Comparison of various density functional methods for distinguishing stereoisomers based on computed1H or13C NMR chemical shifts using diastereomeric penam β-lactams as a test set

Article abstract of DOI:10.1002/mrc.2045

Full ¹H and ¹³C NMR chemical shift assignments were made for two sets of penam β-lactams: namely, the diastereomeric (2S,5S,6S)-, (2S,5R,6R)-, (2S,5S,6R)-, and (2S, 5R, 6S)-methyl 6-(1,3-dioxoisoindolin-2-yl)3,3-dimethyl-7-oxo-4-thia-1-aza-bicyclo[3.2.0] heptane-2-carboxylates (1-4) and (2S,5R,6R)-, (2S,5S,6R)-, and (2S,5R,6S)-6-(1,3-dioxoisoindolin-2-yl)-3,3-dimethyl-7-oxo-4-thia-1-aza- bicyclo[3.2.0]heptane-2carboxylic acids (6-8). Each penam was then modeled as a family of conformera obtained from Monte Carlo searches using the AMBER* force field followed by IEFPCM/B3LYP/6-31G(d) geometry optimization of each conformer using chloroform solvation. ¹H and ¹³C chemical shifts for each conformer were computed at the WF04, WC04, B3LYP, and PBE1 density functional levels as Boltzmann averages of IEFPCM/B3LYP/6-311+G(2d,p) energies over each family. Comparisons between experimental and theoretical chemical shift data were made using the total absolute error (|Δδ|_T) criterion. For the ¹H shift data, all methods were sufficiently accurate to identify the proper stereoisomers. Computed ¹³C shifts were not always successful in identifying the correct stereoisomer, regardless of which DFT method was used. The relative ability of each theoretical approach to discriminate among stereoisomers on the basis of proton shifts was also evaluated. Copyright

