Further studies on the buffer-catalyzed cleavage and isomerization of uridyluridine. Medium and ionic strength effects on catalysis by morpholine, imidazole, and acetate buffers help clarify the mechanisms involved and their relationship to the mechanism used by the enzyme ribonuclease and by a ribonuclease mimic

Article abstract of DOI:10.1021/ja9526933

The cleavage and isomerization of 3',5'-uridyluridine catalyzed by morpholine buffers and by imidazole buffers has been reinvestigated. The key evidence for a previously proposed partitioning mechanism - in which the buffer acid BH⁺ converts the substrate to a phosphorane monoanion intermediate which then partitions either to the 2',5' isomer of the substrate or (with buffer base catalysis) to the cleavage product - is confirmed. The negative catalytic effect of the buffer base on the isomerization reaction is not due to a medium effect. Indeed the medium effect on this reaction is in the opposite direction, strengthening the catalytic evidence. However, this branching mechanism with sequential bifunctional catalysis of the cleavage reaction is accompanied by an additional cleavage path using the buffer base only. This additional path, for which several alternative mechanisms are possible, is required by the results of improved studies on the imidazole catalysis. These show that the previously reported decrease in rate at a low BH⁺/B ratio is due to ionic strength effects. The relative importance of these two pathways - one with kinetic dependence on the buffer acid and one without such dependence - depends on the buffer basicity/acidity. With the acidic buffer acetate/acetic acid, a buffer-acid-catalyzed mechanism for the cleavage and the isomerization is dominant. A bifunctional mechanism, in which one step involves simultaneous acid-base catalysis, seems most likely. The medium effects of added dioxane on all these reactions are sensible in terms of the detailed mechanisms proposed. The relationship of these results to the mechanisms of catalysis by ribonuclease and by an enzyme mimic is discussed.

Full text of DOI:10.1021/ja9526933

+
+
6588
J. Am. Chem. Soc. 1996, 118, 6588-6600
Further Studies on the Buffer-Catalyzed Cleavage and
Isomerization of Uridyluridine. Medium and Ionic Strength
Effects on Catalysis by Morpholine, Imidazole, and Acetate
Buffers Help Clarify the Mechanisms Involved and Their
Relationship to the Mechanism Used by the Enzyme
Ribonuclease and by a Ribonuclease Mimic
Ronald Breslow,* Steven D. Dong, Yael Webb, and Ruo Xu
Contribution from the Department of Chemistry, Columbia UniVersity,
New York, New York 10027
ReceiVed August 8, 1995X
Abstract: The cleavage and isomerization of 3′,5′-uridyluridine catalyzed by morpholine buffers and by imidazole
buffers has been reinvestigated. The key evidence for a previously proposed partitioning mechanismsin which the
buffer acid BH+ converts the substrate to a phosphorane monoanion intermediate which then partitions either to the
2′,5′ isomer of the substrate or (with buffer base catalysis) to the cleavage productsis confirmed. The negative
catalytic effect of the buffer base on the isomerization reaction is not due to a medium effect. Indeed the medium
effect on this reaction is in the opposite direction, strengthening the catalytic evidence. However, this branching
mechanism with sequential bifunctional catalysis of the cleavage reaction is accompanied by an additional cleavage
path using the buffer base only. This additional path, for which several alternative mechanisms are possible, is
required by the results of improved studies on the imidazole catalysis. These show that the previously reported
decrease in rate at a low BH+/B ratio is due to ionic strength effects. The relative importance of these two
pathwayssone with kinetic dependence on the buffer acid and one without such dependencesdepends on the buffer
basicity/acidity. With the acidic buffer acetate/acetic acid, a buffer-acid-catalyzed mechanism for the cleavage and
the isomerization is dominant. A bifunctional mechanism, in which one step involves simultaneous acid-base catalysis,
seems most likely. The medium effects of added dioxane on all these reactions are sensible in terms of the detailed
mechanisms proposed. The relationship of these results to the mechanisms of catalysis by ribonuclease and by an
enzyme mimic is discussed.
Introduction
We have been studying the cyclization/cleavage of ribonucleic
acids catalyzed by buffers. We wanted to determine the
preferred mechanism used by normal nucleotides, not those more
reactive analogs with such good leaving groups as p-nitro-
phenoxide ion. Such reactive leaving groups make the kinetic
analyses easier, but they can certainly change the mechanism
used. With the less reactive normal nucleotides, rather con-
centrated buffers are needed to achieve reasonable rates.
Our earliest work involved imidazole buffers, imitating the
imidazole groups that are catalytic in the enzyme ribonuclease
A. In this work we examined polyuridylic acid,1,2 3′,5′-
uridyluridine (3′,5′-UpU, 1),3-8 and adenosyladenosine (ApA).9
We found that imidazole buffers catalyzed the cyclization/
cleavage of all these substrates (Figure 1). We also saw that
X Abstract published in AdVance ACS Abstracts, July 1, 1996.
(1) Corcoran, R.; Labelle, M.; Czarnik, A. W.; Breslow, R. Anal.
Biochem. 1985, 144, 563-568.
(2) Breslow, R.; Labelle, M. J. Am. Chem. Soc. 1986, 108, 2655-2659.
(3) Breslow, R.; Huang, D.-L.; Anslyn, E. Proc. Natl. Acad. Sci. U.S.A.
1989, 86, 1746-1750.
(4) Anslyn, E.; Breslow, R. J. Am. Chem. Soc. 1989, 111, 4473-4482.
(5) Breslow, R.; Anslyn, E.; Huang, D.-L. Tetrahedron 1991, 47, 2365-
2376.
(6) Breslow, R.; Huang, D.-L. Proc. Natl. Acad. Sci. U.S.A. 1991, 88,
4080-4083.
(7) Breslow, R. Acc. Chem. Res. 1991, 24, 317-324.
(8) Breslow, R. J. Mol. Catal. 1994, 91, 161-174.
(9) Breslow, R.; Huang, D.-L. J. Am. Chem. Soc. 1990, 112, 9621-
9623.
Figure 1. Cyclization/cleavage of a dinucleotide accompanied by
isomerization.
these buffers catalyzed the isomerization of 3′,5′-UpU to 2′,5′-
UpU (2), and vice versa, and similarly in the ApA series.
S0002-7863(95)02693-X CCC: $12 00 © 1996 American Chemical Society

