Axial-selective H/D exchange of glycine-derived 1 H -benzo[ e ][1,4]diazepin-2(3 H)-ones: Kinetic and computational studies of enantiomerization

Article abstract of DOI:10.1021/jo1010608

Glycine-derived 1H-benzo[e][1,4]diazepin-2(3H)-ones (BZDs) 5d-g featuring C9- and N1- substitution exhibit enantiomerization barriers too high to be measured by ¹H NMR coalescence experiments. To address this problem, we found that room-temperature H/D exchange of these compounds is remarkably selective, affording only the axial-d₁ isotopomers. ¹H NMR spectroscopy was then employed to measure the rate of conformational inversion of these d₁-compounds at elevated temperatures. These studies reveal the highest enantiomerization barriers (up to 28 kcal/mol) ever determined for a BZD. Density functional theory calculations match the experimental enantiomerization barriers within 1.2 kcal/mol.

Full text of DOI:10.1021/jo1010608

pubs.acs.org/joc
Axial-Selective H/D Exchange of Glycine-Derived
1H-Benzo[e][1,4]diazepin-2(3H)-ones: Kinetic and
Computational Studies of Enantiomerization
Paul R. Carlier,* Yang-Sheng Sun, Danny C. Hsu, and Qiao-Hong Chen
Department of Chemistry, Virginia Tech, Blacksburg, Virginia 24061
pcarlier@vt.edu
Received July 11, 2010
Glycine-derived 1H-benzo[e][1,4]diazepin-2(3H)-ones (BZDs) 5d-g featuring C9- and N1- substitution
exhibit enantiomerization barriers too high to be measured by H NMR coalescence experiments.
To address this problem, we found that room-temperature H/D exchange of these compounds is
1
1
remarkably selective, affording only the axial-d1 isotopomers. H NMR spectroscopy was then
employed to measure the rate of conformational inversion of these d1-compounds at elevated
temperatures. These studies reveal the highest enantiomerization barriers (up to 28 kcal/mol) ever
determined for a BZD. Density functional theory calculations match the experimental enantiomeriza-
tion barriers within 1.2 kcal/mol.
Introduction
near perfect stereoselectivity, we describe these reactions
as examples of “self-regeneration of stereocenters via
stereolabile axially chiral intermediates”.8-12 To establish
the stereochemical course of these reactions, we demon-
strated by X-ray crystallography that deprotonation of
(S)-1d with KHMDS at -78 °C, followed by treatment with
TBS-Cl, gave enantiopure, axially chiral TBS enol (M)-4d
(Scheme 2).7,13
Benzo[e][1,4]diazepin-2(3H)-ones (BZDs) are well-known
for their importance in medicinal chemistry.1-3 We have
shown that enantiopure BZDs 1b-d undergo enantioselec-
tive deprotonation/alkylation reactions (Scheme 1).4-7 The
successful retentive benzylations of (S)-1b-d depicted in
Scheme 1 are believed to proceed via enantiopure enolates
that are chiral by virtue of the conformation of the BZD
ring. Since deprotonation trigonalizes the original stereo-
genic center at C3, and enolate alkylation occurs with
(8) Carlier, P. R.; Hsu, D. C.; Bryson, S. A. In Stereochemical Aspects of
Organolithium Compounds; Gawley, R. E., Ed.; Topics in Stereochemistry;
€
Siegel, J. S., Series Ed.; Verlag Helvetica Chimica Acta: Zurich, 2010; Vol. 26,
(1) Evans, B. E.; Rittle, K. E.; Bock, M. G.; DiPardo, R. M.; Freidinger,
R. M.; Whitter, W. L.; Lundell, G. F.; Veber, D. F.; Anderson, P. S.; Chang,
R. S. L.; Lotti, V. J.; Cerino, D. J.; Chen, T. B.; Kling, P. J.; Kunkel, K. A.;
Springer, J. P.; Hirschfield, J. J. Med. Chem. 1988, 31, 2235–2246.
(2) Ellman, J. A. Acc. Chem. Res. 1996, 29, 132–143.
(3) Butini, S.; Gabellieri, E.; Huleatt, P. B.; Campiani, G.; Franceschini,
S.; Brindisi, M.; Ros, S.; Coccone, S. S.; Fiorini, I.; Novellino, E.; Giorgi, G.;
Gemma, S. J. Org. Chem. 2008, 73, 8458–8468.
p 53-92.
(9) Wolf, C. In Dynamic Stereochemistry of Chiral Compounds: Principles
and Applications; Royal Society of Chemistry: Cambridge, UK, 2008; pp
282-289.
(10) Kawabata, T.; Suzuki, H.; Nagae, Y.; Fuji, K. Angew. Chem., Int.
Ed. 2000, 39, 2155–2157.
(11) Kawabata, T.; Moriyama, K.; Kawakami, S.; Tsubaki, K. J. Am.
Chem. Soc. 2008, 130, 4153–4157.
(4) Carlier, P. R.; Zhao, H.; DeGuzman, J.; Lam, P. C.-H. J. Am. Chem.
Soc. 2003, 125, 11482–11483.
(5) Carlier, P. R.; Lam, P. C.-H.; DeGuzman, J.; Zhao, H. Tetrahedron:
Asymmetry 2005, 16, 2998–3002.
(6) Carlier, P. R.; Zhao, H.; MacQuarrie-Hunter, S. L.; DeGuzman, J. C.;
Hsu, D. C. J. Am. Chem. Soc. 2006, 128, 15215–15220.
(7) Hsu, D. C.; Lam, P. C.-H.; Slebodnick, C.; Carlier, P. R. J. Am. Chem.
Soc. 2009, 131, 18168–18176.