Full text of DOI:10.1002/mrc.2045

MAGNETIC RESONANCE IN CHEMISTRY
Magn. Reson. Chem. 2007; 45: 819–829
Published online in Wiley InterScience
(www.interscience.wiley.com) DOI: 10.1002/mrc.2045
Comparison of various density functional methods for
distinguishing stereoisomers based on computed 1H or
13C NMR chemical shifts using diastereomeric penam
b-lactams as a test set
Keith W. Wiitala, Christopher J. Cramer and Thomas R. Hoye∗
Department of Chemistry and Supercomputer Institute, University of Minnesota, 207 Pleasant St. SE, Minneapolis, MN 55455-0431, USA
Received 7 May 2007; Accepted 30 May 2007
Full 1H and 13C NMR chemical shift assignments were made for two sets of penam b-lactams: namely, the
diastereomeric (2S, 5S, 6S)-, (2S, 5R, 6R)-, (2S, 5S, 6R)-, and (2S, 5R, 6S)-methyl 6-(1,3-dioxoisoindolin-2-yl)-
3,3-dimethyl-7-oxo-4-thia-1-aza-bicyclo[3.2.0]heptane-2-carboxylates (1–4) and (2S, 5R, 6R)-, (2S, 5S, 6R)-,
and
(2S, 5R, 6S)-6-(1,3-dioxoisoindolin-2-yl)-3,3-dimethyl-7-oxo-4-thia-1-aza-bicyclo[3.2.0]heptane-2-
carboxylic acids (6–8). Each penam was then modeled as a family of conformers obtained from Monte Carlo
searches using the AMBER∗ force field followed by IEFPCM/B3LYP/6-31G(d) geometry optimization of
each conformer using chloroform solvation. 1H and 13C chemical shifts for each conformer were computed
at the WP04, WC04, B3LYP, and PBE1 density functional levels as Boltzmann averages of IEFPCM/B3LYP/6-
311+G(2d,p) energies over each family. Comparisons between experimental and theoretical chemical shift
1
data were made using the total absolute error (j1djT) criterion. For the H shift data, all methods were
sufficiently accurate to identify the proper stereoisomers. Computed 13C shifts were not always successful
in identifying the correct stereoisomer, regardless of which DFT method was used. The relative ability
of each theoretical approach to discriminate among stereoisomers on the basis of proton shifts was also
evaluated. Copyright  2007 John Wiley & Sons, Ltd.
Supplementary electronic material for this paper is available in Wiley InterScience at http://www.interscience.wiley.com/
jpages/0749-1581/suppmat/
KEYWORDS: NMR; 1H; 13C; computed versus experimental shifts; diastereomers; penams
for the computation of chemical shifts, or the use of linear
regression approaches to correct systematic errors associated
with smaller basis sets and/or inaccurate functionals.32–38
Application of these approaches to stereochemical
problems, including those involving issues of relative
configuration,39 significantly extends their practical value.
In order to distinguish among stereoisomers, it is imper-
ative that correct experimental chemical shifts first be in
hand. Fortunately, the assignment of chemical shifts, even
in quite complex natural structures (e.g., natural products)
has become more routine, given recourse to coupling data,40
increasingly high-field magnets, novel pulse sequences, and
multidimensional NMR experiments.41 It is worth noting that
one need not have 100% of the resonances assigned (although
they happen to be for 1–4 and 6–8); rather, it is only essential
that 100% of the portion of resonances that are assigned (best
when that portion is large) be done so with 100% certainty.
An open question with respect to a computed chemical
shifts is whether a given method is sufficiently accurate to
capture the magnetic effects associated with stereochemical
differences. High accuracy is required for definitive stereo-
chemical assignment, particularly in cases where not all of
INTRODUCTION
The use of ab initio calculations to support the interpre-
tation and assignment of NMR spectra has become very
prevalent in recent years. In the last decade, 1H and 13C
chemical shifts of organic molecules have frequently been
computed using density functional theory (DFT) and other
approaches.1–15 Such calculations can provide a crucial link
between structural and stereochemical details and the corre-
sponding empirical NMR chemical shifts.16–29 For molecules
of moderate to large size, DFT offers the most economical
combination of accuracy and efficiency amongst quantum
chemical models, and a number of functionals for the pre-
diction of chemical shift values have been evaluated.30
Accurate computation of NMR chemical shifts using
DFT requires some attention to technical detail. In general,
modern functionals with large basis sets provide results
of reasonable quantitative accuracy.31 Accuracy can be
improved by the use of DFT methods empirically optimized
ŁCorrespondence to: Thomas R. Hoye, Department of Chemistry,
University of Minnesota, 207 Pleasant St. SE, Minneapolis, MN
55455-0431, USA. E-mail: hoye@chem.umn.edu
Copyright  2007 John Wiley & Sons, Ltd.

