Acid-induced exchange of the imino proton in
G[middot]C pairs
Acid-induced exchange of the imino proton in G[middot]C pairs
Sylvie
Nonin
1,2
,
Jean-Louis
Leroy
1
and
Maurice
Guéron
1,
*
1
Groupe de Biophysique de l'Ecole Polytechnique et de l'URA 1254 du CNRS, 91128
Palaiseau
,
France
and
2
CEA-Service de Biologie et Génétique Moléculaire, DBCM/DSV, CEN Saclay, 91191
Gif-sur-Yvette
,
France
Received November 9, 1995;
Revised and Accepted January 8, 1996
ABSTRACT
Acid-induced catalysis of imino proton exchange in G
.
C pairs of DNA duplexes is surprisingly fast, being nearly as fast as for the
isolated nucleoside, despite base-pair dissociation constants in the range of 10
-
5 at neutral or basic pH. It is also observed in terminal G.C pairs of duplexes
and in base pairs of drug-DNA complexes. We have measured imino proton exchange in deoxyguanosine
and in the duplex d(ATATAGATCTATAT) as a function of pH. We show that acid-induced exchange can be assigned to proton transfer from N7-protonated guanosine to cytidine in the open state of the pair. This
is faster than transfer from neutral guanosine (the process of intrinsic
catalysis previously characterized at neutral pH) due to the lower imino proton
pK of the protonated form, 7.2 instead of 9.4. Other interpretations are
excluded by a study of exchange catalysis by formiate and cytidine as exchange
catalysts. The cross-over pH between the regimes of pH-independent and acid-induced exchange rates is more basic in the case of base pairs
than in the mononucleoside, suggestive of an increase by one to two decades in
the dissociation constant of the base pair upon N7 protonation of G. Acid-induced catalysis is much weaker in A.T base pairs, as expected in view of
the low pK for protonation of thymidine.
INTRODUCTION
Imino proton exchange from base pairs at basic and neutral pH is fairly well
understood. At basic pH, OH
-
accepts the imino proton from guanosine or thymidine, and the exchange rate is
therefore proportional to OH
-
concentration. Below pH 8, this process is slower than acceptance by the imino
nitrogen of the opposite nucleoside of the pair, the pH-independent process of `intrinsic' catalysis (
1
).
In both cases, proton exchange requires opening of the pair. At basic pH, the
ratio of exchange rates in the paired nucleoside and in the monomer is equal to
the dissociation constant of the pair, except for a correction factor. In this
way, one determines the dissociation constant
K
diss
, typically 10
-5
to 10
-6
for internal base pairs at room temperature (
2
). Proton acceptors such as ammonia may also be used to accelerate exchange, to
determine the base-pair dissociation constant and, using high acceptor concentrations, the
base-pair lifetime.
Other processes take over below neutral pH. In the guanosine monomer, the proton
exchange rate increases in proportion to the proton concentration. This is
explained by transfer to water of the imino proton of N7-protonated guanosine, whose imino proton pK is lower than in guanosine:
this accelerates the capture by H
2
O (
3
).
Protonation on N7 is also possible in a paired guanosine (Fig.
3
3), so that the same process should occur from G
.
C pairs. As in the case of catalysis by OH
-
at basic pH, the exchange rate would be slower than in the monomer in the ratio
of the dissociation constant. However, this is not at all what is observed.
There is indeed an acid-induced mechanism, but the exchange rate is nearly the same as in the
monomer, i.e. larger than expected by five or six orders of magnitude (
2
)! In the present work, we show that the acid-induced mechanism, which also affects terminal G
.
C pairs (
4
) and G
.
C pairs of DNA-drug complexes (
5
,
6
) is nothing but intrinsic catalysis operating on N7-protonated guanosine. Indeed, the pK of cytidine, the intrinsic proton
acceptor, is six units higher than that of the H
2
O acceptor.
As usual, the guanosine monomer provides the benchmark for the analysis of
exchange in base pairs. We supplement earlier work (
3
) by measurements at lower pH and by a study of the effect of formiate and
cytidine acting as exchange catalysts at acid pH. Next, we examine in detail
the instance of paired G in the d(ATATAGATCTATAT) duplex, and, more briefly,
other cases, including DNA-drug complexes. In the discussion, we show that the observations are
indeed explained by intrinsic catalysis, and not by other mechanisms which we
examine.
MATERIALS AND METHODS
Methods of sample preparation and of NMR were those described previously (
4
,
7
).
PROTON EXCHANGE THEORY (7,8)
Exchange from mononucleosides
Proton transfer from a nucleoside, nuH, to an acceptor, acc, proceeds
via
the formation, by diffusion-controlled collision of an H-bonded complex [nu
..
H
..
acc]:
nuH + acc <====> [nu
..
H
..
acc] <====> nu
-
+ accH
+
(1)
The exchange rate is given by:
k
ex,acc
= q
.
[acc] (1+10
pKnu-pKacc
)
-1
(2)
where [acc] is the concentration of the acceptor, and q is the rate constant of
diffusion-controlled collisions. The last factor in eq 2 is the proton transfer
yield, related to the difference between the pKs of the acceptor and of the
nucleoside donor.
When water itself is the proton acceptor, the notion of a collision is
inappropriate in view of the permanent hydration of the donor, and the product
(q
.
[acc]) in eq 2 should be replaced by a rate constant R(H
2
O). Still, it is convenient to express this constant in the form (q
.
