01

0.1

1

L20

G1A G1T

G1C T3G

T3A

T3C

A4G

A4T

A4C

T5G

T5A

T5C

C6T

K11A

0.001

0.01

0.1

1

L20

G1A G1T

G1C T3G

T3A

T3C

A4G

A4T

A4C T5G

T5A

T5C

C6T

R145A

0.001

0.01

0.1

1

L20

G1A G1T

G1C T3G

T3A

T3C

A4G

A4T

A4C T5G

T5A

T5C

C6T

R145A

0.001

0.01

0.1

1

L20

G1A G1T

G1C T3G

T3A

T3C

A4T

A4C

T5G

T5A

T5C

C6T

P138A

0.001

0.01

0.1

1

L20

G1A G1T

G1C T3G

T3A

T3C

A4T

A4C

T5G

T5A

T5C

C6T

P138A

0.01

0.1

1

L20

G1A

G1T

G1C T3G

T3A

T3C

A4G

A4T

A4C

T5G

T5A

T5C

C6T

R128A

0.01

0.1

1

L20

G1A

G1T

G1C T3G

T3A

T3C

A4G

A4T

A4C

T5G

T5A

T5C

C6T

R128A

K18A

0.001

0.01

0.1

1

L20

G1A G1T

G1C T3G

T3A

T3C

A4G

A4T

A4C

T5G

T5A

T5C

C6T

K18A

0.001

0.01

0.1

1

L20

G1A G1T

G1C T3G

T3A

T3C

A4G

A4T

A4C

T5G

T5A

T5C

C6T

A B C D| | | |W X Y Z

A B C □| | |W X Y Z

recognition of Z is lost

A B C □| |

□ X Y Zrecognition of W, X and

Y is more stringent(synergistic effects)

A)

B) C)

D)

24

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from

Fig. 3

A) K11A B) Q13A

C) K18A

E) P138A

0.01

0.1

1

10

100

G1A G1T

G1C T3G

T3A

T3C

A4G A4T

A4C T5G

T5A

T5C

C6T

Spe

cifc

ityfa

ctor

(S)

0.01

0.1

1

10

100

G1A G1T

G1C T3G

T3A

T3C

A4G A4T

A4C T5G

T5A

T5C

C6T

Spe

cifc

ityfa

ctor

(S)

0.01

0.1

1

10

100

G1A

G1T

G1C T3G

T3A

T3C

A4G A4T

A4C

T5G

T5A

T5C

C6T

Spec

ifcity

fact

or (S

)

0.01

0.1

1

10

100

G1A

G1T

G1C T3G

T3A

T3C

A4G A4T

A4C

T5G

T5A

T5C

C6T

Spec

ifcity

fact

or (S

)

0.01

0.1

1

10

100

G1A G1T

G1C T3G

T3A

T3C

A4G A4T

A4C T5G

T5A

T5C

C6T

Spec

ifcity

fact

or (S

)

0.01

0.1

1

10

100

G1A G1T

G1C T3G

T3A

T3C

A4G A4T

A4C T5G

T5A

T5C

C6T

Spec

ifcity

fact

or (S

)

0.01

0.1

1

10

100

G1A G1T

G1C T3G

T3A

T3C

A4T

A4C T5G

T5A

T5C

C6T

Spe

cifc

ityfa

ctor

(S)

0.01

0.1

1

10

100

G1A G1T

G1C T3G

T3A

T3C

A4T

A4C T5G

T5A

T5C

C6T

Spe

cifc

ityfa

ctor

(S)

D) R128A

0.01

0.1

1

10

100

G1A G1T

G1C T3G

T3A

T3C

A4G A4T

A4C T5G

T5A

T5C

C6T

Spe

cifc

ityfa

ctor

(S)

* * * *

D) R128A

0.01

0.1

1

10

100

G1A G1T

G1C T3G

T3A

T3C

A4G A4T

A4C T5G

T5A

T5C

C6T

Spe

cifc

ityfa

ctor

(S)

* * * *

F) R145A

0.01

0.1

1

10

100

G1A G1T

G1C T3G

T3A

T3C

A4G A4T

A4C T5G

T5A

T5C

C6T

Spe

cifc

ityfa

ctor

(S)

F) R145A

0.01

0.1

1

10

100

G1A G1T

G1C T3G

T3A

T3C

A4G A4T

A4C T5G

T5A

T5C

C6T

Spe

cifc

ityfa

ctor

(S)