+
+
CleaVage and Isomerization of Uridyluridine
J. Am. Chem. Soc., Vol. 118, No. 28, 1996 6589
Figure 2. Our proposed mechanism for the sequential acid-base catalyzed cleavage of 1 by imidazole or morpholine buffers, with branching from
the phosphorane intermediate 5 that leads to isomerization. This simplified scheme omits contributions that are made by nonbuffer species such as
H+ and OH-, but they are included in Figure 3 and in previous treatments.11,12 Also, as discussed previously, the pseudorotation of 5 may well
involve its prior protonation to a neutral phosphorane. Note that the reversible protonation, followed by B catalysis, is just the specific acid/general
base version of BH+ catalysis. Its overall rate depends on [BH+], but not on [H+].
Our early work was performed without control of ionic
strength. We saw apparent bell-shaped pH vs rate profiles for
the cleavage process, indicating that both imidazole (Im) and
imidazolium ion (ImH+) were involved in the catalysis, and in
a sequence with a phosphorane intermediate. Such a pH vs
rate profile does not indicate which is the first catalyst and which
the second in such a sequence. Evidence on this was obtained
from a study of the isomerization reactions. They showed
catalysis by ImH+, and not by Im. In fact, with a constant
concentration of ImH+ the rate of isomerization actually
decreased as Im was added.7,9
Such slowing of the rate by a catalyst, which we referred to
as negatiVe catalysis,10 is consistent with a sequential process
in which an intermediate branches: one branch leads to the
isomer, while the other leads to cleavage with Im catalysis
(Figure 2). Thus increasing [Im] increases the fraction of the
intermediate that partitions along the cleavage path, leading to
a decrease in the observed rate of isomerization.
concentrations with the same pH’s.2,4,9 However, there were
ambiguities in this study as well. (1) Ionic strength was not
controlled, and it could be important. (2) The correction for
the changing pH may not have been successfully made simply
by the subtraction we did. In addition, (3) we saw that the
isomerization reaction slowed on addition of Im; because of
concern about medium effects, described further below, it was
important to show that this slowing was due to catalysis of one
leg of a partitioning mechanism, as in Figure 2, not simply
because added Im made the medium less polar.
To address the first two points, we studied the cleavage and
isomerization of 3′,5′-UpU by morpholine buffers.11,12 Mor-
pholine is more basic than is imidazole, so the negative catalytic
(10) Much ado has been made about our use of the phrase “negative
catalysis”. This term was first used by L. P. Hammett (Physical Organic
Chemistry, 1st ed.; McGraw-Hill Book Company: New York, 1940; p 398).
It simply means that a species that is functioning as a catalyst (i.e., it is not
changed at the end of the reaction, and it is not simply an inhibitor of another
catalyst) is observed to slow a reaction. Of course it does it by speeding
another reaction in a branching process. I do not retract the use of this
perfectly apt phrase by us or by Hammett.
In such a study the pH changes as buffer ratios change, and
we corrected for this by subtracting the rates at lower buffer

+
+
6590 J. Am. Chem. Soc., Vol. 118, No. 28, 1996
Breslow et al.
Figure 3. A condensed and simplified version of Figure 2 by which
the substrate 1 of Figure 2 is reversibly converted to the phosphorane
monoanion 5 with catalysis by the buffer acid BH+ or nonbuffer species
(H+, OH-, H2O), and 5 is then converted to isomer 2 or, with catalysis
by B or nonbuffer species, to cleavage products P1 and P2. The steady-
state (in 5) equations for cleavage (a) and for isomerization (b) are
shown. The constants kn, k-n, and k′n include concentrations of such
nonbuffer species as OH- and H+, so they are functions of pH. The
isomerization reaction shows negatiVe catalysis (cf. ref 10) by B, which
appears only in the denominator of eq b.
Figure 4. Pseudo-first-order rate constants as a function of buffer
concentration for the isomerization of 3′,5′-UpU (1) to 2′,5′-UpU (2),
from ref 12. All runs were at 80 °C, performed at least in duplicate
with 9/1 morpholine/morpholine hydrochloride, with the buffer con-
centration being varied and the ionic strength being maintained constant
at 0.2 M with NaCl. As described previously (ref 12), the measured
pH is constant under these conditions. The solid line is that predicted
(ref 12) from the kinetic treatment of the (expanded) mechanism of
Figure 2. The dashed line shows an increase at high buffer concentra-
tions that we ascribe to a medium effect (see Results and Discussion).
effect of the buffer base should be accentuated. Furthermore,
we kept the buffer ratio constant and controlled ionic strength,
while simply increasing the concentration of a given buffer. We
saw that the pH was constant under these conditions. The
cleavage rate simply increased with increasing buffer concentra-
tion at various ratios, but the isomerization rate showed negatiVe
catalysis again. That is, increasing the concentration of mor-
pholine buffer led to a decrease in the isomerization rate with
a 9/1 B/BH+ buffer at constant pH and ionic strength, and also
at 95/5 B/BH+, but the decrease was smaller with an 8/2 buffer.
This showed that it was the B component of the buffer that
slowed the reaction (or a kinetically equivalent OH-/BH+-
combination), as our partitioning mechanism would predict. The
essence of the argument is presented in Figure 3. The detailed
curves (e.g., Figure 4) could be fitted to the equations that such
a mechanism requires,11,12 but some questions remained.
First of all, the equations predicted that the initial decrease
in rate of isomerization with increasing concentration of
morpholine buffer would be followed by a plateau, as was
observed (Figure 4). However, at higher buffer concentration
we saw an increase in rate. This increase requires either an
additional mechanism at higher buffer concentration or a
medium effect13 as high buffer concentration makes the solvent
less polar.12 The other question was whether a medium effect
could explain even the initial decrease in rate, point 3 above.
Recently Kirby and Marriott have reported a study of medium
effects on the cyclization/cleavage of compound 3.14 As they
indicated, 3 is hundreds of times more reactive than is UpU,
because of the better leaving group, and such increased reactivity
could well change the mechanism. We15 and others16 have
studied the imidazole buffer catalyzed cyclization/cleavage of
the even more reactive compound 4 and see that only the Im
plays a catalytic role. Thus whatever medium effects there are
for substrate 3 might not be relevant to processes involving
normal nucleotides such as UpU. However, the point certainly
needed to be addressed.
With improved instrumentation the otherwise tedious kinetic
studies with normal nucleotides such as UpU are more readily
done. We have now performed new studies of ionic strength
and medium effects on the imidazole and morpholine catalyses
of cleavage and isomerization of 3′,5′-UpU. We have also
examined the cleavage and isomerization of 3′,5′-UpU by acetate
buffer. From these studies have emerged two conclusions: (1)
the originally proposed BH+/B sequence for cleavage of UpU
by imidazole buffers (Figure 2) seems to play a role in these
reactions. However, (2) this mechanism is accompanied by a
cleavage process that has kinetic dependence only on B.
Some of the original evidence for our earlier mechanistic
conclusions does not stand up to a careful re-examination with
attention to the medium, but other evidence is actually strength-
ened when all the factors are fully controlled. The bifunctional
mechanism is consistent with our findings with a synthetic
ribonuclease artificial enzyme,7,17-19 and with our proposals for
the enzyme itself.7 The two parallel processes are also related
to theoretical treatments done by others.
(11) Breslow, R.; Xu, R. Proc. Natl. Acad. Sci. U.S.A. 1993, 90, 1201-
1207.
(12) Breslow, R.; Xu, R. J. Am. Chem. Soc. 1993, 115, 10705-10713.
(13) First suggested by Charles Perrin as an anonymous referee of ref
12.
(14) Kirby, A. J.; Marriott, R. E. J. Am. Chem. Soc. 1995, 117, 833-
834.
(15) Chapman, W. H.; Breslow, R. Unpublished work.
(16) (a) Davis, A. M.; Hall, A. D.; Williams, A. J. Am. Chem. Soc. 1988,
110, 5105-5108. (b) Davis, A. M.; Regan, A. C.; Williams, A. Biochemistry
1988, 27, 9042-9047.
(17) Anslyn, E.; Breslow, R. J. Am. Chem. Soc. 1989, 111, 5972-5973.
(18) Anslyn, E.; Breslow, R. J. Am. Chem. Soc. 1989, 111, 8931-8932.
(19) Breslow, R.; Graff, A. J. Am. Chem. Soc. 1993, 115, 10988-10989.

