Using time-resolved fluorescence depolarization measurements, the global rotational correlation times of RepA free in solution and bound to the operator and to two distinct iteron dsDNA o
Trang 1RepA in complex with operator and iteron sequences
and free in solution
Rutger E M Diederix1,2, Cristina Da´vila1,2, Rafael Giraldo2and M Pilar Lillo1
1 Departamento de Biofı´sica, Instituto de Quı´mica Fı´sica ‘Rocasolano’, CSIC, Madrid, Spain
2 Departamento de Microbiologı´a Molecular, Centro de Investigaciones Biolo´gicas, CSIC, Madrid, Spain
RepA is the DNA replication initiator protein of the
Pseudomonas plasmid pPS10 It is representative of a
family of plasmid replication initiators active in many
Gram-negative bacteria, including the initiators from
plasmids such as pSC101, F and R6K [1] The
opera-tor region preceding the repA gene contains a partially
palindromic sequence (inverted repeat, IR) to which
RepA can bind, which acts as an autogenous repressor
of transcription [2] The plasmid also carries an origin
of replication, containing a sequence with four conti-guous tandem repeats (direct repeats, DR; termed iterons) of the same 6 bp sequence found inverted in the operator region of RepA RepA thus has dual DNA-binding activity: it can bind as a dimer to its operator region, in which case it functions in trans-cription repression; and it can bind in a highly cooper-ative fashion to the four directly repeated iterons, in which case it functions in replication initiation [3]
Keywords
anisotropy; DNA replication; fluorescence;
hydrodynamics; RepA
Correspondence
M P Lillo, Departamento de Biofı´sica,
Instituto de Quı´mica Fı´sica ‘Rocasolano’,
CSIC, Serrano 119, 28006 Madrid, Spain
Fax: +34 91 564 2431
Tel: +34 91 561 9400, ext 1027
E-mail: pilar.lillo@iqfr.csic.es
(Received 26 June 2008, revised 8 August
2008, acccepted 5 September 2008)
doi:10.1111/j.1742-4658.2008.06669.x
RepA, the replication initiator protein from the Pseudomonas plasmid pPS10, regulates plasmid replication and copy number It is capable of autorepression, in which case it binds as a dimer to the inverted repeat oper-ator sequence preceding its own gene RepA initiates plasmid replication by binding as a monomer to a series of four adjacent iterons, which contain the same half-repeat as found in the operator sequence RepA contains two domains, one of which binds specifically to the half-repeat The other is the dimerization domain, which is involved in protein–protein interactions in the dimeric RepA–operon complex, but which actually binds DNA in the monomeric RepA–iteron complex Here, detailed fluorescence studies on RepA and an N-(iodoacetyl)aminoethyl-8-naphthylamine-1-sulfonic acid-labeled single-cysteine mutant of RepA (Cys160) are described Using time-resolved fluorescence depolarization measurements, the global rotational correlation times of RepA free in solution and bound to the operator and to two distinct iteron dsDNA oligonucleotides were determined These provide indications that, in addition to the monomeric RepA–iteron complex, a stable dimeric RepA–iteron complex can also exist Further, Fo¨rster reso-nance energy transfer between Trp94, located in the dimerization domain, and N-(iodoacetyl)aminoethyl-8-naphthylamine-1-sulfonic acid-Cys160, located on the DNA-binding domain, is observed and used to estimate the distance between the two fluorophores This distance may serve as an indica-tor of the orientation between both domains in the unbound protein and RepA bound to the various cognate DNA sequences No major change in distance is observed and this is taken as evidence for little to no re-orienta-tion of both domains upon complex formare-orienta-tion
Abbreviations
(I)AEDANS, N-(iodoacetyl)aminoethyl-8-naphthylamine-1-sulfonic acid; FRET, Fo¨rster resonance energy transfer.
