Kinetic and mechanistic insight into the thermodynamic degradation of saxagliptin

Article abstract of DOI:10.1021/jo202052a

The dipeptidyl peptidase-IV inhibitor saxagliptin (Onglyza) can undergo a thermodynamically favored cyclization to form the corresponding cyclic amidine. The kinetics and mechanism of this conversion were examined to develop a commercial synthesis that afforded saxagliptin with only trace levels of this key byproduct. Important findings of this work are the identification of a profound solvent effect and the determination of an autocatalytic pathway. Both of these phenomena result from transition structures involving proton transfer (Figure presented).

Full text of DOI:10.1021/jo202052a

Note
pubs.acs.org/joc
Kinetic and Mechanistic Insight into the Thermodynamic
Degradation of Saxagliptin
G. Scott Jones,* Scott A. Savage, Sabrina Ivy, Patrick L. Benitez,† and Antonio Ramirez
Chemical Development, Bristol−Myers Squibb Company, One Squibb Drive, New Brunswick, New Jersey 08903, United States
S
* Supporting Information
ABSTRACT: The dipeptidyl peptidase-IV inhibitor saxagliptin (Onglyza)
can undergo a thermodynamically favored cyclization to form the
corresponding cyclic amidine. The kinetics and mechanism of this conversion
were examined to develop a commercial synthesis that afforded saxagliptin
with only trace levels of this key byproduct. Important findings of this work
are the identification of a profound solvent effect and the determination of an
autocatalytic pathway. Both of these phenomena result from transition
structures involving proton transfer.
ipeptidyl peptidase-IV (DPP-IV) inhibitors are an
important new class of oral hyperglycemic medications.
With regards to cyclic amidines 2 and 3, it is worth noting
that standard solution reaction conditions afford exlusively
epimer 3. The body of evidence supports Scheme 1, namely,
that rate-limiting cyclization occurs first, followed by rapid
epimerization. First, monitoring the reaction in the solid state,
where the rate of epimerization of 2 is reduced, revealed the
transient presence of intermediate 2, which subsequently
converted to epimer 3.6 Second, the nitrile epimer of 1 was
observed to cyclize to 3 at a rate that was indistinguishable from
the rate of saxagliptin (1) cyclization.7 Finally, the proton alpha
to the amidine in 3 readily exchanges with deuterium in D2O
solutions, whereas a similar exchange does not occur at this
position for solutions of saxagliptin (1) in D2O. Together, these
data support the proposed mechanism in Scheme 1 and serve
as evidence against an alternative mechanism, in which rate-
limiting epimerization of saxagliptin (1) is followed by rapid
cyclization to amidine 3. The hydrolysis of amidine 3 to
diketopiperazine 4 occurs in the presence of water and is not
critical to understanding the mechanism and kinetics of the
cyclization but is presented here for completeness.
The initial rates for the formation of 3 were monitored in the
process solvents isopropanol (i-PrOH), dichloromethane, and
ethyl acetate at 40 °C and [1] = 0.04 M. Under these
conditions, the cyclization displays a clean first-order depend-
ence on saxagliptin concentration. The studies showed that the
cyclizations in i-PrOH were 5-fold faster than the cyclizations in
ethyl acetate and 3-fold faster when compared to dichloro-
methane. Further examination revealed that the rates correlated
well with the solvent’s propensity to participate in proton
transfer reactions, namely, (a) the solvent’s pKs, where Ks is the
autoprotolysis equilibrium constant of the protic or protogenic
solvent, and (b) the solvent’s proton affinity, which represents
the solvent’s absolute gas phase basicity (Figure 1).8
D
2-Cyano pyrrolidine derivatives such as saxagliptin1 (1) have
been studied extensively for activity as DPP-IV inhibitors
because of the enzyme’s specificity for substrates with an
amino-terminal proline. However, many members of this
chemotype suffer from chemical instability, rendering them
impractical for drug development because of an inherent
inability to be synthesized or formulated at the scales required
for commercial pharmaceutical manufacturing.2 As demon-
strated in Scheme 1, these compounds are prone to undergo
cyclization, whereby the N-terminal amine attacks the nitrile,
forming inactive cyclic amidines or their diketopiperazine
hydrolysis products.