+
+
CleaVage and Isomerization of Uridyluridine
J. Am. Chem. Soc., Vol. 118, No. 28, 1996 6591
This shows that the negative rate effect of morpholine on
the isomerization reaction is a true catalytic effect, and indeed
our raw data underestimated the effect. If we correct for the
medium effect, as indicated by the dioxane results, then the
negative catalytic effect of morpholine is even larger. This is
support for our proposal that the isomerization and the cleavage
reaction branch from a common intermediate, with the cleavage
branch being catalyzed by B. It does not rule out a kinetically
equivalent alternative pathway for the cleavage reaction, as will
be discussed below.
In our earlier work we had seen that the rate of isomerization
showed a complex behavior as a function of buffer concentra-
tion.12 As Figure 4 shows, the rate first decreases, then levels,
then rises again at higher buffer concentration. The decrease
and then leveling are as expected from our mechanism, from
the equations that describe our mechanism (cf. Figure 3).
However, the rise in rate at buffer concentrations higher than
1.0 M required an additional explanation. We pointed out12
that this could be either a medium effect or the addition of
further catalytic terms at high buffer concentrations, but with
our new evidence we prefer the explanation14 in terms of a
medium effect.
As we show in Figure 5B, the addition of dioxane as a mimic
of the medium effect of morpholine leads to a significant rate
increase. After the rate drops to a plateau at 1.0 M in Figure
4, in accord with the kinetic equations, the medium effect keeps
building with more buffer. This then must lead to an increase
in rate after the plateau, as we observe.
These medium effects are sensible. In the isomerization
reaction a substrate monoanion and a catalyst monocation first
neutralize each other by reversible proton transfer, and then the
phosphorane monoanion is formed in a step that restores charge
separation. However, if the transition state occurs during that
second step, the proton transfer will not yet be complete, so
the net result is a decrease in charge on going from the original
reactants to the transition state. This should lead to an increase
in the rate when the medium is made less polar, as we observe.
If isomerization involves protonation of 5 to form a neutral
phosphorane, this will also lead to an increased rate with a less
polar medium. Thus the medium effect supports both our
evidence for this aspect of the mechanism and our particular
interpretation of that evidence.
The medium effect on the cleavage reaction requires further
comment. Even if the only pathway for cleavage were that
shown in Figure 2, one would still expect the observed decrease
in cleavage rate as the medium is made less polar. In the
cleavage branch we invoke the conversion of a B of the buffer
to a BH+, and the phosphorane intermediate to a dianion. This
increase in polarity could explain the slowing with a less polar
medium.
However, evidence discussed later indicates that there is an
additional pathway for the cleavage reaction, one that has kinetic
dependence only on the buffer base B. With a strongly basic
buffer like morpholine, this is the major cleavage path (with
acetate buffer, described later, the buffer acid dominates the
catalysis). As we will describe, this B-catalyzed cleavage
process (one possible mechanism is shown in Figure 6; others
are discussed later) also has a transition state more polar than
the starting materials, and it should also be slowed by the
addition of dioxane. Since this second path is the major one
with basic buffers, the medium effect on cleavage reflects rate
changes in this second path.
Figure 5. Pseudo-first-order rate constant for (A) the cleavage of 3′,5′-
UpU and (B) the isomerization of 3′,5′-UpU to 2′,5′-UpU, both vs the
buffer concentration at 60 °C: (Os) 9/1 morpholine/morpholine
hydrochloride; (4- - -) 0.2 M morpholine buffer plus an amount of
dioxane that brings the total concentration to the amount shown. The
ionic strength was maintained at 0.2 M for all runs. The medium effect
decreases the rate of the cleavage reaction, but it increases the rate of
isomerization, opposite to the effect of the buffer.
Results and Discussion
Medium Effects on Morpholine Catalysis. With morpho-
line buffer we had already operated at constant ionic strength
and constant pH.11,12 Here the major question was the effect
of the change in medium as buffer concentration increases. The
increase in [BH+] is compensated by adjusting the ionic strength
with NaCl, but the increase in [B] makes the medium less polar.
With our new instrumentation that permits HPLC kinetic studies
with automatic sampling, we have repeated the study of the
cleavage and the isomerization of 3′,5′-UpU catalyzed by 9/1
morpholine/morpholinium ion buffer at constant ionic strength,
as a function of total buffer concentration. For instrumental
reasons the study was performed at 60 °C, not the 80 °C of the
previous work. We saw that as before the cleavage rate
increased as the buffer concentration increased (Figure 5A), and
again we saw that the isomerization rate decreased with
increasing buffer concentration (Figure 5B).
Kirby and Marriott added dioxane or pyrrole as mimics of
imidazole that change the medium, but not the catalyst
concentration.14 Thus we examined the effect of added dioxane
on the rate of cyclization/cleavage and the rate of isomerization
of 3′,5′-UpU. The rate constants are listed in the supporting
information and plotted in the various figures. The addition of
dioxane to a morpholine buffer system slightly decreases the
rate of the cleavage reaction (Figure 5A). However, Figure 5B
reconfirms the negative rate effect of a basic morpholine buffer
on isomerization and shows that the addition of dioxane leads
to an increase in the rate of isomerization.
Re-Examining Imidazole Catalysis. In the studies with
imidazole buffer catalysis, the original work did not involve
control of ionic strength, nor did it examine the possible medium

+
+
6592 J. Am. Chem. Soc., Vol. 118, No. 28, 1996
Breslow et al.
Figure 6. A cleavage mechanism with simple catalysis by the buffer base B. The intermediate dianion 6 is not a stable species, and it may not be
fully formed before departure of the leaving group.
effects of nonpolar solvents that mimic imidazole. We have
now addressed these deficiencies. We find that some of the
original evidence for our sequential mechanism does not stand
up to this more careful examination, but that other evidence
for our mechanism is actually strengthened when these effects
are taken into account. The sequential mechanism of Figure 2
still seems to be correct. However, there is also a pathway
catalyzed by Im alone.
First of all, our new studies confirm that the cleavage process
is catalyzed both by Im and by ImH+. The rate constants are
listed in the supporting information and plotted in the figures.
Figure 7A shows the cleavage rate when [ImH+] is held at 0.5
M and [Im] is varied. We also show the data in Figure 7A
with buffer concentration at 10% of the above values. In all
cases ionic strength was held constant at 1.0 M with NaCl, so
the pH’s are the same for the two buffers at any given buffer
ratio. The major catalyst is the buffer Im, not for example OH-.
Assuming that both lines correspond to the simplistic equation
1, the nonbuffer catalysis of the upper line contributes 6% at
the low [Im], increasing to 7% at the high [Im] where [OH-] is
higher.
kc ) kIm[Im] + knonbuffer
(1)
As Figure 8A shows, holding [Im] constant at 0.5 M while
increasing [ImH+] increases the cleavage rate. We do not have
the data for this case at 10% of the buffer concentration, but
from the low buffer data of Figure 7A the situation is clear. At
a 1/1 ratio of [Im] to [ImH+], the rate at 10% of the high buffer
concentration, the same concentration as in Figure 8A, was only
15% of the high buffer rate. Furthermore, the nonbuffer
contribution in Figure 7A was only a few percent of the high
buffer values and increased as the pH was increased. Thus in
Figure 8A the increasing rate as [ImH+] is increased is the result
of catalysis by ImH+. A pH effect would be negligible and
would go in the opposite direction.
From these data, it is clear that cleavage is catalyzed both
by Im and by ImH+. From the slopes of the lines in Figures 7A
and 8A, Im catalysis is ca. 1.5 times as effective as is that of
ImH+ at equal concentrations, a relationship that will be seen
again in the discussion of Figure 9. (Morpholine buffer has a
stronger base, and a weaker acid, so the preference for base
catalysis is even larger.)
Figure 7. Observed pseudo-first-order rate constant for (A) the
cleavage of 1, followed by observing the formation of uridine, and (B)
the isomerization of 1, followed by observing the formation of 2, at 60
°C and ionic strength 1.0 M (adjusted with NaCl) with [ImH+] held
constant and [Im] varying. The pH varies (see Results and Discussion)
as the buffer ratio changes: (Os) [ImH+] ) 0.5 M; (4) [ImH+] )
0.05 M; (]- - -) (in panel B) the difference between the 0.5 M and the
0.05 M data. Also shown in panel B (b- -) are two values when
[Im] and [ImH+] were held at 0.5 M, and 0.2 and 0.3 M dioxane was
added. This shows that the medium effect on this reaction goes in the
opposite direction to the effect of [Im]. The lines are best fits, not
theoretical.
contrast to the results with morpholine catalysis, the cleavage
rate showed at most a slight decrease when dioxane was added.
This medium effect may well explain the slight downward
curvature seen in Figure 7A as Im is added to the medium. Its
We also examined the result of increasing the concentration
of nonpolar solvent by adding dioxane to the medium. In