820
K. W. Wiitala, C. J. Cramer and T. R. Hoye
the stereoisomers may be available for comparison against
one another. So, when complete or nearly complete sets
of such stereoisomers are available, they can serve as use-
ful benchmarks for evaluation, and this work addresses a
particular case in point.
The penam ˇ-lactams 1 through 4 (esters) and 5 through
8 (acids) shown in Fig. 1 were chosen as substrates to test
the ability of the B3LYP,42–45 PBE1,46,47 WP04,32 and WC0432
density functional models to distinguish each of the four
stereoisomers using experimental and theoretical chemical
shift data. The statistical criteria used were total absolute
error (jυjT), mean unsigned error (MUE), and mismatch to
match ratios of jυjT.
were straightforward to assign. Protons H-2, H-5, and H-6
and their ˛ (down and cis to the carboxyl) or ˇ (up and
trans to the carboxyl) orientations were assigned on the basis
of a combination of HMQC, long-range 1H–1H coupling,
and/or NOE experiments, as detailed below for each case.
Carbon chemical shifts were assigned on the bases of the
HMQC and HMBC analyses. In addition, one-bond 1H–13
C
coupling values were useful for distinguishing C-5 and C-6
(as previously observed for similar penam derivatives52)–all
of the 1JCꢀ6ꢀHꢀ6 values fell in the range of 148–168 Hz while
the 1JCꢀ5ꢀHꢀ5 values were all between 174 and 182 Hz. Two-
and three-bond 1H–13C coupling distinguished the three
types of carbonyl resonances present in each spectrum: (i) the
phthaloyl carbonyl carbons were coupled to H-6 but not H-5
for each compound, (ii) the ester carbonyl resonance (in
1–4) included a quartet from coupling to the methyl group,
(iii) the ester and acid (6–8) carbonyl carbon resonances all
showed a doublet from coupling to H-2 and (iv) the lactam
carbonyl was coupled to H-2, H-5, and H-6. In the following
discussion, the preparation of each compound (1–4, 6–8)
is briefly summarized, and specific, noteworthy spectral
features that support the shift assignments are presented.
The (2S, 5S, 6S)-penam methyl esters 1 and 4 were pre-
pared starting from (2S, 5R, 6R)-6-aminopenicillanic acid (6-
Penam ˇ-lactams contain a challenging variety of func-
tional groups and chemical bonding situations. While some
of 1–8 have been previously prepared and reported in the
literature,48,49 we independently synthesized the full com-
plement of stereoisomers (except for 5, which did not prove
to be stable in our hands) to most meaningfully evaluate
various computational approaches. We used, primarily, the
methods of Kukolja,50 Schofield,49 and Fisher and Trinkle51
to prepare these compounds. The 1H and 13C NMR chemical
shifts for each compound were then fully assigned using
1D, 2D, and NOE NMR experiments, with guidance from
existing NMR analyses in the penam family. The details sup-
porting those assignments are presented next; readers more
interested in the computational aspects are directed to the
section ‘Computational Methods’ further below.
APA,
Aldrich).
The
6-APA
was
phthaloylated
(N-carbethoxyphthalimide), esterified (MeI, K2CO3, DMF,
rt), and base epimerized (DBU, CH2Cl2, rt) at C-6 to give
the methyl ester 4 (100% trans isomer). Subsequent epimer-
ization of the thiazolidine ring at C-5 was achieved by use
of the Kukolja procedure50 (Cl2, CCl4 or CH2Cl2, rt; SnCl2,
THF, rt), through which oxidative cleavage of the sulfide
and reductive reclosure gave a mixture (1.8 : 1) of 4 (major)
and 1 (minor). These were separated by silica gel column
chromatography.
NMR CHEMICAL SHIFT ASSIGNMENTS
For all compounds (1–4, 6–8) the aromatic (phthaloyl),
methyl ester, carboxylic acid, and gem-dimethyl protons
The 1H NMR resonances for H-2, H-5, and H-6 in
the spectrum of 1 were assigned using coupling constant
analysis, NOE, and HMQC experiments. The value of the
coupling constant between the H-6 (5.63 ppm) and H-5
(5.29 ppm) protons ( J D 4.2 Hz) was consistent with their
cis orientation on the lactam ring.53 The ˇ-oriented H-2
resonance (4.00 ppm) showed an NOE to the singlet at
1.71 ppm, which was therefore assigned as the protons from
the ˇ-methyl group, and not to that at 1.73 ppm, which was
assigned to the ˛-methyl protons.
In the proton-coupled 13C NMR spectrum of 1 the
phthalimido carbonyl carbon at 166.8 ppm appeared as a
dd ( J D 4, 4 Hz) due to the couplings with H-6 and
a phthalimido ortho-proton; the ester carbonyl carbon at
165.7 ppm displayed coupling to H-2 and the ester methyl
group and appeared as a dq (J D 8, 4 Hz); the lactam carbonyl
carbon (C-7, 165.3) appeared as a ddd due to coupling with
H-2 ( J D 10 Hz), H-5 ( J D 7 Hz), and H-6 ( J D 7 Hz).
Each of the C-5 and C-6 resonances at 63.6 and 59.1 ppm,
respectively, was a dd ( J D 174, 4 Hz and J D 149, 4 Hz,
respectively); these observations are consistent with earlier
observations made for similar penicillin derivatives.52
The (2S, 5R, 6R)-penam methyl ester 2 was prepared by
phthaloylation and esterification of 6-APA, retaining the
original (2S, 5R, 6R)-configuration. The J value of 4.1 Hz
Figure 1. Structures and configurations of the ˇ-lactams
investigated in this study.
Copyright  2007 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2007; 45: 819–829
DOI: 10.1002/mrc

