A
B
Figure 2. Dependencies of
logK
I
on the length of inhibitor (
n) for ss and ds
deoxyribooligonucleotides (a) and ribooligonucleotides (b). (a) d(pT)
n
(crosses), d(pA)
n
(filled squares), d(pC)
n
(open circles), d(pG)
n
(triangles),
d[(pF)
n
pT] (open squares), d(pT)
n
d(pA)
n
(filled circles). (b) (pU)
n
(crosses), (pC)
n
(open circles), (pA)
n
(filled squares), (pU)
n
(pA)
n
(closed
circles); the curves for d(pA)
n
(crosses) and d(pT)
n
d(pA)
n
(diamonds) are
given for comparison.
Figure 3. Logarithmic dependencies of factor f for APE1 on the relative
hydrophobicity of nucleotide bases of homo-d(pN)
n
estimated from isocratic
reverse phase chromatography of different nucleosides according to ref. (46).
Extrapolation of the curve to zero hydrophobicity corresponding to
orthophosphate gives an electrostatic factor
e
= 1.51.
Nucleic Acids Research, 2004, Vol. 32, No. 17
5139
Downloaded from https://academic.oup.com/nar/article/32/17/5134/1334164 by guest on 16 February 2023
9 out of 10 links of d[(pF)
9
pT] interact with APE1 with
virtually the same efficiency as the deoxyribose phosphate
structural elements of the backbone of nonspecific d(pN)
n
.
A 1.53–1.66-fold change in affinity on d(pN)
n
elongation by
one nucleotide unit (corresponding to a change in
DG
of
0.26 to 0.31 kcal/mol) is lower than would be expected
for strong electrostatic contacts (up to
1.0 kcal/mol)
or hydrogen bonds (
2 to 6 kcal/mol) (41), but is compar-
able with the values for weak hydrophobic, ion–dipole and
dipole–dipole interactions (41). The crystal structure of human
APE1 reveals that the enzyme possesses a preexisting posi-
tively charged surface for DNA binding and inserts loops into
both grooves of DNA (14,15). This strip of positive potential
probably underlies APE1 interactions with internucleotide
phosphate groups of specific and nonspecific DNA. Negatively
charged internucleotide groups of nonspecific ODNs could
interact with the DNA-binding groove of APE1 through dipo-
lar electrostatic forces rather than electrostatic interactions
between point charges. Thus, the interaction between APE1
and DNA may resemble interaction between surfaces of oppo-
site charge (6,7). It seems reasonable that APE1 could use a
specific distribution of charged and neutral amino acid resi-
dues in the DNA-binding site for interactions with internucleo-
tide phosphates and nucleobases, respectively. Therefore, in
addition to weak hydrophobic or van der Waals interactions
between DNA bases and amino acids of inserted protein loops,
transition of DNA bases from water to an even slightly more
hydrophobic environment of DNA-binding subsites can also
lead to a favorable gain in energy during complex formation.
The increase in APE1 affinity for DNA per base by a factor of
1.01–1.1 (
DG
= 0.26 to 0.31 kcal/mol) is comparable with
a gain in energy upon transfer of nucleobases from water to
1–3 M aqueous methanol (6,7).
From the data discussed it can be concluded that the sugar–
phosphate moiety of dNMPs [or each nucleotide of d(pN)
n
]
interacts with the active center of APE1 through relatively
strong nonspecific contacts with their phosphate groups and
significantly weaker contacts with bases. ODNs containing
two or more nucleotides can form several thermodynamically
comparable microscopic complexes with APE1; the number of
such complexes increases with increasing ODN length when
n < 5 and decreases when n > 6, until d(pN)
10
, which can
form only one complex with the enzyme (Figure 4). All
interactions of APE1 with the nucleotide units of ODNs,
except one unit that presumably fits directly into the active
center, are weak and additive.
Additive interaction of APE1 with nucleotide units of
ribooligonucleotides
According to structural data, APE1 introduces a kink into the
helix of specific DNA (15). Structural characteristics of RNA
and DNA differ markedly in solution: ds RNA usually exists in
the A form and ds DNA in the B form, while ss r(pN)
n
adopt
much more rigid nonflexible structures as compared with
d(pN)
n
(47). Therefore, it was interesting to compare APE1
interaction with d(pN)
n
and r(pN)
n
(Tables 1 and 3). The
affinity of the APE1 active center for AMP (373
mM) and
CMP (447
mM) was 2–2.7-fold lower than that for dNMPs
(163–165
mM), while the affinity for UMP (1873 mM) was
11.4-fold lower. The log-dependencies for (pA)
n
and (pU)
n
were linear for 1 <
n < 8–9, and only for (pC)
n
was the curve
linear up to
n
= 10 (Figure 2B). The values of f factors for
d(pC)
n
(1.53), d(pT)
n
(1.58) and d(pA)
n
(1.66) are slightly
higher compared with those for the respective ribo-
oligonucleotides (pC)
n
(1.29), (pU)
n
(1.38) and (pA)
n
(1.40). Thus, not only can the active site of APE1 distinguish
between conformationally different ribonucleotides and deoxy-
ribonucleotides, but other subsites of the enzyme can interact
with 9 out of 10 nucleotides less efficiently as well. At the level
of decanucleotides, the difference between interaction of
APE1 with 9 nt units of d(pN)
10
and r(pN)
10
was estimated
as factors of 7.1, 5.3 and 4.9 for dA/rA, dT/rU and dC/rC,
respectively. Overall, d(pN)
10
and r(pN)
10
interact with APE1
due to superposition of the same nonspecific interactions with
internucleotide phosphates and bases. Most probably, (pN)
n
cannot be kinked by APE1 in the same way as d(pN)
n
, result-
ing in their lower affinity. Interestingly, the log-dependencies
are linear up to 10 residues only for (pC)
n
, which, of all r(pN)
n
,
possesses the highest conformational flexibility (47), while
the affinity of (pA)
n
and (pU)
n
increases only up to
n
= 8
(Figure 2B). One possible explanation is that the two terminal
nucleotides of (pA)
n
and (pU)
n
accommodated in the DNA-
binding cleft of APE1 could lie far away from its positively
charged region. Thus, the increased affinity for d(pN)
n
could
stem from bringing the oppositely charged surfaces of APE1
and ODNs closer together as a result of easier conformational
changes in the DNA and APE1 structures.