+
+
CleaVage and Isomerization of Uridyluridine
J. Am. Chem. Soc., Vol. 118, No. 28, 1996 6593
strength, but only a modest effect is expected. In the scheme
of Figure 2 one expects no observable catalysis by ImH+ if the
first step is rapid and reversible compared with the subsequent
step that leads to isomerization. That is, both the forward and
reverse steps from 1 to 5 are catalyzed by BH+, so one can see
BH+ in the kinetics only if that first step is partially rate
determining. Since with imidazole buffers the cleavage is ca.
10 times as fast as the isomerization, the catalysis of the first
step by ImH+ makes a larger contribution to the overall cleaVage
rate than to the isomerization rate.
The data on isomerization, in Figure 7B, show a small (10%)
decrease in rate as [Im] is increased at the high buffer
concentration, and essentially no rate change at a buffer
concentration 10-fold lower. This is what we first referred to
as a negative catalytic effect, the slowing of isomerization by
added Im.9,10 We now see that this slowing is not a medium
effect but is indeed evidence for a sequential mechanism with
partitioning of an intermediate, as in Figures 2 and 3.
The critical data are shown in Figure 7B. Addition of dioxane
causes an increase in the isomerization rate. As we discussed
above with respect to the same findings in the morpholine case,
this is quite reasonable for a reaction whose transition state is
less polar than are the starting materials. As in the morpholine
case, this shows that the decrease in rate actually caused by
the basic component of the buffer is not a medium effect, but
is the result of a partitioning mechanism in which the cleavage
leg is catalyzed by B. As with the morpholine case, it is clear
that the negative catalytic effect is larger than the raw data of
Figure 7B show, compensated to some extent by a medium
effect that works in the opposite direction. Thus if we corrected
for the medium effect, the modest downward slope of the rate
data in Figure 7B with increasing [Im] would be a considerably
larger downward slope resulting from the catalytic partitioning
effect.
Figure 8. Observed pseudo-first-order rate constant for (A) the
cleavage of 1, followed by observing the formation of uridine, and (B)
the isomerization of 1, followed by observing the formation of 2, at 60
°C and ionic strength 1.0 M (adjusted with NaCl) with [Im] ) 0.5 M,
as a function of [ImH+]. The pH varies (see Results and Discussion)
as the buffer ratio changes. The lines are best fits, not theoretical.
Even when the medium effect is taken into account, the
decrease in rate is small. It seems fairest to say that this new
data does not refute the previously reported negative catalytic
effect in the imidazole casesa negative effect that was much
more pronounced in the morpholine casesbut it is not strong
support.
Our early studies leaned heavily on the observation of bell-
shaped curves for the cleavage rate as a function of the imidazole
buffer ratio. That is, we saw a rate maximum when both Im
and ImH+ were present, but the rate of cleavage decreased when
the Im/ImH+ ratio was high or low; the rate maximum occurred
near a 1/1 ratio. However, these early studies were done without
control of the ionic strength, and this could be a problem.
Increasing Im while decreasing ImH+ leads to a decrease in
ionic strength, and this could have rate effects. Indeed, Perrin
suggested that a medium effect or an ionic strength effect could
account for our observed rate maximum.20 We now find that
the latter is indeed the case.
As Figure 9 shows, we can confirm the original finding4 that
the cleavage rate essentially levels off after a maximum when
the Im/ImH+ ratio is increased (when we subtracted the non-
buffer-catalyzed rate, which is faster at the higher pH with a
more basic buffer, this had led to a bell-shaped rate vs buffer
composition curve4), but only if we do not adjust the ionic
strength. When NaCl is used to maintain constant ionic strength,
no such leveling is seen and the rate simply increases as the
concentration of imidazole base increases. Thus the original
evidence in favor of a sequential bifunctional mechanism for
imidazole buffer catalysis of the cleavage reaction was flawed.
However, extrapolation of the correct curve in Figure 9 to pure
Figure 9. Observed pseudo-first-order rate constants for the cleavage
of 3′,5′-UpU vs the state of protonation (BH+/[BH+ + B]) of 0.8 M
imidazole buffer at 60 °C. The pH varies (see Results and Discussion)
as the buffer ratio changes. The reaction was examined under conditions
of both (0- - -) uncontrolled ionic strength and (bs) ionic strength
controlled at 1.0 M with NaCl. The lines drawn are to emphasize the
trends, not to fit a particular kinetic model.
catalytic effect is diminished at high concentrations by the
medium effect.
Figures 7B and 8B show the same studies on the isomeriza-
tion rate. These new data with control of ionic strength confirm
our earlier conclusion that the isomerization reaction is catalyzed
by ImH+ (Figure 8B). The ki shows a modest 50% increase
over the range from 0.2 to 0.8 M [ImH+] at constant ionic
(20) Perrin, C. L. J. Org. Chem. 1995, 60, 1239-1243.

