Serine protease mechanism-based mimics. Direct evidence for a transition state bridge proton in stable potentials

Article abstract of DOI:10.1021/jo702003q

(Chemical Equation Presented) We have synthesized 1-(2-hydroxyacetyl) piperidine-2-one (2) and 1-(2-hydroxyacetyl)azepan-2-one (3). Equilibrium (K_f) between the free alcohol (open form) and the tetrahedral intermediate (cyclol) is readily estab

Full text of DOI:10.1021/jo702003q

Serine Protease Mechanism-Based Mimics. Direct Evidence for a
Transition State Bridge Proton in Stable Potentials
Loengrid Bethencourt and Oswaldo Nu´n˜ez*
Laboratorio de Qu´ımica Ambiental y Fisicoqu´ımica Orga´nica, UniVersidad Simo´n Bol´ıVar,
Caracas 89000, Venezuela
onunez@usb.Ve
ReceiVed September 12, 2007
We have synthesized 1-(2-hydroxyacetyl)piperidine-2-one (2) and 1-(2-hydroxyacetyl)azepan-2-one (3).
Equilibrium (Kf) between the free alcohol (open form) and the tetrahedral intermediate (cyclol) is readily
1
established, and both forms are observed in the D2O H NMR spectra of 2 and 3. Therefore, their
interconversion can be considered as an almost thermoneutral non-identical one. Pseudo-first-order rate
1
constants (kobs) were obtained by simulating the AB H NMR system observed for the cyclol. By best
fitting the experimental points of a kobs versus pD profile to the equation kobs ) 0.5k0r + 0.5kr Kac/(Kac +
[D+]) + 0.5kfKao/(Kao+ [D+]), the parameters involved were obtained: rate constants of rupture and
formation (k0r and k0f ) Kfk0r) catalyzed by water, rate constants of rupture (kr) and formation (kf) from
the conjugated bases of the cyclol form and the open form, and their acidity equilibrium constants Kac
and Kao. The system studied mimics the serine alcohol attack on the peptide bond and its reverse reaction
in serine protease enzymes. In fact, the reaction rates are similar or perhaps even faster than the ones
obtained for enzymatic reactions. The results also show the participation of water molecules forming
catalytic proton bridges in stable potentials with the two interconverted forms. The position change of
the bridged proton is sensitive to lactam ring size, and it is manifested by considerable change in the pKa
values of both cyclol and open forms. Other evidence such as kinetics, ∆S°, ∆Sq, and proton inventory
experiments and semiempirical molecular calculations support this proposal.
Introduction
macrocycles formed from N-(2-aminoacetyl)-2-lactams. We
have also reported3 the rate of formation of the corresponding
stable tetrahedral intermediates of N-heteroethylphthalimide
(hetero: hydroxy, amino, and thioxy) and their rates of rupture
We have been interested in measuring rates of intramolecular
transformations in stable tetrahedral intermediates. For instance,
we have reported1,2 results on transannular rates in bislactam
(2) Guedez, T.; Nu´n˜ez, A.; Tineo, E.; Nu´n˜ez, O. J. Chem. Soc., Perkin
Trans. 2 2002, 2078-2082.
(3) Nu´n˜ez, O.; Rodr´ıguez, J.; Angulo, L. J. Phys. Org. Chem. 1993, 7,
80-89.
(1) Tineo, E.; Pekerar, S. V.; Nu´n˜ez, O. J. Chem. Soc., Perkin Trans. 2
2002, 244-246.
10.1021/jo702003q CCC: $40.75 © 2008 American Chemical Society
Published on Web 02/22/2008
J. Org. Chem. 2008, 73, 2105-2113
2105

Bethencourt and Nu´n˜ez
SCHEME 1a
a Hydrolysis paths for compound 1 (n ) 5). Only the open form of compound 1 and its hydrolysis products (2-pyrrolidone, hydroxyacetic acid, and NBA)
are observed in the 1H NMR at 7 < pD < 11. The cyclol and macrocycle are barely observed at pD > 11. For compounds 3 (n ) 7), the macrocycle is also
observed at any pD, and for 2 (n ) 6), only the cyclol and open forms are observed. At pD > 12, the corresponding hydrolysis products are observed for
compounds 2 (n ) 6) and 3 (n ) 7).
SCHEME 2a
1
a Compound 3 forms observed in D2O by H NMR. For compound 2, only the open form and cyclol are observed.
to diacylimides. Recently, we have synthesized N-(2-hydroxy-
acetyl)-2-pyrrolidone (1), expecting to observe chemical be-
havior similar to that observed in the N-(2-aminoacetyl)-2-
lactams. However, we found4 at pD > 7.5 an irreversible
cleavage of the exocyclic and endocyclic C-N bond of 1 to
yield, in the latter case, N-(4-hydroxyacetyl)butanoic acid (NBA)
(Scheme 1). This product corresponds to a lactam ring opening,
a process of wide interest in bioorganic chemistry. As shown
in Scheme 1, the exocyclic cleavage of 1 yields 2-pyrrolidone.
Its formation is first-order-catalyzed by OD-, whereas NBA
production is second-order with respect to [OD-]. In fact, NBA
is produced from the cyclol form of 1 that cleaves, breaking its
amide bond, to produce the macrocycle that finally opens to
yield NBA. The cyclol and macrocycle forms are detected only
at high pD (pD > 12). Therefore, these two forms are not stable
enough at neutral pD. However, the pKa value (11.9) for the
cyclol form of 1 could be obtained experimentally4 from product
ratios and estimated (pKa ) 11.7) from linear free energy
relationships where only inductive effects are taken into account.
The low stability of the cyclol and macrocycle forms of 1 has
been attributed4 to the resonance of the planar conformation of
the five-membered ring lactam that retards the intramolecular
attack of the 2-hydroxyacetyl moiety. Strain in forming the
cyclol (tetrahedral intermediate) may also play a role.
thesized 1-(2-hydroxyacetyl)piperidin-2-one (2) and 1-(2-hy-
droxyacetyl)azepan-2-one (3). In fact, for these compounds, the
cyclol (tetrahedral intermediate) is the most stable form at any
pH. Therefore, equilibrium constants Kf ) [cyclol]/[open form]
have been measured as well as thermodynamic parameters at
neutral pH.