Trang 2Interestingly, in the latter case, the protein binds as
a monomer [2–6]
Free in solution, the protein is essentially dimeric,
but it dissociates and binds as a monomer in the
pres-ence of even a single iteron sequpres-ence [2,3] The
mecha-nism by which this occurs is unclear, but it involves
considerable conformational changes in RepA [3,4]
judged by comparison of crystal structures of
(trun-cated) RepA dimer [5] and the monomeric RepA–
iteron complex that was modeled on the complex
structure of the close homolog RepE from the F
plas-mid [2,7–9] For the latter protein, the crystal
struc-tures of both the monomer–iteron and dimer–operator
complexes are available, indicating secondary
struc-tural changes in the linker connecting the dimerization
and DNA-binding domains, and rearrangement of the
relative orientation of the two domains [7,9] The
con-formational change upon iteron binding may expose a
recognition site for protein–protein interaction,
enabling coupling of recently replicated origins from
different plasmid molecules [10,11] This so-called
handcuffing is thought to be the mechanism for
repli-cation inhibition in iteron-containing plasmids [12]
Following our series of biophysical studies of RepA
[3–6], we report hydrodynamic and structural studies
on RepA and its complexes with operator and
single iteron sequences Global rotational correlation
times were determined by fluorescence anisotropy
decay experiments using the extrinsic fluorophore
N-(iodoacetyl)aminoethyl-8-naphthylamine-1-sulfonic
acid (AEDANS), specifically coupled to Cys160 in the
single-cysteine mutant C160–RepA The AEDANS
probe was also used as a Fo¨rster resonance energy
transfer (FRET) acceptor to monitor putative
interdo-main movements in RepA upon binding the various
DNA sequences We show that, despite the extensive
structural rearrangement that is known to occur upon
monomerization and DNA binding to the iteron
sequence [3–6], an appreciable change in the
inter-domain organization is not actually observed Finally,
it appears that monomerization does not occur
effi-ciently in very short oligonucleotides that contain few
bases more than the iteron sequence, and RepA binds
as a dimer instead
Results
Labeling and characterization of C160–RepA
C160–RepA is a double-mutant of His6-tagged
wild-type RepA [4] in which two wild-wild-type Cys residues
(C29 and C106) have been changed to Ser The single
remaining Cys160 is located on the C-terminal
DNA-binding domain of RepA, also called the WH2 domain, which specifically recognizes the operator and iteron sequences [1,2] Most C160–RepA is expressed
in inclusion bodies, and the His6-tagged protein was purified from solubilized inclusion bodies using Ni(II)-affinity chromatography under denaturing conditions
As shown previously [3,4], the His6-tag does not inter-fere with protein function or structure, and it was not removed after purification Refolding of C160–RepA is
by rapid 20-fold dilution in buffer (0.15 m (NH4)2SO4,
15 mm Na-acetate, 0.03 mm EDTA, 3% glycerol,
pH 6.0) Almost all the protein is recovered and is present as a single, soluble species Refolded C160– RepA is dimeric, as judged by size-exclusion chromato-graphy, where it elutes at exactly the same volume as wild-type RepA (not shown)
Labeling of the single Cys of native C160–RepA with IAEDANS gives very low yields (< 5%) The yield can be improved significantly (to 50%) by per-forming the labeling reaction under conditions where the protein is unfolded, i.e in the presence of 5.6 m guanidinium hydrochloride Presumably, this poor reactivity is related to the low solubility of the native protein (up to 10–20 lm) Under denaturing condi-tions, RepA can easily be concentrated 10- to 100-fold, thus favoring the bimolecular labeling reaction greatly under the conditions of 15-fold excess IAEDANS The CD spectrum of unlabeled or AEDANS-labeled C160–RepA is indistinguishable from that of wild-type RepA at 5 C (Fig 1A), indicating that the secondary structure is not affected by the mutation or by AEDANS labeling Thermal denaturation analysis of the protein variants suggests a lower stability of the mutant (Fig 1B) The C160–RepA variants show a lower melting temperature than wild-type RepA (60 versus 67C for wild-type RepA), and the thermal transition of unlabeled C160–RepA has a substantially lower slope (reduced co-operativity) than wild-type RepA and the labeled variant However, room temper-ature is well below the melting transition, and as the experiments described here have been performed at or below this temperature, it can safely be assumed that the mutant protein is fully folded This is supported by the observation that the fluorescence emission spec-trum of the unique Trp residue (W94), a sensitive indi-cator of the folding state of the dimerization domain
of RepA [4], is unchanged in the mutant with respect to that of wild-type RepA (Fig 1C) Finally, AEDANS C160–RepA and wild-type RepA show identical binding to the operator sequence (Fig 1D), confirming that mutation and labeling do not affect the function, and by implication therefore also the structure, of RepA
Trang 3Figure 2A shows the emission spectrum of AEDANS
C160–RepA excited at 295 and 375 nm, respectively
When excited at 295 nm, fluorescence contributions
from both AEDANS and W94 are visible Figure 2B
shows the excitation spectrum of the acceptor
(kem= 480 nm) There is a clear contribution from
W94 visible as a shoulder at 280–290 nm, which is
assignable to FRET from W94 to C160-AEDANS
A
B
Fig 2 (A) Fluorescence emission spectra of AEDANS-labeled C160–RepA, excited at 295 nm (solid line) and 375 nm (dashed line) The spectra are normalized with respect to the emission intensity at 484 nm (B) Excitation spectrum of AEDANS C160– RepA, measured at 480 nm The arrow indicates the contribution of Trp fluorescence The spectra were recorded at 23.5 C, in 0.15 M
(NH4)2SO4, 15 m M NH4-acetate, 0.03 m M EDTA, 3% glycerol;
pH 6.0 [RepA] was 2 l M.