DFT methods for distinguishing stereoisomers via chemical shifts: penam test set 821
and Ugi54 and then Baldwin et al.55 using different methods,
was formed instead.
between the H-6 and H-5 protons at 5.68 and 5.61 ppm,
respectively, confirmed their cis arrangement on the lactam
ring. Again, an NOE experiment involving irradiation of the
H-2 ˇ-methine singlet (at 4.68 ppm) allowed the assignment
of the ˇ- versus the ˛-methyl group resonances at 1.83 and
1.51 ppm, respectively.
In the proton-coupled 13C NMR spectrum of 2 the
phthalimido carbonyl carbon at 166.8 ppm again appeared as
a dd ( J D 3, 3 Hz) due to coupling with the H-6 and the ortho
phthalimido protons. The ester and lactam (C-7) carbonyl
resonances appeared as overlapping multiplets centered at
168.6 ppm. The C-5 resonance at 67.1 appeared as a ddd
( J D 176, 4, 3 Hz) from coupling to H-5, H-6, and H-2,
respectively, while the C-6 resonance at 58.6 ppm was a dd
( J D 150, 3 Hz) from coupling to H-6 and H-5.
The 1H and 13C NMR shift assignments for acids 6
through 8 were made using similar analyses to those detailed
above for esters 1 through 4. Other than the small differences
in the values themselves, the chemical shift trends among
sets of related compounds were maintained. Of course, for
each the resonance for the carboxyl carbon in the proton-
coupled 13C NMR spectrum now appeared just as a doublet
(rather than a dq) because of coupling only to H-2.
COMPUTATIONAL METHODS
Each ˇ-lactam was modeled as a family of conformers
characterized by different dihedral angles about the C-
6–N, the ester C-2–C bond, and the ester C–O bonds. By
varying the torsional angles about the N-1–C-7, C-3–S-4,
and C-5–S-4 bonds, the conformations of the thiazolidine
ring56 were also considered (i.e. the C-3 envelope and the
S-4 envelope forms). Conformer families were obtained
from Monte Carlo conformational searches (50 000 steps;
the degrees of freedom are illustrated in Fig. 2(b)) and
The (2S, 5S, 6R)-penam methyl ester 3 was prepared
by phthaloylation and esterification of 6-APA followed by
Kukolja epimerization at C-5, which gave a mixture (7.4 : 1)
of 3 (major) and 2 (minor). These were separated by silica
gel column chromatography. The J value of 2.3 Hz between
the H-6 and H-5 resonances at 5.56 and 5.44 ppm indicate
that these protons have a trans orientation on the lactam
ring. Irradiation of the H-2 ˇ-methine proton at 3.90 ppm
leads to an NOE enhancement of the C-3 ˇ-methyl resonance
at 1.69 ppm. By default, the C-3 ˛-methyl group was at
1.49 ppm.
geometry optimization with PRCG57,58 (500 steps). The force
field AMBERŁ
(as implemented in Macromodel 6.0) and
59,60
the GB/SA solvation model61 for chloroform were used
throughout the study.
In the proton-coupled 13C NMR spectrum, of 3 the
phthalimido carbonyl carbon at 166.8 ppm again appeared
as a dd ( J D 4 and 4 Hz) due to coupling with the H-6 and the
