DNA-Protein Interactions Unit, Department of Biochemistry,

1
Co-operative DNA Binding by the SfiI Endonuclease
Yana S. Kovacheva, Stuart R.W. Bellamy and
Stephen E. Halford
DNA-Protein Interactions Unit, Department of
Biochemistry, School of Medical Sciences,
University of Bristol, Bristol BS8 1TD, UK
2
Introduction
The Type II SfiI endonuclease recognises an
interrupted palindromic sequence of 13 bp (shown
right, where n is any base). The point of
cleavage is shown by the arrows.
SfiI binds two copies of its recognition site, at
separate locations in the DNA. In the presence
of Mg2, it then cleaves four phosphodiester
bonds, at both sites in both strands, in a highly
concerted reaction (1,2). The native protein is a
tetramer of identical subunits (1).
The crystal structure of the protein bound to two
cognate duplexes has recently been determined (3
Fig. 1). Two subunits bind one copy of the
recognition sequence, with each monomer binding
one GGCC half-site. The two DNA-binding clefts
are located on the opposite sides of the protein,
yet DNA cleavage occurs only when both clefts are
filled (2,3).
Figure 1. Crystal structure of SfiI bound to two
15-mers (3)
3
Project Aims

• The aim of the project is to use
  fluorescence-based methods to study the binding
  of two specific DNA duplexes to the SfiI
  tetramer, the conformational rearrangements of
  the protein, and its DNA looping reactions.
• Binding can be followed in Ca2 buffer, which can
  substitute for Mg2, as no cleavage occurs in
  presence of Ca2.

Previously proposed binding mechanism (2)
Figure 2. SfiI titrations of 10 nM DNA (2)

• Upon titrating D with increasing E, ED2 is formed
  directly, even at low E, whereas ED1 is only
  formed when E is added in large molar excess of
  over D (Fig. 2).
• This poses the problem can binding to one site
  be observed in real time before binding to the
  second site?
• How does the second binding occur?

Ca2 Binding buffer 2 mM CaCl2 25 mM NaCl 20 mM
Tris-HCl (pH 7.5) 5 mM ß-ME 100 ?g/ml BSA
4
DNA-DNA and DNA-Protein Distances
In order to study DNA binding to SfiI, an attempt
was made previously to measure DNA-DNA FRET.
However, at that time, the crystal structure was
not yet solved and the chosen FRET pair
(Fluorescein-Rhodamine Fl-Rh, R049 Å) was
inefficient. The crystal structure has now
revealed that the DNA-DNA vertical distance is 72
Å, and the diagonal 85 Å (Fig. 3). Moreover, both
DNA molecules are orientated at 67º to each other
and each DNA is bent by 25º (3). Therefore, the
distance between the DNA ends is too great to
measure by FRET using the Fl-Rh pair.
Trp41,42
Trp85
DNA to Cys230 10 Å DNA to Trp41,42,85 20 - 50
Å
Cys230
Figure 3. DNA-DNA distances measured between
5-ends of two 15-mers bound to SfiI (3)
Figure 4. DNA-Protein distances
Because of the long DNA-DNA distance, we decided
to use FRET between the DNA and protein residues
that have fluorescent properties, such as
tryptophan (3 per monomer, Fig. 4). Another
possibility is to label the cysteine residue (one
per monomer) with a quencher and use
donor-quencher FRET.
5
Fluorescent Methods - 1
The DNA-binding step has been monitored by FRET,
first at equilibrium and second by stopped-flow
kinetics. DNA carrying a 5'- fluorescent label,
Alexa350 (Molecular Probes), was used as the
acceptor of emission from the donor tryptophan
residues in the protein

• FRET requirements
• Donor and acceptor spectra must overlap
• Donor quantum yield must be high enough
• Donor and acceptor must be in close vicinity
• Donor and acceptor dipole moments must be in
  proper orientation