A
B
C
D
Fig 1 (A) Near- and far-UV CD spectra of wild-type RepA (solid line) and C160–RepA both unlabeled (dashed line) and AEDANS-labeled (dash-dots) The spectra were recorded at 5 C with
3.5 l M wild-type and unlabeled C160–RepA, and 7 l M AEDANS C160–RepA The buffer spectrum is subtracted and the spectra have been transformed to mean residual ellipticity units (B) Ther-mal denaturation curves for wild-type RepA (solid lines) and C160– RepA unlabeled (dashed line) and AEDANS-labeled (dash-dots) The temperature dependence of the ellipticity at 220 nm is shown, nor-malized to help compare the different proteins (C) Fluorescence emission spectra (k ex = 295 nm) of wild-type RepA (solid line), C160–RepA both unlabeled (dashed line) and AEDANS labeled (dash-dots), recorded at 23.5 C with 2 l M protein and with intensities normalized with respect to their emission maximum at
327 nm (D) Binding of wild-type RepA ( ) and AEDANS C160– RepA (s) to 10 nm Alexa568-labeled 1IR, monitored by Alexa568 fluorescence anisotropy (k ex = 535 nm, k em = 605 nm) Data for both proteins were fitted (see Eqns 3 and 5) together (solid line) to
a 2 : 1 RepA : 1IR binding equilibrium using the quadratic equation This yielded Kd= 5 ± 2 nm , compatible with previous reports [3] Experiments were carried out in 0.15 M (NH 4 ) 2 SO 4 , 15 m M NH 4 -acetate, 0.03 m M EDTA, 3% glycerol; pH 6.0.
Trang 4Binding of C160–RepA to operator and iteron
sequences followed by AEDANS fluorescence
The fluorescence of AEDANS–C160 was studied as a
function of DNA concentration for the operator and
two distinct iteron sequences (described in Table 1)
RepA binding to 1IR and 1DR has been studied in
detail previously [3,6] When increasing amounts of
1IR, 1DR or 1DR-short are added to AEDANS
C160–RepA, no effect is seen on the shape or intensity
of the ‘pure’ AEDANS fluorescence, i.e the emission
spectrum excited at 375 nm (not shown) There is,
however, a clear increase in the fluorescence anisotropy
for each of the sequences (Fig 3D–F), indicating a
decrease in the rotational mobility of AEDANS C160– RepA The anisotropy increase is slightly different for each of the three sequences, and relates to an increased global rotational correlation time for the AEDANS probe caused by C160–RepA binding to DNA (see below) Addition of DNA also induces a change in the shape of the excitation spectra This is caused by a decrease in the Trp contribution to AEDANS fluores-cence, as illustrated by the difference spectra between free and bound RepA, which are typical of Trp (Fig 3A–C)
The increase in directly excited AEDANS anisotropy matches very well with the decrease in W94 contribu-tion to AEDANS fluorescence for each of the three
Table 1 Sequence of the oligonucleotides used in this study IR (operator, half sites in bold), 1DR (single iteron underlined, with the half site also present in the operator in bold, purported DnaA box dashed underlined), 1DR-short (single iteron underlined, with the half site also present in the operator in bold).
A D
B E
C F
Fig 3 Excitation spectra (kem= 480 nm) of AEDANS C160–RepA with increasing con-centrations of 1IR, 1DR and 1DR-short (A, B and C, respectively), causing changes in the direction of the arrows The spectra are inner filter corrected and normalized to the intensity at 340 nm Difference spectra between free and DNA-bound RepA are shown as dashed lines RepA was 1.25 l M
and 0, 0.2, 0.4, 0.6 and 1 l M 1IR (A), 0, 0.4, 1.2, 2.4 and 4 l M 1DR (B), and 0, 0.8, 1.8, 3.2 and 6 l M 1DR-short (C) (D) Fluores-cence intensity (k ex = 280 nm,
kem= 480 nm), corrected and normalized as
in (A) ( ), and AEDANS fluorescence anisot-ropy (k ex = 375 nm, k em = 480 nm) of AEDANS C160–RepA as a function of [1IR] (s) Data were fit using the quadratic bind-ing equation (see Eqns 3–4) (E) and (F) as (D), except they refer to titrations with 1DR and 1DR-short, respectively Experiments were performed at 23.5 C, in 0.15 M
(NH 4 ) 2 SO 4 , 15 m M NH 4 -acetate, 0.03 m M
EDTA, 3% glycerol, pH 6.0.