The unusual stability of the tetrahedral intermediate (cyclol
form) makes these systems unique since their rates of cleavage
and formation, in these almost thermoneutral non-identical
reactions, are barely influenced by the difference in stability
between reactants and products. In fact, for compounds 2 and
3, the cyclol and open forms (Scheme 2) are both detected by
1H NMR in an ample pH range, making it possible to measure
1
their interconversion rates via dynamic H NMR.
Direct observation and trapping of tetrahedral intermediates
have been reported5,6 previously in the hydrolysis of the N,O-
trimethylenephthalimidium ion in a system similar to the one
studied herein. However, two main aspects make these systems
chemically different: First, the attacking and leaving groups in
the former system are not in the same molecule; that is, the
reaction is intermolecular with reactants and products with quite
different stability; therefore, thermodynamic enthalpic and
entropic factors influence the observed rates. As mentioned
above, this is not the case in the system described in this work.
Second, in Gravitz’s and Jencks’s system, there is a weak N
Expecting to observe the cyclol and/or macrocycle forms in
other less strained 1-(2-hydroxyacetyl)lactams, we have syn-
(4) Garcia Borboa, L.; Nu´n˜ez, O. J. Phys. Org. Chem. 2006, 19, 737-
743.
(5) Gravitz, N.; Jencks, W. P. J. Am. Chem. Soc. 1974, 96, 489-499.
(6) Gravitz, N.; Jencks, W. P. J. Am. Chem. Soc. 1974, 96, 507-515.
2106 J. Org. Chem., Vol. 73, No. 6, 2008

Serine Protease Mechanism-Based Mimics
SCHEME 3a
were added. After these reactants were mixed, recently distilled
thionyl chloride (1.75 mL, 24 mmol) was added dropwise while
stirring. The mixture was heated at 70-75 °C for 3 h. After this
time, the excess of thionyl chloride was removed by distillation.
The remaining product was used directly in the next reaction step
without further purification.
a The system shown in Scheme 2 can be considered as serine protease
mechanism-based mimics of the acyl-enzyme tetrahedral intermediate
(EATI) formation and its corresponding partitioning.
1-(2-Benzyloxyacetyl)azepan-2-one and 1-(2-Benzyloxyacetyl)-
piperidine-2-one. In a two-neck flask provided with an argon
atmosphere, either 1.27 g (10.8 mmol) of azepan-2-one or 1.07 g
(10.8 mmol) of piperidin-2-one was mixed with dry pyridine (0.9
mL, 10.8 mmol) and 5 mL of toluene. This mixture was cooled to
0 °C, and the benzyloxyacetyl chloride, previously prepared and
maintained at 0 °C, was added slowly. The product solution took
on a yellow color. The reaction mixture was stirred at 0 °C for 2
h. After this time, the mixture was refluxed under an argon
atmosphere for 12 h. At the end of this time period, the mixture
turned a brown color. It was cooled at 0 °C, and 25 mL of ice/
water mixture was added. The two phases were separated, and the
aqueous one was extracted twice with 20 mL of toluene. The
organic phases were combined and treated with sodium bicarbonate
and dried with anhydrous magnesium sulfate. Finally, the solvent
was evaporated to yield a dark brown oil.
(amide) electron pair driving force for the expulsion of the
leaving group instead the O electron pair(s) of the cyclol form
(Scheme 2).
The almost thermoneutral reactions of this work may also
mimic enzyme reactions. For instance, the open form in Scheme
2 could be envisioned as a serine protease active site where
peptide (endocyclic amide CdO bond) and serine enzyme
residues (exocyclic hydroxymethyl group) coexist in the same
molecule. We are then dealing with mechanism-based mimics7
where the substrate-enzyme binding process has been bypassed.
The formation (k2) and rupture (k-2) of the cyclol form in
Scheme 2 mimics the formation (k2) and cleavage (k-2) of the
acyl-enzyme tetrahedral intermediate (Scheme 3), whereas the
eventual rupture of the endocyclic C-N bond to yield the
macrocycle form mimics its formation (k3) and its reverse
reaction (k-3).
1
1-(2-Benzyloxyacetyl)azepan-2-one: 2.52 g, 70%; H NMR
(400 MHz, CDCl3) δ 7.35 (m, 5H, Ph), 4.64 (s, 2H, -O-CH2-
Ph), 4.61 (s, 2H, -CH2-O), 3.92 (t, 2H,CH2-N, J ) 5.13 Hz),
2.67 (t, 2H, lactam -CH2-C-(O), J ) 4.03 Hz), 1.72 (m, 6H,
lactam -(CH2)3); 13C NMR (100 MHz, CDCl3) δ 177.8, 173.5,
137.8, 128.5, 128.3, 128.1, 127.9, 125.4, 73.3, 72.9, 43.1, 39.5,
29.2, 28.5, 23.6; MS m/z (fraction, relative intensity) 261 (M+, 5),
154 (M+ - 107, 6 × 105), 91 (M+ - 170, 1 × 106).
It has been pointed out8-10 that perfectly evolved Michaelian
enzymes should exhibit equilibrium constants close to unity for
the interconversion of enzyme-bound substrate and product.
Instead of perfection, it has been affirmed11 that the unity
condition corresponds to an optimal state of operation of
presently existing enzymes since the corresponding intrinsic
barriers are involved. This equilibrium constant close to unity
is in fact the one found in the models studied in this work.
Therefore, these models may be considered as mimics of an
evolved serine protease.
1-(2-Benzyloxyacetyl)piperidine-2-one: 2.95 g, 99%; 1H NMR
(400 MHz, CDCl3) δ 7.33 (m, 5H, Ph), 4.64 (s, 2H, -CH2-O),
4.63 (s, 2H, -O-CH2-Ph), 3.74 (t, 2H, -CH2-N, J ) 6.22 Hz),
2.52 (t, 2H, lactam -CH2-C(O), J ) 5.86 Hz), 1.81 (m, 4H, lactam
-(CH2)2); 13C NMR (100 MHz, CDCl3) δ 174.3, 173.3, 137.7,
128.5, 128.1, 127.9, 73.3, 73.0, 44.1, 34.5, 22.3, 20.2; MS m/z
(fraction, relative intensity) 156 (M+ - 91, 5 × 104), 140 (M+
-
Moreover, recently,12 it has been reported that tetrahedral
intermediates formed in the peptide cleavage in HIV proteases
can be trapped at the active site. Therefore, the use of
compounds 2 and 3 as tetrahedral intermediate inhibitors of HIV
proteases is worth exploring.