(12) Note that we no longer favor the term “memory of chirality,” in part
because it is the configuration, not the “chirality” of the original stereogenic
center that is “memorized.” We believe that the term “self-regeneration of
stereocenters via stereolabile axially chiral intermediates” is more accurate
and descriptive. For additional justification of this term, see refs 8 and 9.
(13) The helical descriptors (M) and (P) are used to assign the sense of
chirality of the ring, according to the sign of the C2N1C10C11 dihedral angle
(M = minus, P = positive).
6588 J. Org. Chem. 2010, 75, 6588–6594
Published on Web 09/09/2010
DOI: 10.1021/jo1010608
r
2010 American Chemical Society

Carlier et al.
JOCArticle
SCHEME 1. Enantioselective Deprotonation/Benzylation
Reactions of (S)-Alanine-Derived BZDs (S)-1b-da
SCHEME 3. Enantiomerization and Synthesis of
Glycine-Derived BZDs
aNote that due to a Cahn-Ingold-Prelog priority switch, the config-
uration descriptors at C3 change following these retentive transforma-
tions.
SCHEME 2. Stereochemical Course of Retentive
Deprotonation/Benzylation of (S)-1d
importance of steric interactions between the R1 and R2
group to impart a low racemization rate to the BZD
enolates.4,14 Consistent with this proposal, the potassium
enolate derived from 1d was experimentally determined to
have an enantiomerization barrier of 19.9 ( 0.1 kcal/mol
(257 K),7 whereas the enantiomerization barrier of the
enolate of 1a is estimated at less than 12 kcal/mol.4,5 Thus
replacement of the C9-H in 1a with methyl increased
the enolate enantiomerization barrier by at least 8 kcal/mol.
Consequently, deprotonation/benzylation of (S)-1d oc-
curs in greater than 99:1 er. In this paper, we describe
experimental and computational studies of the effect of C9
substitution on the enantiomerization barriers of glycine-
derived BZDs. We furthermore show that slow ring inver-
sion in such compounds allows a remarkable axial-selective
H/D exchange at C3.
Results and Discussion
Glycine-derived BZDs 5 are axially chiral and, under
conditions of slow ring inversion, exhibit diastereotopic pro-
tons at C3 (Scheme 3). Since enantiomerization (ring
inversion) of BZDs 5 exchanges the axial and equatorial
orientation of the C3 protons (Scheme 3), 1H NMR coales-
cence experiments can in principle be used to determine the
barrier to this process. Previous application of this technique
to 5a and 5b indicated free energies of enantiomerization
(ΔGqenant) of 18.0 ( 0.2 (391 K) and 21.3 ( 0.2 kcal/mol (437
K), respectively.14 BZDs 5d-g were then prepared from the
known N-H analogues 5h and 5i,15 as described in Scheme 3.
Variable-temperature NMR studies of 5d and 5f were per-
formed in DMSO-d6. However, up to the maximum tem-
perature of the probe (453 K), coalescence of the
diastereotopic C3 protons was not observed, nor was sig-
nificant line broadening. Thus, enantiomerization of 5d and
We therefore deduce that the enolate intermediate 3d is
similarly axially chiral, and (M)-configured. Furthermore,
1H NMR spectroscopy of the benzylation product immedi-
ately following cold work up demonstrated that 2d is formed
exclusively in the Bn-axial conformation, indicating concave
face alkylation of (M)-3d (Scheme 2).7 Note that on standing
at room temperature, 2d equilibrates to a mixture of Bn-axial
and Bn-equatorial conformers. In the context of these re-
sults, the racemizing deprotonation/benzylation of (S)-1a at
-78 °C (Scheme 1) suggests that the enolate derived from 1a
completely racemizes before addition of benzyl bromide. The
disparate outcomes from (S)-1a and 1d thus point to the
(15) Sternbach, L. H.; Fryer, R. I.; Metlesics, W.; Reeder, E.; Sach, G.;
Saucy, G.; Stempel, A. J. Org. Chem. 1962, 27, 3788–3796.
(14) Lam, P. C.-H.; Carlier, P. R. J. Org. Chem. 2005, 70, 1530–1538.
J. Org. Chem. Vol. 75, No. 19, 2010 6589

Carlier et al.
JOCArticle
SCHEME 4. Room-Temperature Preparation of Axial-d1
Isotopomers of 5d-g and Their Equilibration to the
Equatorial-d1 Conformers at Elevated Temperaturesa
aAll compounds are racemic: (()-5d-g are arbitrarily depicted in
the (M)-conformation; (()-eq-d1-5d-g is consequently drawn in the
(P)-conformation.
5g is much slower than that of 5a, signifying that replacement
of the C9-H in 5a with Me (5d) or Cl (5f) significantly
increased the barrier to inversion.
As a possible way to determine the barrier to inversion in
5d-g we realized that the axial-d1 isotopomers, if they could
be prepared, would be useful for 1H NMR kinetic studies. On
the basis of the aforementioned studies of the stereochemis-
try of deprotonation and alkylation of (S)-1d (Scheme 2),7
such a transformation seemed possible. As shown in Scheme
4, only the axial C3 proton of 5d-g possesses the proper
alignment with the C2-O π* orbital to allow deprotonation.
Furthermore, based on our previous studies of the stereo-
chemistry of deprotonation/alkylation of (S)-1d,7 we ex-
pected that deuteration of enolates 6d-g should occur on
the concave face only, giving ax-d1-5d-g initially. Finally, if
conformational inversion of ax-d1-5d-g to eq-d1-5d-g is
slow under the deprotonation conditions, then selective
monodeuteration would occur.