Trang 5tested oligonucleotides (Fig 3D–F) The change in
flu-orescence and anisotropy were fit simultaneously for
each titration In the fits, the protein concentration
was left free, to serve as an indicator of stoichiometry
In the case of 1IR, the fit resulted in a binding
stoichi-ometry of 2 : 1, i.e dimer binding The reactant
con-centrations were too high to obtain relevant
information on the binding affinity For binding to
1DR, the best fit yielded a binding stoichiometry of
1 : 1, i.e monomer binding, with a Kd between 0.2
and 0.6 lm With 1DR-short, a reliable estimate for
the stoichiometry of binding could not be made
Assuming binding as monomer or as dimer,
respec-tively, the dissociation constants obtained were
2.1 ± 0.2 and 2.9 ± 0.2 lm, without an apparent
dif-ference in goodness of fit However, in a separate
experiment involving inter-monomeric homoFRET
(see below) the binding stoichiometry was confirmed as
dimeric RepA to the 1IR and IDR-short sequences,
and monomeric RepA to 1DR The binding affinity
under these conditions is thus 2.9 lm
FRET between Trp94 and the AEDANS
As mentioned above, DNA binding induces an
appar-ent decrease in FRET efficiency between W94 and
AEDANS–C160 Along with this decrease, there is
also a considerable degree of quenching of W94
fluo-rescence This residue has a relatively high quantum
yield for Trp [13] that is strongly quenched upon
bind-ing to its cognate DNA sequences (see Table 2) This
quenching is unrelated to FRET, as it also occurs with
unlabeled RepA Furthermore, it does not decrease the
lifetime of W94 fluorescence significantly (see Table S1),
indicating that it is static in nature We do not have
an unequivocal interpretation of the origin of the static quenching However, judging from the binding stoichi-ometry together with the shape of the binding curves (Fig 3), it is safe to conclude that the quenching does not affect the RepA–DNA binding equilibria, and thus that dark state(s) of W94 are present in the RepA– DNA complexes Because the fraction of non-fluores-cent donor molecules does not contribute to the Trp fi AEDANS energy transfer process, a correc-tion of the FRET efficiencies for the presence of non-fluorescent W94 is required (see Eqn 1, Experimental procedures) After doing so, it appears that the differ-ence in FRET efficiency between free RepA and its DNA complexes is actually relatively minor (see Table 2) Accordingly, the resulting distance calculated between W94 and AEDANS–C160 does not display large variations between the different species
However, there are a number of caveats that should
be taken into account First, there are several tyrosine residues in RepA As the fluorescence was excited at
280 nm, there is the possibility that some of the five tyrosines present in the W94-containing N-terminal domain of RepA also contribute to the experimental FRET efficiency, by Tyr fi Trp energy transfer As the distance between W94 and the nearest Tyr residue
is 15 A˚ [5], this contribution is negligible, however This conclusion is well supported by the apparent lack
of contribution of Tyr to the excitation spectrum of acceptor AEDANS indicated in the excitation differ-ence spectra seen in Fig 3A–C Second, the Fo¨rster and donor-acceptor distances determined here, relate
to the R0 value determined assuming hj2i = 2 ⁄ 3,
R0(2/3) (see Experimental procedures) This value was calculated to be 25 ± 1 A˚ The value of hj2i is not known exactly, leading to additional uncertainty The maximum and minimum limits of the value ofhj2i for the W94⁄ AEDANS–C160 couple in RepA were esti-mated as described previously [14,15], from the depo-larization factors determined from time-resolved fluorescence anisotropy recorded for wild-type RepA W94 and AEDANS C160–RepA (see below, and Table S1) It appears that the factor hj2i for RepA– DNA complexes would have a value between 0.06 and 3.51, which in turn yields an uncertainty in the abso-lute distance between 0.7 and 1.3 times the value of R(2/3), presented in Table 2
Nevertheless, the R(2/3) value in the complex with 1DR is in excellent agreement with the distance mea-sured between the Cb atoms of both residues in the structural model of RepA [2] based on the monomer– iteron complex structure of the homologous RepE pro-tein [7] W94 and C160 are each located on one of the
Table 2 Fluorescence and FRET parameters of the W94–
AEDANS–C160 pair and resulting average inter-probe distances, in
free RepA and RepA bound to various cognate DNA sequences.
FRET efficiency was determined using Eqn (1), and assuming
e W 94
280 nm
e AEDANS
340 nm = 1 and e AEDANS
280 nm
e AEDANS
340 nm = 0.17 (see Experimental procedures) The apparent quantum yield of W94 (F W94
) was deter-mined both for wild-type RepA and unlabeled C160–RepA The
degree of quenching, i.e the ratio of F W94 in free and DNA-bound
RepA was used to determine the fraction of fluorescent donor
(d + in Eqn 1).