The results reported herein indicate that the water molecule
plays a special role in the reaction by hydrogen bonding the
cyclol and open forms. In agreement with the proposal13 for
the role of hydrogen bonds during proton transfer in general
catalytic transition state stabilization in enzyme catalysis, the
proton bridges found in this work are strong hydrogen bonds
in stable potential. The participation of water at the reactant
stage is supported by pKa, equilibrium entropic values, rate
constants, entropic activation parameters, proton inventory
experiments, and semiempirical molecular calculations.
107, 2.5 × 105), 91 (M+ - 170, 3 × 105).
1-(2-Hydroxyacetyl)azepan-2-one (3). In a high-pressure re-
sistance flask, 1.53 g (5.86 mmol) of 1-(2-benzyloxyacetyl)azepan-
2-one in 125 mL of ethyl acetate was added. To this solution was
added 2.18 g of 5% Pd/C catalyst. The mixture was reduced under
continuous stirring and hydrogen pressure (40 psi) for 7 h. The
mixture was filtered, and the residual solvent was eliminated by
vacuum at room temperature.
A white hygroscopic solid (0.68 g, 70% yield) corresponding to
1-(2-hydroxyacetyl)azepan-2-one was obtained: 1H NMR (400
MHz, CDCl3) δ 4.61 (s, 2H, -C(O)-CH2-O-), 4.44 (br s, 1H,
-OH), 3.97 (t, 2H, CH2-N, J ) 5.49 Hz), 2.71 (t, 2H, lactam
CH2-C(O), J ) 4.39 Hz), 1.20-1.90 (m, 6H, lactam -(CH2)3);
13C NMR (100 MHz, CDCl3) δ 177.6, 176.8, 65.9, 35.5, 29.1, 27.9,
23.6, 22.1.
1-(2-Hydroxyacetyl)piperidine-2-one (2): Using the same
procedure as indicated above for 1-(2-benzyloxyacetyl)azepan-2-
one (3), 2.37 g (9.6 mmol) of 1-(2-benzyloxyacetyl)piperidine-2-
one in 140 mL of ethyl acetate was reduced with 3.05 g of 5%
Pd/C. A hygroscopic white solid (1.38 g, 91% yield) was
obtained: 1H NMR (400 MHz, CDCl3) δ 4.64 (s, 2H, -C(O)-
CH2-O-), 3.78 (t, 2H, CH2N, J ) 5.86 Hz), 3.39 (br s, 1H, -OH),
2.55 (t, 2H, lactam -CH2-C(O)), 1.86-1.70 (m, 4H, lactam
-(CH2)2); 13C NMR (100 MHz, CDCl3) δ 177.5, 173.1, 66.8, 38.4,
24.3, 21.2, 20.0.
Signal Identification. The assignments of the equilibrium forms
of compounds 2 and 3 were made based on previous identification1,2
of N-(2-aminoacetyl)-2-lactams, 2D 1H NMR COSY, and NOESY
experiments and 13C NMR.
Sample Preparation. Samples of 1-(2-hydroxyacetyl)piperidine-
2-one (2) and 1-(2-hydroxyacetyl)azepan-2-one (3) in D2O at
different pD were prepared using phosphate buffers and at constant
Experimental Section
Synthesis. Benzyloxyacetyl Chloride. In a two-neck flask
provided with a condenser and a positive pressure of argon, 1.7
mL (12 mmol) of benzyloxyacetic acid and 15 mL of dry toluene
(7) Kirby, A. J. Angew. Chem., Int. Ed. Engl. 1994, 33, 551-553.
(8) Alberty, W. J.; Knowles, J. R. Biochemistry 1976, 15, 5631-5640.
(9) Alberty, W. J.; Knowles, J. R. Angew. Chem., Int. Ed. Engl. 1977,
16, 285-293.
(10) Knowles, J. R.; Alberty, W. J. Acc. Chem. Res. 1977, 10, 105-
111.
(11) Pettersson, G. Eur. J. Biochem. 1991, 195, 663-670.
(12) Veerapandian, B.; Cooper, J.; Sali, A.; Blundell, T.; Rosati, R.;
Dominy, B.; Damos, D.; Hoover, D. Protein Sci. 1992, 1, 322.
(13) Showen, K. B.; Limbach, H. H.; Denisov, G. S.; Schowen, R. L.
Biochim. Biophys. Acta 2000, 1458, 43-62.
J. Org. Chem, Vol. 73, No. 6, 2008 2107

Bethencourt and Nu´n˜ez
TABLE 1. Equilibrium Constants, Rate Constants, and Thermodynamic and Activation Parameters Obtained in This Work for Compound 2
(n ) 6) and Compound 3 (n ) 7)
pKa
cyclol,
open form
b,c
k0r
k0f
krd
kf
∆Gq
∆Gq
∆Gq
∆Gq
f
(kJ/mol)
0r
0f
r
n
Kfa
(s-1
)
(s-1
)
(s-1
)
(s-1
)
(kJ/mol)
(kJ/mol)
(kJ/mol)
6
7
12 ( 0.6
36 ( 3
7.3, 10.5
5.1, 9.8
4
2
48
72
5.0
8.5
20
64
64
66
56
55
63
61
60
57
(47)e (-40)f
(30)e (-70)f
a Neutral pH. b Standard deviation from simulations: 5%. c Obtained by simulation using eq 1 and equilibrium constants. d Obtained by simulation using
eq 1. e ∆Hq (kJ/mol). f ∆Sq (J/mol‚K).
ionic strength. In a typical sample, 0.2 mol/dm3 of compounds 2
or 3, 1 mol/dm3 of phosphate buffer (H3PO4 or KH2PO4), and 2
mol/dm3 of KCl were used. Samples were prepared directly into
the NMR tube by adding 25 mg of compounds 2 or 3, 0.12 g of
KCl, 0.7 dm3 of D2O, and the corresponding amount of buffer.