[acc]). One factor of this product may be chosen arbitrarily, and since the
proton must finally be transferred to bulk water, we set [acc] = 55 M by
convention, and we designate
q
H2O
a pseudo-collision rate. The acceptor pK is that of the H
3
O
+
/H
2
O couple, -1.7 (see below).
N7-protonation lowers the pK of the N1 (imino) nitrogen of guanosine,
accelerating imino proton exchange (
3
). The pK [designated pK(N7)] and the enthalpy for N7-protonation of guanosine have been determined by
15
N NMR titration (
9
). The residence time of the HN7 proton is equal to the diffusion-controlled collision time when pH is equal to pK(N7). This is ~10
-11.
10
pK(N7)
, or ~1 ns for a pK of 2.15 (Table
2
), a time short compared to the characteristic NMR times. The observed imino
proton exchange rate for isolated guanosine, k
i
, is therefore the average of the rates corresponding to neutral G and N7-protonated G:
k
i
= (1 - f
+
)[Sigma]k
ex,acc
+ f
+
[Sigma]k
+
ex,acc
(3)
where k
ex,acc
and k
+
ex,acc
are the exchange rates by proton transfer to the acceptor, from neutral G and
from N7-protonated G respectively, and where f
+
is the molar fraction of N7-protonated guanosine:
f
+
= (1 + 10
pH-pK(N7)
)
-1
(4)
The acceptors include H
2
O, OH
-
and any added acceptors.
For example, at high pH (e.g. pH 11), exchange in guanosine proceeds mainly by
proton transfer from the neutral nucleoside to the base OH
-
, a process designated N
OH
- (Table
1
). Eqs 2 and 3 give:
k
i
= q
OH
-
.
[OH
-
] = q
OH
-
.
10
pH-pKi
(5)
where q
OH
- is the rate constant of the collision of OH
-
with the monomer and
K
i
is the ionization product of water (Table
2
). In contrast, exchange at low pH is mostly by proton transfer from the
protonated nucleoside HN7G
+
to water (the P
H2O
process). By eqs 2-4:
k
i
= f
+.
q
+
H2O
.
55
.
(1+10
pK(N1;HN7G+)-(-1.7)
)
-1
(6)
where q
+
H2O
is the pseudo-collision rate constant of water with the protonated nucleoside. Proton
transfer from neutral G to water (the N
H2O
process) is significant around neutral pH. The rate is:
k
i
= q
H2O
.
55
.
(1+10
pK(N1;G)-(-1.7)
)
-1
(7)
Let us designate by pNB (for neutral and basic) the cross-over between a pH-independent exchange regime and one which is base-controlled. Similarly, pAN will designate the pH for crossover
with an acid-controlled regime. If the relevant processes are those of eqs 5 and 7 for
pNB, and of eqs 6 and 7 for pAN, one has:
pNB = pKi - pK(N1;G) + log(q
H2O
/q
OH
-)
(8)
pAN = pK(N7) - [pK(N1;HN7G
+
) - pK(N1;G)]
+ log (q
+
H2O
/q
H2O
)
(9)
The value of pAN may be determined from experiment, and pK(N7) and pK(N1;G) are
tabulated. Assuming that the pseudo-collision rate constant of water is the same for neutral and for N7-protonated guanosine, eq 9 provides a means for the determination of
the imino proton pK of N7-protonated guanosine. The values in Table
2
for both G and T have been determined in this manner.
.
Nomenclature of the principal imino proton exchange processes operating at
different pHs on guanosine and in G
.
C pairs
acid pHneutral pHbasic pHguanosineHN7G
+
/H
2
O
b
G/H
2
OG/OH
-
P
H2O
N
H2O
N
OH
-
G@C pairHN7G
+
/cytidineG/cytidineG/OH-
P
C
N
C
N
OH
-
a
See Figure 4.
b
Each process is described as DONOR/ACCCEPTOR, and designated by a DONOR
ACCEPTOR
symbol, where P stands for N7-protonated guanosine, N for neutral guanosine and C for the cytidine
acceptor.
Imino proton exchange from the open state of a base pair
a
Values are given for the four temperatures indicated, the symbol # being used as
filler.
b
Except for the protonated nucleosides, the pK values are either taken from the
literature (for some temperatures) or derived from them for use in Table 3,
using the enthalpy value which is also from the literature.
c
The
q
values are deduced from the measured exchange time according to eq 2, using the
stated pK values, except for the values in brackets, which are explained in
Note
e
. In the case of the water acceptor, the use of
q
does not imply a diffusion-controlled collision (see text).
d
Ts'o,P.O.P. (1974)
Basic Principles of Nucleic Acid Chemistry vol.1
, Acad. Press, New York and London, p. 462.
e
Dawson, R.M.C., Elliott,D.C., Elliot,C., and Jones,K.M.,
Data for Biochemical Research
, Oxford University Press, Ely House W1, 1969, p. 140.
f
The imino proton pK values for the protonated nucleosides are derived from the
measured exchange times at acid pH, using eq 2. For the derivation, the
collision rate constant of water for the protonated nucleoside,
q
+
(H
2
O), is assumed equal to that for the neutral nucleoside,
q
(H2O). Using values at two temperatures, one obtains the enthalpy and the pK
values quoted for other temperatures.
g
Values derived from the ionization constant of water at 0 and 24oC,
CRC Handbook of Chemistry and Physics, 50th Edition
, Editor R.C. Weast, The Chemical Rubber Publishing Company, Cleveland, p. D-120.
h
See `Proton exchange theory'.