In the case of saxagliptin, substantially increased stability was
designed into the molecule via side-chain bulk with an
adamantyl group as well as by addition of the methano-bridge
on the proline moiety.3 Although these structural modifications
address the cyclization issue effectively from a physiological
perspective,4 minimizing the formation of the cyclic amidine 3
impurity remained the primary challenge of designing a process
to synthesize pure 1 on a commercial manufacturing scale.5
This paper discusses the roles of protic solvents and
autocatalysis in the cyclization of 1 to 3 and the related
development of a kinetic model that was utilized to guide
process development.
The cyclization reactions were monitored by HPLC analysis
of reaction streams, and the concentration of species 1, 3, and 4
was determined by direct comparison to fully characterized
samples prepared independently. Unless otherwise noted, all
experiments utilized pure 1, containing <0.1% of amidine 3 and
no diketopiperazine 4. As saxagliptin is isolated as the
crystalline monohydrate,5 a molar equivalent of water was
present in all solutions, and thus the relatively slow hydrolysis
of 3 to 4 was observed during reactions. The reported rate
constants and activation energies were regressed from the
concentration profiles of species 1, 3, and 4 using the
commercially available software, DynoChem.
Received: October 3, 2011
Published: November 4, 2011
© 2011 American Chemical Society
10332
dx.doi.org/10.1021/jo202052a | J. Org. Chem. 2011, 76, 10332−10337

The Journal of Organic Chemistry
Note
Scheme 1. Undesirable Conversion of Saxagliptin to Amidine and Diketopiperazine Byproducts
Figure 1. Measured rate constant versus solvent’s (a) pKs, where Ks is the autoprotolysis equilibrium constant of the protic or protogenic solvent and
(b) proton affinity.
Figure 2. Plot of relative rate versus deuterium atom fraction.
To ascertain the participation of i-PrOH in the reaction
coordinate and its role in the manufacturing process, we carried
out proton inventory studies by monitoring the kinetic isotope
effect, kn/k0, as a function of the mole fraction of i-PrOD in i-
PrOH, n, at 40 °C and [1] = 0.04 M. The Gross−Butler
equation was employed for purposes of evaluating the results
(eq 1).9 The polynomial fit of the experimental data to a
derived form of eq 110 (Figure 2) reveals a number of insights
about the reaction of interest. First, a normal kinetic isotope
effect kH/kD ∼ 3 indicates that a proton transfer involving i-
PrOH is indeed rate limiting. Second, the curvature of the line
that fits the data (solid line in Figure 2) provides insight into
the number of protons involved in the transition state.9,11 If
only a proton from the amine group in 1 was transferred in the
transition state, the rates would vary linearly with deuterium
content (dashed line in Figure 2). The fact that the data are
best fit with a quadratic implies that two protons are moving in
the transition state. Solving the quadratic equation that best fits
TS
the data in Figure 2 gives two fractionation factors Φ1TS ∼ Φ2
∼ 0.55. The ΦiTS values and the overall kinetic isotope effect
reported here are in line with proton inventories reported by
others for two simultaneous proton transfers.9,11
(1)
The simplest and most general transition state that is
consistent with the experimental observations is transition
10333
dx.doi.org/10.1021/jo202052a | J. Org. Chem. 2011, 76, 10332−10337

The Journal of Organic Chemistry
Note
structure 5a. A search for transition structures with DFT
calculations at the B3LYP/6-31+G(d)/PCM12 level using
MeOH instead of i-PrOH and a model substrate (R = t-Bu)
for simplification afforded stucture 5b, which displays a double
proton transfer between the alcohol solvent and the 2-cyano
pyrrolidine with concomitant N−CN bond formation. The
calculations overestimate the ΔG‡298 value for 5b (149 kJ/mol)
relative to the experimental results shown in Table 2 but are in
qualitative agreement with the higher stabilization imparted by
the autocatalytic pathway (vide infra).13
Table 1. Rate Equations Corresponding to Figure 3a
reaction
rate expression
rate constant
1 → 3
k [1]
0.002 1/h
1 + 3 → 3 + 3
k [1] [3]
k [3] [H2O]
0.158 L/mol·h
0.068 L/mol·h
3 + H2O → 4 + NH3
Two hypotheses were considered as possible explanations for
the autocatalysis. First, a simple change in the concentration of
basic species during the reaction might be responsible for the
increasing rates of cyclization.14 Alternatively, amidine 3 could
promote a double proton transfer transition state such as 6a
analogous to transition state 5a. Similar complexes have been
proposed in the 2-pyridone catalyzed mutarotation of
tetramethyl glucose,15 the lactim−lactam tautomerism of 2-
pyridone,16 and the reactions of various amidines.17
Interestingly, monitoring the formation of 3 at higher
temperatures and concentrations of 1 in i-PrOH (i.e., 75 °C
and [1] = 0.85 M) afforded a sigmoidal profile instead of the
exponential shape expected for a first-order dependence in
substrate. The sigmoidal profile suggested the existence of
autocatalysis, which was supported by kinetic modeling
demonstrating that the best fit required an autocatalytic
component in amidine 3 (Figure 3a). The simple three-
reaction model is shown in Table 1. At the maximum
concentration of amidine 3, the rate of the autocatalytic
pathway is approximately 20-fold greater than the unimolecular
conversion of 1 to 3. To further support the autocatalysis
hypothesis, a series of reactions was conducted, in which
increasing amounts of cyclic amidine 3 were spiked into a 0.375
M solution of 1 in i-PrOH at 75 °C. Analysis of the initial rates
from these reactions confirmed that the cyclization rates are
directly proportional to the initial concentrations of amidine 3
(Figure 3b). Taken together, these data unequivocally
demonstrate autocatalysis as a key factor in the decomposition
of 1 at high concentrations and temperatures.