The sample pH was measured directly into the NMR tube before
and after taking the corresponding spectrum. The pH values did
not vary more than 0.1 pH units; pD values were obtained using
the relation:14 pD ) pH + 0.40.
In order to explore the presence of general catalysis, samples of
compound 3 at two different total phosphate buffer concentration
and constant pH were prepared. Total buffer concentrations of 0.75,
1.00, 1.50, and 2.00 mol/dm3 were used while maintaining a pD
constant at 6.3 and 9.0. A 400 MHz instrument was used to run
the different spectra.
Additionally, a sample of 25 mg of compound 3 in 0.7 cm3 of
D2O was used to obtain 1H NMR spectra at 18.5, 35, 45, and 55 °C.
A similar sample was used to perform COSY and NOESY and
EXSY 2D 1H NMR experiments. In the last case, two mixing times
of 100 and 300 µs were used. A 500 MHz instrument was used to
perform these experiments. For the proton inventory experiments,
volume fractions of H2O in D2O in the range of 0.02-0.20 were
used.
A line shape analysis program15 was used to simulate each
experimental spectrum and to obtain the pseudo-first-order rate
1
constants. The H NMR AB system assigned to the cyclol form
was used to evaluate rates. Best match between the experimental
and simulated spectrum was obtained by minimizing the line shape
difference between the two spectra while changing kobs values. A
1
FIGURE 1. D2O H NMR spectra of the AB system of the cyclol
form of compound 2 at pD ) 6.3 (top) and pD ) 9.9 (bottom). The
standard deviation no higher than 5% could be achieved performing
intraconversion between the A and B is catalyzed when increasing pD.
simulations.
times the corresponding one for compound 2 at neutral pD and
Results
at 19 °C (see Table 1). The equilibrium constant for compound
1 could not be measured4 since, at neutral pH, only the open
In Figures S1 and S2 (Supporting Information), the 1H NMR
spectra of compound 2 in CDCl3 and D2O are shown. In the
same figures, the assignments of the different signals corre-
sponding to the cyclol and open forms are depicted. In Figure
form is detectable by 1H NMR. Nevertheless, it can be remarked
that for compounds 2 and 3 the relative stability of the three
forms in D2O is cyclol > open form > macrocycle. The
1
macrocycle form of compound 3 is also observable in the H
S3 (Supporting Information), the assignment of signals in CDCl3
NMR. Yet, its contribution in the spectrum is barely detectable
for the three forms observed for compound 3 is shown. From
as well its participation in the exchange with the cyclol forms.
the relative integration of the open form exocyclic methylene
Two-dimensional 1H NMR EXSY experiments at 300 ms
protons (ca. 4.6 ppm, see Figure S2 in Supporting Information)
confirmed the last observation (not shown). The plot of ∆G°
and the AB 1H NMR system observed for the same protons in
(kJ/mol) versus T (K) (four points; see Experimental Section)
the cyclol (ca. 4.3 ppm, see Figure S2 in Supporting Informa-
was a straight line (∆G° ) 0.0295T - 16.924; R2 ) 0.98). From
tion), it is found that the equilibrium constants for the cyclol
this plot, ∆H° ) -16.9 kJ/mol and ∆S° ) -29.5 J/mol‚K are
form in D2O at neutral pD and room temperature for compounds
obtained.
2 and 3 are 12 and 36, respectively. However, this equilibrium
1
In Figure 1, the experimental H NMR spectrum of the AB
constant decreases considerably in CDCl3 (see Figures S1 and
system of the cyclol form of 2 in D2O at two different pD values
S3 in Supporting Information). For instance, for compound 2 it
is shown. A specific base catalysis or a kinetically equivalent
is equal to 0.9 and for compound 3, it is equal to 2. However,
one is observed since the AB system changes toward coales-
in D2O, for compound 3, the cyclol/open form equilibrium is 3
cence when pD increases. Similar changes are also observed
for compound 3.
(14) Glosoe, P.; Long, F. J. Phys. Chem. 1960, 64, 188-190.
Rate constants for the AB exchange at each pD were obtained
(15) Reich, H. J. WINDNMR: NMR Spectrum Calculations, version
7.1.11; Wisconsin University, 2005.
by simulating15 the line shape of each spectrum. In order to
2108 J. Org. Chem., Vol. 73, No. 6, 2008

Serine Protease Mechanism-Based Mimics
SCHEME 4a
1
a Cyclol proton exchange of the observed H NMR AB system. As it is shown, the probability of the AB exchange is half the rate of rupture (k0r) and
formation (k0f).
FIGURE 2. Plot of kobs versus pD for the AB 1H NMR system signal
FIGURE 3. Plot of kobs versus pD for the AB 1H NMR system signal
exchange of the cyclol form of compound 2. Dots: experimental points.
Solid lines: best fitting of the experimental points to eq 1, from which
the following constants were obtained: k0r ) 4 s-1; pKa cyclol ) 7.3;
kr ) 5 s-1; pKa open form ) 10.5 and kf ) 20 s-1. Error bars: 5% due
to error in simulation.
exchange of the cyclol form of compound 3. Dots: experimental points.
Solid lines: best fitting of the experimental points to eq 1, from which
the following constants were obtained: k0r ) 2 s-1; pKa cyclol ) 5.1;
kr ) 8.5 s-1; pKa open form ) 9.8 and kf ) 64 s-1. Error bars: 5% due
to simulation error.
observe exchange between the A and B protons of the 1H NMR
AB system corresponding to the cyclol form, it is required that
its C-O bond cleaves to yield the open form and that this form
re-closes to form the cyclol with the AB protons exchanged
(Scheme 4).