.
Thermodynamic and kinetic parameters of imino proton exchange in mononucleosides
(-7; 0; 15; 25oC)
a
When the dissociation constant is much less than 1, the kinetics is first order,
and the exchange time is:
[tau]
ex
=[tau]
0
+ ([tau]
ex,acc,open
/K
diss
)
(10)
where [tau]
0
is the base-pair lifetime and [tau]
ex,acc,open
is the exchange time (the inverse of the exchange rate k
ex,acc,open
) from the open pair.
If transfer to a given acceptor proceeds from the open state as in the case of
the monomer, the exchange rates in the open state are given by eqs 2 and 3. Any
difference between the case of the monomer and that of the open state is
summarized by a factor [alpha]:
k
ex,acc,open
= [alpha]k
ex,acc
(11)
where k
ex,acc
refers to the monomer (eq 2). Similar relations hold for the k
+
rates which apply to protonated G. As previously discussed (10), [alpha] is not strongly catalyst-sensitive, and not too far from unity.
An important contribution to proton exchange is intrinsic catalysis, which
involves indirect transfer of the imino proton of guanosine (or thymidine) via
a water bridge to the imino position of the cytidine (or adenosine) opposite,
in the open pair. As in the case of direct proton transfer to water, intrinsic
catalysis does not involve a diffusion-controlled collision, but the collision formalism may still be used. In
view of the final transfer of the proton to bulk water, we again use the
conventional value, 55 M, for the C (or A) acceptor concentration.
Due to the existence of both neutral and N7-protonated G, the exchange kinetics can be complicated, and it is not even
first order in general. However, a first order description should suffice for
our NMR measurements, due to one of the following properties: the fast rate of
N7-deprotonation, the long characteristic times of the NMR measurements, the
fact that in most conditions, one of the two forms of G (neutral or N7-protonated) is responsible for most of the exchange, and the fact that the
exchange times in the conditions of measurement are much longer than the base-pair lifetimes. As a result, the exchange rate can be expressed as an
average over neutral and N7-protonated species, using averaged values for the base pair lifetime, for
the dissociation constant and for the transfer rate in the open state.
Labelling the averages with the prime symbol, we have for the base-pair lifetime [tau]'
0
:
1/[tau]'
0
= (1 - f
+
)/[tau]
0
+ f
+
/[tau]
+
0
(12)
Similarly, the open-state lifetime [tau]'
open
is given by:
1/[tau]'
open
= (1 - f
+
)/[tau]
open
+ f
+
/[tau]
+
open
(13)
and:
K'
diss
= [tau]'
open
/[tau]'
0
(14)
Each acceptor, whether external or intrinsic, brings its contribution to the
exchange rate from the open state, both for neutral and for N7-protonated G. In analogy with eq 3, the exchange rate k'
ex,open
from the open states is:
k'
ex,open
=(1 - f
+
)([Sigma]k
ex,acc,open
+ k
ex,int
)
+ f
+
([Sigma]k
+
ex,acc,open
+ k
+
ex,int
)
(15)
where we have singled out the intrinsic catalyst acting on neutral G and on N7-protonated G, and where each contribution is computed according to eq 2.
The pK and
q
values of relevant acceptors are given in Table
2
.
Lastly, the imino proton exchange time [tau]
ex
is given by:
[tau]
ex
[equivalent to] 1/k
ex
= [tau]'
0
+ 1/(K'
diss
.
k'
ex,open
)
(16)
The solution pH influences [tau]
ex
through the concentration of OH
-
and the extent of N7-protonation of G (the factor f
+
, which influences both [tau]'
0
and K'
diss
).
For example, at high pH (e.g
.
pH 11), exchange proceeds mainly by proton transfer from the neutral species to
the base OH
-
(the N
OH
- process), and the exchange time is much longer than the base-pair lifetime. By eqs 2, 5 and 11, one obtains:
1/[tau]
ex
= K
diss
.
k
ex,OH
-
,open
(17)
or:
1/[tau]
ex
= K
diss
.
[alpha]q
OH
-
.
10
pH-pKi
(18)
where q
OH
- is the rate constant of the collision of OH
-
with the monomer.
As a second example, in the
A
bsence of
A
dded
C
atalyst, the exchange time of the imino proton of a G
.
C pair at neutral pH is controlled by intrinsic catalysis, i.e. proton transfer
to cytidine. In eq 16, we can neglect both the first term and the contribution
of N7-protonated G. Hence:
1/[tau]
AAC
[approx] K
diss
.
k
ex,int
[approx] K
diss
.
(q
int
.
55)
.
10
pK(N3;C)-pK(N1;G)
(19)
where q
int
is the pseudo-collision rate constant corresponding to the conventional 55 M
concentration of the cytidine acceptor. As compared to direct transfer to water
(eq 7), the pK-dependent term is enhanced a million times by the larger pK of the
cytidine acceptor, 4.3 versus -1.7 for water (Table
2
).
As a third example, consider the exchange from a G
.
C pair by proton transfer to cytidine from N7-protonated G, whose proportion is given by eq 4. This is expected to be a
major exchange process at acid pH. The exchange time [tau]
+
AAC
is a function of K
+
diss
,
q
+
and the imino proton pK of N7-protonated G. In eq 15, the contribution of neutral G is negligible.