6
FRET Titrations of SfiI with DNA
Firstly, to test whether DNA labelled with
Alexa350 gave an efficient FRET signal with the
Trp residues, the Trp was excited at 280 nm. The
decrease of its emission and the increase in
Alexa350 emission at 440 nm were both monitored
(Fig. 5).
Figure 5. DNA Titrations by FRET raw data (a)
and normalised data (b)
a)
b)
Alexa Em
Trp Em
Control
3e5
Trp
2e5
Fluorescence (rel)
Fluorescence
1e5
Alexa350
0
350
400
450
500
Wavelength, nm
DNA, nM
Experimental conditions E0 100 nM, DNA 0
- 275 nM, in Ca2 - Binding Buffer
The titrations with DNA caused background
problems as the Alexa350 emission also increases
with the DNA concentration. In addition, the
protein contains 12 Trp residues in total, so it
is not clear which contributes to the FRET
signal.
7
FRET Titrations of DNA with SfiI
Secondly, we tried titrating a fixed DNA
concentration with increasing enzyme. The Trp was
excited at 280 nm, and the increase of Alexa350
emission was monitored at 440 nm (Fig. 6a).
Figure 6. Enzyme Titrations (a), raw data (b),
normalised data fitted to a binding curve.
a)
b)
Binding curve fit to 1-DNA KED
42.8 ? 3.9 nM 2-DNA KED2 32.2 ? 6.2
nM KED1 is indeterminate
Fluorescence ( rel)
Fluorescence
Wavelength, nm
Enzyme, nM
Experimental conditions E0 0 - 300 nM, DNA
30 nM, in Ca2 Binding Buffer, raw data (a). In
(b), the blue and pink lines are the best fits to
a binding curve for 1 or 2 DNA molecules with
protein.
The data was fitted to theoretical binding curves
for the association of either 1 or 2 molecules of
DNA with the protein (Fig. 6b). The data matched
both the 1- and the 2-DNA models but failed to
define a value for the first binding constant in
the 2-DNA model. The single KD values observed in
both cases, of 36 nM, thus corresponds to the
co-operative formation of ED2, indicating that
essentially none of the ED1 complex accumulates
during the titration. Hence, the formation of
ED1 (in high E) cannot be determined by this
method (Fig. 6b).
8
Stopped-Flow Kinetics
The binding of the DNA to SfiI was observed in
the stopped-flow apparatus. The Trp was excited
at 280 nm and a cut-off filter (395 nm) was used
to observe the Alexa350 DNA Emission (Fig. 7a).
The increase in FRET followed a single
exponential. The rate of increase in FRET was
measured at fixed DNA, varying E, obtaining
the apparent rate constants, which were then
plotted against the enzyme concentration (Fig.
7b).
Figure 7. Measuring the binding reaction over
time E0 200 nM, DNA 20 nM in
Ca2-Binding Buffer (Fig. 7a). Determination of
KD for this reaction (Fig. 7b).
a)
b)
Rate, sec-1
kon 43.2 s-1 koff 3.9x108 M-1s-1 KD 110.5
nM
Enzyme, nM
The KD measured corresponds to KED2 which
confirms that only the formation of ED2 complex
is observed, and the ED1 formation cannot be
detected. The reaction is therefore highly
co-operative.
9
Fluorescent Methods - 2
Other studies have used derivatives of SfiI
carrying a fluorescent quencher (QSY9, Molecular
Probes) at the single cysteine residue (Cys230)
in each subunit, which can quench the emission
from appropriate fluorescent labels on the DNA,
10 Å away (Fig. 9). The QSY9 can quench donor
emission in the range 480 -620 nm.
The maleimide derivative of QSY9 was used to
label all four SfiI cysteine residues. The
labelled protein was purified by extensive
dialysis and quantified by Maldi-Tof MS. The
labelling efficiency was 75 . The activity of
the labelled protein was found to be comparable
to WT SfiI, therefore this method can be used for
future experiments.
10 Å
Figure 9.
10
FRET by Quenching
Initial experiments were carried out to test
whether the labelled SfiI-QSY9 can quench
fluorescein (Fl) emission upon DNA binding in
trans (Fig. 10a). For this purpose, a DNA
molecule carrying a 5'- fluorescein label was
used as a donor and excited at 490 nm. Its
emission at 515 nm was decreased by the quencher
label on the cysteine residues in the protein as
both labels are in close vicinity and their
spectrums overlap as required (Fig. 10b).
The FRET reaction caused a 35 quenching of the
signal (Fig. 10c). Using a quencher instead of
acceptor is an advantage because the quencher
cannot emit and therefore does not give
background problems.
11
DNA-DNA FRET with Bulged DNA
The DNA-DNA distances in the doubly bound
SfiI-DNA complex are too great to measure
directly (Fig. 3). Therefore we decided to extend
the DNA ends and bring them closer together. The
DNA molecules were designed by molecular
modelling the DNA duplex contains a bulge and
makes a turn of 90 (4). Upon binding, the ends
of the bulged DNA (Fig. 11) will be closer to
each other than if using normal B-DNA duplexes
(Fig. 3). First, FRET will be tested by binding
two DNA molecules to SfiI (in trans). However,
statistically, only 25 of the bound complexes
will have the orientation/combination of D-Q
which can give a FRET signal. This problem can be
avoided if the two sites are in cis. In
addition, sites in cis will permit studies of DNA
looping.
Two complimentary oligonucleotides were used 1)
Top strand 46-mer (with extra bases) 2) Bottom
strand 41-mer with 5'-label donor (Alexa488) or
quencher (QSY7)
1) 5-AATAGGCCTTGTTGGCCACATGCATCGAAdUAAGCTACGATAAA
CTG -3 2) 3-TTATCCGGAACAACCGGTGTACGTAGC------CG
ATGCTATTTGAC-X-5
For the control experiment two straight
complimentary oligonucleotides were used 3) Top
strand 41-mer (no extra bases) 4) Bottom
strand 41-mer with 5'-label donor (Alexa488) or
quencher (QSY7)
3) 5-AATAGGCCTTGTTGGCCACATGCATCGGCTACGATAAACTG
-3 4) 3-TTATCCGGAACAACCGGTGTACGTAGCCGATGCTATTTGA
C-X-5
To check whether the modelled oligonucleotides
are in the correct orientation, the Alexa488
emission can alternatively be quenched with the
QSY9 label on the cysteine in the protein.
12
Bulged DNA in trans
Initially, reactions were carried out in trans to
test whether the modelled bulged DNA molecules
give a FRET signal as expected 12.5
theoretical FRET when 25 DNA are bound in the
correct orientation
I) DNA-DNA FRET (donor Alexa488-quencher
QSY7) II) DNA-Protein FRET (donor
Alexa488-quencher QSY9)
b)
b)
a)
50 nM Alexa488-DNA 50 nM QSY7-DNA Titrations
with E0 - 400 nM
50 nM Alexa488-DNA, Titrations with E0 -120 nM
c)
c)
Control
Control
Fluorescence (rel)
Fluorescence (rel)
Bulged DNA-Protein Quenching
Bulged DNA-DNA FRET
Figure 12.
Figure 13.
SfiI-WT, nM
SfiI-QSY9, nM
The DNA donors emission can be quenched by QSY9
from SfiI when bulged DNA is used (Fig. 13a,c).
A slight decrease in the emission is observed
with the control DNA (Fig. 13b,c). So the
bulged DNA is close enough to the protein for
FRET.
The DNA-DNA FRET showed 10 signal using
bulged DNA molecules (Fig. 12a,c) and no FRET
occurs when straight DNA duplexes are used (Fig.
12b,c). Therefore, the DNA ends are in correct
orientation and close to each other.
13
Conclusions