To investigate the two scenarios, we examined the effect of
additives morpholine (7) and 2-aminothiazoline (8) upon the
cyclization rates. The pKa values for the conjugated acids of 7
and 8, 8.3 and 8.7 respectively, are similar to that of amidine 3
(8.3),18 but 7 and 8 differ in their capacity to mediate proton
transfer-based transition states akin to 6a. Interestingly,
addition of 0.1 equiv morpholine (7) resulted in rates
comparable to those measured in the absence of additive,
whereas addition of 0.1 equiv 2-amino-2-thiazoline (8) afforded
faster rates that reproduced those measured in the presence of
0.1 equiv amidine 3 (Figure 4). The excellent agreement
between the amidine- and thiazoline-spiked experiments and
the absence of any effect in the morpholine-spiked reaction
Figure 3. (a) Representative concentration profiles for the cyclization of saxagliptin ([1] = 0.85 M in i-PrOH at 75 °C). (b) Plot of initial cyclization
rates versus concentration of amidine 3 ([1] = 0.375 M in i-PrOH at 75 °C).
10334
dx.doi.org/10.1021/jo202052a | J. Org. Chem. 2011, 76, 10332−10337

The Journal of Organic Chemistry
Note
Figure 4. Molar conversion versus time for spiking experiments comparing additives: amidine 3, thiazoline 8, morpholine 7, with a control
experiment with no additive.
support the participation of the amidine functionality as
proposed in transition structure 6a. A simplified transition
structure 6b (R = t-Bu) could be located using DFT
computations at the B3LYP/6-31G+(d)/PCM level.12 6b
displays a double proton transfer between the 2-cyano
pyrrolidine and the amidine with simultaneous formation of
and obtain activation energies from the rate constants.19 The
experimental data across these conditions and in time were well
represented by the two simple rate expressions shown in Table
Table 2. Rate Equations Corresponding to Figure 5
ΔG‡ (kJ/mol)
the incipient N−CN bond. The calculations predict a ΔG‡
rate expression
Ea (kJ/mol)
298
298
k [1] [i-PrOH]
k [1] [3]
60
49
98
82
value of 96 kJ/mol for the autocatalytic transition state 6b,
which is fairly close to the experimental estimate of 82 kJ/mol
presented in the following section.
2. Consistent with transition states 5 and 6, both rate
expressions involve bimolecular interactions between 1 and i-
PrOH20 or between 1 and 3, respectively. As demonstrated by
the parity plot shown in Figure 5, the simple model based on
the mechanistic studies described here does an excellent job of
predicting reaction performance in the commercial manufactur-
ing process.
In summary, amidine 3 was the key impurity that required
control during process development of the DPP-IV inhibitor
saxagliptin. Amidine 3 is formed by a cyclization of saxagliptin
that involves a double proton transfer. Hydrogen bonding and
proton transfer in the transition state are important aspects of
both the uncatalyzed and the autocatalytic routes to amidine 3.
Detailed understanding of these factors played an essential role
during the development of a robust commercial manufacturing
process for saxagliptin.
The goal of this work was to improve our ability to control
the undesired formation of 3 during the synthesis of saxagliptin.