FIGURE 1. Room-temperature 1H NMR spectra (400 MHz, CDCl3)
of (A) 5d; (B) ax-d1-5d, immediately following H/D exchange; (C)
an equilibrium mixture of ax- and eq-d1-5d, resulting after heating
ax-d1-5d for 4 h at 328 K.
As hoped, simple exposure of 5d-g to KO-t-Bu in CD3OD
gave very selective axial monodeuteration in near quantita-
tive weight recovery (96-98%). We have previously used a
similar procedure to retentively deuterate (S)-1a;5 the inno-
vation here is that axial deuteration of a racemic glycine-
derived BZD is achieved, and because of slow ring inversion
at room temperature, the pure ax-d1 isotopomers can be
isolated. As shown in Figure 1A, 5d has two diastereotopic
protons at C3 that couple each other: axial at δ 3.75 ppm, and
equatorial at δ 4.72 ppm. Upon treatment with KO-t-Bu in
CD3OD, the signal for the axial proton diasappears, and the
signal for the equatorial proton becomes an apparent singlet,
consistent with selective H/D exchange of the axial proton in
5d (Figure 1B). After 4 h at 328 K, an equilibrium mixture of
ax-d1- and eq-d1-5d is formed, as evidenced by apparent
singlets of equal integration at δ 3.75 and δ 4.72 ppm
(Figure 1C). The appearance of these signals as singlets,
rather than as 1:1:1 triplets, reflects inability to resolve the
FIGURE 2. First-order kinetic analysis of equilibration of ax-d1-5d
to the mixture of ax- and eq-d1 conformers at 328 K (CDCl3).
X represents mole fraction.
6590 J. Org. Chem. Vol. 75, No. 19, 2010

Carlier et al.
JOCArticle
TABLE 1. Measured Rates for Equilibration of ax-d1-Isotopomers of 5d-g, Experimentally Derived Enantiomerization Barriers for 5d-g, and
Computed (B3LYP/6-31G*) Enantiomerization Barriers for 5d-g
B3LYP/6-31G* (PCM)
ΔGqenant(T)c (kcal/mol)
BZD
T (K)
solvent
104 ꢀ kequila (s-1
)
exptl ΔGqenant(T)b (kcal/mol)
5d
5e
5f
328
371
328
373
CDCl3
1.33 ( 0.03
1.98 ( 0.09
0.99 ( 0.01
3.45 ( 0.12
25.1 ( 0.2
28.2 ( 0.2
25.3 ( 0.2
27.9 ( 0.2
26.3
28.2
25.1
27.5
d6-DMSO
CDCl3
d6-DMSO
5g
aFirst-order equilibration rate constants of ax-d1-5d-g, equal to k1 þ k-1 (Scheme 4). bFor protio-5d-g, based on the assumption that kenant(5) =
k1(ax-d1-5) = k-1(eq-d1-5). Error estimate based on indicated uncertainties in rate and (3 °C uncertainty in temperature. cCalculated from B3LYP/
6-31G* (PCM) electronic energies in the indicated (protio) solvent of the B3LYP/6-31G* equilibrium geometries 5d-g and enantiomerization transition
structures 5d*-g*, incorporating free energy corrections derived from the vacuum frequencies (scaled by 0.9826) at the indicated temperature.
small two bond H-D coupling constant (expected to be
1.6 Hz)16 under these conditions. However, as expected, C3 in
ax-d1-5d appears as a 1:1:1 triplet in the 13C NMR spectrum,
due to one-bond 13C-D coupling (1JCD = 20 Hz). Moreover,
exactly analogous results were obtained when we applied the
same procedure to 5e-g (see the Supporting Information for
1H NMR spectra overlays and 13C NMR spectra).
If our analysis is correct, selective monodeuteration of
5a should not be possible under these conditions, since
unlike 5d-g, it has a small R2 substituent (H). Although
H/D exchange of 5a should initially give ax-d1-5a with
high selectivity, rapid conformational inversion to eq-d1-5a
will occur at room temperature (equilibration t1/2 = 0.9 s at
25 °C, based on ΔGqenant(5a, 298 K) = 18.0 kcal/mol).14
Thus, axial deprotonation of eq-d1-5a should ensue, followed
by deuteration, to give d2-5a. As expected, exposure of 5a to
KO-t-Bu in CD3OD at room temperature gave complete
conversion to d2-5a within 20 min. When the experiment was
repeated at 0 °C, 1H NMR monitoring within 5 min showed
a mixture of 5a, ax-d1-5a, and eq-d1-5a.
The distinct 1H NMR signatures of the ax-d1- and eq-d1-
5d-g conformers and their equilibration at elevated tem-
peratures enabled kinetic studies. As depicted in Scheme 4
for ax-d1-5d, the first-order equilibration rate constant
kequil = k1 þ k-1. Equilibration of ax-d1-5d was observed
order reaction was applied:7,17 a plot of the data is shown in
Figure 2. Because ax- and eq-d1-5d are pseudoenantiomeric
FIGURE 3. B3LYP/6-31G* geometries and relative free energies
(371 K) of (M)-5e (equilibrium geometry) and 5e* (enantiomeriza-
tion transition structure). The (M)-enantiomers are depicted arbitrarily.
over 4 h at 328 K, and a standard analysis for reversible first-
isotopomers of 5d, we assume kequil = k1 þ k-1 = 2kenant
,
much higher than previously characterized compounds 5a,
5b (18.0 ( 0.2 and 21.3 ( 0.3 kcal/mol, respectively).14
Comparing 5a and 5f, replacement of the C9-H in 5a with
Cl increased the barrier by 7.3 kcal/mol. Comparing 5b and
5g, replacement of the C9-H in 5b with Cl increased the
barrier by a similar amount (6.6 kcal/mol).
where kenant is the enantiomerization rate constant of 5d.