Therefore, the measured rate constants (kobs) are half the rate
constants for cleavage and formation. The kobs in the ample pD
range studied is then given by
Kac
Kac + [D+]
kobs ) 0.5k0r +
×
Kao
Kao + [D+]
0.5kr +
× 0.5kf (1)
FIGURE 4. Proton inventory experiment performed in the range of
0.8-1 atom fraction of deuterium. Error bars on the experimental points
correspond to the maximum deviation (5%) in rate constants according
to the line shape simulations. Solid line: kn/ko ) (1 + n(æ - 1))Zn, æ
) 0.98, Z ) 0.95.
where, k0r is the pseudo-first-order rate constant for water-
catalyzed rupture of the cyclol; kr and kf are the corresponding
pseudo-first-order rate constants of rupture and formation of
the conjugate bases of the cyclol and open forms; and Kac and
Kao are the acidity equilibrium constants for the cyclol and the
open form. In Figures 2 and 3, plots of kobs versus pD for
compounds 2 and 3 are shown.
These were obtained using the four temperatures indicated in
the Experimental Section. A proton inventory experiment
(Figure 4) performed on compound 3 at neutral pH and room
temperature shows only a small or no isotope effect (æ ) 0.98,
Z ) 0.95) in the range of the atom fraction of deuterium of
0.80-1.0. General catalysis by phosphate buffer at pH ) 5.9
and 8.6 was not observed either.
The solid line corresponds to the best fitting of the experi-
mental points to eq 1. From the best fit, the rate constants and
acidity constants involved in eq 1 have been obtained. These
parameters are shown in Table 1. In this table, the corresponding
∆Gq values are also shown. For the reaction of compound 3
catalyzed by water, the ∆Hq and ∆Sq values are also shown.
J. Org. Chem, Vol. 73, No. 6, 2008 2109

Bethencourt and Nu´n˜ez
SCHEME 5a
SCHEME 6a
a D2O-cyclol-D2O-open form interconversion scheme and rate and
equilibrium constants involved according to eq 1 and plots of Figures 3
and 4 for compounds 2 and 3. The three mechanistic routes (1, 2, and 3),
pointed out with dashed lines, correspond, respectively, to the three terms
of eq 1. The arrows are placed just after the corresponding rate-limiting
steps. To the left and right side of the marked square, participation of a
second water molecule is shown. Therefore, the pKa values obtained
correspond to the water hydrogen-bonded complexes.
a Water molecules hydrogen-bonded to the cyclol (c), the open form (a),
and the transition state (b) of compounds 1 (n ) 5, pKac ) 11.9), 2 (n )
6, pKac ) 7.3, pKao ) 10.5), and 3 (n ) 7, pKac ) 5.1, pKao ) 9.8). The
pKa values depend on the relative strain stability introduced by the hydrogen-
bonded cycle. For n ) 7, cyclol, δ ) 0.4 (see text), the D is located in the
middle of the two donors: the O of the cyclol and the O of water. For n )
5, δ ) 0 (see text), the hydrogen bond is weak. The difference in pKa
between the two cyclols is ca. 7. A similar situation occurs with the open
form -OD alcohol moiety, However, in this case, the hydrogen-bonded
cycle is more flexible, and small difference is observed between the two
pKa values (ca. 2). Therefore, the ∆δ between them is ca. 0.1. For n ) 6,
there is an intermediate behavior. The reaction at higher pH can be
envisioned through the same structures as those shown above, but with full
negative charge on alcohol oxygen instead of charge fractions (δ) with a
second water participation (mechanism 2 in Scheme 5) or the D2O-alcohol
form D2O complex interacting with a OD- molecule (mechanism 3 in
Scheme 5).
Discussion
We have been able to measure directly the rate constants for
cleavage and formation of a C-O bond in tetrahedral intermedi-
ates. As shown in Figures 2 and 3, three plateaus are detected
in the kobs versus pD profiles for compounds 2 and 3. These
correspond to the partial contribution to the rate of each term
in eq 1. The first one at neutral pD corresponds to the water-
catalyzed open form/cyclol interconversion, where the D2O-
cyclol rupture (k0r) is rate-limiting. The rate constants were
obtained from the plateau value (kobs ) k0r) and the equilibrium
constant obtained by 1H NMR integration of the cyclol and open
form signals: k0f ) Kfk0r. This mechanism is pointed out in
Scheme 5 with a horizontal dashed line. When the pD value is
increased, the second term in eq 1 becomes important due to
the low pKa value of the cyclol forms. Therefore, the main
contribution to the rate, at this stage, corresponds to the cyclol
conjugate base rupture. This mechanism is marked with a
vertical dashed line with the arrow just after the rate-limiting
step in Scheme 5. When Kac > [D+], the second level figures
are reached. Then, the kr and pKa values of the cyclol forms
are obtained. At higher pH, the third term in eq 1 predominates.
The corresponding mechanism is depicted with a diagonal
orientation in Scheme 5. In this mechanism, the cleavage of
the D2O-cyclol form is promoted by OD-, but this step is fast
since the cleavage of the C-O bond is also catalyzed by the
water molecule that is forming a hydrogen bond with the cyclol
form. Therefore, is the fraction (f) of the -OD-open form
complex what becomes relevant in this mechanism. However,
this species could bifurcate to the option of returning to the
D2O-cyclol form or go to the -OD-cyclol species. Therefore,
(Table 1). These results are in agreement with the experimental
observation of an equilibrium ([cyclol]/[open form]) slightly
shifted toward the cyclol form. The cyclol/open form equilibrium
depends on the solvent and the lactam cycle size. For instance,
in the case of compound 2 in CDCl3, the equilibrium becomes
1 order of magnitude lower than that in D2O, and for 3, it is ca.