+
+
6594 J. Am. Chem. Soc., Vol. 118, No. 28, 1996
Breslow et al.
Figure 10. A possible pathway for the cleavage reaction that has kinetic order only in B. As discussed, the BH+ catalysis would be kinetically
invisible.
Im or pure ImH+ indicates that there is an appreciable rate
catalyzed by ImH+, and a larger one catalyzed by Im. This is
the relationship we saw in Figures 7A and 8A.
Calculations have indicated that such a phosphorane dianion
will likely not be a stable intermediate,22,23 and its fragmentation
to a product alkoxide ion and a phosphate anion should be
exothermic. (The monoanion 5 is predicted to be stable.) This
version of cyclization without prior protonation can be thought
of as a base-catalyzed displacement reaction, possibly with some
addition character to it.
Perrin has suggested an alternative base-catalyzed mechanism
to accommodate our published data (Figure 11).20 In his
scheme, the buffer base operates first to convert the substrate
to the phosphorane dianion 6, and this then rapidly protonates
to form the phosphorane monoanion 5. Both this scheme and
that of Figure 2 convert the substrate to the same monanion 5,
and using a proton and a buffer base B, but in a different
sequence. We had considered schemes related to that of Figure
11, but rejected them since they did not accommodate the
apparent bell-shaped rate vs buffer composition curve. Since
we have now shown that the bell-shaped curve was an artifact
caused by variations in ionic strength, the scheme of Figure 11
must be reconsidered.
The ionic strength effect on the cleavage rate is reasonable.
If ionic strength is not maintained with NaCl, decreasing the
ImH+ Cl- concentration decreases the polarity of the medium,
just as adding dioxane does. Thus the same slowing of the
cleavage rate with decreasing polarity of the medium is seen.
Why is a bell-shaped curve not seen when the experiments
are done correctly? The branching mechanism of Figure 2
should generate a bell-shaped curve. This is clear from Figure
3, where eq a has the product of B and BH+ in the numerator.
We propose that the bell-shaped curve is not seen in our case
because it is hidden by an additional process: cleavage can also
occur catalyzed by B alone. This extra processswhich is the
major one and even more important with a more basic buffer,
as the buffer has more B and less BH+sprevents the drop in
rate when [BH+] becomes small at one end of the pH vs rate
profile. A drop should occur if the sequential mechanism alone
operated.
According to this proposal, cyclization to form a phosphorane
can occur either with or without prior protonation. The reaction
of the protonated phosphate will of course be intrinsically faster,
but the operating pH is so far above the phosphate pKa that
only a small fraction of the phosphate is protonated. Thus the
overall rate of cleavage catalyzed by base alone is greater than
that for the bifunctional sequential process; it hides the rate
decrease expected from the latter mechanism alone as [BH+]
decreases while [B] increases.
Mechanism of the Reaction Catalyzed by B Alone. One
possible mechanism is shown in Figure 10, with B converting
substrate to an intermediate phosphorane dianion 6 and BH+
carrying it forward to the cleavage products. As we have
described elsewhere,21 such a mechanism will show catalysis
only by B, not by the BH+ that is used in the later step. This
mechanism cannot by itself explain the negative catalytic effects
that we observe with buffer bases in the isomerization process,
effects that go opposite to the medium effect and that require
that some of the cleavage process go by a base-catalyzed branch
from a common intermediate also involved in the isomerization
process.
If the scheme as shown involved only buffer species, it alone
could not explain our finding of negative catalysis of the
isomerization by buffer base. That is, the rate of isomerization
would follow a kinetic equation (eq c of Figure 11) in which
the buffer base appears in the numerator as well as in some
terms of the denominator. If those terms were completely
dominant in the denominator, that could lead to a rate
independent of [B], but never to a rate that decreases with
increasing [B] as is observed. However, Perrin points out the
possibility that our negative catalysis could be observed if the
first step is catalyzed not only by B but also by OH-. (He also
suggests that the cleavage step in his scheme can be catalyzed
both by BH+ and by solvent species.) With the inclusion of
the additional catalyst terms, the equation governing the
isomerization rate is now eq d of Figure 11.
As eq d shows, one can account for the negative rate effect
of [B] on the isomerization reaction if the second term in the
numerator of the equation is significant; that is, if the catalysis
by OH- is comparable to the catalysis by B. From the data
(22) (a) Haydock, K.; Lim, C.; Bru¨nger, A. T.; Karplus, M. J. Am. Chem.
Soc. 1990, 112, 3826-3831. (b) Lim, C.; Karplus, M. J. Am. Chem. Soc.
1990, 112, 5872-5873. (c) Dejaegere, A.; Lim, C.; Karplus, M. J. Am.
Chem. Soc. 1991, 113, 4353-4355. (d) Dejaegere, A.; Karplus, M. J. Am.
Chem. Soc. 1993, 115, 5316-5317.
(23) Wladkowski, B.; Krauss, M.; Stevens, W. J. Am. Chem. Soc. 1995,
117, 10537-10545.
A related mechanism was shown in Figure 6. There the first-
formed phosphorane dianion 6 fragments to products without
further catalysis.
(21) Breslow, R. Proc. Natl. Acad. Sci. U.S.A. 1993, 90, 1208-1211.

+
+
CleaVage and Isomerization of Uridyluridine
J. Am. Chem. Soc., Vol. 118, No. 28, 1996 6595
Figure 11. Scheme suggested by Perrin (ref 18) for the cleavage and isomerization of UpU. In this scheme the substrate is converted to the
phosphorane monoanion 5 by a first step with B catalysis followed by a fast favorable protonation. The phosphorane anion 5 then goes on to the
isomer 2 as in our scheme of Figure 2, or 5 is converted to the cleavage product by a sequential deprotonation by OH-, then protonation by BH+.
This cleavage sequence is equivalent to B catalysis, so the overall process shows no kinetic order in BH+, only in B. Equation c is the steady-state
rate equation for isomerization of 1 to 2 if only buffer species play catalytic roles, and if the pH is such that 6 is rapidly and essentially completely
protonated to form 5. In this scheme the rate of formation of 5 is simply the rate of formation of 6, which is rapidly protonated to 5. Partitioning
of 5, which contributes the denominator of the equation, involves the isomerization reaction and also reversible conversion of 5 to 6. It is shown
as an equilibrium using H+ and OH-, but the situation does not change if BH+ and B are used instead. k′ is just a composite of the rate constants
k-1 and k2 with K. The cleavage leg with rate constant k2 can also have a contribution from nonbuffer catalysts, but this does not affect the
conclusions. This scheme in its simplest form (eq c) cannot account for the observed negative catalysis of isomerization by B, as discussed in the
text, but it can account for it if the B in the first step is OH- as well as buffer catalyst, as Perrin suggested, if that OH- plays a significant catalytic
role (eq d), and if 6 is stable enough to be a true intermediate (see Results and Discussion). In this equation, kOH is the constant for such OH-
catalysis of the conversion of 1 to 6, and k′′ reflects the reversal of the first step by nonbuffer species. This is one of several schemes that account
for the part of the cleavage catalyzed simply by B, but it does not account for the observed BH+ process or the evidence that isomerization is
catalyzed by BH+ and H+, not B and OH-.
plotted in Figure 7A, the cleavage rate with imidazole buffer
has a dependence on [Im] that is ca. 7 times as large for the
upper curveswith [ImH+] at 0.5 M and varying [Im]sas it is
for the lower curve where all the buffer concentrations are 10%
of those for the points above them. The pH’s vary, but the pH
at each buffer ratio is the same for the high and the low buffer
concentrations. Thus for the cleavage reaction, the kinetic
contribution of [OH-] is about 40% that of [Im] at the low buffer
concentration, but only 4% of that for [Im] at the high buffer
concentration. From the data in Figure 7B it is clear that both
the buffer and the nonbuffer species make a significant
contribution to the rate of isomerization, but the catalyst is BH+
or H+, not B or OH- as the Perrin scheme requires.
Our scheme of Figure 2 and the Perrin scheme of Figure 11
both postulate that the same intermediate phosphorane monoan-
ion 5 is being formed in the isomerization and also in the
cleavage reaction. The data of our Figures 7 and 8sand the
data we have published previouslysare consistent with the
alternative of Figure 11 if a nonbuffer species makes a
significant contribution to the first step, forming the intermediate,
and if the dominant role of the basic buffer species B on the
cleavage rate reflects mainly its role in the conversion of the
intermediate to cleavage product, as our schemes both show.
Of course the Perrin scheme seems to use BH+ for the
cleavage branch, not B as in our scheme, but this is illusory. In
our scheme the B is used to deprotonate the intermediate species
5, and then the resulting BH+ is used to protonate a leaving
group. In the Perrin scheme the intermediate 5 is deprotonated
by OH-, and then BH+ is used to protonate the leaving group.
The two are kinetically equivalent. The rate of a process
depending on the product [OH-][BH+] depends on [B], so either
version could be used in either scheme equally well. In his
scheme the dianion 6 could be protonated to the monanion 5
by BH+, and the process then reversed by B. This would make
the cleavage branch simply B catalyzed. Similarly, our scheme
in Figure 2 could be modified to include reversible deprotonation
of 5 to the dianion 6 by OH-, then protonation of the leaving
group by BH+, without changing the kinetics.
The Perrin scheme of Figure 11 postulates that the phospho-
rane dianion 6 is a true intermediate with enough lifetime to
protonate before fragmentation. Calculations suggest that this
is unlikely,22,23 that such a dianion cannot be a true intermediae