Therefore, we treat the equilibration of ax-d1-5d as a pseu-
doenantiomerization. By application of the Eyring
equation18 to the calculated kenant, we experimentally deter-
mine ΔGqenant (328 K) of 5d to be 25.1 ( 0.2 kcal/mol (Table
1). Equilibration of the other N-Me BZD ax-d1-5f was also
observed at 328 K, giving an enantiomerization barrier of
25.3 ( 0.2 kcal/mol (Table 1). The equilibration of N-i-Pr
analogues ax-d1-5e and 5g was very slow at 328 K but
progressed rapidly at higher temperatures. Thus equilibra-
tion of these compounds was observed at 371 and 373 K,
giving enantiomerization barriers of 28.2 ( 0.2 and 27.9 (
0.2 kcal/mol, respectively (Table 1). The experimentally
determined enantiomerization barriers of 5d-g are indeed
The nearly identical enantiomerization barriers seen for 5d
and 5f and for 5e and 5g suggest that the C9-CH3 and C9-Cl
substituents provide a similar degree of steric hindrance to
the N1 substituent in the enantiomerization transition struc-
tures. To the best of our knowledge, 5e and 5g have the
highest enantiomerization barriers ever determined for a
BZD. Gilman and co-workers demonstrated that BZD 5j
had an enantiomerization barrier ΔGq
> 24 kcal/mol
enant
1
(variable-temperature H NMR spectroscopy; coalescence
not achieved at 473 K);19 this barrier was high enough to
allow preparative resolution. However, the author’s obser-
vation of complete racemization of (M)-(-)-5j in solution
(16) Given that the gyromagnetic ratio of D is 6.5-fold less that of 1H, the
two-bond coupling constant of the H and D at C3 in ax-d1-5d should be 6.5-
fold less than the two-bond coupling constant of the axial and equatorial
protons in 5d (10.5 Hz).
(17) Connors, K. A. In Chemical Kinetics; VCH Publishers: New York,
1990; pp 60-61.
(18) Eyring, H. Chem. Rev. 1935, 17, 65–77.
(19) Gilman, N. W.; Rosen, P.; Earley, J. V.; Cook, C.; Todaro, L. J.
J. Am. Chem. Soc. 1990, 112, 3969–3978.
J. Org. Chem. Vol. 75, No. 19, 2010 6591

Carlier et al.
JOCArticle
TABLE 2. B3LYP/6-31G* ΔEqenant Values for 5a,b, d-g, and
C12N1C10C9 Dihedral Angles for the Corresponding Equilibrium
Geometries and Enantiomerization Transition Structures
dihedral
dihedral
C12N1C10C9
enant TS
C12N1C10C9
eq geometry
5a-db (deg)
a
ΔEq
(kcal/mol)
enant
BZD
R1
R2
5a-d*c (deg)
5a
5b
5f
5d
5g
5e
Me
i-Pr
Me
Me
i-Pr
i-Pr
H
H
Cl
Me
Cl
Me
17.3
20.0
23.9
24.2
26.5
26.9
33.6
43.5
53.6
54.6
63.6
65.6
6.0
18.3
30.5
32.0
43.2
45.3
aEnergies are ZPVE-corrected to calculate ΔEqenant. bAbsolute values
of dihedral angles for 5a,b,d-g: note the dihedral angles are <0 for the
(M)-enantiomers. cAbsolute values of dihedral angles for 5a*,b*,d*-g*:
note the dihedral angles are <0 for the (M)-enantiomers.
FIGURE 4. Selected cyclic ketones examined in the literature for
axial-selective deprotonation.
within 3 days at room temperature also suggests a barrier less
than 24.5 kcal/mol.20
equilibrium geometries and enantiomerization transition
structures of 5a,b,d-g in detail: an extensive tabulation of
bond angles and dihedral angles can be found in the Sup-
porting Information. From these studies, we noted that the
absolute value of the C12N1C10C9 dihedral (see Figure 2),
which reflects torsional interactions between the N1 substit-
We have previously been successful in matching experi-
mental enantiomerization barriers of benzodiazepines14 and
benzodiazepine enolates7 with calculated values of ΔGq
enant
derived from density functional calculations. We therefore
located the minimum energy equilibrium geometries of 5d-g
and the corresponding enantiomerization transition struc-
tures 5d*-g* at B3LYP/6-31G*.21 Representative structures
(M)-5e and (M)-5e* are depicted in Figure 3.
uent and C9, is correlated well to the calculated ΔEq
enant
values22 (Table 2). Specifically, as ΔEq
increases from
enant
17.3 kcal/mol (5a) to 26.9 kcal/mol (5e), there is a steady
increase in the C12N1C10C9 dihedral angle, both in the
equilibrium geometries and in the enantiomerization transi-
tion structures. Comparison of 5a and 5b, 5f and 5g, 5d and
5e shows that as the size of the R1 substituent increases, so
does the C12N1C10C9 dihedral (both equilibrium geome-
try and enantiomerization transition structure) and the
As noted in our previous studies, BZD equilibrium geo-
metries adopt a boat conformation. In contrast, BZD en-
antiomerization transition structures are considerably but
not completely flattened and, therefore, exist as a pair of
(M)- and (P)-enantiomers.14 To calculate ΔGq
from
enant
these structures, we applied free energy and polarized con-
tinuum model (PCM) solvent corrections at the experimen-
tal temperature and solvent (CHCl3 or DMSO). As can be
seen in Table 1, excellent agreement is seen between experi-
ment and theory for 5d-g, with the largest deviation being
1.2 kcal/mol. Since the equilibrium geometries and enantio-
merization transition structures of 5a and 5b had not been
calculated previously, we also located these structures, and
again applied the appropriate free energy and PCM correc-
tions. The enantiomerization free energies obtained for 5a and
5b (17.5 and 21.5 kcal/mol) also match experimental values
very well (18.0 ( 0.2 and 21.3 ( 0.3 kcal/mol, respectively).14
To gain further insight into the role that the R1 and R2
substituents play in determining the experimental free energy
of activation ΔGqenant, we inspected the B3LYP/6-31G*
ΔEq
value. A similar trend is seen for increasing the
enant
size of the R2 substituent (cf. 5a vs 5f,d; 5b vs 5g,e). These
trends in dihedral angles suggest the need to relieve steric
interactions between R1 and R2 as these substituents increase
in size. Since in every case the dihedral angle decreases on
proceeding from an equilibrium geometry to the correspond-
ing enantiomerization transition structure (e.g., 33.6° in 5a to
6.0° in 5a*), steric interaction between the R1 and R2 substit-
uents must increase during enantiomerization. Thus the en-
ergetic consequences of increasing the size of the R1 and/or R2
substituents are greater for the enantiomerization transition
structures than for the equilibrium geometries, leading to the
9.6 kcal/mol increase in ΔEqenant seen from 5a to 5e.