Instead of eq 19, we have:
1/[tau]'
AAC
[approx] f
+
/[tau]
+
AAC
[approx] f
+
K
+
diss
.
k
+
ex,int
[approx] f
+
K
+
diss
.
(q
+
int
.
55)
.
10
pK(N3;C)-pK(N1;HN7G+)
(20)
The pH dependence of this acid-induced process is contained in the fraction f
+
of N7-protonated G (eq 4).
Using eqs 18-20, one can determine the pH values pNB and pAN corresponding to the two
cross-overs from the pH-independent process to the basic, and to the acidic processes. They
are given by eqs 20 and 21, which may be compared with eqs 8 and 9:
pNB = pK
i
- pK(N1;G) + [pK(N3;C) + log(55)]
+ log(q
int
/([alpha]q
OH
-))
(21)
pAN = pK(N7) - [pK(N1;HN7G
+
) - pK(N1;G)]
+ log[(q
+
int
K
+
diss
)/(q
int
K
diss
)]
(22)
If the exchange rates are controlled by the processes of eqs 18-20, the exchange parameters can be determined from the measured rates.
First, one obtains [alpha]K
diss
by substituting the rate at basic pH in eq 18. Next, substituting in eq 21 the
experimental value of pNB (Fig.
4
4), one obtains the exchange rate ratio q
int
/([alpha]q
OH
-). Lastly, substituting pAN in eq 22, one obtains the ratio (q
int
K
diss
)/(q
+
int
K
+
diss
) which provides for a comparison of base-pair opening properties of neutral and protonated G. Such procedures were
used for computing Table
3
. Alternatively, the same parameters can be derived by substitution of directly
measured exchange rates in eqs 19 and 20.
a
1/[tau]
AAC
is the pH-independent exchange rate due to intrinsic catalysis.
b
pAN is the pH for which the rates of exchange
via
the pH-independent pathway and
via
the acid-induced pathway are equal.
c
pNB is the pH for which the rates of exchange
via
the pH-independent pathway and
via
OH
-
-catalysis are equal.
d
The product [alpha]K
diss
is the ratio of the rate of OH
-
-catalyzed exchange (or ammonia-catalyzed exchange, in cases II and VI) to that for 2'-3'cGMP (measurements II-V) or for deoxyguanosine (others).
e
This is the difference pAN - pAN(monomer), eq 30. The pAN(monomer) values at 0 and 25oC are obtained from those listed by linear intra-and extrapolation
vs.
1/T. The collision factors relative to H
2
O are those of the monomer.
f
This is the difference pNB - pNB(monomer), eq 29. The pNB(monomer) values at 0 and 25oC are obtained from those listed by linear intra-and extrapolation
vs.
1/T. The collision factor relative to H
2
O is that of the monomer.
g
The roman numbers correspond to those in Figure 5.
h
This work.
i
G
1
is the nucleoside paired with C1, etc.
j
Guéron,M., and Leroy,J.-L. (1992) In Eckstein,F., and Lilley,D.M.J. (eds)
Nucleic Acids and Molecular Biology
. Springer Verlag, Berlin, Vol. 6, pp. 1-22.
k
Kochoyan,M., thesis, unpublished. Thèse de doctorat de l'Université Paris VII, spécialité Biophysique, 1987; p. 80, p. 110.
l
J.-L. Leroy, unpublished data.
m
Non-symmetrical complex with chromomycin, which is bound to G3. G3
*
designates G3 of the other strand. Gao,X., and Patel,D.J. (1990)
Biochemistry
29, 10940-10956.
n
Complex with chromomycin. Leroy,J.-L., Gao,X., Guéron,M., and Patel,D.J. (1991)
Biochemistry
30, 5653-5661.
o
Complex with echinomycin. Leroy,J.-L., Gao,X., Misra,V., Guéron,M., and Patel,J. (1992)
Biochemistry
31, 1407-1415.
.
Parameters of imino proton exchange in G
.
C pairs
The pKs for protonation of OH
-
and of H
2
O
The transfer of a proton from H
3
O
+
to an acceptor A is written as:
A + H
3
O
+
<-> AH
+
+ H
2
O
(23)
The constant K
ma
(A) of the mass action law is given by:
K
ma
(A) = [AH
+
][H
2
O]/([A][H
3
O
+
])
(24)
or:
[AH
+
]/[A] = (K
ma
(A)/[H
2
O]) 10
-pH
(25)
whence:
pK = log
10
(K
ma
(A)/[H
2
O])
(26)
When the acceptor is OH
-
, eq 24 becomes:
K
ma
(OH
-
) = [H
2
O]
2
/([OH
-
][H
3
O
+
]) =[H
2
O]
2
/K
i
(27)
hence:
pK = log
10
(K
i
/[H
2
O])
(28)
where K
i
is the ionization constant of water, which is a function of temperature. For
instance, at 25oC, the value of K
i
is 1
.
10
-14
, and the pK is therefore 15.7.
When the acceptor is H
2
O, K
ma
is equal to 1 by eq 24. The pK of the H
3
O
+
/H
2
O couple is therefore simply log
10
(1/[H
2
O]), independent of temperature (Table
2
).
RESULTS
Imino proton exchange in mononucleosides
Proton abstraction from deoxyguanosine versus pH.