Species
F W94
(± 0.02)
d +
(± 0.08)
FRET efficiency (± 0.15)
R( 2 / 3 ) (A ˚ ) (± 7) b
a
Values based on extrapolations from binding curves and as such
not experimentally confirmed b Using R0( 2 / 3 ) = 25 ± 1 A ˚
Trang 6two different domains of RepA, and therefore changes
in distance between both residues can be interpreted in
terms of domain movements Because no relevant
change is observed, it can be concluded that no
signifi-cant reorientation takes place between the two domains
of RepA upon binding to the operator or iteron DNA
or as a result of the monomerization of RepA that
accompanies binding to 1DR We can not currently
exclude a rotation centered about C160, as this will also
not affect the distance between both residues Also,
note that, in theory, inter-monomeric FRET may occur
in the case of RepA dimers This is unlikely however,
considering the distance between both W94 residues
( 20 A˚) and that both DNA binding domains
con-taining the AEDANS probes are located roughly on
opposing ends of the dimerization domains [5]
Time-resolved fluorescence depolarization and
rotational correlation times of RepA and its DNA
complexes
Time-resolved depolarization measurements were
per-formed to obtain information on global and local
dynamics of the AEDANS and W94 probes in free and DNA-bound RepA The decay of the total fluores-cence intensity was recorded, as well as the decays of its vertically and horizontally polarized components The anisotropy decay of the fluorophore can be described in terms of its slow and fast components, i.e
of global and local re-orientational motions, respec-tively This was carried out for both W94 in wild-type RepA and AEDANS-labeled C160–RepA AEDANS has a much longer fluorescence lifetime than Trp, allowing a much greater level of confidence in the determination of correlation times pertaining to the global rotational motion Nevertheless, the global rota-tional information obtained from Trp fluorescence anisotropy decays (see Fig S1 and Table S1) shows a trend in agreement with the data from the AEDANS experiments Furthermore, despite the relatively poor photon-counting statistics, the local dynamics of W94 have been characterized from the Trp decays In Fig 4, anisotropy decays (kem= 480 nm) are shown for the different AEDANS C160–RepA species, together with best fits assuming a bi-exponential function for r(t) (see Experimental procedures) The
A B
C D
Fig 4 Anisotropy decays R(t) (k ex = 375 nm, k em = 480 nm) of AEDANS C160–RepA free in solution (A) and bound
to 1IR (B), 1DR (C) and 1DR-short (D) Experiments were performed at 23.5 C in 0.15 M (NH4)2SO4, 15 m M NH4-acetate, 0.03 m M EDTA, 3% glycerol, pH 6.0 Experi-mental data (s) were reconstructed from the fluorescence decays that were polarized parallel and perpendicular to the polarization plane of the excitation beam, after subtract-ing their respective dark counts Fits to the data are shown as solid gray lines AEDANS C160–RepA was 2 l M in each experiment and with 2.5 l M 1IR, 8 l M 1DR and 12 l M
1DR-short, respectively Weighted residuals for the fits between experimental and calcu-lated R(t) are shown below the curves.
Trang 7analogous decays with kem= 530 nm, with
corre-sponding best fits and tabulated parameters, are
sup-plied in Fig S2 and Table S1
The AEDANS data confirm the presence of discrete
complexes under the conditions of the experiment,
and that binding is complete, in agreement with the
binding curves (Fig 3), except for the case of the
complex with 1DR-short, which under these
condi-tions should contain 20% free RepA As expected,
the global rotational correlation time, /2, increases
upon binding of RepA to its cognate DNA Apart
from the RepA–1DR-short complex, the observed
val-ues easily fall within the range reasonably expected
from molecules of this size and shape (Table 3) The
expected values for free RepA and the dimeric RepA–
1IR complex were calculated on basis of
hydro-dynamic shapes and volumes corresponding to prior
[3] sedimentation velocity measurements as shown in
Fig 5 Both can be characterized as rigid elongated
shapes For the monomer RepA–1DR complex, the
structure modeled on the homologous mRepE–DNA
crystal structure [7] was used directly to calculate the
expected global rotational correlation time The
calcu-lated values for the RepA–1DR-short complex pertain
to a monomer, i.e the modeled structure as above,
but with a truncated oligonucleotide having 30 bp
instead of the 45 bp of 1DR This purported complex
of monomeric RepA with 1DR-short is not shown,
but it is easily imagined that this complex is quite
spherical and that it should have a relatively short
global rotational correlation time This is clearly not
what is observed Note that because the orientation of
the AEDANS probe in the complex is not known, we
provide a range of calculated values, i.e the minimum
and maximum of the correlation times corresponding
to the complex (see Experimental procedures)
Never-theless, even given this significant uncertainty, the
measured value of the complex with 1DR-short clearly
exceeds the maximum value that was calculated for a hypothetical complex involving RepA monomer By contrast, a correlation time around 100 ns fits well with a complex involving dimeric RepA and a 30 bp oligonucleotide It should further be noted that the presence of 20% free RepA in the case of the 1DR-short complex will lead to a slight underestimation of the rotational correlation time There appears to be linear correlation between oligonucleotide size (zero for free RepA) and experimental correlation time for the complexes involving dimeric RepA (Table 3) Only the complex between 1DR and RepA does not fit this
Table 3 Fluorescence lifetimes, time-resolved and steady-state anisotropy parameters for AEDANS–C160 in free RepA and RepA bound to various cognate DNA sequences.a
Sample
hr ss i
± 0.002
hsi c
(ns)
± 0.4
b 1
± 0.05
/ 1 (ns)
± 3
b 2
±0.05
/ 2 (ns)
± 10
/ 2 calc (ns) (max–min)
a Estimated errors at the 67% confidence level [30] b Steady-state anisotropy from fits to the data in Fig 3 c k ex = 375 nm, k em = 480 nm;
r 0 (from the fits) = 0.31 ± 0.015); T = 23.5 C d
Minimum and maximum calculated rotational correlation times assuming a prolate ellipsoid shape, and using shape factors from frictional ratios previously [3] determined using sedimentation velocity experiments e Minimum and maximum calculated rotational correlation times calculated using the HYDROPRO program [17] using as input homology models of the RepA– 1DR and 1DR-short complexes, respectively, based on the crystal structure [7] of monomeric RepE in complex with iteron DNA.