ortho phthalimido protons. The ester carbonyl at 167.3 ppm
was a dq ( J D 4 and 4 Hz) arising from coupling to H-2 and
the methoxy protons. The lactam (C-7) carbonyl carbon at
165.9 ppm was a ddd (J D 7, 3, and 2 Hz), due to coupling
with H-2, H-5, and H-6, respectively. As with 1 and 2, the
C-5 resonance at 66.6 ppm was a ddd ( J D 177, 4, and 4 Hz)
and C-6 at 60.1 ppm a d ( J D 152 Hz).
All the minimum energy conformer geometries obtained
were then fully optimized at the density functional level of
theory employing the hybrid generalized gradient approxi-
mation (GGA) functional B3LYP42–45 with the 6-31G(d) basis
set.62 Chloroform solvation effects were included via the
integral equation formalism polarized continuum model
(IEFPCM).63–66 Solute cavities were constructed using default
united-atom radii66 (UA0).
For each optimized conformer geometry, 1H and 13C
atomic chemical shielding tensors s were computed30 at the
density functional level using the gauge independent atomic
orbital (GIAO) formalism,67–69 the 6-311 C G(2d,p) basis
set, and chloroform solvation as modeled by IEFPCM70,71
employing solute cavities built from Bondi radii.72 Isotropic
atomic chemical shifts (υ) in units of ppm were computed
as differences between the atomic isotropic shieldings of
In the 1H NMR spectrum of the (2S, 5R, 6S)-penam
methyl ester 4, the H-5 to H-6 J value of 1.8 Hz (at 5.58
and 5.41 ppm, respectively) indicated their trans orientation
on the lactam ring. Irradiation of the H-2 methine proton
produced a NOE enhancement of the C-3 ˇ-methyl singlet at
1.66 ppm. By default the C-3 ˛-methyl group singlet resided
at 1.49 ppm.
In the proton-coupled 13C NMR spectrum the phthalim-
ido carbonyl carbon at 166.6 ppm appeared as a dd ( J D 4,
4 Hz) due to the couplings with the H-6 and the ortho
phthalimido protons. The ester carbonyl carbon at 167.8 ppm
displayed coupling to H-2 and the ester methyl group and
appeared as a dq ( J D 5, 4 Hz). The lactam carbonyl carbon
(C-7, 167.4 ppm) was a ddd ( J D 6, 4, and 4 Hz) due to
the coupling with H-2, H-5, and H-6, respectively. The C-5
resonance (69.2 ppm) was a ddd ( J D 178, 7, 4 Hz) and C-6
(64.5 ppm) a d ( J D 152 Hz).
We then attempted to prepare the entire set of stereoiso-
meric carboxylic acids 5 through 8 from the corresponding
methyl esters 1 through 4 by the ester cleavage methodology
of Fisher and Trinkle (LiI, EtOAc, reflux).51 This was suc-
cessful for 6–8, but failed in the case of 5. Imine 9 (Fig. 2(a)),
the enantiomer of which was previously prepared by Schutz
Figure 2. (a) Imine 9, byproduct formed during attempted
synthesis of acid 5 from ester 1. (b) Degrees of freedom
surveyed by Monte Carlo conformational searches. Arrows
indicate viable rotations, dotted lines torsion sites, and wavy
lines ring-closure sites.
Copyright  2007 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2007; 45: 819–829
DOI: 10.1002/mrc