Except as noted, we focus on the -7oC data for guanosine (Fig.
1
1, left panel). Three processes can be distinguished (Table
1
): (i) At high pH, the exchange rate is proportional to OH
-
concentration. This is assigned to imino proton abstraction from neutral
guanosine by OH
-
, the N
OH
- process. Substituting the observed rate in eq 5, we compute the
collision rate constant q
OH
-. The result, indicated in Table
2
, is independent of the exact value of the imino proton pK, because the latter
is much smaller than that of OH
-
: each collision with OH
-
results in abstraction of the imino proton. (ii) Acid-induced catalysis is assigned to imino proton abstraction by water from
the fraction of guanosine which is protonated on N7 (
3
). From the observed acid-induced catalysis, and with the knowledge of the fraction of N7-protonated guanosine, eq 6 provides the value of the product q
+
H2O
.
10
-pK(N1;HN7G)
, where q
+
H2O
is the pseudo-collision rate constant and the pK is that of the imino proton of N7-protonated G. (iii) The experimental exchange time around pH 5 is
shorter than expected from the two processes just mentioned (Fig.
1
1). This is assigned to a third process, imino proton abstraction by water from
neutral guanosine, the N
H2O
process. The exchange time corresponding to this process, 84 ms at -7oC, is indicated by a horizontal line in Figure
1
1. Introducing this value in eq 7, we obtain for the collision parameter the
value given in Table
2
, q
H2O
= 104
.
10
9
s
-1
M
-1
.
Imino proton exchange in a G
.
C pair
The d(ATATAGATCTATAT) imino proton spectrum; the effect of formiate.
Figure
3
3 displays the spectra at different pH values. In the neutral pH spectrum,
assigned to the Watson-Crick duplex, the 12.6 p.p.m. peak was assigned to the G6 imino proton by
the NOESY cross-peaks with cytidine amino protons (mixing time 250 ms, 15oC, data not shown).
At lower pH, two imino proton peaks shift to high field. They are assigned to
bases T
1
and T2 of the terminal pairs, whose dissociation constant increases due to
protonation at the imino position (N1) of adenine (4).
At pH 5 and below, imino proton peaks appear at ~11 p.p.m. The integrated intensity is 15% of that of the main spectrum at
pH 4.5 and 50% at pH 3.5. They may arise from a frayed duplex or from a single
strand. Hoogsteen pairs, in which the N3 imino proton of protonated C is
hydrogen-bonded to N7 of G, and in which the G imino proton is exposed to the
solvent, could also contribute. We frequently observe such peaks in low-pH spectra of self-complementary oligo-deoxynucleotides. They have also been reported in 200 base-pair fragments of calf thymus DNA (
12
).
The bottom spectrum shows the effect of a proton acceptor (formiate, 0.17 M) on
proton exchange at pH 4.17, a pH where acid-induced exchange is manifest (Fig.
4
4). There is no discernible change in the line-broadening, nor in the longitudinal relaxation (not shown), of the G6
imino proton in the duplex (at 12.6 p.p.m.). This shows that catalysis of
exchange from the pair is >= 400 times slower than for exchange from the monomer (Fig.
2
2), as expected if the dissociation constant of the pair is small. In contrast,
the effect of formiate on the line width is obvious in the case of the weak
spectrum, which corresponds to protons presumably exposed to the solvent, and
for the peak, in the duplex spectrum, of the imino proton of the terminal pair
A
.
T
1
, whose dissociation constant, measured at neutral pH, is large. This point will
be discussed below.
Proton exchange from the G
.
C pair of [d(ATATAGATCTATAT)]
2
. The exchange time of the G6 imino proton versus
pH is plotted in Figure
4
4. At high pH, the straight line with a slope of -1 is characteristic of catalysis by OH
-
. It is displaced vertically by a factor of 1.3
.
10
-5
from the corresponding line for the monomer: this number is equal to [alpha]K
diss
according to eq 5 and 18. Around neutral pH, the pH-independent exchange process is only 13 times slower than the N
H2O
process acting in the monomer. Below pH 5, the plus one slope corresponds to an
acid-induced process, which, similarly, is only 4.6 times slower than the acid-induced process P
H2O
in the monomer.
The unexpectedly fast exchange at neutral and acid pH is readily explained by
intrinsic catalysis which is more efficient than direct transfer to water, due
to the larger pK of the intrinsic acceptor, cytidine N3, as discussed in the
comparison of eqs 7 and 19. The fit in terms of transfer from G (the N
C
,process, eq 19) and from HN7G+ (the P
C
process, eq 20) is indicated by full lines. It provides the parameters pAN, pNB,
and those derived from them, which are listed in Table
3
.
Acid-induced exchange of the G imino proton in other conditions.
We have collected data on acid-induced imino proton exchange in Watson-Crick G
.
C pairs, including pairs from DNA duplexes (pairs I-VII) and from DNA complexes with chromomycin, a minor groove drug, and
with echinomycin, a bis-intercalator (pairs VIII-XII). The parameters relevant for exchange are presented in Table
3
, and the exchange times are plotted schematically in Figure
5
5.
DISCUSSION
Proton exchange from G
.