Fig 5 Prolate ellipsoids equivalent to (non-hydrated) free RepA (upper) and RepA–1IR complex (lower), with axial ratio and volumes corresponding to frictional ratios determined from prior sedimenta-tion velocity analysis (3) and molecular mass (23) respectively The modeled structure of monomeric RepA–1DR is shown in two orien-tations (center) For clarity, the purported structure of RepA mono-mer with 1DR-short is not shown The length of 1DR-short only allows for five nucleotides (half a helical turn) to protrude from either end of the protein–DNA interface.
Trang 8correlation, in line with the fact that it is the only
complex involving monomeric RepA
Finally, we note that the range of global rotational
correlation times calculated for the dimer–operator
complex of the F plasmid RepE protein, which is
highly homologous to RepA and of which the crystal
structure is known [9], is shorter (57–83 nucleotides)
than observed here for the RepA–1IR complex This
could mean that there are significant differences
between the RepE– and RepA–operator complexes,
which are possibly related to the different spacing
between the half repeats in both operator DNA
sequences [9]
Oligomerization state of free and complexed
RepA determined by homoFRET
In order to understand the oligomerization state of
RepA in the different DNA complexes better,
homo-FRET experiments were carried out Herein, use is
made of C160–RepA specifically labeled with Atto532
In a double-labeled sample, FRET is expected to occur
between the two Atto532 moieties whenever the
inter-probe distance is not greater than 1.5 times the
Fo¨rster distance The calculated R0(2/3) = 55 A˚ for
Atto532–Atto532 homoFRET, and thus energy
trans-fer is expected to occur in double-labeled RepA
dimers Thus, no FRET is expected when RepA is
monomeric, or in single-labeled Atto532–RepA dimers
HomoFRET between the fluorophores is detected
through depolarization of their emission, but note that
this occurs only if they do not happen to be aligned
more-or-less parallel to each other in the dimer
C160–RepA samples labeled with 60% Atto532 (i.e
with 43% of Atto532 residing on double-labeled RepA
dimers) show clearly different excitation anisotropy
spectra from C160–RepA samples labeled with only
10% Atto532, i.e with very little double-labeled RepA
dimers (< 5%) This is shown in Fig 6A, where there
is an evident decrease in anisotropy for the sample
containing the double-labeled C160–RepA dimers,
which is less pronounced at longer excitation
wave-lengths (red-edge excitation) The enhanced
fluores-cence depolarization in the double-labeled dimers with
respect to the single-labeled samples is a clear
indica-tion of homoFRET in the double-labeled samples [16]
The increase in steady-state anisotropy observed upon
decreasing the degree of Atto532-labeling from 60% to
10% is also observed when excess 1DR is added to
60% labeled Atto532 C160–RepA, but not upon the
addition of excess 1IR and 1DR-short (Fig 6B) This
means that addition of 1DR abolishes the homoFRET,
by inducing RepA monomerization In fact, the
addi-tion of 1IR and 1DR-short causes a small decrease in anisotropy which may be related to enhanced homo-FRET caused by slight rearrangement of the mono-mers in the RepA dimono-mers or by minor aggregation Thus, RepA is dimeric free in solution and when bound to its operator sequence, but also when bound
to 1DR-short In the presence of excess 1DR, mono-merization of RepA takes place
Discussion
One of the striking properties of RepA is that it is able
to recognize two types of DNA sequence, either the operator – with inverted repeats – or the iteron, in which the same 6 bp sequence half-site found in the operator is specifically recognized Upon binding to the operator, RepA remains dimeric; it thus retains its symmetry matching the inverted repeats of the oligo-nucleotide When this symmetry is absent, i.e for the
A
B
Fig 6 (A) Excitation anisotropy spectra of Atto532–C160 RepA labeled to different degrees (solid line: 60% label, dashed line: 10%) [RepA] is 0.5 l M in either case, and conditions are 0.5 M
(NH 4 ) 2 SO 4 , 50 m M NH 4 -acetate pH 6.0, 30 l M EDTA, 10% glycerol,
T = 6 C (B) Average changes in steady state fluorescence anisot-ropy between 60% Atto532–C160 RepA and, from left to right, 10% Atto532–C160 RepA, 60% Atto532–C160 RepA in the pres-ence of 2 l M 1IR, 1–4 l M 1DR-long and 8–16 l M 1DR-short Condi-tions: 0.15 M (NH4)2SO4, 15 m M NH4-acetate pH 6.0, 10 l M EDTA, 3% glycerol, T = 6 C In the experiments with DNA, [RepA] = 15 nm
Trang 9iteron sequence, RepA binds as a monomer instead of
a dimer
When operator or iteron DNA is added to
AE-DANS C160–RepA, discrete complexes are formed
(Fig 3), characterized by higher AEDANS
fluores-cence anisotropy values and decreased apparent
Trp-AEDANS FRET (see below) RepA binds operator
DNA (1IR) with a clear stoichiometry of 2 : 1, i.e the
protein binds as a dimer With the iteron sequence
1DR, which includes an additional stretch of bases
(see Table 1), a stoichiometry of 1 : 1 is found, i.e