Figure 4. Schematic structure of DNA binding site of APE1. The DNA-binding
site of the enzyme consists of two sets of ten subsites each, but only one set of
subsites interacting with the cleaved strand, shown in the figure, contains a
specific subsite (‘0’ subsite) with increased affinity for one specific or
nonspecific nucleotide unit of DNA. Lengthening of nonspecific d(pN)
n
(1 <
n < 10) leads to the formation of several alternative thermo-
dynamically comparable complexes of these ODNs with different subsites
on the enzyme.
5140
Nucleic Acids Research, 2004, Vol. 32, No. 17
Downloaded from https://academic.oup.com/nar/article/32/17/5134/1334164 by guest on 16 February 2023
Affinity of APE1 for nonspecific DNA duplexes
Some enzymes, such as UDG, partially melt ds d(pN)
10
and
contact both strands of this relatively short ODN almost inde-
pendently (28). In contrast, DNA polymerases and Topo I
interact with both base-paired DNA strands (24–26,32,33).
However, the contribution of the second strand to the affinity
of any enzyme for ds DNA is usually much smaller than that of
the first strand. A remarkable feature of the behavior of Topo I
and DNA polymerases is the ‘assembly’ and subsequent
stabilization of correct duplexes for which the melting
temperature (
T
m
) in solution is substantially lower than the
reaction temperature (6,7,24–26,32,33).
Figure 2 and Tables 1 and 3 show that the minimal ligand
exhibiting duplex properties toward APE1 is d(pT)
6–8
d(pA)
6–8
, and for an octamer duplex the
T
m
in solution
[21
C, calculated according to ref. (48)] is lower than the
reaction temperature (37
C). Shorter duplexes with
T
m
signif-
icantly lower than the reaction temperature behave as ss ODNs
and not as duplexes under the present reaction conditions
(Tables 1 and 3). Thus, short duplexes are weakly stabilized
by their interaction with APE1. Similar to ss ODNs, a linear
increase in log
K
I
for duplexes was found up to
n
= 10. The
affinity of APE1 for d(pT)
n
d(pA)
n
is
5-fold higher than that
for ss d(pA)
n
. The change in APE1 affinity for d(pT)
n
d(pA)
n
(
n > 6) is described formally by the same equation as for ss
ODN (see above), but the factor
f increases from 1.58 [for
d(pT)
n
] or 1.66 [for d(pA)
n
] to 2.44. The ratio of these
f factors
(characterizing an increase in affinity due to the addition of a
single unit of the second strand) is 1.47–1.54 (
DG
= 0.23 to
0.26 kcal/mol). Note that the formation of a single A:T or
G:C pair in solution is characterized by
DG
values of
1.2 to
1.9 kcal/mol and 2.0 to 2.8 kcal/mol, respectively (6,7).
Interestingly, the affinity of an ORN duplex (pA)
10
(pU)
10
is
only 2.7-fold higher than that for ss (pA)
10
(Table 3), whereas
the ratio of
K
I
values for d(pA)
10
and d(pA)
10
d(pT)
10
is 5
(Table 1). The addition of d(pT)
n
to a complementary (pA)
n
strand does not lead to an increase in the affinity of the mixed
d(pT)
10
(pA)
10
duplex (
K
I
= 2.5 mM) compared with that for
d(pT)
10
(
K
I
= 2.5 mM) (Tables 1 and 3; Figure 2B). Thus, the
contribution of the second strand is much lower than that of
the first strand. In addition, APE1 seems to be unable to distort
the solution structure of the RNA–RNA and RNA–DNA
duplexes.
Thermodynamic model of APE1 interaction with
nonspecific DNA
The contribution of interactions of any unit of nonspecific
d(pN)
10
(
K
I
= 163–167 mM, DG
= 5.2 kcal/mol) with
APE1 does not depend on the particular base (Table 1).
The relative contribution of a phosphate group can be approxi-
mately estimated (
DG
= 4.77 kcal/mol) from the K
I
value
for orthophosphate (360
mM). Thus, the contributions of the
nucleoside moiety of any dNMP unit of an ODN can be esti-
mated from the difference in
DG
for the nucleotides and
orthophosphate as
0.43 kcal/mol. Since nine d(pA) nucleo-
tide units of one strand of ds d(pA)
10
interact with APE1
through weak additive contacts (
f
= 1.66; K
d
= 0.6 M;
DG
= 0.31 kcal/mol), the net relative contribution
of these nine nucleotides may be estimated as
DG
=
2.76 kcal/mol, DG
of the nine internucleotide phosphates as
2.23 kcal/mol and that of the nine bases as 0.53 kcal/mol.
Thus, all contacts of APE1 with the poly(dA) strand inter-
acting with the enzyme’s DNA binding groove provide
DG
of
7.96 kcal/mol. From the ratio of K
d
values (equal to
5, or
K
d
= 0.2 M), characterizing the increased affinity for ds
d(pA)
10–16
d(pT)
10–16
compared with d(pA)
10–16
, the contribu-
tion of the 10 nt units of the second strand to the affinity of ds
DNA may be estimated as
DG
= 0.97 kcal/mol. Extrapola-
tion of structural data to APE1 complexed with undamaged
DNA (3,15) suggests that, in order to search for lesions, the
enzyme severely distorts and possibly melts DNA locally.
Taking this into account, all interactions of APE1 with non-
specific DNA can be summarized using the thermodynamic
model shown in Figure 5.
Contribution of a specific AP site in DNA to its affinity
for APE1
The relative contributions of an AP site to the total affinity of
APE1 can be estimated for specific DNA. The
K
I
values for
duplexes corresponding to specific 14X8 and 24X8 ODNs
(Table 2) were determined by using them as inhibitors of
the APE1 cleavage of apurinized ds polymeric [
3
H]DNA.
The increase in affinity on transition from nonspecific ss
d(pT)
14
(Table 1) and ss 14X8 (14C8 and 14G8) to specific
ss 14X8 and ss 24X8 varied from 6.4 to 8.6 depending on the
ODN sequence and length. The ratio of
K
I
values for ss non-
specific 24A8 and specific 24R8 was equal to 6.0 (Table 2).
Transition from nonspecific ds d(pT)
14
(Table 1) and ds
14N8 (14C8 and 14G8) to specific ds 14R8 and ds 24R8
Figure 5. Thermodynamic model of APE1 interactions with nonspecific DNA.