20 times lower, indicating a better water stabilization of the
cyclol form in this solvent. On the other hand, the equilibrium
constant increases when increasing the ring size. For instance,
for compound 3 (n ) 7), K ) 36, for compound 2 (n ) 6), K
) 12, and for compound 1 (n ) 5), K , 1. Therefore, the ring
strain is an important factor that regulates the relative stabilities
of these stable tetrahedral intermediates. We were able to
measure the equilibrium (cyclol/open form) thermodynamic
parameters for compound 3 at neutral pH. For instance, ∆S°,
∆H°, and ∆G° values (18.5 °C) are -29.5 J/mol‚K, -16.9 kJ/
mol, and -8.4 kJ/mol, respectively. Although the ∆S° is
negative, this value is lower than expected since in the open
form there are three free rotation modes at the external
2-hydroxyacetyl group worth ca. 21 J/mol‚K each that are lost
in going to the cyclol form. Therefore, in the open form, there
must exist some rotation restriction that increases the entropy
value to -29.5 J/mol‚K. We attribute this rotation restriction
to a water hydrogen bound to the -OH of the 2-hydroxyacetyl
pendant substituent (see Scheme 6a). The low pKa value (5.1)
kinetically determined for the cyclol form of 3 is also a result
that we have attributed to the water hydrogen bond participation
via an intramolecular cyclic interaction, as shown in Scheme
6c. Through this interaction, the cyclol -OH bridges form a
strong hydrogen bond that increases the negative charge on the
the rate law for this step is kobs
) f
-OD-open
form (f D2O-open form × k0r + kf). At relatively low pD, the
first term in parentheses predominates since k0f has a high value;
see Table 1. This is then an alternative route to return to the
D2O-cyclol form without involvement of OD-, but as soon as
the pD increases, the second term in the last parentheses
predominates since the fraction of the D2O-open form decreases
and the last equation is simplified to the third term in eq 1.
Once the -OD-open form fraction becomes equal to 1, the third
level in Figures 2 and 3 occurs and the kf and pKa values of the
open form are obtained. In Scheme 5, the cyclol-open form
interconversion in the ample pD range studied is shown.
The rate constants of formation of the cyclol (C-O bond
formation) are faster than the corresponding C-O cleavage
2110 J. Org. Chem., Vol. 73, No. 6, 2008

Serine Protease Mechanism-Based Mimics
oxygen and makes the alcohol more acidic (and not more basic
since the bridge is established with a water molecule instead of
another intramolecular heteroatom). It has been suggested13 that,
in this proton bridge interaction, the proton is in a stable potential
and the location of the bridging proton will shift toward the
more basic residue. In fact, in Scheme 6c, water becomes the
more basic center since its pKa increases from the hydronium
ion one toward the water one since another of its protons is
also strongly hydrogen bound to a second water molecule (the
third hydrogen forms a weak hydrogen bond since the ether
counterpart has a low pKa). Therefore, the induced change of
charge on the active center is drained and balanced at long
distance through water hydrogen bonding.
Scheme 6 hydrogen bonds with the water molecule are in a
downward potential (perpendicular coordinate in a two-
dimensional energy diagram) as predicted by the alternative
model16,17 rather than the canonical18 one in which the proton
transferred in concert with the main reaction coordinate in a
upward energy potential (parallel coordinate in a two-dimen-
sional energy diagram).
According to the alternative proposal,16,17 bond order will be
conserved at unity about the bridging proton so that changes in
the proton location will generate shifts in electrical charge within
the bonding partners. Furthermore, it has been estimated13 from
cross-correlation plots that the expected sensitivity of the
electrical shift charge is 0.06 units of charge/unit change in pK
of a bonding partner. This proposal has been used13 to explain
the reported19,20 change of pKa of the Asp-His dyad from 7 to
12 when going to a transition state due to long distance proton
bridges induced by the serine -OH pKa change induced by its
nucleophilic attack on the peptide carbonyl group. This kind of
long distance effect has been used13 to explain, for instance,
the nonlocal character of mutation and to understand enzyme
evolution. In our system, the change in pKa is introduced by
molecular strain and not by changing the pKa of one of the
heteroatoms involved. These results represent a new contribution
to the understanding of hydrogen bond interactions in the sense
that the hydrogen bond is also sensitive to changes in strain
introduced, for instance, in these models, by the size of the
lactam ring. We predict then that other strains at the enzymatic
active site will also produce sensitive changes in the position
of the bridged proton perhaps bringing about a change in the
enzyme activity.
The alternative model16,17 is operating in this work since a
greater stabilization of the hydrogen bond makes the center more
acidic and moves the position of the bridged proton away from
the acidic center. This geometric response is expected only if
the proton is in a downward potential in which stabilization of
products induces a proton movement away from the reactants
in a typical “anti-Hammond” sense. Therefore, the more stable
this interaction is, the more acidic the cyclol form becomes (the
same argument is valid for the open form, Scheme 6a). In fact,
for compound 2, the cyclol pKa value is 7.5, ca. 2 orders of
magnitude less acidic than the cyclol form of compound 3 due
to ring strain restriction. The latter is considerably more sensitive
to the cyclol acidity than to its stability relative to the open
form where only a factor of 3 is involved between compound
3 and compound 2. This difference in sensitivity is due to the
significant change of charge density (0.0613 × 2 ) 0.12) of
positive charge at the O alcohol center (∆δ ) 0.12 in Scheme
6c, cyclol form). This difference also contributes to the relative
stability of the cyclol form with respect to the open form, but
the last form is also stabilized by a cyclic hydrogen bond water
interaction (Scheme 6a); therefore, the equilibrium constant does
not show the same sensitivity as the pKa values. The sensitivity
of the pKa of compounds 1, 2, and 3 is quite related to the
stability of the structures arrived at by molecular calculation21
shown in Scheme 6c. Optimization with molecular mechanics
and MOPAC minimization using the AM1 method gives the
relative heat of formation of the hydrogen-bonded cyclol forms
of compounds 1, 2, and 3. When these heats of formation
(-181.7, -190.8, and -193.1 kcal/mol) are plotted versus the
experimental pKa values of these compounds, a good correlation
is obtained: y ) 0.57x + 115.69, R2 ) 0.98. The three pKa
values used are the ones shown in Table 1. Notice that the
estimated pKa (11.7) of the cyclol form of 1 is in good agreement
with the experimental one4 (11.9), and it takes into account only
inductive effects. This value decreases as predicted by the above
correlation due to the third cyclic hydrogen bond strain relief
introduced by increasing the lactam ring size. Therefore, in the
cyclol form of 1, the hydrogen bond is weak or does not exist
at all due to strain arising from the five-membered ring. In fact,
when semiempirical AM1 molecular calculations21 are per-
formed on the cyclols’ conjugate bases interacting with oxonium
ion (“cycloxide”-H3O+), the stabilities of the three “cycloxide”-
H3O+ correlate well with the experimental pKa values. More-
over, in the calculation, an important elongation of the ether
C-O bond (ether) is observed that follows the order: n ) 5 .
n ) 6, n ) 7. Therefore, for n ) 7, the stability of the anion
may be due to strain, anomeric effect22 induced by the ether
oxygen electron pairs, or both (no change in the C-N bond is
observed among the three compounds). In any case, as the
experimental data show, the calculation confirms the stability
order of the anion-D3O+ of n ) 7 > n ) 6 . n ) 5, so that the
pKa order is n ) 5 . n ) 6 > n ) 7.