C pairs at acid pH is nearly as fast as from isolated guanosine. This seems at
first in conflict with the well-established result that exchange catalysis by transfer to a
given
proton acceptor occurs in the case of a base pair as it does from the
mononucleoside, except for a slowing down by a factor equal to the dissociation
constant: this is the case for instance when exchange is catalyzed by OH
-
, as happens at basic pH, or by ammonia etc. When the exchange rate is found to
be proportional to H
+
concentration, one could naively expect the same behaviour. But even though H
+
acts in the same way on monomer and open pair (it produces the N7-protonated G from which exchange occurs predominantly), nevertheless the
proton acceptor is not the same: for the monomer it is H
2
O, whereas for the open-state nucleoside, it is the complementary cytidine. This intrinsic
catalyst is more efficient than H
2
O by a factor whose logarithm is the difference in pK of the two acceptors,
about six units.
The intrinsic catalyst also operates at neutral pH, where its activity was first
detected (
1
), and where proton exchange (now from the non-protonated G) is again nearly as fast as from the monomer (Fig.
4
4).
In the present work, we provide evidence for this interpretation of acid-induced catalysis, based on exchange measurements in the monomers and in
numerous base pairs. The thesis is that acid-induced exchange of the G imino proton in base pairs is due to intrinsic
catalysis of exchange from protonated guanosine, occurring in the open state,
like exchange from neutral guanosine at neutral or basic pH, and that
furthermore, the properties of the base pair and of the open state are not
drastically changed by the N7-protonation of G.
The observations support this thesis, and furthermore they exclude tentative
interpretations of acid-induced exchange which do not rely on intrinsic catalysis: exchange from
the open state of a pair with a very large dissociation constant, and exchange
from a modified structure carrying Hoogsteen-paired G or unpaired G, in which the G imino proton is exposed.
Acid-induced imino proton exchange in the nucleoside monomers
It has been proposed previously that acid-induced imino proton exchange is due to abstraction by water of the imino
proton from N7-protonated guanosine (
3
), in contrast to abstraction from neutral guanosine which is the process
responsible for pH-independent exchange. Since the proton acceptor, H
2
O is the same in acid-induced and in pH-independent exchange, one may assume that the pseudo-collision factor is the same in the two processes. From their
relative efficiencies (e.g
.
the value of pAN), one determines by eq 9 that the pK of the imino proton
changes by ~2.3 units upon N7 protonation, leading to the pK values for HN7G
+
listed in Table
2
.
If the source of the pH dependence is N7 protonation, the log of the exchange
time should be a linear function of pH with slope 1 when the pH is larger than
the pK of N7 (~2.3), and should level off at lower pH. A fit of the data to this
interpretation is shown by the full line in Figure
1
1. There is a good fit down to pH 2, and the expected levelling off may be
discernible. However the exchange times <pH 2 are too short, due perhaps to experimental error (caused by the large
linewidth) or to the influence of other protonated species of G (for instance G
protonated on the amino group) on proton exchange.
The marginal acid-induced imino proton exchange in thymidine (Fig.
1
1) is similarly ascribed to protonation at O2 or O4, which occurs with a pK
around 0 (
11
), as pointed out in the `Results' section.
The notion that acid-induced exchange corresponds to a change in the properties of the proton
donor
may be tested with other proton acceptors than water, such as formiate and
cytidine. The latter is of particular interest since it is the intrinsic
acceptor in G
.
C pairs. The effect of these acceptors varies with pH as does the effect of
water (Fig.
2
2), and the simplest explanation is that, like water, they abstract the imino
proton of G or of N7-protonated G, depending on pH. The corresponding fits to eqs 2-4 are shown in Figure
2
2. Their quality supports the model, and notably the features of exchange
catalyzed by the acceptor and not by the conjugate acid.
In fitting the data, one must use for consistency the imino proton pK value of
HN7G
+
determined in the study of the H
2
O acceptor. The only adjustable parameters are the collision factors of the
acceptor with G and with HN7G
+
. The fitting values are given in Table
2
. Those for neutral G may be compared to the values compiled for other acceptors
(
10
). The values for HN7G
+
are similar if not identical to those for G (Table
2
). The enhancement of the collision factor of formiate with protonated G may be
due to the negative charge of formiate. The reduction in the case of cytidine
may be due to lowering of the effective acceptor concentration by trapping of
cytidine in C
.
C
+
pairs.
Discounting these considerations, one could determine the imino proton pK in
HN7G
+
by the exchange catalysis by formiate or cytidine, instead of H
2
O. Assuming in each case the same collision factor for G and for HN7G
+
, the pK values would differ by no more than 0.5 from those derived from
catalysis by H
2
O (Table
2
). This gives a measure of the uncertainties of the model.
Overall, the experimental results are in agreement with the proposition of
McConnell and Politowski. Furthermore they provide arguments against
alternative explanations of the acid-induced catalysis which we now consider, and which would assign the pH-dependence of exchange catalysis by formiate or cytidine to the
conjugate acid, whose concentration is pH-dependent. For instance, N3-protonated cytidine could bind to N7 of guanosine, forming a
Hoogsteen pair, in which the G imino proton would remain accessible to attack
by H
2
0. This process can be excluded, because its efficiency should level off when
the pH is less than the pK of cytidine, ~4.5, contrary to observation (Fig.