For the enzyme subsites interacting with the cleaved strand, the
DG
values
characterizing their contacts with the d(pA)
n
chain of d(pA)
n
d(pT)
n
are given;
for the subsites interacting with the noncleaved strand, the
DG
values refer
to their contacts with the d(pT)
n
chain. All types of nonspecific additive
interactions of APE1 with the d(pA)
n
d(pT)
n
duplex provide
DG
7.96 kcal/mol.
Nucleic Acids Research, 2004, Vol. 32, No. 17
5141
Downloaded from https://academic.oup.com/nar/article/32/17/5134/1334164 by guest on 16 February 2023
(Table 2) led to a decrease in
K
I
by a factor of 6.4–11.5. The
affinity for specific ds 24R8 is only 3.8-fold higher than that
for nonspecific 24A8 duplexes (Table 2). Interestingly, APE1
bound free deoxyribose-5
0
-phosphate 6.6-fold more effec-
tively than various dNMPs. Thus, a relative contribution of
specific interactions of APE1 with the natural AP site is com-
parable at the level of minimal ligands (6.6-fold), ss d(pN)
n
and ds ODNs (3.8–11.5-fold) (Table 2).
The affinity of APE1 for the 14F8:d(pA)
14
duplex (0.16
mM;
Table 2) was 2.1-fold higher than that for d(pT)
14
:d(pA)
14
(0.33
mM) (Table 1). A very similar 2.5-fold difference in
the affinity was observed for the phosphorylated tetrahydro-
furan analogue d(pF) and various dNMPs. Thus, the contribu-
tion of specific and nonspecific interactions of different
nucleotides of specific DNA to its total affinity for APE1 is
nearly additive. The same situation occurs for two other repair
enzymes, UDG and Fpg (6,7,28,30).
Thermodynamic model of APE1 interaction with
specific DNA
According to X-ray crystallographic data, APE1 electrostati-
cally orients a rigid, preformed DNA-binding face and inserts
loops into the DNA helix through both the major and the minor
groove, stabilizing the target AP site in an extrahelical con-
formation. APE1-bound DNA is severely distorted, with the
DNA bent at about 35
and the helical axis kinked by
5 s.
Figure 6 presents a summary of APE1 contacts with specific
DNA (3,15). Immediately 3
0
to the AP site, APE1 forms sev-
eral bonds with two phosphates (p2 and p3) and braces the AP
DNA backbone for the double loop insertion. At a position
opposite to the everted AP site, Met-270 is inserted through the
minor groove to pack against the orphaned base partner of the
abasic site and occupy the space where it would be found in
regular B-DNA. Above the abasic site, Arg-177 is inserted
through the major groove and provides a hydrogen bond to the
AP site 3
0
phosphate (p1). Interactions in the major groove are
unusual for base excision repair enzymes and, as the sequence
and conformation of the Arg-177 loop is unique to APE1, it
probably reflects specific APE1 functions. On the 5
0
side of the
lesion, the side chains of several amino acids residues contact
the p-1 and p-2 phosphates of the damaged strand and the p-1,
p-3, p-4 and p-5 phosphates of the undamaged strand, which
results in a widening of the minor groove by
2 s (3,15).
These 5
0
contacts may anchor the DNA for the kinking caused
by the loop insertion at a position 3
0
of the extrahelical
abasic site.
Specific binding of extrahelical AP sites occurs in a hydro-
phobic pocket bordered by Phe-266, Trp-280 and Leu-282,
which pack against the hydrophobic face and edge of the
abasic deoxyribose. All listed interactions between APE1
and AP DNA stabilize the extrahelical AP site conformation
and effectively lock APE1 onto the AP DNA.
As discussed above, the active site of APE1 can efficiently
interact with different nucleotide units of DNA. At the same
time, tight packing of the abasic deoxyribose against Phe-266,
Trp-280 and Leu-282 should prevent productive binding of
normal deoxynucleotides (3,15). We have shown previously
that some sequence-specific enzymes have increased affinity
for free deoxynucleotides compared with the same deoxy-
nucleotide units within DNA (6,7,30,31). This may result
from the absence of steric hindrance to free nucleotide bind-
ing, or from the restrictions imposed in a longer d(pN)
n
on
nucleotide eversion or on a particular conformational change
necessary for productive interaction of a unit of d(pN)
n
with the
catalytic center of the enzyme. As free dNMPs, deoxyribose-
5
0
-phosphate, deoxyribose and orthophosphate are the least
restrained in their search for optimal binding interactions,
their
K
d
values may put an upper estimate on the affinity of the
respective elements of long DNA for the active site of APE1.
The affinity of APE1 for deoxyribose-5
0
-phosphate [d(pR);
K
I
= 25 mM] and its tetrahydrofuran analogue [d(pF),
A
B
Figure 6. (A) Schematic representation of contacts between APE1 and specific
ds DNA revealed by X-ray crystallography (3,15). Arrows indicate interactions
between the various amino acid residues and structural elements of DNA,
assisting the sharp DNA kinking (see text for details). (B) Thermodynamic
model of APE1 interactions with specific DNA, displaying
DG
values
characterizing different contacts and strengthening of some contacts in
comparison with nonspecific DNA (see Figure 5). The total
DDG
value
characterizing a change in all types of interactions upon transition from
nonspecific to specific DNA can be estimated at
1.1 to 1.5 kcal/mol.
5142
Nucleic Acids Research, 2004, Vol. 32, No. 17
Downloaded from https://academic.oup.com/nar/article/32/17/5134/1334164 by guest on 16 February 2023
K
I
= 59 mM] is 6.6- and 2.8-fold higher than that for non-
specific dNMPs (
K
I
= 163–167 mM; Table 1). This increase in
the affinity of APE1 for d(pR) in comparison with dNMPs can
be a result of better interaction of the enzyme with the sugar
moiety of d(pR). On the other hand, removal of the base from
dNMPs could also lead to a remarkable strengthening of the
enzyme’s contacts with both sugar and phosphate groups.