The lack of general base catalysis by phosphate buffer at pH
6.3 for the interconversion of the cyclol and open form of 3 is
another piece of evidence for the efficiency of the role of water
general catalysis via a strong hydrogen bond. In the alcohol
formation from the adducts of N,O-trimethylenephthalimidium
cation, Gravitz and Jencks6 have found general acid catalysis.
However, in this system, the driving force for the C-O cleavage
comes from a weak N (amide) electron pair; therefore, general
acid catalysis is required. This is not the case for the cyclol
forms of this work where the driving force for the C-O cleavage
comes from an oxygen (alcohol or alkoxide) electron pair.
Furthermore, a water-hydrogen bond as depicted in Scheme 6
is not predicted to be formed in Gravitz’s and Jencks’s system
since there is an entropic factor against the bimolecular
expulsion of the associated alcohol. A substitution of water by
a general acid-base buffer in the system described in this work
is also improbable since the charge dispersion shown in Scheme
6c through a second water molecule makes it possible for the
directly attached water molecule to match its pKa with the cyclol
pKa and form a stronger hydrogen bond. This charge transfer
(16) Cordes, E. H. Prog. Phys. Org. Chem. 1967, 4, 1-44.
(17) Eliason, R.; Kreevoy, M. M. J. Am. Chem. Soc. 1978, 100, 7037-
7041.
(18) Jencks, W. P. Chem. ReV. 1972, 72, 705-718.
(19) Liang, T. C.; Abeles, R. H. Biochemistry 1987, 26, 7603-7608.
(20) Finucane, M. D.; Malthouse, J. P. G. Biochem. J. 1992, 286, 889-
900.
(21) Cambridgesoft. CSMOPAC PRO. 1996-2001. Semiempirical Cal-
culations.
(22) Kirby, A. J. The Anomeric Effect and Related Stereolectronic Effects
at Oxygen; Springer: New York, 1983.
J. Org. Chem, Vol. 73, No. 6, 2008 2111

Bethencourt and Nu´n˜ez
efficiency is limited in the case of other bifunctional buffers,
such as for instance NaH2PO4.
the open form that does not require any base catalysis assistance,
but the formed alkoxide group is assisted by water and not by
oxonium ion. When both are present, as is the case of the first
step of the mechanism 3 in Scheme 5, the rupture and formation
of the C-O bond are very fast compared to the rates measured
in this work and probably even higher than the reported
enzymatic values. Although the intramolecular distances may
resemble those at the enzyme active site, neither the orientation
between the serine hydroxyl group and the protein carbonyl
group nor the inductive effect may be the same. Therefore,
changes in these maximum k2 and k-2 values may be observed.
For instance, k2 values26 for trypsin and modified Cys191-
Cys220 trypsin hydrolysis of D-ValLeuLys-AMC are 280 and
1.4 s-1, respectively.
The cyclol/open form increase when going from CDCl3 to
D2O also supports the role of a hydrogen bond with water. In
agreement with the strain introduced by the lactam ring, the
latter increase is more important for compound 3 (n ) 7) than
for compound 2 (n ) 6).
For compound 3, we have also obtained the corresponding
activation parameters at neutral pD from a plot of 1/kobs versus
1/T (see Table 1). As shown in Table 1, the ∆Sq values for
rupture and formation of the cyclol form are both negative. This
is not the expected result for the rupture since the transition
state is less ordered than the cyclol form. However, if water
participates in the rupture, a more ordered transition state
(Scheme 6b) as compared to the water-bonded cyclol form is
indeed predicted since even if a strong hydrogen bond is formed
some rotational movement is allowed in the reactants. In fact,
in the case of the ammonium ion23 and water, the strength of
the hydrogen bond only prevents a nearly free rotation. Even
where the hydrogen bond appears to be quite strong due to
similarities in pKa between donor and acceptor, there seems to
be hardly any barrier to rotation.24 This movement becomes
more restricted at the transition state due to the orientation
imposed by the reaction coordinate.
Finally, it is important to point out that the rate of water
exchange of the proton forming hydrogen bond should be quite
fast. For instance, according to the Swain-Grunwald mecha-
nism27 for the cyclol form of compound 3 (pKa ) ca. 5), this
rate is ca. 105 s-1 since the rate of dissociation of the hydrogen-
bonded form is ca. 109 s-1, so that the rate-limiting step for the
exchange is the proton transfer to the water molecule. Therefore,
the proton-bridged signal is averaged with the water one in the
1H NMR spectrum. As has been established,28 this averaging
makes the direct detection of the hydrogen bond not possible.
1
More chances to observe the hydrogen bond by H NMR are
Regarding the symmetry and mobility of the hydrogen bond,
due to the asymmetry of the involved donors, a double-well
potential hydrogen jumping between donors is expected.
Recently, it has been shown25 that, even in the cases of
symmetric donors and counterion, the H bond is asymmetric
and there is an equilibrium between solvatomers. If it is the
case, the change in pKa observed in this work may also be
interpreted as the average contribution of the two donors’ pKa,
for instance, in the case of the cyclol form: H3O+ (pKa ) ca.
-1) and cyclol-OH (pKa ) ca. 12), that for compound 3,
assuming 50% contribution of each donor, the final pKa would
be 0.5 × (-1) + 0.5 × 12 ) 5.5 (experimental pKa ) 5.1).
for the bridge -OH of the open form where the pKa is ca. 9-10.