In closing this section, we wish to provide further context
for the very high axial selectivity seen in both the deproto-
nation of 5d-g and in the deuteration of the correspond-
ing enolates. The term “stereoelectronic control”, as first
introduced by Corey and Sneen in 1956,23 concerned deproto-
nation of carbonyl compounds, and protonation of the corre-
sponding enols or enolates. As a class, cyclohexanone
derivatives might be expected to show high deprotonation
stereoselectivity, due to chair conformational preference
and consequent differentiation of axial and equatorial R-
protons relative to the CdO π* orbital. However to date,
axial selectivities for deprotonation of such compounds
(20) Calculation of the krac for 5j from ΔGqenant = 24.5 kcal/mol (298 K)
and assumption of first-order racemization would predict <3% ee after 72 h.
(21) Gaussian 03, Revision B.05: Frisch, M. J.; Trucks, G. W.; Schlegel,
H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Montgomery, J. A.,
Jr.; Vreven, T.; Kudin, K. N.; Burant, J. C.; Millam, J. M.; Iyengar, S. S.;
Tomasi, J.; Barone, V.; Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.;
Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda,
R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.;
Klene, M.; Li, X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Adamo, C.;
Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.;
Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.;
Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Zakrzewski, V. G.; Dapprich,
S.; Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.;
Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.;
Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz,
P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.; M. A. Al-Laham,
Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M. W.; Johnson,
B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A. Gaussian, Inc.,
Pittsburgh, PA, 2003.
(22) We chose to compare ΔEqenant values, to avoid any possible bias due
to differences in calculated TΔSq terms. However, the same trend is seen in
calculated ΔGqenant(T) values, with and without PCM corrections.
(23) Corey, E. J.; Sneen, R. A. J. Am. Chem. Soc. 1956, 78, 6269–6278.
6592 J. Org. Chem. Vol. 75, No. 19, 2010

Carlier et al.
JOCArticle
(and protonation of their enolates) have been very low.
Corey reported selectivities less than 2-fold for axial vs
equatorial deprotonation of 3-β-acetoxycholestan-7-one 6
at C6; selectivities for protonation of the corresponding
enols were also less than 2-fold (Figure 4).23 Simple 4-tert-
butylcyclohexanone 7 is reported to show only 5.5-fold
selectivity for deprotonation of the axial proton relative to
the equatorial proton,24 and a number of other cyclohex-
anone derivatives show similarly low selectivity.25 One
important counter-example is provided by twistan-4-one 8,
which is reported to show a 290-fold preference for
Experimental Section
General Procedure A: N1-Alkylation of Gly-Derived N-H 1,4-
Benzodiazepin-2-ones. The Gly-derived N-H 1,4-benzodiazepin-
2-one 5h or 5i (0.60 g, 2.2 mmol, 1 equiv) dissolved in anhydrous
THF (15 mL) was added to potassium hydride or sodium
hydride (1.6 equiv) at 0 °C under nitrogen. The resulting
solution was stirred at 0 °C for 30 min. The alkylating agent
(MeI, i-PrOTs, or i-PrOTf28) was added dropwise to the solu-
tion, the reaction mixture was stirred for 6 h at room tempera-
ture, quenched by the addition of satd NH4Cl (aq), and
extracted with CH2Cl2 (3 ꢀ 30 mL). The combined organic
extracts were dried over Na2SO4, filtered, and concentrated. The
crude product was purified by flash column chromatography on
silica gel (2:1 EtOAc/hexanes).
deprotonation of pseudoaxial proton Hψax relative to
26
pseudoequatorial proton Hψeq
.
The authors attribute
the selectivity to better orbital overlap of the C-Hψax bond
with the carbonyl group. Finally, C2-symmetric cyclohep-
tanone 9 was shown to exhibit a 73-fold preference for H/D
exchange of pseudoaxial proton Hψax relative to pseudoe-
quatorial proton Hψax, a result again interpreted in terms of
better orbital overlap during deprotonation.27 It should be
noted that for both 8 and 9, selectivities in the enolate
deuteration step must also be high. On the basis of the
similar structural features of 5d-g and 9, it is tempting to
speculate that slow rotation of the biaryl bond in the enolate
derived from 9 is also critical to achieving high facial
selectivity in the deuteration step.
General Procedure B: Axial-Selective H/D Exchange of Gly-
Derived 1,4-Benzodiazepin-2-ones. The N-alkyl-Gly-derived
BZD 5d-g (20 -30 mg, 1 equiv) was dissolved in CD3OD
(2 mL), and KO-t-Bu (1 equiv) was added; note that dissolution
of 5g required addition of a few drops of CDCl3. After dissolu-
tion, the mixture was transferred to an NMR tube, and the
l
reaction was monitored by H NMR for 2-10 h at room
temperature to ensure complete H/D exchange of the axial
C3-proton. The reaction was quenched by addition of satd
NH4Cl (aq, 0.25 mL), and the CD3OD was removed in vacuo.