25

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from

Fig. 4

0.9

0.92

0.94

0.96

0.98

1

µM/µ

E

wt K11A Q13A R128AA) B)wt Q13A R128AK11A

FreeDNA

Specific complex

Unspecificcomplex

E M E M E M E M

*

** *

26

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from

Fig. 5

0

50

100

150

200

250

1 2 3 4 5 6 7 8 9 10 11 12 13 14

M.EcoRV

n=699 segments (1=end, 14=middle)

Anu

mbe

r of c

ompl

exes

0

20

40

60

80

100

1 2 3 4 5 6 7 8 9 10 11 12 13 14n=399 segments (1=end, 14=middle)

K11AB

num

ber o

f com

plex

es

0

20

40

60

80

1 2 3 4 5 6 7 8 9 10 11 12 13 14

R128A

n=294 segments (1=end, 14=middle)

C

num

ber o

f com

plex

es

27

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from

Fig. 6

M.EcoRV n=693A

num

ber o

f com

plex

es

030

6090

120150

180

0-5

10-1

5

20-2

5

30-3

5

40-4

5

50-5

5

60-6

5

70-7

5

80-8

5

90-9

5

100-

105

110-

115

120-

125

130-

135

140-

145

K11A n=399B

num

ber o

f com

plex

es

0

20

40

60

80

100

0-5

10-1

5

20-2

5

30-3

5

40-4

5

50-5

5

60-6

5

70-7

5

80-8

5

90-9

5

100-

105

110-

115

120-

125

130-

135

140-

145

C

num

ber o

f com

plex

es

bending angle (°)

D

num

ber o

f com

plex

es

R128A n=294

0

2040

60

80100

120

0-5

10-1

5

20-2

5

30-3

5

40-4

5

50-5

5

60-6

5

70-7

5

80-8

5

90-9

5

100-

105

110-

115

120-

125

130-

135

140-

145

0

10

20

30

40

50without enzyme n=149

0-5

10-1

5

20-2

5

30-3

5

40-4

5

50-5

5

60-6

5

70-7

5

80-8

5

90-9

5

100-

105

110-

115

120-

125

130-

135

140-

145

28

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from

Fig. 7

M.EcoRV

n=70 segments (1=end, 14=middle)

0

5

10

15

20

25

30

1 2 3 4 5 6 7 8 9 10 11 12 13 14

K11A

n=94 segments (1=end, 14=middle)

B

num

ber o

f com

plex

esnu

mbe

r of c

ompl

exes

A

0

15

30

45

60

75

1 2 3 4 5 6 7 8 9 10 11 12 13 14

29

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from

Fig. 8

EcoDam

G A T CK P R

G A T A T C

K P R

M.EcoRV

K P R

M.EcoRV

G A T A T C

K

PR

M.EcoRVK

PR

M.EcoRV

G A T

A T C

K

PR

M.EcoRVG A T

A T C

K

PR

M.EcoRVK

PR

M.EcoRV

N N N A T C

K

PR

M.EcoRVK

PR

M.EcoRV

N N N

A T C

K

PR

M.EcoRVN N N

A T C

K

PR

M.EcoRVK

PR

M.EcoRV

Base flipping, DNA methylationand product release

G A T A T C N N N A T C

K

PR

M.EcoRVK

PR

M.EcoRV

A B

30

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from

Supplemental information

Supplemental Table 1: Summary of the SFM results.

M.EcoRV (wt) (699 molecules analyzed, cM.EcoRV=250 nM)

DNA molecules bound by protein 46% (of all DNA molecules observed)

Protein bound to EcoRV site (Segment 14) 30% (of all bound protein)

Average of bending angle in the second peak (50-70°) 59.4° (±4.3°)

Complexes showing bending of 50-70° at the EcoRV site 54% (of protein bound to segment 14)

Complexes showing bending of 50-70° not at EcoRV site 5% (of all proteins bound to segments 1-13)

average bending angle in segments 1-13 14.7° (±23.6°)

average DNA bending angle at the EcoRV binding site (segment 14) without protein

10° (± 16.5°)

K11A (399 molecules analyzed, cK11A=300 nM)

DNA molecules bound by protein 12% (of all DNA molecules observed)

Protein bound to EcoRV site (Segment 14) 22.5% (of all bound protein)

Average of bending angle in the second peak (50-70°) 60.9° (± 3.7°)

Complexes showing bending of 50-70° at the EcoRV site 42% (of protein bound to segments 14)