2
2). Furthermore, in the Hoogsteen pair, N7 would be protonated less than one
percent of the time (considering the difference between the pK of cytidine and
that of N7, Table
2
), so that the exchange contribution of cytidine would be <1% of the maximum effect of water. By reference to Figure
1
1, this corresponds to an exchange time of 5 ms at 15oC, much longer than the exchange time at pH 3 in 123 mM of cytidine, which
is 65 [mu]s, as deduced from the right bottom panel in Figure
2
2.
Another possibility is to combine binding of the conjugate acid to N7 with
attack of the imino proton by the acceptor instead of by H
2
O. This leads to a rate which varies as the product of the concentrations of the
acceptor and of its conjugate acid, whereas the good fit of the data points to
the model proposed above (bottom panels of Fig.
2
2) shows that the exchange time is independent of the conjugate acid
concentration.
Acid-induced imino proton exchange in G[middot]C pairs
Interpretation of exchange as from a unique open state.
We now show that base-catalyzed, pH-independent, and acid-induced imino proton exchange from base pairs can all be
explained as occurring from the same open state of the pair, with only modest
differences between the neutral G
.
C pair and the HN7G
+.
C pair.
The simplest characterization of the three exchange regimes is by the cross-over parameters pNB and pAN, whose experimental values are given in Table
3
. Assuming that the proton acceptors are OH
-
and H
2
O for the monomer, and OH
-
and cytidine for base pairs, we have by the combination of eqs 8 and 21, and of
eqs 9 and 22:
pNB(pair) = pNB(guanosine) + [pK(N3;C) + log(55)]
+ log(q
int
/([alpha]q
H2O
))
(29)
where [alpha] refers to OH
-
catalysis (eq 18); and:
pAN = pAN(guanosine) - log(q
+
H2O
/q
H2O
) + log(q
+
int
/q
int
)
+ log(K
+
diss
/K
diss
)
(30)
where the
q
factors in the first log term are those of the monomer.
The value of pNB is a measure of the relative efficiencies of attack by OH
-
and of concerted proton transfer, both from the open state of the pair.
Concerted transfer could be more sensitive to the structure of the open state,
and a change in its rate would shift pNB. As an example, intrinsic catalysis is
much slowed down in the case of terminal pairs, presumably because of wider
opening (
4
), and this shifts pNB down. Indeed, the smallest pNB value in Table
3
is for a terminal pair (case II, pNB = 7.8).
With respect to the monomer, pNB is shifted by about +6 units (the terms in
brackets) due to the pK difference between the proton acceptors cytidine and
water. The log term, which involves the ratio of the pseudo-collision factors for transfer to cytidine in the open state and for
transfer to water in the monomer, also contributes to the shift. We have no
independent estimate for this ratio. The experimental values of pNB, ranging
from 7.8 to 9.8, correspond to values of the log term between -4.35 and -2.22 (Table
3
, last column).
If the exchange parameters of the HN7G
+.
C and G
.
C pairs were the same, all the log terms in eq 30 would be zero, and pAN would
be the same as for the monomer, ~4.4. (In the representation of Figure
4
4, this would correspond to equal vertical distances between the horizontals and
between the lines of slope 1 on the acid side). In fact, the average pAN in
Table
3
is ~1.2 units above that of the monomer, indicating some change in properties
of the base pair (open or closed) upon G protonation. The increase in pAN could
be due to an equal increase of the logarithm of the dissociation constant when
G is protonated, which could be rationalized if base-pair opening brings the N7 group near to a phosphate group, a conformation
that could be stabilized in the case of HN7G
+
by the proton charge and/or by H-bonding, direct or through a water molecule. Opening towards the major
groove of B-DNA does indeed bring GN7 closer to the 3'-oxygen of the 5'-neighbor of the same strand.
More important than the origin of the pAN shift is its relatively small value
and variation, <2 units across a collection of base pairs whose dissociation constant spans
seven orders of magnitude (Table
3
). The range of pAN, a function of the properties of two different base pairs,
HN7G
+.
C and G
.
C, is in fact smaller than that of pNB, a parameter which compares processes
involving the single pair G
.
C. This small range suggests that exchange takes place from an open state of the
pair with similar properties across the entire pH range. If instead the acid-induced process occurred in the closed state, or in an open state
unrelated to the open state responsible for the pH-independent process, pAN would be expected to be strongly affected by the
latter's dissociation constant, and this is contrary to observation.
Acid-induced proton exchange without intrinsic catalysis.
In this section, we examine tentative explanations of efficient acid-induced catalysis which are not based on intrinsic catalysis.
Since the acid-induced exchange rate is comparable to that of the monomer, one
explanation would be that the base-pair dissociation constant is very large when G is protonated, for
instance 1/4.6 in the case of the pair of Figure
4
4.
There are three arguments against this mechanism. One is provided by the case of
terminal pairs, such as case II (Table
3
, Fig.
5
5) in which acid-induced exchange is even faster than that of the monomer: this requires a
more efficient catalyst than H
2
O, as in a similar observation which provided one of the original arguments for
intrinsic catalysis at neutral pH (
1
). A second argument is that changing to a large dissociation constant should
change the chemical shift, as observed for terminal base pairs upon heating or
upon going to basic pH (
4
). In the duplex spectra in Figure
3
3, there is a large shift of the T
1
imino proton peak at acid pH, and of T2 as well, which we ascribe to the
increase in the rather large dissociation constant, upon protonation of the
complementary adenosine. But the G6 peak does not move. The third argument,
based on the studies of the external catalysts formiate and cytidine, is the
strongest. It is simply that, as stated in the `Results' section, these
catalysts are >= 400 times less efficient in the duplex than in the monomer. This is
incompatible with a dissociation constant of 0.1 or larger for the HN7G
+.