The aqueous suspension was extracted with CH2Cl2 (3 ꢀ 5 mL).
The combined organic extracts were dried over anhydrous
Na2SO4, filtered, and concentrated. The monodeuterated prod-
ucts were obtained in 96-98% weight recovery and needed
no further purification. As discussed in this paper and below, the
products isolated in this way featured exclusive axial deutera-
tion. Furthermore, 1H NMR integrals and HRMS analysis were
consistent with <10% d2-byproduct. No attempts were made to
more accurately quantitate this potential contaminant, since it
would not interfere with our 1H NMR kinetic studies. However,
according to our current mechanistic understanding, d2-byproduct
would acrue only from deprotonation/deuteration of eq-d1-5d-g.
Based on our measured ΔGqenant values (Table 1 of manuscript),
we estimate that ax-d1-5d and 5f would equilibrate to a 95:5
mixture of ax-d1 and eq-d1 conformers over 6 h at room tempera-
ture. For N-i-Pr analogues 5e and 5g, we estimate that the ax-d1
conformers would still be >99% populated after 6 h at room
temperature. These calculations are consistent with our estimation
of <10% d2-byproduct in ax-d1-5d-g.
Conclusion
In conclusion, we demonstrated that the C3 H/D exchange
of 5d-g at room temperature is remarkably selective, afford-
ing the ax-d1 isotopomers exclusively. In contrast, when 5a
was subjected to the same conditions, it afforded d2-5a within
20 min at room temperature. The selectivity seen for 5d-g is
attributed to three factors: exclusive axial deprotonation,
exclusive concave-face deuteration, and slow conforma-
tional inversion of ax-d1-5d-g to eq-d1-5d-g under the
deprotonation conditions. In the case of 5a, the presence of
a small R2 substituent (H) allows fast conformational inver-
sion of ax-d1-5a to eq-d1-5a under the deprotonation condi-
tions, resulting in a second deprotonation and deuteration.
The ax-d1-isotopomers of 5d-g were then used to experi-
mentally determine the enantiomerization barriers of glycine-
derived BZDs 5d-g by observing their equilibration to a
mixture ofthe ax- and eq-d1 conformersby1H NMR spectros-
copy at elevated temperatures. These studies revealed
enantiomerization barriers of 25-28 kcal/mol. Density func-
tional theory calculations matched these experimental
barriers to within 1.2 kcal/mol, and demonstrated a cor-
relation between the enantiomerization barrier and the
C12N1C10C9 dihedral angle in both the equilibrium geo-
metries and enantiomerization transition structures. These
studies and the insights gained herein should prove useful for
designing other systems in which to achieve enantioselective
reaction of stereolabile intermediates.
1,7,9-Trimethyl-5-phenyl-1H-benzo[e][1,4]diazepin-2(3H)-one
(5d). The following were used in general procedure A above: 5h
(0.30 g, 1.1 mmol, 1.0 equiv), KH (0.24 g, 1.8 mmol, 1.6 equiv,
30% suspension in mineral oil), methyl iodide (2.1 mL, 33 mmol,
30 equiv). Purification with flash column chromatography on
silica gel (1:2 EtOAc/hexanes) provided 5d (0.29 g, 91%) as pale
l
yellow solid: mp 102-103 °C; H NMR (CDCl3, 500 MHz): δ
7.67 (br d, J = 7.5 Hz, 2H), 7.45 (br t, J = 7.5 Hz, 1H), 7.39 (br
d, J = 7.5 Hz, 2H), 7.20 (br s, 1H), 6.90 (br s, 1 H), 4.72 (d, J=
10.5, 1H), 3.75 (d, J= 10.5 Hz, 1H), 3.18 (s, 3H), 2.36 (s, 3H),
2.28 (s, 3H); 13C NMR (CDCl3, 125 MHz) δ 170.5, 170.2, 140.8,
139.1, 135.3, 134.9, 132.7, 130.3, 129.5, 128.3, 127.8, 56.9, 36.6,
20.9, 19.8; HRMS (EI) calcd for C18H18N2O (Mþ) 278.141913,
found 278.143077 (-4.2 ppm, -1.2 mDa).
Axial 3-Deuterio-1,7,9-trimethyl-5-phenyl-1H-benzo[e][1,4]-
diazepin-2(3H)-one (ax-d1-5d). General procedure B was applied
twice to 5d on a 20-30 mg scale. H/D exchange was complete
within 2 h, and the desired axially deuterated product ax-d1-5d
was obtained in an average 97% weight recovery: 1H NMR
(CDCl3, 500 MHz): δ 7.67 (dt, J = 7.2, 1.6 Hz, 2H), 7.45 (tt, J =
7.2, 2.4 Hz, 1H), 7.39 (tt, J = 7.2, 1.6 Hz, 2H), 7.20 (br s, 1H),
6.90 (br s, 1H), 4.69 (s, 1H), 3.18 (s, 3H), 2.36 (s, 3H), 2.28 (s,
3H); 13C NMR (CDCl3, 125 MHz) δ 170.5, 170.2, 140.8, 139.1,
135.3, 134.8, 132.8, 130.34, 130.32, 129.5, 128.3, 127.8, 56.6 (t,
(24) Trimitsis, G. B.; Van Dam, E. M. J. Chem. Soc., Chem. Commun.
1974, 610–611.
(25) Pollack, R. M. Tetrahedron 1989, 45, 4913–4938.