K
I
for the internucleotide phosphates in abasic DNA may be esti-
mated at
100 mM by extrapolation of the line for d[(pF)
n
pT]
to
n
= 0 (Figure 1). Thus, a difference in the affinity for dNMPs
and d(pF) (2.8-fold) is comparable with the ratio of
K
I
values
for the internucleotide phosphates of d(pN)
n
and d[(pF)
n
pT]
(3.0-fold). Therefore, the transition from d(pN)
n
to d[(pF)
n
pT]
leads mainly to strengthening of the enzyme active center
contacts with only one internucleotide phosphate, while the
contribution of the tetrahydrofuran moiety to the affinity for
ODNs is very low. A similar situation probably occurs at the
AP site unit in the AP DNA; detectable inhibition of APE1 by
free deoxyribose was observed only at very high concentra-
tions of this ligand (IC
50
> 0.5 M; K
I
> 0.17 M). Assuming
comparable contributions of the phosphate groups of d(pF) and
d(pR) to their affinity for APE1 (
K
I
= 100 mM; DG
=
5.54 kcal/mol), the contribution of the deoxyribose moiety of
d(pR) can be estimated at
K
I
= 0.25 M (DG
= 0.83 kcal/mol),
a value that is in agreement with inhibition by free deoxyri-
bose. Thus, the difference in APE1 affinity for deoxyribose
itself and deoxyribose plus the adenine base within dAMP can
be estimated at
DDG
= 0.4 kcal/mol (Figure 6B).
A significant overall 6.6-fold increase in the d(pR) affinity
as compared with that for dNMP was observed, which is in
good agreement with the steric restrictions imposed by
Phe-266, Trp-280 and Leu-282 discussed above. However,
the efficiency of APE1 interactions with AP sites in DNA
is likely to depend strongly on the everted conformation of
this nucleotide, which allows its 5
0
-phosphate to form stronger
contacts with the enzyme (Figure 6). This strengthening may
result from DNA backbone compression (3,15) and from dis-
placement of the p-1 to a position where it can form more
efficient contacts with four amino acid residues of the enzyme
(Figure 6).
Depending on the sequence of ODNs used, a transition from
nonspecific ss to different specific ss AP ODNs led to an
increase in their affinity by a factor of 6.0–7.7 (Table 2).
This difference remained nearly the same (3.8–11.5-fold)
for different ds nonspecific versus specific AP ODNs (Table
2), and all these increase factors were comparable with the
ratio of the
K
I
values for dNMPs and d(pR) (6.6). Thus, the
affinity improvement for different ss and ds AP sites contain-
ing specific compared with the respective nonspecific substrate
(Table 2) is 6.0–11.5-fold (
K
d
= 0.087–0.26 M, DG
= 0.81
to
1.47 kcal/mol). The increase in affinity of APE1 for
specific ds ODNs compared with that for specific ss ODNs
is 2.7–4.2 (Table 2), which is comparable with the 2–5-fold
difference between
K
I
values for ss and ds nonspecific ODNs
(Tables 1 and 2).
It is quite possible that some of the nonspecific contacts
between APE1 and the internucleotide phosphate groups or
nucleobases of the cleavable strand of specific ds ODNs are
different from the contacts arising in nonspecific d(pN)
n
duplexes. Given the drastic APE1-dependent changes in
the structure of specific DNA (Figure 6A), there could be a
weakening of some contacts and strengthening of others formed
by enzymes at the stage of primary complex formation. How-
ever, our data suggest that overall there is no remarkable
thermodynamic difference between the majority of these con-
tacts in specific and nonspecific ODNs. Taking into account a
comparable difference in the affinity of APE1 for specific and
nonspecific ligands at the level of a single dNMP DNA element,
ss and ds DNAs, the contribution of all nonspecific contacts
can be approximately put at
DG
3.3 kcal/mol (Figure 6B).
Transition from nonspecific to specific ODNs is probably
also accompanied by some reorganization of contacts between
APE1 and the second strand as well as between both DNA
strands. Nevertheless, the average additional increase in the
affinity for nonspecific (
DG
= 0.97 kcal/mol) and specific
DNAs due to the presence of the complementary strand may be
characterized by similar values of
K
d
= 0.2–0.5 M and
DG
= 0.2 to 1.0 kcal/mol. The contributions of the AP
site and the second strand of ds DNA to the affinity depend to
some extent on the DNA sequence. However, at the level of
d(pR), ss and ds AP ODNs, the affinity of APE1 for specific
ligands in comparison with nonspecific ones usually increases
by a factor of 3.8–6.6 (
DG
= 0.8 to 1.1 kcal/mol), most
probably reflecting the contribution of the d(pR) unit to the
affinity of these ligands for the active center of APE1. The
recognition of specific AP DNA by APE1 can be generally
described
using
the
thermodynamic
model
shown
in
Figure 6B.
The efficiency of specific contact formation by APE1, as in
the case of all studied DNA-dependent enzymes (6,7,31), does
not exceed one to two orders of affinity, while the relative
contribution of nonspecific interactions to the total affinity is
four to five orders of magnitude greater (6,7). Formation of the
enzyme–DNA complex cannot alone explain the observed
specificity of enzyme catalysis. All the enzymes investigated
to date, including APE1, interact with noncognate RNA–RNA
and RNA–DNA duplexes with affinities comparable with
those for DNA–DNA duplexes, and the affinity for such com-
plexes is still only one to two orders of magnitude lower than
that for specific DNA–DNA duplexes (6,7). However, the
enzymes do not catalyze conversion of noncognate duplexes
even at their saturating concentrations. The specificity of
DNA-dependent enzymes lies in the
k
cat
term; the rate is
usually elevated by four to eight orders of magnitude upon
transition from nonspecific to specific DNAs (6,7).
Kinetic factors: reaction rate and the specificity of
APE1 action
Previous studies showed that APE1 could in principle cleave
DNA at nonmodified nucleotides but only at high enzyme
concentrations and longer incubation times (10
4
–10
7
-fold)
compared with AP DNA (35,39). APE1 cannot hydrolyze
nonspecific DNA with noticeable efficiency: the rate of non-
specific enzyme action decreases by six to eight orders of
magnitude (35,39). The catalytic stage appears significantly
more sensitive to the DNA structure than the stage of the
enzyme–DNA complex formation. The rate of APE1-
dependent hydrolysis of ODNs notably depends on the AP
site structure; the rate for ds 14R8 decreased 10–14-fold
when the natural aldehydic AP site was replaced with an AP
site bearing a hydroxy group (NaBH
4
-reduced deoxyribose).