monomer binding When the number of bases flanking
the iteron sequence is considerably shorter, as with
1DR-short, the binding affinity is significantly
decreased (2.9 lm), and nears that of non-specific
DNA [6] Still, a discrete complex is formed in this
case, as corroborated by fluorescence anisotropy decay
measurements
Fluorescence anisotropy decay analysis is a potent
method to obtain information on the local and global
dynamics of species in solution Here, it is used to
characterize the discrete species discussed above For
free RepA and RepA in complex with 1IR or 1DR,
experiments were performed with AEDANS The
anal-ysis, summarized in Table 3, yields global rotational
correlation times for free RepA and the complex with
1IR corresponding to species involving dimeric RepA,
as expected In the case of the complex with 1DR, a
fair correlation is also found between the experimental
and calculated global rotational correlation times For
the latter, the hydropro program was employed,
which is able to extract hydrodynamic parameters
using the protein’s atomic co-ordinates [17] A
homo-logy model based on the RepE–iteron structure was
used as input Note that the bending angle of the 1DR
as observed by EMSA (52) is significantly larger than
in the crystal structure (20) which was used for the
homology model [6,7] Furthermore, the crystal
stru-cture has a much shorter DNA oligonucleotide than
the 1DR sequence: the latter is 3–4 times longer
than the protein itself and may thus form a source of
significant flexibility, difficult to account for in model
building
However, using the same structure as a basis to
construct a potential complex between 1DR-short and
monomeric RepA is not realistic The observed global
rotational correlation time for the RepA–1DR-short
complex cannot conceivably be justified assuming a
complex similar to the RepE–iteron complex
How-ever, the purported dimeric RepA–1DR-short complex
fits very well into the linear correlation between
oligo-nucleotide size and experimental correlation time for
the complexes involving dimeric RepA The complex
between 1DR and RepA does not fit this correlation,
in line with the fact that it involves monomeric RepA
It is thus tempting to assume that dimeric RepA is actually involved in binding the 1DR-short sequence, despite the fact that it contains the full 22 bp iteron This last conclusion is corroborated by the observa-tion that inter-monomeric homoFRET is observable with 1DR-short, but not 1DR (Fig 6) That dimer-binding to iterons is, in principle, possible has previ-ously been shown by us According to EMSAs carried out under crowded conditions, a fraction of RepA dimers was observed to bind to the 1DR sequence [6] This fraction is obviously much larger in the case of 1DR-short, and the extra bases on the longer, mono-mer-binding, oligonucleotide 1DR seem to play a role
in aiding monomerization The presence of bases downstream of the iteron sequence has also previously been shown to promote binding of Rep to pSC101 [18]
The related replication initiator protein p from R6K plasmid is a well-documented case where not only monomers, but also dimers, are known to bind to the iteron [19] Interestingly, dimers of p protein occupy a much shorter stretch of the iteron sequence than do monomers; whereas almost the entire 22 bp iteron sequence is occupied by the p monomers, only half of this – notably including the specific 6 bp recognition sequence (repeat) – is occupied when dimeric p protein
is bound [19] This may occur here as well As there is only one half of the inverted repeat of the operator sequence present in 1DR-short, it is likely that only one of two WH2 DNA-binding domains in RepA dimers is involved in binding This also makes sense energetically, the RepA dimer binds operator DNA with Kd= 0.7 nm i.e DG =)21.2 kJÆmol)1 [6] Sub-tracting from this a penalty of 7.8 kJÆmol)1 for the DNA bending [20] induced by dimer binding (61), a free energy of ()21.2 to 7.8) ⁄ 2 = )14.5 kJÆmol)1 is expected for binding of a single DNA-binding domain without the need to force DNA bending This trans-lates to Kd= 1.3 lm, which is reasonably close to the value of 2.9 lm observed here for 1DR-short
It is clear that in vitro, the effect of decreased length
of the iteron flanking sequence is to weaken the iteron-binding affinity of monomeric RepA so that, at high [RepA], dimer binding occurs In vivo, this effect may
be comparable in the sense that monomer-binding affinity is attenuated by the length or identity of the flanking sequence It is well established that Rep pro-tein dimers do not act as initiators in plasmid replica-tion [21] A positive effect on monomer-binding affinity thus provides a way of selecting against dimer binding, favoring monomer binding and thus
Trang 10initiation It should be mentioned that the four iterons
in pPS10 are contiguous, thus limiting the degree to
which the flanking sequences may contribute to
bind-ing In other replicons, however, there are spacer
sequences between the iterons, which in addition may