It is worth noting that even in i-PrOH, the rate of the undesired
conversion is rather slow. However, within the context of the
long processing times encountered in commercial manufactur-
ing and the stringent purity requirements of an active
pharmaceutical ingredient, even such relatively slow con-
versions require detailed understanding. The cyclization kinetic
studies represent ideal systems, using very pure starting
materials and single-solvent reaction matrices that did not
change in time. Despite the fact that these conditions are
somewhat removed from those encountered in the commercial
manufacturing process, the knowledge gained in these studies
formed the foundation of a kinetic model for the commercial
process.
In the commercial process, the unit operation that required
the greatest understanding to minimize the cyclization was the
distillation, or concentration, that took place following the
formation of 1. This operation was of particular interest
because a number of parameters with potential to influence
cyclization kinetics (i.e., temperature, solvent composition, and
concentration) change in time as the operation proceeds. A set
of experiments was undertaken to determine the influence of
these parameters on the rate of cyclization during the
concentration. Samples of the processing stream were taken
at the beginning, middle, and end of the concentration,
representing discrete conditions encountered in the manufac-
turing process. Each stream was further subdivided and held at
different temperatures in order to monitor the cyclization rate
EXPERIMENTAL SECTION
■
(1aS,4S,6aR,7aS)-4-((1R,3R,5R,7S)-3-Hydroxyadamantan-1-
yl)-6-iminohexahydro-1H-cyclopropa[4,5]pyrrolo[1,2-a]-
pyrazin-3(1aH)-one (3). A mixture of saxagliptin (1, 66.6 g, 200
mmol) and n-BuOH (250 mL) was stirred at 90 °C. Then, n-BuOAc
(350 mL) was added to the resulting suspension, and the mixture was
heated to reflux until a pale yellow solution was obtained. To remove
water from the mixture, 120 mL of n-BuOH/n-BuOAc azeotrope were
collected and discarded. After 3 h at reflux, the amidine 3 started to
precipitate out. The suspension was refluxed for 5 h, cooled to rt, and
stirred for 16 h at rt. The mixture was filtered, and the filter cake was
washed with EtOAc (200 mL) and dried with suction. The crude
material (44.6 g) was dissolved in i-PrOH (900 mL) at reflux, and the
solution was polish filtered hot. The solution was concentrated at
atmospheric pressure to ca. 300 mL to afford a suspension that was
cooled to rt and filtered with suction. The solid was washed with i-
PrOH (150 mL) and dried under vacuum (∼3 mmHg, 55−60 °C) to
constant weight for 1 h to yield the desired amidine 3 as white crystals
1
(24.2 g, 38.4%): mp > 209.7 °C; H NMR (300 MHz, MeOH-d4) δ
10335
dx.doi.org/10.1021/jo202052a | J. Org. Chem. 2011, 76, 10332−10337

The Journal of Organic Chemistry
Note
Figure 5. Parity plot of predicted versus experimentally observed data.
0.77 (m, 1H), 0.83 (dd, J = 10.5, 6.0 Hz, 1H), 1.47−1.70 (m, 11H),
2.03 (m, 1H), 2.16 (br s, 2H), 2.52 (m, 1H), 3.56 (br s, 1H), 3.89 (br
t, J = 7.5 Hz, 1H), 4.08 (ddd, J = 6.0, 6.0, 3.0 Hz, 1H); 13C NMR (75
MHz, CDCl3) δ 7.9, 12.1, 31.8, 32.3, 35.2, 36.5, 39.1, 45.0, 45.2, 47.4,
53.7, 69.2, 70.9, 160.9, 168.9; HRMS calculated for C18H26N3O2 [M +
H+] 316.2025, found 316.2029.
ASSOCIATED CONTENT
* Supporting Information
General methods, rate data, copies of NMR spectra, and
molecular modeling coordinates. This material is available free
of charge via the Internet at http://pubs.acs.org.
■
S
Isolation of (1aS,4S,6aS,7aS)-4-[(1R,3R,5R,7S)-3-Hydroxyada-
mantan-1-yl]-6-iminohexa-hydro-1H-cyclopropa[4,5]pyrrolo-
[1,2-a]pyrazin-3(1aH)-one (2). A series of saxagliptin (1) tablets
stressed at 50 °C in a stability study were crushed and extracted with
0.1 N HCl. The resulting extract was lyophilized, and amidine 9 was
AUTHOR INFORMATION
Corresponding Author
*E-mail: scott.jones@bms.com.