Nucleic Acids Research, 2004, Vol. 32, No. 17
5143
Downloaded from https://academic.oup.com/nar/article/32/17/5134/1334164 by guest on 16 February 2023
This situation is similar to human UDG, where even minimal
modifications of uracil, deoxyribose or internucleotide phos-
phate (e.g. introduction of a fluorine atom at certain positions)
of a dUMP unit of specific ODNs often do not change the
affinity for this ligand but result in a decrease in
k
cat
that is less
than three to four orders of magnitude, sometimes abolishing
uracil excision altogether (49). However, the gross DNA struc-
ture does not seem to influence
k
cat
, since its value was similar
for high-molecular-weight plasmid DNA and the oligonucleo-
tide substrates, both measured in this study and reported in the
literature (42). The independence of
k
cat
on DNA length was
also observed for DNA repair glycosylases such as Fpg (50).
From the structure of the specific APE1–DNA complex, it
is evident that enzyme-dependent DNA conformation adjust-
ment involves pronounced kinking of both strands (15). It is
known that the AP site significantly increases the ability of
DNA to be kinked (51,52). However, ss d[(pT)
7
pR(pT)
6
]
was a relatively poor substrate for APE1 and was not sig-
nificantly hydrolyzed after 1 h of incubation, whereas ss
hetero-ODNs of the same length containing AP sites
were effectively cleaved after 10 min (35). The duplex of
d[(pT)
7
pRd(pT)
6
] with d(pA)
14
was cleaved 7-fold better
(35). At the same time, the
V
max
values of hetero ss AP
ODNs were 10–15-fold higher than that for ds homo AP
ODNs, while the
V
max
values of hetero ss and ds AP ODNs
differed by only 2–3-fold depending on the sequence (35).
These data indicate that APE1 can distort ss as well as ds
DNA, but the efficiency of DNA adjustment to the
conformation optimal for catalysis depends on the DNA
sequence.
Examples of many DNA-dependent enzymes show that the
adjustment of DNA structure to the optimal conformation
depends both on the initial structure in solution and on its
flexibility in the enzyme-driven direction (6,7). The ability
of different ds ODNs to be kinked and partially melted, neces-
sary for DNA distortion by APE1, depends on several struc-
tural characteristics of DNA (33). DNA kinking and bending
is notably facilitated in pyrimidine–purine sequences, which
favor bending towards the major groove, and in regions with
sterically unfavorable minor groove interactions between N3
and NH
2
of guanine and N3 of adenine (33). Hetero-ODNs
with the AP site incorporated in the context of a more flexible
and easily kinked trinucleotide ARC demonstrated the highest
affinity and
V
max
values. Thus, in contrast to a quite rigid ss
d(pT)
n
, the structure of ss specific hetero-ODNs can probably
be changed by the enzyme much more easily. As shown above,
addition of a complementary strand even to intrinsically rigid
homo ss AP ODNs can convert such a ligand into a good APE1
substrate (35). This result agrees well with the important role
of the second DNA strand in productive DNA distortion by
APE1, as evident from the structural data (3,15). Similar results
have been observed for human UDG (28) and for Fpg (53). All
these data suggest that the second strand can be actively
involved in attaining the optimal DNA conformation in com-
plexes with repair enzymes. The capability of both strands of
specific ds DNA to be distorted by APE1 may be very impor-
tant for more productive formation of all APE1–DNA contacts
revealed by X-ray crystallography (3,15). Introduction of an
additional AP site into the second strand of a 24mer hetero-
ODN leads to an 8–10-fold increase in affinity over two alter-
native duplexes containing a single AP site in either of the
strands (35). Such an increase in the affinity, however, did not
lead to a significant increase in the cleavage rate. Therefore, it
cannot be excluded that the formation of a limited number of
strong contacts between APE1 and the two strands of AP DNA
is not obligatory for the productive eversion of the AP site
from DNA.
Comparison of APE1 with other DNA-dependent
enzymes
APE1, like many other DNA-depending enzymes (UDG, Fpg,
Topo I, EcoRI, HIV integrase), interacts efficiently with both
specific and nonspecific ss and ds ODNs (27,28,30–34)
through contacts with the internucleotide phosphate groups
and bases of DNA. The factor
e (1.51) for APE1 (reflecting
its interaction with one internucleotide phosphate of nonspe-
cific DNA) is comparable with
e factors for other enzymes:
UDG (1.35), DNA polymerases (1.52), Fpg (1.54), RNA heli-
case (1.61), Topo I (1.67), EcoRI (2.0) and DNA ligase (2.14)
(6,7,24–28,30–34). The factor
h (1.01–1.10) for APE1 is
remarkably lower than that for other enzymes interacting
with bases of ss or ds DNA: DNA polymerases (1.03–1.32),
Topo I (1.04–1.4), UDG (1.04–1.41), RNA helicase (1.05–
1.59) and DNA ligase (1.1–1.62) (6,7,24–28,30–34). The rela-
tive contribution of the second strand to the affinity of APE1
for ds DNA is much lower than that for the first strand, again
similar to other enzymes analyzed. Recognition of nonspecific
DNA by sequence- and structure-specific DNA-dependent
enzymes may be considered a first stage of specific DNA
recognition. This stage of the primary complex formation
due to nonspecific interactions between DNA and enzymes
provides high affinity of any DNA-dependent enzyme for any
DNA (6,7). High affinity of the enzymes to nonspecific DNA
allows their ‘sliding’ along DNA to the site containing a
specific sequence, lesion or structural element (6,7). The posi-
tively charged DNA-binding grooves of the enzymes and the
negatively charged DNA sugar–phosphate backbone can inter-
act during primary complex formation through many weak
additive contacts. Since all these contacts are thermodynami-
cally nearly equal, the enzymes can easily slide along DNA
in search of specific elements, which are then recognized in
unique enzyme-specific ways (6,7).
APE1 binds the DNA minor groove via a conserved minor-
groove widening loop (3,15), suggesting that the enzyme could
search for AP sites by using this loop to slightly distort DNA.
Minor-groove widening is probably a conserved function of
the four-layered
a,b-sandwich fold, as similar interactions are
also seen in bovine DNase I (54,55) and in
E.coli Xth (17). The
penetration of the DNA minor groove anchors one half of
APE1 to the DNA, while the electrostatic attraction between
the positive APE1 DNA-binding groove and the negative
DNA phosphodiester backbone ensures that the entire enzyme
molecule remains properly oriented. In this half-bound con-
figuration APE1 can slide progressively along DNA, scanning
for regions that can accommodate the kinking induced by the
enzyme (3). Only abasic DNA can be deformed in this manner
within the constraints of the APE1 abasic nucleotide-binding
pocket (Figure 6) (3,15).