Complexes showing a bending of 50-70° not at EcoRV site 40% (of all proteins bound to segments 1-13)

Complexes showing a bending of 50-70° in segments 2,3,9,10 (NNNATC containing segments)

43% (of all proteins bound to 2,3,9,10)

average DNA bending angle of complexes in segments 1,4-8,11-13

27.0° (± 34.5°)

Complexes showing a bending of 50-70° in segments 1, 4-7,11-13

20% (of all protein bound to segments 1, 4-7, 11-14)

R128A (294 molecules analyzed, cR128A=800 nM)

DNA molecules bound by protein 43% (of all DNA molecules observed)

Protein bound to EcoRV site (Segment 14) 21.4% (of all bound protein)

Average bending angle of all compexes segments 1-14 14.5° (±23.7°)

Average bending angle of complexes bound to seg 14 22° (± 23.8°)

31

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from

Supplemental Figure S1: Example of the purified proteins used in this study. All protein preparations were pure ≥95% as determined by scanning of Coomassie stained SDS-gels.

MW wt K11A Q13A K18A R128A P138A R145A

66.2 kDa

45 kDa

25 kDa

GST M.EcoRV

GST tag

32

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from

Supplemental Figure S2: Examples of the kinetic data used to calculate the data shown in Figs. 2

and 3.

GATATC

C6TG1AT3A

0

250

500

750

1000

1250

0 5 10 15 20

Time, min

CPM

0

1000

2000

3000

4000

5000

6000

0 5 10 15 20Time, min

CPM

0

1000

2000

3000

4000

5000

6000

0 5 10 15 20

Time, min

CPM

0

1000

2000

3000

4000

5000

6000

0 5 10 15 20Time, min

CPM

GATATC

A4CC6TT5G

0

40

80

120

160

200

240

0 5 10 15 20

Time, min

CPM

0

1000

2000

3000

4000

5000

6000

0 5 10 15 20Time, min

CPM

0

300

600

900

1200

1500

0 5 10 15 20Time, min

CP

M

GATATC

G1AT3AC6T

GATATC

G1AT3AC6T

GATATC

T3AG1AC6T

GATATC

C6T

T5GA4C

GATATC

A4C

C6TT5G

wt wt

K11A R128A

P138AQ13A

R145A

33

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from

Supplemental Figure S3: DNA bending of K11A at EcoRV sites half sites (NNNATC). The

distribution of bending angles observed with K11A is compared for enzyme molecules bound at

the specific site (A), bound at NNNATC sites (B) and bound at unrelated sites (C).

0

20

40

60

80

100

0 10 20 30 40 50 60 70 80 90 100

110

120

130

140

n=316 bending angle (°)

DNA bending in segments 1-13

0

20

40

60

80

100

0 10 20 30 40 50 60 70 80 90 100

110

120

130

140

n=316 bending angle (°)

DNA bending in segments 1-13

0

10

20

30

40

0 10 20 30 40 50 60 70 80 90 100

110

120

130

140

n=143 bending angle (°)

Segments containing NNNATC sites (2,3,9,10)

0

10

20

30

40

0 10 20 30 40 50 60 70 80 90 100

110

120

130

140

n=143 bending angle (°)

Segments containing NNNATC sites (2,3,9,10)

0

15

30

45

60

75

0 10 20 30 40 50 60 70 80 90 100

110

120

130

140

n=173 bending angle (°)

Segments without NNNATC sites (1, 4-8, 11-13)

0

15

30

45

60

75

0 10 20 30 40 50 60 70 80 90 100

110

120

130

140

n=173 bending angle (°)

Segments without NNNATC sites (1, 4-8, 11-13)

A

B

C

34

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from

Tomasz P. Jurkowski, Nils Anspach, Liliya Kulishova, Wolfgang Nellen and Albert Jeltschrecognition of an expanded ecodam recognition site

The M.EcoRV DNA-(adenine N6)-methyltransferase uses DNA bending for

published online October 31, 2007J. Biol. Chem. 

  10.1074/jbc.M706933200Access the most updated version of this article at doi:

 Alerts:

  When a correction for this article is posted• 

When this article is cited• 

to choose from all of JBC's e-mail alertsClick here

Supplemental material:

  http://www.jbc.org/content/suppl/2007/11/08/M706933200.DC1

by guest on June 26, 2018http://w

ww

.jbc.org/D

ownloaded from