■
1
Present Address
isolated using semipreparative HPLC: H NMR (400 MHz, DMSO-
†Department of Bioengineering, Stanford University, 318
Campus Drive, Stanford, California 94305, United States.
d6) δ 0.42 (ddd, J = 5.0, 5.0, 2.5 Hz, 1H), 1.06 (ddd, J = 8.3, 5.0, 5.0
Hz, 1H), 1.45−1.70 (m, 10H), 1.60−2.00 (m, 2H), 1.77 (m, 1H), 1.96
(ddd, J = 13.3, 8.7, 2.0 Hz, 1H), 2.12 (br s, 2H), 2.76 (ddd, J = 13.3,
7.9, 7.9 Hz, 1H), 3.35 (m, 1H), 3.74 (br s, 1H), 5.20 (t, J = 7.9 Hz,
1H), 8.62 (s, 1H), 9.18 (s, 1H), 9.38 (s, 1H); 13C NMR (100 MHz,
CDCl3) δ 14.8, 21.0, 31.2, 36.3, 60.7, 61.3, 66.4, 163.5, 164.6; HRMS
calculated for C18H26N3O2 [M + H+] 316.2025, found 316.2035.
(1aS,4R,6aR,7aS)-4-[(1S,3S,5R)-3-Hydroxyadamantan-1-yl]-
tetrahydro-1H-cyclopropa-[4,5]pyrrolo[1,2-a]pyrazine-3,6-
(1aH,6aH)-dione (4). A suspension of amidine 3 (12.5 g, 40 mmol)
in water (125 mL) was stirred at 85 °C in a flask equipped with a
bubbler to allow the escape of ammonia formed by the reaction. After
heating the mixture for 1 h, additional water (50 mL) was added. The
suspension was heated for 18 h, at which time LC−MS analysis
indicated that the mixture contained approximately equal proportions
of amidine 3 and diketopiperazine 4. The suspension was then cooled
to rt and filtered. The filtrate containing 4 was evaporated, and the
residue was azeotroped with i-PrOH (1 L). The residue was dissolved
in MeOH (300 mL) and decolorized with 3 g of charcoal. The
resulting suspension was filtered through Celite, and the solvent was
evaporated at reduced pressure to give an off-white solid that was then
dissolved in i-PrOH (350 mL). The solvent was switched at normal
pressure from i-PrOH to EtOAc, and a total of 2 L of distillate was
collected. Diketopiperazine 4 precipitated during the distillation, which
was stopped when the final volume of the suspension reached ca. 300
mL. The mixture was cooled to rt, and the solid was filtered and
washed with EtOAc (50 mL). The collected white solid was dried
under vacuum (2 mmHg, 60 °C) to constant weight for 1 h to afford
ACKNOWLEDGMENTS
■
The authors would like to thank Venkatramana Rao for
contributing the deuterium exchange work referenced here and
for useful discussions during this research. The authors would
also like to thank Yande Wang for the isolation and
characterization of amidine 2 during drug product stability
studies.
REFERENCES
■
(1) FDA News Release: FDA Approves New Drug Treatment for
Type 2 Diabetes. http://www.fda.gov/NewsEvents/Newsroom/
PressAnnouncements/ucm174780.htm (July 31, 2009).
(2) Havale, S. H.; Manojit, P. Bioorg. Med. Chem. 2009, 17, 1783.
(3) Magnin, D. R.; Robl, J. A.; Sulsky, R. B.; Augeri, D. J.; Huang, Y.;
Simpkins, L. M.; Taunk, P. C.; Betebenner, D. A.; Robertson, J. G.;
Abboa-Offei, B. E.; Wang, A.; Cap, M.; Xin, L; Tao, L.; Sitkoff, D. F.;
Malley, M. F.; Gougoutas, J. Z.; Khanna, A.; Huang, Q.; Han, S.-P.;
Parker, R. A.; Hamann, L. G. J. Med. Chem. 2004, 47, 2587.
(4) Fura, A.; Khanna, A.; Vyas, V.; Koplowitz, B.; Chang, S.-Y.;
Caporuscio, C.; Boulton, D. W.; Christopher, L. J.; Chadwick, K. D.;
Hamann, L. G.; Humphreys, W. G.; Kirby, M. Drug Metab. Dispos.
2009, 37, 1164.
(5) Savage, S. A.; Jones, G. S.; Kolotuchin, S.; Ramrattan, S. A.; Vu,
T.; Waltermire, R. E. Org. Process Res. Dev. 2009, 13, 1169.