APE1 belongs to a group of highly specific DNA-dependent
enzymes that catalyze the conversion of specific DNA four
to eight orders of magnitude more effectively than that for
5144
Nucleic Acids Research, 2004, Vol. 32, No. 17
Downloaded from https://academic.oup.com/nar/article/32/17/5134/1334164 by guest on 16 February 2023
nonspecific DNA ([(6,7) and references therein]. The increase
in affinity of such enzymes for specific ds ODNs compared
with nonspecific ones was estimated at 7–10-fold (UDG, Fpg),
50–70-fold (HIV integrase), 50–100-fold (EcoRI), 200–250-
fold (Topo I) (6,7) (27–34) and 6–11-fold for APE1 (this
study). Thus, the efficiency of specific contact formation
between such enzymes and DNA does not exceed one to
two orders of affinity and the relative contribution of nonspe-
cific interactions to the total affinity is four to six orders of
magnitude greater than that of specific interactions. Although
these enzymes do not act on nonspecific DNA, the formation
of a primary complex cannot alone explain their specificity. At
the same time, the low affinity of enzymes for specific parts of
their substrates can be of biological significance. An increase
in the affinity for specific sequences limited to one to two
orders of magnitude ensures a relatively short lifetime for a
specific complex. The specificity of enzyme action can thus be
provided by the impossibility of productive enzyme-depen-
dent deformation of nonspecific DNA during the short exis-
tence time of the complex. For several DNA-dependent
enzymes (including APE1), the conformational adjustment
step of the reaction, in contrast to DNA binding, is extremely
sensitive for specific DNA elements, and it is this step that
determines the reaction rates for different DNAs (6,7,24–34).
According to structural data, APE1 cannot promote productive
eversion of a normal nucleotide into the enzyme active site
pocket (3,15) and therefore a satisfactory orbital overlap and
high reaction rate cannot be achieved. The formation of
specific bonds between the extrahelical abasic site and
amino acid residues in the active site (Figure 6) is most
probably one of the final stages in the selection of specific
DNA by APE1. After formation of such contacts, the reaction
can be accelerated by six to seven orders of magnitude.
Specific contacts between APE1 and the sugar moiety of
d(pR) can provide, at most, a 6.6-fold increase in the affinity
for specific DNA. Experimentally determined increase in
affinity for AP DNA, compared with nonspecific DNA,
does not exceed 3.8–11-fold. Moreover, this small increase
arises not only from APE1-specific interaction with the sugar
moiety of the AP site but also from strengthening of the
enzyme contacts with other parts of the cleaved and non-
cleaved strands of AP DNA (Figure 6). Thus, the actual ther-
modynamic contribution of APE1-specific interaction with the
extrahelical AP site is remarkably low. In general, recognition
of small ligands by enzymes is based on the formation of
several strong contacts (hydrogen bonds, electrostatic con-
tacts, stacking interactions, etc.) with specific structural
elements. Interestingly, during formation of a specific complex
of ds DNA with EcoRI, 12 hydrogen bonds are formed, pro-
viding in total only about two orders of affinity (27). This
means that the energy of each of these 12 bonds is rather
low (
DG
0.23 kcal/mol) and comparable with the
energy of weak additive nonspecific interactions (6,7,27).
Only one order of affinity (
DG
0.28 to 0.36 kcal/mol)
is accounted for by five pseudo-Watson–Crick hydrogen
bonds formed by a uracil residue with UDG (28). Similar
weak specific contacts with nucleotides of DNA were
observed for all other investigated enzymes (6,7), indicating
that formation of specific contacts between enzymes and
DNA is not very important at the stage of protein–DNA
complexation. This hydrogen bond energetic summary
does not take into account the solvation reorganization
energies (enthalpy and entropy) of hydrogen-bond networks
such as these and must thus be a lower limit for the isolated
hydrogen bond contributions in these cases. On the contrary,
such contacts are extremely important at the stage of adjust-
ment of DNA and enzyme conformations, and only in the
case of specific DNA do specific contacts provide a very precise
alignment of electronic orbitals of the reacting atoms.
ACKNOWLEDGEMENTS
This research was made possible in part by grants from the
Wellcome Trust UK (070244/Z/03/Z), Presidium of the
Russian Academy of Sciences (Physicochemical Biology
Program 10.5), Russian Foundation for Basic Research (01-
04-48892, 02-04-49605), Russian Ministry of Education
(PD02-1.4-469), Award no. NO-008-X1 of the U.S. Civilian
Research & Development Foundation for the Independent
States of the Former Soviet Union (CRDF), Russian Science
Support Foundation (to D.O.Z.) and funds from the Siberian
Division of the Russian Academy of Sciences.
REFERENCES
1. Friedberg,E.C., Walker,G.C. and Siede,W. (1995)
DNA Repair and
Mutagenesis. ASM Press, Washington, DC.
2. Atamna,H., Cheung,I. and Ames,B.N. (2000) A method for detecting
abasic sites in living cells: age-dependent changes in base excision
repair.
Proc. Natl Acad. Sci. USA, 97, 686–691.
3. Mol,C.D., Hosfield,D.J. and Tainer,J.A. (2000) Abasic site recognition
by two apurinic/apyrimidinic endonuclease families in DNA base
excision repair: the 3
0
ends justify the means.
Mutat. Res., 460, 211–229.
4. Eisen,J.A. and Hanawalt,P.C. (1999) A phylogenomic study of DNA
repair genes, proteins, and processes.
Mutat. Res., 435, 171–213.
5. Aravind,L., Walker,D.R. and Koonin,E.V. (1999) Conserved domains in
DNA repair proteins and evolution of repair systems.
Nucleic Acids Res.,
27, 1223–1242.
6. Nevinsky,G.A. (1995) Important role of weak interactions in long DNA
and RNA molecule recognition by enzymes.
Mol. Biol. (Moscow), 29,
6–19.
7. Bugreev,D.V. and Nevinsky,G.A. (1999) Possibilities of the method of
step-by-step complication of ligand structure in studies of protein–nucleic
acid interactions: mechanisms of functioning of some replication, repair,
topoisomerization, and restriction enzymes.
Biochemistry (Moscow), 64,
237–249.
8. Phillips,S.E.V. and Moras,D. (1999) Protein–nucleic acid interactions.
Curr. Opin. Struct. Biol., 9, 11–13.