+
+
6596 J. Am. Chem. Soc., Vol. 118, No. 28, 1996
Breslow et al.
Figure 12. A cleavage mechanism that would show only BH+ catalysis. As discussed in the text, it is excluded for the morpholine and imidazole
buffers by the observed negative rate effect of B on the isomerization process.
but will fragment during its formation. The isomerization
reaction requires a true intermediate that can protonate so that
isomerization is possible. Our observed negative rate effects
of buffer bases on isomerization require that such a protonated
intermediate also be on the path to cleavage. This favors our
mechanism of Figure 2 over the Perrin mechanism of Figure
11, but only if it is true that dianion 6 cannot be a true
intermediate. At this point it seems best to consider both
mechanisms as possible.
Subject to the above concerns, the Perrin scheme may be a
viable alternative for the reaction catalyzed by B alone, but it
does not deal with the reactions catalyzed by BH+. Our data
show that the cleavage reaction and, even more so, the
isomerization reaction are catalyzed by BH+. In those reactions
the first step must be protonation of the substrate, in contrast
to the Perrin scheme in which such protonation occurs after
phosphorane formation and is kinetically invisible (since it is
fast and essentially complete).
accompanied24,25 by a path using base catalysis alonesthe
schemes of Figures 6, 10, and 11 are all possiblesthat is of
course unrelated to a bifunctional mechanism such as that seen
with our mimic of ribonuclease, or with ribonuclease itself.
However, when the leaving group is much more stable, in
substrate 4, base catalysis is all that occurs.15,16 With such a
good leaving group the more or less direct displacement reaction
of Figure 6 will be favored.
Acetate Buffer. The balance between the path of Figure 2
and that of Figures 6, 10, and 11 will depend on the relative
basicity and acidity of the buffer components, of course. We
have examined the situation with sodium acetate and acetic acid
as buffer catalysts. The pKa of morpholine is 8.33, that of
imidazole is 7.0, and that of acetic acid is 4.75. Thus a buffer-
acid-catalyzed path should be favored with acetate buffer. This
is what we observe.
In Figure 13 we plot the rate of cleavage and of isomerization
of 3′,5′-UpU by various ratios of acetic acid to sodium acetate,
both at a high buffer concentration and at a low one. (With
this acidic buffer the isomerization rate is somewhat faster than
the cleavage rate, in contrast to the situation with imidazole or
morpholine buffers where the isomerization is considerably
slower than the cleavage.) As the plots show, both the cleavage
rate (Figure 13A) and the isomerization rate (Figure 13B)
increase with increasing concentration of HOAc. In contrast
to the data with morpholine or imidazole, there is no significant
catalysis contributed by the buffer base, which is of course much
weaker. That is, the rates extrapolated to [HOAc] ) 0, with
all the buffer as [AcO-], are not significantly above zero. Thus
mechanisms in which the first step is protonation of the
substrate, as in the scheme of Figure 2, are the dominant ones
here.
Are There Alternative Mechanisms for the BH+-Catalyzed
Reactions? We must also consider possible alternatives for
the BH+-catalyzed process. One of them is our scheme of
Figure 2. It was originally based on the observation of a bell-
shaped curve that now is seen to have an alternative explanation,
so there is no longer direct evidence for catalysis of the cleavage
leg of the partitioning scheme by B. The negative catalysis by
B might involve only the scheme of Figure 11. Thus we must
also consider paths in which BH+ alone plays a role, without
participation of B in a later step.
Such a path is shown in Figure 12. Here the BH+ is used
both to form the intermediate 5, as in our Figure 2, and also to
catalyze its decomposition to a cyclic neutral phosphate without
deprotonation. However, in the same medium, that intermediate
5 is mainly being sent forward to cleavage products with
catalysis by B, not BH+ (as in our scheme, or in the equivalent
and indistinguishable cleavage branch of the Perrin scheme).
Otherwise one would not see negative effects of B on the rate
of isomerization. One cannot postulate that the same intermedi-
ate 5 undergoes mainly a base-catalyzed decomposition mech-
anism if it is formed by one path, but uses a different acid-
catalyzed decomposition mechanism in the same medium if it
is formed by a different path. The path with BH+ in its first
step must be using B on the cleavage leg, not BH+.
The data in Figure 13A,B show upward curvature and roughly
fit a theoretical curve for second-order catalysis by HOAc.
However, we must also consider medium effects. We have
(24) Haim (Haim, A. J. Am. Chem. Soc. 1992, 114, 8384-8388) argued
that the isomerization and cleavage reactions did not proceed through a
common intermediate. Our conclusion is that this could be partly correct:
the cleavage reaction has an additional path that may bypass the isomer-
ization intermediate 5sunless it involves the scheme of Figure 11sas well
as the one that passes through it.
(25) Kosonen and Lo¨nnberg (Kosonen, M.; Lo¨nnberg, H. J. Chem. Soc.,
Perkin Trans. 2 1995, 1203-1209) describe relevant studies on the cleavage
and isomerization of a uridine phosphate triester. Since their intermediate
phosphorane anion does not carry a proton, it cannot show the base-catalyzed
decomposition we invoke for our cases.
Thus we conclude that the scheme that best explains our
observed BH+ catalysis of cleavage and isomerization by
imidazole buffers is Figure 2, as we originally proposed. It is