have some sequence variability [22] It would be
inter-esting to see whether our findings for RepA can be
extrapolated to other Rep proteins
An attractive feature of using AEDANS as an
extrinsic label is that, besides its use to analyze the
global rotational correlation times of
macromole-cules, it is useful as a FRET acceptor for intrinsic
Trp residues RepA fortunately has only one Trp,
making this use of the AEDANS probe more
mean-ingful and helping interpretation of the FRET in
terms of distances between the two fluorophores
Moreover, W94 and C160 are located on the
dimer-ization and DNA-binding domains of RepA,
respec-tively, allowing us to interpret any observed changes
in FRET in terms of relative movements between
the two domains
It emerges that the average estimated distance
between the C160–AEDANS and W94 is 22 A˚ in
the free RepA dimer, and this distance decreases by a
few angstroms upon binding either the 1IR, or
1DR-short oligonucleotides and increases slightly upon
monomerization and binding to 1DR (see Table 2)
The average distance observed in the complex with
1DR is in very good agreement with the value
mea-sured between the Cbatoms of residues W94 and C160
in the homology model of RepA, supporting the
esti-mated value It is interesting that within the error, the
distance between the AEDANS and indole moieties
apparently does not change significantly between
unbound RepA and RepA bound to either 1IR (as a
dimer with both DNA-binding domains involved in
binding), or 1DR-short (as a dimer, but presumably
with only one domain involved), or indeed when
bound as a monomer to 1DR This suggests that
bind-ing to both inverted half-repeats, as in the operator
sequence, does not trigger large conformational
rear-rangements with respect to the free dimeric protein or
to the dimer purportedly bound via one DNA-binding
domain (1DR-short) Although significant structural
rearrangements of RepA occur upon monomerization
[3–5], these do not appear to grossly alter the relative
orientation of the two domains with respect to each
other Naturally, it should be noted that manifold
rela-tive orientations of the two domains may exist,
satisfy-ing the observed distance, but which are still
significantly different We are currently working
towards a more comprehensive understanding of
inter-domain orientations using FRET
Experimental procedures
Cloning, expression and purification of wild-type RepA and C160–RepA
In all cases, the concentration of protein is expressed in monomer units What is referred to as wild-type RepA is the His6-tagged variant of RepA, which was expressed and purified as described previously [4] This protein is indistin-guishable from that without His-tag, except that it has a higher solubility [3,4] It was therefore used without sub-sequent removal of the tag C160–RepA also has the His6-tag and is a single-cysteine variant of wild-type RepA
in which two of the three wild-type Cys residues (C29, C106) have been successively replaced by Ser using the PCR-based QuickChange Kit (Stratagene, Cedar Creek, CA, USA) Mutations were verified by sequencing C160–RepA was expressed as wild-type RepA [4] but almost all C160–RepA was present in the form of insoluble aggregates The protein was isolated by solubilization of the inclusion bodies and purification by Ni(II)-affinity chromatography under dena-turing conditions, similarly as described previously [4] This results in pure protein, exhibiting a single band on SDS⁄ PAGE After purification, the protein was reduced by addition of 2 mm 2-mercaptoethanol and exchanged to unfolding buffer (5.6 m guanidinium hydrochloride, 0.56 m (NH4)2SO4, 0.2 m NH4-acetate, 0.2 mm EDTA, 1.2% Chaps, pH 6.0) Immediate refolding is achieved by fast 20-fold dilution in 0.15 m (NH4)2SO4, 15 mm NH4-acetate, 0.03 mm EDTA, 3% glycerol, pH 6.0 A small amount of precipitate generated by the refolding procedure was spun down at 14 000 g for 20 min The latter buffer was used both for storage ()80 C) and experiments
Protein labeling C160–RepA was labeled with IAEDANS (Molecular Probes, Leiden, The Netherlands) under denaturing condi-tions, as follows C160–RepA was concentrated to
150 lm in unfolding buffer by ultrafiltration (10 kDa cut-off) A small amount of 1 m Tris⁄ HCl (pH 8.5) was added to increase the pH to 7.2, and Tris(2-carboxyethyl) phosphine to keep the single Cys reduced (1 mm) The end volume was 1.6 mL IAEDANS was dissolved (40 mm) in unfolding buffer and quickly mixed with the reduced pro-tein to a final concentration of 2 mm The reaction was allowed to proceed for 2 h at room temperature, and then 12.3 mg of glutathione was added to quench the reaction The reaction mixture was exchanged for fresh unfolding buffer by extensive ultrafiltration The labeling efficiency was close to 50%, as judged from UV⁄ Vis spectroscopy AEDANS C160–RepA was refolded in the same way as unlabeled RepA Similarly, C160–RepA was labeled with maleimide Atto532 (Atto-Tec, Siegen, Germany), with 60% labeling efficiency Here, the degree of labeling in the folded