9. Tainer,J.A. and Friedberg,E.C. (2000) Dancing with the elephants:
envisioning the structural biology of DNA repair pathways.
Mutat. Res.,
460, 139–141.
10. Luscombe,N.M., Austin,S.E., Berman,H.M. and Thornton,J.M. (2000)
An overview of the structures of protein-DNA complexes.
Genome Biol.,
1, 1–37.
11. Pingoud,A. and Jeltsch,A. (2001) Structure and function of type II
restriction endonucleases.
Nucleic Acids Res., 29, 3705–3727.
12. Rice,P.A. and Baker,T.A. (2001) Comparative architecture of
transposase and integrase complexes.
Nature Struct. Biol., 8, 302–307.
13. Aggarwal,A.K. and Doudna,J.A. (2003) Protein–nucleic acid
interactions.
Curr. Opin. Struct. Biol., 13, 3–5.
14. Gorman,M.A., Morera,S., Rothwell,D.G., de La Fortelle,E., Mol,C.D.,
Tainer,J.A., Hickson,I.D. and Freemont,P.S. (1997) The crystal
structure of the human DNA repair endonuclease HAP1 suggests the
recognition of extra-helical deoxyribose at DNA abasic sites.
EMBO J.,
16, 6548–6558.
15. Mol,C.D., Izumi,T., Mitra,S. and Tainer,J.A. (2000) DNA-bound
structures and mutants reveal abasic DNA binding by APE1 DNA
repair and coordination.
Nature, 403, 451–456.
Nucleic Acids Research, 2004, Vol. 32, No. 17
5145
Downloaded from https://academic.oup.com/nar/article/32/17/5134/1334164 by guest on 16 February 2023
16. Beernink,P.T., Segelke,B.W., Hadi,M.Z., Erzberger,J.P.,
Wilson,D.M.,III and Rupp,B. (2001) Two divalent metal ions in
the active site of a new crystal form of human apurinic/apyrimidinic
endonuclease, Ape1: implications for the catalytic mechanism.
J. Mol.
Biol., 307, 1023–1034.
17. Mol,C.D., Kuo,C.F., Thayer,M.M., Cunningham,R.P. and Tainer,J.A.
(1995) Structure and function of the multifunctional DNA-repair enzyme
exonuclease III.
Nature, 374, 381–386.
18. Hosfield,D.J., Guan,Y., Haas,B.J., Cunningham,R.P. and Tainer,J.A.
(1999) Structure of the DNA repair enzyme endonuclease IV and its DNA
complex: double-nucleotide flipping at abasic sites and three-metal-ion
catalysis.
Cell, 98, 397–408.
19. Lesser,D.R., Kurpiewski,M.R. and Jen-Jacobson,L. (1990) The energetic
basis of specificity in the EcoRI endonuclease-DNA interaction.
Science,
250, 776–786.
20. Engler,L.E., Welch,K.K. and Jen-Jacobson,L. (1997) Specific
binding by
EcoRV endonuclease to its DNA recognition site GATATC.
J. Mol. Biol., 269, 82–101.
21. Engler,L.E., Sapienza,P., Dorner,L.F., Kucera,R., Schildkraut,I. and
Jen-Jacobson,L. (2001) The energetics of the interaction of
BamHI
endonuclease with its recognition site GGATCC.
J. Mol. Biol., 307,
619–636.
22. Jen-Jacobson,L. (1997) Protein-DNA recognition complexes:
conservation of structure and binding energy in the transition state.
Biopolymers, 44, 153–180.
23. Tsodikov,O.V., Holbrook,J.A., Shkel,I.A. and Record,M.T.,Jr (2001)
Analytic binding isotherms describing competitive interactions of a
protein ligand with specific and nonspecific sites on the same DNA
oligomer.
Biophys. J., 81, 1960–1969.
24. Kolocheva,T.I., Nevinsky,G.A., Levina,A.S., Khomov,V.V. and
Lavrik,O.I. (1991) The mechanism of recognition of templates by DNA
polymerases from pro- and eukaryotes as revealed by affinity
modification data.
J. Biomol. Struct. Dyn., 9, 169–186.
25. Ljach,M.V., Kolocheva,T.I., Gorn,V.V., Levina,A.S. and Nevinsky,G.A.
(1992) The affinity of the Klenow fragment of
E. coli DNA-polymerase 1
to primers containing bases noncomplementary to the template and
hairpin-like elements.
FEBS Lett., 300, 18–20.
26. Kolocheva,T.I., Maksakova,G.A., Zakharova,O.D. and Nevinsky,G.A.
(1996) The algorithm of estimation of the K
m
values for primers in DNA
synthesis catalyzed by human DNA polymerase
a. FEBS Lett., 399,
113–116.
27. Kolocheva,T.I., Maksakova,G.A., Bugreev,D.V. and Nevinsky,G.A.
(2001) Interaction of endonuclease
EcoRI with short specific and
nonspecific oligonucleotides.
IUBMB Life, 51, 189–195.
28. Vinogradova,N.L., Bulychev,N.V., Maksakova,G.A., Johnson,F. and
Nevinskii,G.A. (1998) Uracil DNA glycosylase: interpretation of X-ray
data in the light of kinetic and thermodynamic studies.
Mol. Biol.
(Moscow), 32, 400–409.
29. Ishchenko,A.A., Koval,V.V., Fedorova,O.S., Douglas,K.T. and
Nevinsky,G.A. (1999) Structural requirements of double and single
stranded DNA substrates and inhibitors, including a photoaffinity label,
of Fpg protein from
Escherichia coli. J. Biomol. Struct. Dyn., 17,
301–310.
30. Ishchenko,A.A., Vasilenko,N.L., Sinitsina,O.I., Yamkovoy,V.I.,
Fedorova,O.S., Douglas,K.T. and Nevinsky,G.A. (2002)
Thermodynamic, kinetic, and structural basis for recognition and repair
of 8-oxoguanine in DNA by Fpg protein from
Escherichia coli.
Biochemistry, 41, 7540–7548.
31. Zharkov,D.O., Ishchenko,A.A., Douglas,K.T. and Nevinsky,G.A. (2003)
Recognition of damaged DNA by
Escherichia coli Fpg protein: insights
from structural and kinetic data.
Mutat. Res., 531, 141–156.
32. Bugreev,D.V., Buneva,V.N., Sinitsina,O.I. and Nevinskii,G.A. (2003)
The mechanism of the supercoiled DNA recognition by eukaryotic type I
topoisomerases. I. The enzyme interaction with nonspecific
oligonucleotides [in Russian].