(6) Amidine 2 was observed during solid state stability studies of (1)
amorphous saxagliptin (1) as well as (2) saxagliptin tablets.
(7) A 2:1 solution of saxagliptin (1) and its epimer was prepared in
isopropanol and subsequently monitored for conversion to amidine 3.
After approximately 30% conversion to amidine 3, saxagliptin (1) and
its epimer remained at a 2:1 ratio, and both had converted exclusively
to amidine 3 at the same rate. This observation, together with the fact
that none of amidine 2 is observed when saxagliptin (1) cyclizes in
1
the desired diketopiperazine 4 (6.2 g, 50%): mp 260−261 °C; H
NMR (300 MHz, MeOH-d4) δ 0.79 (m, 1H), 0.84 (dd, J = 9.0, 6.0 Hz,
1H), 1.45−1.72 (m, 11H), 2.16 (m, 1H), 2.21 (br s, 2H), 2.42 (dd, J =
13.5, 7.5 Hz, 1H), 3.48 (br s, 1H), 4.01 (m, 2H); 13C NMR (75 MHz,
CDCl3) δ 8.4, 12.6, 31.6, 36.1, 38.6, 44.2, 45.0, 47.0, 55.6, 67.2, 69.0,
165.4, 171.7; HRMS calculated for C18H25N2O3 [M + H+] 317.1865,
found 317.1865.
10336
dx.doi.org/10.1021/jo202052a | J. Org. Chem. 2011, 76, 10332−10337

The Journal of Organic Chemistry
Note
solution, refutes the possibility of slow epimerization of saxagliptin (1)
followed by rapid cyclization.
(8) Marcus, Y. The Properties of Solvents; John Wiley & Sons:
Chichester, U. K., 1998; Wiley Series in Solution Chemistry Vol. 4.
(9) For a detailed description of the technique and the application of
the Gross-Butler equation, see: Venkatasubban, K. S.; Schowen, R. L.
Crit. Rev. Biochem. 1984, 17, 1.
(10) For simplicity, we assume that the fractionation factors for
exchangeable protons in the reactants, ΦiR, are unity. Considering that
the primary amine 1 and isopropyl alcohol are involved in the reaction,
this assumption simply means that the distribution of deuterium in
these species is equal to the distribution of deuterium in the bulk
solvent, isopropyl alcohol. For alcohols and amines, the assumption
that ΦiR ∼ 1 is generally considered valid as these protons are readily
exchangeable.11 With this assumption, the denominator of eq 1 is
unity.
(11) Schowen, K. B. J. In Transition States of Biochemical Processes;
Gandour, R. D., Schowen, R. L., Eds.; Plenum Press: New York, 1978.
(12) The calculations incorporate thermal corrections to Gibbs free
energies, polarized continuum model (PCM) adjustments for i-PrOH
as the bulk solvent, and corrections to basis set superposition errors
(BSSE) using the counterpoise method as implemented in
Gaussian’09. See the Supporting Information for details and
references.
(13) For a discussion on the limitations of replacing discrete
solvation by a continuum in H-bonded systems, see: Continuum
Solvation Models in Chemical Physics: From Theory to Applications;
Tomasi, J., Mennucci, B., Cammi, R, Eds.; Wiley: Chichester, U. K.,
2007.
(14) During the development of the saxagliptin manufacturing
process, it was observed that the rate of cyclization increases
monotonically with pH.
(15) Swain, C. G.; Brown, J. F. Jr. J. Am. Chem. Soc. 1952, 74, 2538.
(16) Field, M. J.; Hillier, I. H. J. Chem. Soc., Perkin Trans. 2 1987, 617.
(17) (a) Chen, Y.; Gai, F.; Petrich, J. W. J. Am. Chem. Soc. 1993, 115,
10158. (b) Boyle, P. H.; Convery, M. A.; Davis, A. P.; Hosken, G. D.;
Murray, B. A. J. Chem. Soc. Chem. Commun. 1992, 239.
(18) Acidity values in aqueous solution: Bordwell, F. G. Acc. Chem.
Res. 1988, 21, 456.
(19) Arrhenius Ea and Eyring ΔG‡ values were calculated from
298
plots of ln(k) and ln(k/T) as a function of 1/T, respectively.
(20) Within the range of isopropanol concentrations germane to the
saxagliptin process, the rate is first order in isopropanol. For broader
ranges, the rate may demonstrate saturation kinetics.
10337
dx.doi.org/10.1021/jo202052a | J. Org. Chem. 2011, 76, 10332−10337