+
+
CleaVage and Isomerization of Uridyluridine
J. Am. Chem. Soc., Vol. 118, No. 28, 1996 6597
by a much weaker base, so it might need the help that a
hydrogen-bonded acid would provide.
It should also be noted that in the cyclization step an anionic
base is removing a proton, so the transition state has a dispersed
negative charge. Such a change from localized to dispersed
negative charge should be faVored by less polar solvents, in
line with the observed medium effect. A similar change from
localized to dispersed negative charge occurs in the fragmenta-
tion step that leads to cleavage products, again consistent with
the observed medium effect here.
The kinetic expressions for this mechanismsderived by using
the steady-state treatmentsshow that both cleavage and isomer-
ization should be simply second-order in HOAc, as is apparently
observed (acetate ion does not appear in the kinetic expression,
since21 it is formed from HOAc in a preceding step). Thus this
mechanism seems a likely analog of the sequential acid-base
bifunctional mechanism found for the other buffers, but it cannot
be considered fully established until more work is done with
the acetate buffer and other related catalysts. If it is correct, it
has a step with simultaneous acid-base catalysis by the two
buffer components, in direct analogy to the simultaneous
bifunctional catalysis in the enzyme ribonuclease7 and in our
ribonuclease mimic.7,17-19,26
Relevance of These Studies to the Mechanism of Action
of an Artificial Enzyme. We have described the hydrolysis
of the cyclic phosphate 7 by a set of â-cyclodextrins carrying
two imidazole groups attached to the primary carbons of the
cyclodextrins.7,17-19,26 Among the three isomers with imidazoles
on neighboring glucoses (6A, 6B), on glucoses separated by
one (6A, 6C), or on glucoses separated by two other glucose
units (6A, 6D), the best catalyst was the 6A, 6B isomer.17
Furthermore, the pH vs rate profile of this catalyst showed that
it was operating with one basic imidazole ring and one acidic
imidazolium ring. Finally, a proton inventory using deuterium
isotope effects showed that the two catalytic groups were
operating at the same time, in a simultaneous bifunctional
process.18 This differs from the sequential process seen with
simple buffer catalysts (although the seeming second-order
process with acetic acid buffers does require one simultaneous
bifunctional step, as in the scheme of Figure 14).
Figure 13. Observed pseudo-first-order rate constants for (A) the
cleavage of 3′,5′-UpU and (B) the isomerization of 3′,5′-UpU, both vs
the fraction of HOAc in acetate buffer at 60 °C. The pH of course
varies as the buffer ratio changes: (Os) [buffer] ) 1.0 M; (4- -)
[buffer] ) 0.1 M; (]- - -) the difference between the 1.0 M and the
0.1 M data. Ionic strength was maintained at 1.0 M with NaCl. The
curved lines shown are plots vs [HOAc]2.
examined the effect of adding dioxane to a 0.3 M buffer
consisting of 9/1 HOAc/AcO-. When 0.6 M dioxane was
added, the kc went from 0.15 × 10-3 h-1 to 0.22 × 10-3 h-1
.
With the same buffer and addition, ki went from 0.60 × 10-3
h-1 to 0.68 × 10-3 h-1. In this case both the isomerization
and the cleavage reaction are faster when dioxane is added,
although the effect is larger in the cleavage case. This contrasts
with the slowing of cleavage by dioxane when the catalyst was
morpholine or imidazole.
The base group in such a mechanism is surely delivering the
water nucleophile, but there are two possible roles for the ImH+
in the hydrolysis of 7. One could be to protonate the leaving
oxygen, as in the scheme of Figure 15. The other is to protonate
a phosphate anionic oxygen, so as to produce a phosphorane
monoanion intermediate 8, as in Figure 16. The intervention
of such a phosphorane monoanion is of course the main thrust
of our mechanistic conclusions in the buffer-catalyzed cleavage
and isomerizations of UpU. The preferred geometry in the
bifunctional catalyzed hydrolysis of compound 7 indicates that
a phosphorane monoanion is also involved here.
The difference reflects the difference in charge type for the
buffer components. In the acetate buffer the acid form is neutral,
so any process in which it protonates a substrate anion will not
lead to net decreased charge. Similarly, since the acetate anion
is negative, any process in which it removes a proton from the
OH group of an intermediate will not lead to net increased
charge, as such deprotonations did when a neutral base was
involved. The medium effect of acetic acid should be even
smaller than that of dioxane, so the upward curvatures in Figure
13 are probably not due simply to the medium effect.
We tentatively propose a mechanism (Figure 14) that accounts
for the apparent second-order kinetics in HOAc. In the first
step we write the reversible protonation of the substrate by buffer
acid, just as in the mechanism of Figure 2. However, to account
for the second molecule of HOAc that must appear in the
transition state for cleavage, we propose that the cyclization to
form a phosphoranesagain catalyzed by the buffer basesis in
this case also catalyzed by an HOAc molecule hydrogen-bonded
to the neutral substrate. The cyclization would then lead directly
to the fully protonated neutral phosphorane.
Since both the nucleophile and the leaving group must depart
in-line in a simultaneous displacement at phosphorus, a direct
reaction such as that of Figure 15 would be best catalyzed by
the A,D isomer of our catalyst set. Models show that the A,B
isomer cannot simultaneously deliver a water to one face of
the phosphorus while protonating a leaving group on the
opposite face; yet the A,B isomer was the preferred catalyst.
This indicates that the scheme of Figure 16 is preferred, the
proton is being delivered to an anionic phosphate group, which
is within reach, not to the leaving group. Then the resulting
phosphorane monoanion 8 decomposes to product.
This is proposed mainly to handle the apparent second-order
kinetics, but it is not unreasonable. In contrast to the cyclization
process in the mechanism of Figure 2, this cyclization is driven
(26) Breslow, R.; Schmuck, C. J. Am. Chem. Soc. 1996, 118, 6601.

+
+
6598 J. Am. Chem. Soc., Vol. 118, No. 28, 1996
Breslow et al.
Figure 14. Likely mechanism for the cleavage and isomerization of 3′,5′-UpU catalyzed by acetic acid, to explain the apparent second-order
dependence on [HOAc]. The scheme as shown is somewhat compressed: in the first step there is protonation of the substrate followed by hydrogen
bonding of the second HOAc, and in the cleavage of the phosphorane IH2 to the cyclic phosphate, the acetate probably deprotonates the phosphorane
while the HOAc protonates the leaving group, as also occurs in the reversal of the formation of IH2. The scheme differs from that of Figure 2 in
that the cyclization step forms a neutral phosphorane in a simultaneous bifunctional process catalyzed by the buffer acid and the buffer base. This
change reflects the greater acidity of HOAc than of the morpholinium or imidazolium cations, and the weaker basicity of acetate ion than of
morpholine or imidazole. A steady-state treatment of this mechanism indicates that both kc and ki are simply proportional to [HOAc]2, with no
dependence on AcO-. This is because the acetate ions used as catalysts in two of the steps are formed from HOAc in the preceding step and are
thus kinetically invisible (ref 19).
Figure 15. An a priori possible mechanism for the hydrolysis of
substrate 7 by a cyclodextrin bis-imidazole. This excluded mechanism
predicts that the most effective catalyst would have the two catalytic
groups as far apart as possible, but that is not what is observed.
As we have pointed out,7 in such a simultaneous process there
is no longer the question of whether the acid or the base operates
first, as in the simple buffer schemes. Thus both our scheme
of Figure 2 and the Perrin scheme of Figure 11 become identical
when extrapolated to a simultaneous acid-base process. Of
course in the Perrin scheme some proton-transfer steps are
shown using H+ or OH-, but with any significant concentration
of buffer catalysts, in solution or within a catalyst-substrate
complex, all these transfers will be done by the catalyst species
as in our scheme. Substitution of BH+ and B for the H+ and
OH- in the equilibration (Figure 11) between dianion 6 and
monoanion 5 does not affect the validity of his suggestion.
Figure 16. Correct mechanism for hydrolysis of substrate 7 by
cyclodextrin bis-imidazole, which protonates the anionic oxygen and
forms a phosphorane monoanion 8. This explains the finding that the
most effective cyclodextrin bis-imidazole isomer has the two catalytic
groups on neighboring glucose units, mounted only 51° apart. This is
a simultaneous analog of the mechanism in Figure 2.
seriously from the findings with the artificial enzyme that uses
the principal catalytic group of the enzyme, we have suggested
that the enzyme may well follow a related path (Figure 17),
using simultaneous28 acid-base catalysis. In most textbooks
the enzyme mechanism is written as in Figure 18, with the basic
imidazole group delivering the neighboring oxygen to the
phosphate while the acidic imidazolium group protonates the
Mechanism of the Enzyme Ribonuclease A.27 On the basis
of our conclusions from the previous buffer studies, and more
(27) For reviews, see: (a) Richards, F. M.; Wyckoff, H. W. In The
Enzymes; Boyer, P. D., Ed.; Academic Press: New York, 1971; Vol. 4. (b)
Blackburn, P.; Moore, S. In The Enzymes; Boyer, P. D., Ed.; Academic
Press: New York, 1982; Vol. 15. (c) Campbell, C. L.; Petsko, G. A.
Biochemistry 1987, 26, 8579-8584 and references cited therein.
(28) Matta, M. S.; Vo, D. T. J. Am. Chem. Soc. 1986, 108, 5316.