(26) Fraser, R. R.; Champagne, P. J. J. Am. Chem. Soc. 1978, 78, 657–
658.
(27) Fraser, R. R.; Champagne, P. J. Can. J. Chem.-Rev. Can. Chim. 1976,
54, 3809–3811.
(28) Beard, C. D.; Baum, K.; Grakauskas, V. J. Org. Chem. 1973, 38,
3673–3677.
J. Org. Chem. Vol. 75, No. 19, 2010 6593

Carlier et al.
JOCArticle
1JCD= 20.0 Hz), 36.6, 20.8, 19.8. Equilibrium mixture of ax-d1-
5d and eq-d1-5d: 1H NMR (CDCl3, 500 MHz) δ 7.66 (dt, J = 7.0,
1.6 Hz, 2H), 7.44 (tt, J = 7.2, 2.4 Hz, 1H), 7.39 (tt, J = 7.0, 1.6
Hz, 2H), 7.19 (br s, 1H), 6.90 (br s, 1H), 4.69 (s, 0.5 H,
equatorial), 3.73 (s, 0.5 H, axial), 3.10 (s, 3H), 2.36 (s, 3H),
2.28 (s, 3H); HRMS(EI) calcd for C18H17N20OD (Mþ) 279.1482,
found 279.1493 (-4.09 ppm, -1.2 mDa).
δ 7.59 (dt, J = 7.2, 1.7 Hz, 2H), 7.51 (dd, J = 7.2, 2.5 Hz, 1H),
7.47 (dt, J = 7.6, 2.5 Hz, 1H), 7.41 (tt, J = 7.2, 1.6 Hz, 2H), 7.29
(d, J = 7.2 Hz, 1H), 7.28 (d, J = 2.5 Hz, 1H), 3.38 (s, 3H); 13
C
NMR (CDCl3, 125 MHz): δ 170.0, 169.0, 142.7, 138.3, 131.6,
130.8, 130.2, 130.0, 129.6, 129.4, 128.5, 122.6, 56.4 (quintet,
1JCD = 22.1 Hz), 34.9; HRMS(EI) calcd for C16H12ClN20OD2
(M þ H) þ 287.0915, found 287.0941 (þ9.29 ppm, þ2.6 mDa).
Kinetic Studies. Samples of ax-d1-5d-g were dissolved in
CDCl3 or DMSO-d6 and heated to 55 °C (5d,f), 98 °C (5e), or
100 °C (5g), and the mole fraction of the ax-d1-conformer (XA)
was determined by 1H NMR integration. Kinetic runs at 55 °C
were performed with the sample continuously in the probe; the
probe temperature was previously calibrated with reference to
the signals of ethylene glycol. Kinetic runs at 98 and 100 °C were
performed by placing the NMR tube in a heated water bath for
the designated time and then removing the tube and placing it in
an ice bath until NMR analysis could be performed. Equilibra-
tions were followed for 4-5 half-lives, and plots of ln[2XA - 1]
vs time gave linear plots with R2 > 0.98, as illustrated in Figure 2
1-Isopropyl-7,9-dimethyl-5-phenyl-1H-benzo[e][1,4]diazepin-
2(3H)-one (5e). The following were used in general procedure A
above: 5 h (0.30 g, 1.1 mmol, 1 equiv), KH (0.24 g, 1.8 mmol, 1.6
equiv, 30% suspension in mineral oil), isopropyl tosylate (0.94 g,
4.4 mmol, 4 equiv). Purification with flash column chromato-
graphy on silica gel (1:2 EtOAc/hexanes) provided 5e (0.15 g,
44%) as pale yellow solid: mp 160-162 °C; 1H NMR (CDCl3,
500 MHz) δ 7.71 (br d, J = 7.2 Hz, 2H), 7.46 (tt, J = 7.2, 1.4 Hz,
1H), 7.41 (br t, J = 7.2 Hz, 2H), 7.21 (br s, 1H), 6.86 (br s, 1H),
4.57 (d, J = 9.8 Hz, 1H), 3.65 (d, J = 9.8 Hz, 1H), 3.46 (hept,
J = 6.8 Hz, 1H), 2.41 (s, 3H), 2.28 (s, 3H), 1.73 (d, J = 6.8 Hz,
3H), 1.13 (d, J = 6.8 Hz, 3H); 1H NMR (DMSO-d6, 500 MHz) δ
7.65 (br d, J = 7.0 Hz, 2H), 7.52 (tt, J = 7.0, 2.4 Hz, 1H), 7.47 (tt,
J = 7.0, 1.6 Hz, 2H), 7.36 (br s, 1H), 6.82 (br s, 1H), 4.33 (d, J =
9.8 Hz, 1H), 3.56 (d, J = 9.8 Hz, 1H), 3.45 (hept, J = 6.8 Hz,
1H), 2.39 (s, 3H), 2.258 (s, 3H), 1.64 (d, J = 6.8 Hz, 3H), 0.98
(d, J = 6.8 Hz, 3H); 13C NMR (CDCl3, 125 MHz) δ 170.1,
169.6, 141.3, 139.0, 135.5, 134.7, 133.9, 131.3, 130.3, 129.4,
128.4, 127.4, 58.8, 56.5, 21.5, 20.8, 19.9, 18.4; HRMS (EI) calcd
for C20H22N2O (Mþ) 306.1732, found 306.174 (-2.45 ppm,
-1.2 mDa).
for 5d. The slope of these plots corresponds to -kequil
=
-2kenant (Scheme 4). The raw data for 5d-g and kinetic plots
for 5e--g are provided in the Supporting Information.