In this case, the exchange rate will be ca. 1-10 s-1 and non-
coalescence with the dominant water signal might be observed
in the H2O spectrum; we did not observe any -OH signal when
using a 20% H2O/D2O mixture. However, in CDCl3 (Figure
S1, Supporting Information), both -OH signals for the cyclol
form (ca. 4.15 ppm) and the open form (ca. 3.4 ppm) are
observed but the open form signal exchanges (with traces of
water) slower than the cyclol one.
Conclusions
The proton inventory results are particularly germane to the
water participation hypothesis. As shown in Figure 4, there is
a kH/kD isotope effect close to 1 for compound 3. This is, in
fact, what should be expected if the proton transfer does not
occur along the reaction coordinate since the transfer has been
prearranged at the reactants through strong hydrogen bond
proton bridges. Hence, the protons were already bonded at the
reactants, and therefore, there is no isotopic contribution at the
transition state (Scheme 6b).
The kinetic rate constants measured in this work establish a
rate constant range for the serine hydroxyl attack on the peptide
bond in serine proteases: from water-catalyzed activation of
the serine alcohol group to fully unprotonated alkoxide attack.
For instance, the k0f values in these models are 72 and 48 s-1
for n ) 7 and 6, respectively. These rate constants mimic the
minimum value for serine -OH attack to the peptide bond (k2
values) in enzyme proteases since water is acting as the general
base catalyst instead of histidine. Meanwhile, kf (ranging from
64 s-1 (n ) 7) to 20 s-1 (n ) 6)), represents the value that
should be obtained in enzyme proteases in the absence of an
acid catalyst since the attacking group is the alkoxide group of
The relative stability between the cyclol (tetrahedral inter-
mediate) and open form (peptide-enzyme complex) in this
serine protease mechanism-based mimic depends on the ring
size of the lactam cycle. The cyclol stability increases when
increasing the cycle size (n ) 5-7). Water is highly associated
with the cyclol and open form through strong hydrogen bonds
in stable potential proton bridges. This suggestion is supported
by thermodynamic and kinetic entropic values, the strong
sensitivity of pKa values to the lactam ring size (strain effect),
the lack of an observable solvent isotope effect and phosphate
buffer catalysis, and semiempirical molecular calculations. The
general acid-base catalysis by water does not occur in the
reaction coordinate, and the strength of the hydrogen bond is
sensitive to the strain arising from the size of the lactam ring.
This result can be extrapolated to the enzyme active site where,
in general, any strain could cause a significant change in the
stability of the hydrogen bond, reducing the enzymic activity
even at long distance since the effect could be transmitted
through a hydrogen bond.
Evidence for the existence of hydrogen bonds in the active
sites of enzymes with energies in the range of 10-20 kcal/
mol, using intramolecular proton transfer (general acid-base)
(23) Perrin, C. L.; Gipe, R. K. Science 1987, 238, 1393-1394.
(24) Perrin, C. L.; Dwyer, T. J.; Rebeck, J., Jr.; Duff, R. J. J. Am. Chem.
Soc. 1990, 112, 3122-3125.
(25) Perrin, C. L.; Lau, J. S. J. Am. Chem. Soc. 2006, 128, 11820-
11824.
(26) Wang, E. C. W.; Hung, S. H.; Cahoon, M.; Heddstrom, L. Protein
Eng. 1997, 10, 405-411.
(27) Grunwald, E.; Ralph, E. Acc. Chem. Res. 1971, 4, 107.
(28) Perrin, C. L. Science 1994, 266, 1665-1668.
2112 J. Org. Chem., Vol. 73, No. 6, 2008

Serine Protease Mechanism-Based Mimics
catalysis in model systems, has been previously presented29 but
at the transition state level. The results presented in this work
would be the first case, to our knowledge, where evidence is
presented of the existence of an asymmetric (double potential)
strong hydrogen bond with water and at the reactants level.
catalyzed hydrolysis has been previously reported;30 however,
the rates reported are at least 60 times slower than the values
reported in this work. We are now involved in the synthesis
and kinetic study of compound 4 in which a pendant imidazole
group has been attached to the exocyclic methylene group of
the open form. Therefore, we will have in the same molecule
the substrate peptide bond, the hydroxymethyl group that mimics
the serine residue at the active site, and an imidazole group
that mimics the enzymic histidine role.
The system studied mimics the serine alcohol attack to the
peptide bond and its reverse reaction in serine protease enzymes.
A rate constant range for the serine attack can be readily
established: from serine -OH attack without base catalysis and
-O- attack. However, k2 values higher than those shown in
this work have been reported, for instance, for trypsin. It is quite
possible that the five-membered ring formation involved in the
cyclol form of these models introduces a strain that is not
involved in the enzyme reaction. It is also important to remark
that, in the systems studied, the cleavage to the acyl alcohol
(C-N bond) is slower than the reverse cyclol formation (C-O
bond cleavage), making the acyl formation rate-limiting. This
observation can also be specific for these models since opening
the cyclol brings about a strain release when the C-O bond is
broken and the open form is formed. Although, in the formation
of the macrocycle (peptide bond cleavage), there will also be
strain relief, a cyclic peptide is formed. Nevertheless, the kinetics
obtained for these models (ca. 2-70 s-1) are of the same order
of magnitude as those in the enzymatic reactions. The lacton-
ization of 2-hydroxymethylbenzoic acid and the corresponding
benzamide as models for the acylation step in the chymotrypsin-
Acknowledgment. We acknowledge the Environment Man-
agement Unit (UGA) at Simon Bolivar University for financial
support. We also acknowledge the Royal College of Surgeons
in Ireland and, in particular, Professor Kevin Nolan for his
hospitality and support during the sabbatical of one of the
authors (O.N.) where part of this manuscript was written. Author
(O.N.) dedicates this article to the memory of E.N.T.
Supporting Information Available: Additional figures. This
materialisavailablefreeofchargeviatheInternetathttp://pubs.acs.org.
JO702003Q
(29) Hartwell, E.; Hodgson, D. R. W.; Kirby, A. J. J. Am. Chem. Soc.
2000, 122, 9326-9327.
(30) Chiong, K. N. G.; Lewis, S. D.; Shafer, J. A. J. Am. Chem. Soc.,
1975, 97, 418-423.
J. Org. Chem, Vol. 73, No. 6, 2008 2113