+
+
CleaVage and Isomerization of Uridyluridine
J. Am. Chem. Soc., Vol. 118, No. 28, 1996 6599
some calculations on the enzyme mechanism by the Karplus
group22 showed that our proposed process, with protonation to
form a phosphorane intermediate, is energetically reasonable.
Recent calculations by Wladkowski23 come to the stronger
conclusion that our phosphorane process is the preferred path
for the enzyme, although in those calculations the phosphorane
monoanion then goes onswith further protonationsto a neutral
phosphorane. Inclusion of such an extra equlibration of the
phosphorane monoanion with a neutral phosphorane would not
affect the conclusions related to the schemes of either Figure 2
or Figure 11.
Contrary arguments about the enzyme mechanism have
recently been advanced30,31 and will be addressed elsewhere.
Summary of the Mechanistic Arguments
1. The cleavage of 3′,5′-UpU to form the 2′,3′-cyclic
phosphate and its isomerization to 2′,5′-UpU are both catalyzed
by morpholine buffer, imidazole buffer, and acetate buffer.
2. Isomerization with imidazole buffers involves a pathway
in which the buffer acid converts substrate to a phosphorane
intermediate. The intermediate can then branch: pseudorotation
sends it to the isomer, while buffer base sends it to the cleavage
products. This branching mechanism is indicated by the
negatiVe effect of buffer base on the isomerization rate. Studies
of medium effects strengthen this evidence.
3. In addition to this cleavage mechanismswith buffer acid
catalyzing the formation of an intermediate and buffer base
sending the intermediate to cleavage productssthere is an
additional cleavage process in which only buffer base plays a
role. This is the dominant mechanism with morpholine or
imidazole buffers, hiding some of the kinetic expectations for
the [BH+, then B] mechanism.
4. With acetate buffer, acid catalysis dominates. Now both
cleavage and isomerization involve conversion of 3′,5′-UpU to
a phosphorane intermediate by the buffer acid. A proposed
bifunctional process explains the detailed kinetic behavior.
5. The responses of all these reactions to medium effects
are consistent with the proposed mechanisms.
Figure 17. Mechanism we have proposed for the action of ribonuclease
A, in which the first function of the imidazolium ion of His-119 is to
protonate the phosphate anionic oxygen, leading to a phosphorane
monoanion intermediate.
6. The earliest evidence for sequential bifunctional catalysis
by imidazole bufferssa bell-shaped rate of cleavage vs buffer
ratioswas incorrect because of the neglect of ionic strength
effects. Curiously, this early omission was the impetus for
further experiments that actually established a role for this
mechanism.
7. In an enzyme mimic, a simultaneous bifunctional mech-
anism operates that is related to the sequential bifunctional
mechanism of buffer catalysis. Both processes involve conver-
sion of a phosphate anion substrate to a phosphorane monoanion
intermediate.
8. A mechanism has been proposed for the enzyme ribonu-
clease A that is essentially the same as that for the artificial
enzyme.
Figure 18. Classical mechanism for cyclization-cleavage of RNA
by ribonuclease A, still favored by some workers.
leaving group. The resulting cyclic phosphate ester is then
hydrolyzed in a second related process in which the scheme is
reversed, but with water substituting for the nucleoside product.
In at least one text our alternative is now shown as the likely
path.29
In addressing the matter of an enzyme mechanism, analogy
is useful but hardly definitive. However, we pointed out7 that
the determined structures of ribonuclease A with some bound
anionic substrates or inhibitors showed that the imidazolium
group is indeed hydrogen-bonded to an anionic phosphate
oxygen, not to the neutral leaving group oxygen. Thus the
proton transfer we postulate should occur readily. Furthermore,
Experimental Section
Materials. 3′,5′-Uridyluridine, uridine, imidazole, imidazole hy-
drochloride, acetic acid, and sodium acetate were purchased (Sigma)
in the highest purity available. Morpholine (Aldrich) was redistilled,
and morpholine hydrochloride was prepared from it. The 2.0 M stock
solutions of the buffer components were prepared in advance and used
in all the kinetic studies.
Kinetic Studies on UpU. The kinetic assay used in the current
studies was similar to that described previously,12 except that a fully
automated HP1090 HPLC instrument was used in the current studies.
(30) Thompson, J. E.; Raines, R. T. J. Am. Chem. Soc. 1994, 116, 5467-
5468.
(29) Abeles, R. H.; Frey, P. A.; Jencks, W. P. Biochemistry; Jones and
Bartlett: Boston, 1992.
(31) Herschlag, D. J. Am. Chem. Soc. 1994, 116, 11631-11635.

+
+
6600 J. Am. Chem. Soc., Vol. 118, No. 28, 1996
Breslow et al.
The differences from the previous procedures are as follows: (1)
Borosilicate glass 0.1 mL microvials were used as reaction vessels.
(2) The reaction vials were incubated in the thermostated automatic
sampler compartment of the HPLC instrument at 60.0 °C. (3) A 15
µL sample of the reaction solution was withdrawn periodically and
automatically from the reaction vial and injected directly onto the
column for analysis. The HPLC eluent was 4.5 mM pH 7 phosphate
buffer with 4.0% MeOH, as before.12 Since the sampling needle was
part of the solvent delivery system in the chromatographic procedure,
it was completely flushed with solvent between injections. As before,12
an internal standard of potassium p-nitrobenzenesulfonate was used in
the reaction mixture to calibrate the HPLC signals.
Typical Kinetic Run. To a reaction vial were added stock solutions
of 15 µL of the potassium p-nitrobenzenesulfonate internal standard
(200 µM) with 3′,5′-UpU (30 mM), 37.5 µL of imidazole (2.0 M),
37.5 µL of imidazole hydrochloride (2.0 M), 37.5 µL of NaCl solution
(2.0 M), and 22.5 µL of deionized water. After stirring on a Vortex
mixer the vial was incubated in the automatic sampler compartment of
the HPLC instrument at 60 °C, and the instrument was programmed to
withdraw 15 µL of the reaction solution at intervals over 13-50 h (the
longer time for the isomerization study, which was slower with the
basic buffers) and to inject it directly onto a C-18 reverse phase column.
The integrated data furnished by the instrument were then calibrated
with respect to the internal standard and analyzed with a standard
computer program using the initial rate treatment. Only a few percent
of the substrate was consumed.
Kinetic data for the runs are plotted in the figures of this paper and
tabulated (155 rate constants) in the supporting information.
Acknowledgment. This work has been supported by the
NIH, by the Office of Naval Research, and by the NSF. We
thank Charles Perrin for useful interchanges. We also thank
Brian Wladkowski for permission to cite some of his work
before publication.
Supporting Information Available: Tabulated rate con-
stants and pH’s that are plotted in the figures of the current
paper (155 rate constants, 7 pH’s) (6 pages). See any current
masthead page for ordering and Internet access instructions.
JA9526933