Computational Methods. All density functional theory calcu-
lations were performed using the B3LYP29,30 functional and
6-31G* basis set in Gaussian 03.21 All structures were character-
ized by vibrational frequency analysis as minima (no imaginary
frequencies) or transition structures (one imaginary frequency).
For all enantiomerization transition structures, the vector asso-
ciatedwith the imaginary frequency correspondedto an inversion
of the benzodiazepine ring. Note that these enantiomerization
transition structures are not Cs-symmetric and thus exist as a pair
of (M)- and (P)-enantiomers.14 Note that for N-isopropyl struc-
tures, all three possible conformations around the N1-i-Pr bond
were used as starting geometries to ensure that the minimum
energy equilibrium geometries and transition structures were
located. Zero point energy corrections were calculated from
vibrational frequencies scaled by 0.9826;31 free energy corrections
were calculated using the same scaling factor, at the temperature
of the corresponding kinetic experiment. Implicit solvent correc-
tions were calculated by means of PCM32,33 single-point calcula-
tions in the solvent of the corresponding kinetic experiment: the
default radii (UA0) and surface (SES) of Gaussian 03 were used
for these calculations.
Axial 3-Deuterio-1-isopropyl-7,9-dimethyl-5-phenyl-1H-benzo-
[e][1,4]diazepin-2(3H)-one (ax-d1-5e). General procedure B was
applied twice to 5e on a 20-30 mg scale. H/D exchange required
10 h and gave the desired axially deuterated product in an average
96% weight recovery: 1H NMR (CDCl3, 500 MHz) δ 7.71 (dt,
J = 7.2, 1.6 Hz, 2H), 7.46 (tt, J = 7.2, 1.6 Hz, 1H), 7.41 (tt, J =
7.2, 1.6 Hz, 2H), 7.21 (br s, 1H), 6.87 (br s, 1H), 4.54 (s, 1H), 3.46
(hept, J = 6.8 Hz, 1H), 2.41(s, 3H), 2.27 (s, 3H), 1.73 (d, J = 6.8
Hz, 3H), 1.13 (d, J = 6.8 Hz, 3H); 13C NMR (CDCl3, 125 MHz)
170.1, 169.6, 141.3, 139.0, 135.4, 134.7, 133.9, 131.3, 130.3, 129.4,
128.4, 127.4, 58.6(t, 1JCD = 20Hz), 56.5, 21.5, 20.8, 19.9, 18.4; 1H
NMR (DMSO-d6, 500 MHz) δ 7.66 (br d, J = 6.9 Hz, 2H), 7.53
(br t, J = 6.9 Hz, 1H), 7.48 (br t, J = 6.9 Hz, 2H), 7.35 (br s, 1H),
6.83 (br s, 1H), 4.31 (s, 1H), 3.46 (hept, J = 6.8 Hz, 1H), 2.40 (s,
3H), 2.26 (s, 3H), 1.65 (d, J = 6.8 Hz, 3H), 0.99 (d, J = 6.8 Hz,
1
3H). Equilibrium mixture of ax-d1-5e and eq-d1-5e: H NMR
(DMSO-d6, 500 MHz) δ 7.64 (br d, J = 7.9 Hz, 2H), 7.52 (br t,
J= 7.9 Hz, 1H), 7.47 (br t, J= 7.9 Hz, 2H), 7.35 (br s, 1H), 6.82 (br
s, 1H), 4.31 (s, 0.5 H, equatorial), 3.54 (s, 0.5 H, axial), 3.45 (hept,
J = 6.8 Hz, 1H), 2.39 (s, 3H), 2.25 (s, 3H), 1.64 (d, J = 6.8 Hz, 3H),
0.98 (d, J = 6.8 Hz, 3H); HRMS (EI) calcd for C20H21N2OD (Mþ)
307.1795, found 307.1803 (-2.71 ppm, -1.2 mDa).
Acknowledgment. We thank the National Science Foun-
dation (CHE-0750006) and the Department of Chemistry,
Virginia Tech, for financial support of this work.
Supporting Information Available: Synthetic procedures
and tabulations of analytical data for 5f,g and their 3-deuterio
isotopomers; 1H and 13C NMR spectra for all compounds
disclosed in the paper; kinetic data and plots; energies, Cartesian
coordinates, and selected bond and dihedral angles of calculated
structures. This material is available free of charge via the
Internet at http://pubs.acs.org.
3,3-Dideuterio-7-chloro-1-methyl-5-phenyl-1H-benzo[e][1,4]-
diazepin-2(3H)-one (d2-5a). General procedure B was applied
twice on a 10-15 mg scale to 5a,15 giving d2-5a in an average
97% weight recovery. When the exchange was performed at
room temperature, complete conversion to d2-5a was observed
within 20 min. When the CD3OD was precooled to 0 °C and the
reaction mixture was kept at 0 °C for 5 min, the 1H NMR
spectrum indicated a mixture of starting material, ax-d1-5a,
and eq-d1-5a. The reaction mixture was then warmed to room
(29) Becke, A. J. Chem. Phys. 1993, 98, 5648–5652.
(30) Lee, C. T.; Yang, W. T.; Parr, R. G. Phys. Rev. B 1988, 37, 785–789.
(31) Merrick, J. P.; Moran, D.; Radom, L. J. Phys. Chem. A 2007, 111,
11683–11700.
(32) Miertus, S.; Scrocco, E.; Tomasi, J. Chem. Phys. 1981, 55, 117–129.
(33) Tomasi, J.; Mennucci, B.; Cammi, R. Chem. Rev. 2005, 105, 2999–3093.
l
temperature and monitored by H NMR spectroscopy. It was
found that H/D exchange of the both C3-protons was complete
after 40 min of total reaction time: 1H NMR (CDCl3, 500 MHz)
6594 J. Org. Chem. Vol. 75, No. 19, 2010