Bioorg. Khim., 29, 163–174.
33. Bugreev,D.V., Sinitsina,O.I., Buneva,V.N. and Nevinskii,G.A. (2003)
The mechanism of supercoiled DNA recognition by eukaryotic type I
topoisomerases. II. A comparison of the enzyme interaction with specific
and nonspecific oligonucleotides [in Russian].
Bioorg. Khim., 29,
277–289.
34. Bugreev,D.V., Baranova,S., Zakharova,O.D., Parissi,V., Desjobert,C.,
Sottofattori,E., Balbi,A., Litvak,S., Tarrago-Litvak,L. and
Nevinsky,G.A. (2003) Dynamic, thermodynamic, and kinetic basis for
recognition and transformation of DNA by human immunodeficiency
virus type 1 integrase.
Biochemistry, 42, 9235–9247.
35. Beloglazova,N.G., Petruseva,I.O., Bulychev,N.V., Maksakova,G.A.,
Johnson,F. and Nevinskii,G.A. (1997) Isolation and substrate specificity
of apurine/apyrimidine endonuclease from human placenta.
Mol. Biol.
(Moscow), 31, 1104–1111.
36. Takeshita,M., Chang,C.-N., Johnson,F., Will,S. and Grollman,A.P.
(1987) Oligodeoxynucleotides containing synthetic abasic sites. Model
substrates for DNA polymerases and apurinic/apyrimidinic
endonucleases.
J. Biol. Chem., 262, 10171–10179.
37. Castaing,B., Boiteux,S. and Zelwer,C. (1992) DNA containing a
chemically reduced apurinic site is a high affinity ligand for the
E. coli
formamidopyrimidine-DNA glycosylase.
Nucleic Acids Res., 20,
389–394.
38. Cantor,C.R., Warshaw,M.M. and Shapiro,H. (1970) Oligonucleotide
interactions. 3. Circular dichroism studies of the conformation of
deoxyoligonucleotides.
Biopolymers, 9, 1059–1077.
39. Beloglazova,N.G., Lokhova,I.A., Maksakova,G.A., Tsvetkov,I.V. and
Nevinskii,G.A. (1996) Apurine/apyrimidine endonuclease from human
placenta. Recognition of apurinized DNA by the enzyme.
Mol. Biol.
(Moscow), 30, 220–230.
40. Cornish-Bowden,A. (1976)
Principles of Enzyme Kinetics. Butterworths,
London.
41. Fersht,A. (1985)
Enzyme Structure and Mechanism. 2nd edn. W. H.
Freeman & Co., New York.
42. Lucas,J.A., Masuda,Y., Bennett,R.A.O., Strauss,N.S. and Strauss,P.R.
(1999) Single-turnover analysis of mutant human apurinic/apyrimidinic
endonuclease.
Biochemistry, 38, 4958–4964.
43. Wilson,D.M.,III, Takeshita,M. and Demple,B. (1997) Abasic site binding
by the human apurinic endonuclease, Ape, and determination of the DNA
contact sites.
Nucleic Acids Res., 25, 933–939.
44. Nguyen,L.H., Barsky,D., Erzberger,J.P. and Wilson,D.M.,III (2000)
Mapping the protein-DNA interface and the metal-binding site of the
major human apurinic/apyrimidinic endonuclease.
J. Mol. Biol., 298,
447–459.
45. Kolocheva,T.I., Levina,A.S. and Nevinsky,G.A. (1996) Recognition of
the primers containing different modified nucleotide units by the Klenow
fragment of DNA polymerase I from
E. coli. Biochimie, 78, 201–203.
46. Doronin,S.V., Lavrik,O.I., Nevinsky,G.A. and Podust,V.N. (1987)
The efficiency of dNTP complex formation with human placenta
DNA polymerase
a as demonstrated by affinity modification.
FEBS Lett., 216, 221–224.
47. Saenger,W. (1984)
Principles of Nucleic Acid Structure. Springer-
Verlag, New York.
48. Breslauer,K.J., Frank,R., Blocker,H. and Marky,L.A. (1986) Predicting
DNA duplex stability from the base sequence.
Proc. Natl Acad. Sci. USA,
83, 3746–3750.
49. Kubareva,E.A., Volkov,E.M., Vinogradova,N.L., Kanevsky,I.A.,
Oretskaya,T.S., Kuznetsova,S.A., Brevnov,M.G., Gromova,E.S.,
Nevinsky,G.A. and Shabarova,Z.A. (1995) Modified substrates as
probes for studying uracil-DNA glycosylase.
Gene, 157, 167–171.
50. Zaika,E.I., Perlow,R.A., Matz,E., Broyde,S., Gilboa,R., Grollman,A.P.
and Zharkov,D.O. (2004) Substrate discrimination by
formamidopyrimidine-DNA glycosylase: a mutational analysis.
J. Biol. Chem., 279, 4849–4861.
51. Ayadi,L., Coulombeau,C. and Lavery,R. (2000) The impact of abasic
sites on DNA flexibility.
J. Biomol. Struct. Dyn., 17, 645–653.
52. Barsky,D., Foloppe,N., Ahmadia,S., Wilson,D.M.,III and
MacKerell,A.D.,Jr (2000) New insights into the structure of abasic
DNA from molecular dynamics simulations.
Nucleic Acids Res., 28,
2613–2626.
53. Ishchenko,A.A., Bulychev,N.V., Maksakova,G.A., Johnson,F. and
Nevinsky,G.A. (1997) Recognition and conversion of single- and
double-stranded oligonucleotide substrates by 8-oxoguanine-DNA
glycosylase from
Escherichia coli. Biochemistry (Moscow), 62,
204–211.
54. Lahm,A. and Suck,D. (1991) DNase I-induced DNA conformation. 2 s
structure of a DNase I-octamer complex.
J. Mol. Biol., 222, 645–667.
55. Weston,S.A., Lahm,A. and Suck,D. (1992) X-ray structure of the DNase
I-d(GGTATACC)
2
complex at 2.3 s resolution.
J. Mol. Biol., 226,
1237–1256.
5146
Nucleic Acids Research, 2004, Vol. 32, No. 17
Downloaded from https://academic.oup.com/nar/article/32/17/5134/1334164 by guest on 16 February 2023
Dostları ilə paylaş: |