C pair. But it is exactly what is expected if the dissociation constant is <1/400, as it is indeed at neutral or basic pH.
Another mechanism of acid-induced exchange is suggested by the imino proton spectrum which develops
upfield in DNA at acid pH (e.g. in Fig.
3
3), a spectrum compatible with the presence of G
.
C
+
Hoogsteen pairs. In such a pair, the G imino proton exchanges quickly because
it is exposed to the H
2
O acceptor. If the structure which harbors the Hoogsteen pair is in sufficiently
fast exchange with the duplex, this will contribute to proton exchange as
measured on the duplex spectrum. It was tentatively proposed in a previous
publication (
10
) that this could explain acid-induced exchange.
However, this process is not efficient enough. As explained in the discussion of
proton exchange from the monomer, GN7 would be protonated <1% of the time in the Hoogsteen pair. At pH 4.17, with a relative population of
at most 20% (Fig.
3
3), the Hoogsteen-bearing structure will provide an exchange rate which is 2/1000 times that
of the fully protonated monomer, which is ~10
4
s
-1
, according to Figure
1
1. This leads to a rate of 20 s
-1
, or an exchange time of 50 ms, comparable to 30 ms observed (Fig.
4
4). But this rate would be achieved only if exchange between the two forms
corresponding to the two spectra is fast enough. Again, the formiate experiment
proves decisive: broadening of the upfield spectrum by formiate (lowest
spectrum in Fig.
3
3) is not passed on to the low-field (duplex) spectrum, which is unchanged even in its longitudinal
relaxation.
Exchange transferred from the imino proton of unpaired HN7G
+
is excluded by similar arguments.
Application to NOE measurements.
Both the theory and the observations show that efficient acid-induced exchange of the imino proton with water is a general feature of G
.
C pairs. Since such exchange may affect NOE transfer, one should choose for NOE
measurements a pH at which the rate is small. Table
3
and Figures
4
4 and
5
5 show that in the case of base-paired imino protons, pH 7 is optimum in all cases, and is in some cases
better than pH 5.2, the pH for slowest imino proton exchange from the monomer.
The case of A
.
T pairs.
In A
.
T pairs, exchange of the imino proton proceeds as for G
.
C pairs at basic and neutral pH, but acid-induced exchange has not been characterized (
2
). This is understandable, since the pH below which it would be visible should
be two units less than for G
.
C pairs (compare the pKs in Table
2
, and the pAN values in Fig.
1
1). In such conditions, the interpretation of accelerated exchange is dubious
because of extensive duplex dissociation.
CONCLUSION
We have shown that the general and surprisingly efficient acid-induced exchange of the guanosine imino proton in Watson-Crick pairs is caused by intrinsic catalysis in the open state of
the HN7G
+.
C pair, a pair whose dissociation constant is moderately enhanced by the
protonation of G. Together with the case of Watson-Crick A
.
T and G
.
C pairs at neutral pH in the absence of external catalysts, and the case of C
.
C
+
pairs (
13
), this is a third situation of general occurrence where intrinsic catalysis
provides the fastest exchange mechanism, by far.
REFERENCES
1 Guéron, M., Kochoyan, M. and Leroy, J.-L. (1987) Nature 328, 89-92.MEDLINE Abstract
2 Guéron, M., and Leroy, J.-L. (1992) In Eckstein,F. and Lilley,D.M.J. (eds) Nucleic Acids and Molecular Biology. Springer Verlag, Berlin, Vol. 6, pp. 1-22.
3 McConnell, B. and Politowski, D. (1984) Biophysical Chemistry 20, 135-148.
4 Nonin, S., Leroy J.-L., and Guéron, M. (1995) Biochemistry 34, 10652-10659.MEDLINE Abstract
5 Leroy, J.-L., Gao, X., Guéron, M., and Patel, D. J. (1991) Biochemistry 30, 5653-5661.MEDLINE Abstract
6 Leroy, J.-L., Gao, X., Misra, V., Guéron, M., and Patel, D. J. (1992) Biochemistry 31, 1407-1415.MEDLINE Abstract
7 Guéron, M., and Leroy, J.-L. In James,T. D. (ed.) Methods in Enzymology, Vol. 261. Academic Press, San Diego. In the press.
9 Buchner, P., Maurer, W., and Ruterjans, H. (1978) J. Magn. Reson. 29, 45-63.
10 Guéron, M., Charretier, E., Hagerhorst, J., Kochoyan, M., Leroy,J.-L., and Moraillon, A. (1989) In Sarma, R.H., and Sarma,M.H. (eds.) Structure and Methods, Proceedings of the Sixth Conversation in Biomolecular Stereodynamics. Adenine Press, NY, Vol. 2, pp. 113-137.
11 Dawson, R. M. C., Elliott, D. C., Elliot, C., Jones, K. M. (1969) Data for Biochemical Research. Oxford University Press, Ely House, W1, p. 140.
12 Banville, D. L., Marzilli, L. G., and Wilson, W. D. (1986) Biochemistry 25, 7393-7401.MEDLINE Abstract
13 Leroy J.-L., Gehring, K., Kettani, A., and Guéron, M. (1993) Biochemistry 32, 6019-6031.