• DNA Binding to SfiI is a highly co-operative
  reaction, and the binding of each DNA molecule
  was not detected individually.
• The formation only of the doubly bound complex
  ED2 was observed at equilibrium and by
  stopped-flow kinetics - a very rapid process, as
  ED2 is formed in less than 0.02 sec. However, the
  measured KD shows that the binding of the
  fluorescent oligos was not as tight as the
  radioactive oligos previously used in gel-shift
  experiments (2).
• In order to measure FRET between two DNA
  molecules, the DNA ends were brought closer
  together (at 46 Å) by designing extended DNA
  molecules containing extra bases in the duplex
  causing a bend of 90 . The results showed that
  it is possible to measure DNA-DNA FRET. However,
  the initial FRET experiments were done using SfiI
  acting in trans, bridging two separate DNA
  molecules, which is always disfavoured relative
  to those in cis and statistically gives only 25
  of the molecules orientated correctly for FRET.

Future work
1) Ideally, we would like to perform the FRET
experiments in cis where the sites are on the
same DNA molecule, which will give us a method of
measuring DNA looping. PCR products of 100 - 400
bp, with a SfiI site close to each end, can be
constructed with a donor-quencher pair and used
for FRET experiments. This method can explore the
DNA looping reaction in detail and can also give
information about the kinetics of the binding
steps.
2)
1)
2) Moreover, each DNA end can be tagged with a
different fluorescent label (D1 and D2, both
quenched by QSY9 from the protein using different
cut-off filters). Thus, we can measure the
binding rate for each site independently.
References
Acknowledgements
We thank Eva Vanamee and Aneel Aggarwal for the
SfiI crystal structure, Richard Sessions for the
molecular modelling, Mark Dillingham and Frank
Peske for ideas and discussions, and the Wellcome
Trust for funding.
(1) Wentzell et al. (1995) J. Mol. Biol. 248,
581-595 (2) Embleton et al. (1999) J. Mol. Biol.
289, 785-797 (3) Vanamee et al. (2005) EMBO J.
24, 4198-208 (4) Gollmick et al. (2002) NAR, 30,
2669-2677