822
K. W. Wiitala, C. J. Cramer and T. R. Hoye
the solutes and corresponding reference atoms in tetram-
ethylsilane (TMS). Four hybrid GGA functionals were exam-
ined – namely, B3LYP,42–45 PBE1,46,47 WP04,32 and WC04.32
Population-averaged chemical shifts for each family of con-
formers were computed assuming Boltzmann statistics based
on B3LYP/6-311 C g(2d,p) free energies including chloro-
form solvation effects according to20,31
the MUE between computed and experimental shifts (MUE,
ppm)
ꢀ
ꢁ
ꢁ
ꢁ
ꢁ
exp
corr
MUE D
υi ꢀ υi /n
ꢀ4ꢁ
i
where n is the total number of chemical shifts under
consideration; and the ratio of the total absolute error for
all of the stereochemical mismatches to the total absolute
error of the stereochemical matches (Eqn 5).


o








/RT
υieꢀG
ꢀ
i
υ D
ꢀ1ꢁ
ꢀ
o
eꢀG
/RT
j


i
ꢂ
ꢀ
j
miss
match


R D
jυjT
jυjT
ꢀ5ꢁ
where i and j run over conformers, G is the free energy of the
conformer in solution, R is the universal gas constant, and
T is 298 K. The free energy in solution is taken as the sum
of the electronic energy and solvation free energy computed
at the B3LYP level using the 6-311 C G(2d,p) basis set62 and
IEFPCM chloroform solvation free energies.
j
Use of relative configuration of penam methyl
esters to evaluate computed 1H NMR chemical
shifts
The specific numbers of conformers in each family used
for modeling were 12 for 1, 10 for 2, 12 for 3, 12 for 4, 14 for
6, 16 for 7, and 16 for 8. For the penam methyl esters 1–4, 16
protons and 17 carbons were modeled while 13 protons and
16 carbons were modeled for the penams 5–8. The computed
1H shifts of each methyl group were arithmetically averaged
We have previously assessed the influence of stereo-
chemistry on proton and carbon chemical shifts for a
series of monomethylcyclohexanols using small families
of conformers.74 The studies presented here extend that
methodology to larger molecules containing multiple stere-
ogenic centers and a much wider array of functionality
and heteroatoms. Monte Carlo conformational searches fol-
lowed by geometry optimization using the AMBER* force
field is an efficient approach to construct conformer fam-
ilies in the penams. Because of the approximate nature
of the force field and its associated parameters, particu-
larly with respect to potentially challenging combinations
of functional groups like those present in penams 1–8, we
subsequently reoptimized all conformer geometries at the
IEFPCM/B3LYP/6-31G(d) level of theory.
1
owing to their free rotation at 298 K. The H and 13C shifts
of the corresponding ortho, para, and ipso positions of the
phthalimido group were averaged during the Boltzmann
process. A pruned (75, 302) integration grid containing 75
radial shells and 302 angular points per shell (approximately
7000 points for each atom) was used on each atom. Density
functional calculations were carried out using the Gaussian
03 suite of electronic structure programs.73
1
RESULTS AND DISCUSSION
The H NMR chemical shifts were computed using the
B3LYP, PBE1, and WP04 methods described previously.
Various combinations of experimental and theoretically
computed chemical shifts data were evaluated using the
total absolute error (jυjT) criterion. The best correlation
(smallest jυjT value) was used to identify a configurational
match. The correlation data for penams 1–4 is summarized
in Tables 2–4. For example, in Table 2 the experimental shift
data for 1 (1exp) are correlated with the computed shifts
for each of 1–4 (1corr –4corr). The total absolute error is best
(smaller) for the matched pair 1exp versus 1corr and poorer
Linear correction of computed chemical shifts
Linear corrections of computed chemical shifts have been
found to be useful for correcting systematic errors associated
with particular density functionals and thereby improving
the accuracy of their predictions.32–38 In particular, the linear
corrections listed in Table 1, which were optimized over
a diverse training set of small molecules not including
the penams,32 were applied to the 1H and 13C chemical
shift values (υcomp) obtained with each of the functionals
we studied; corrected predictions (υcorr) are determined
according to:
Table 1. Slope (m, unitless) and intercept (b, ppm) values for
linear correction of chemical shifts32 for various density
functionalsa
υcorr D m ð υcomp C b
ꢀ2ꢁ
where m is the slope and b the intercept specific to a particular
functional/basis set combination.
13C
1H
Theory
m
b
m
b
Criteria for evaluating DFT methods
B3LYP
WP04
WC04
PBE1
0.9488
0.9601
1.0032
0.9486
ꢀ2.1134
ꢀ3.0273
ꢀ0.9647
ꢀ1.257
0.9333
0.9587
0.9451
0.9169
0.1203
0.1127
0.1157
0.1895
Methods were evaluated for their ability to make stereochem-
ical distinctions using the following statistical parameters:
the absolute error between computed and experimental
chemical shifts (jυjT, ppm)
ꢀ
ꢁ
ꢁ
ꢁ
ꢁ
a For use with IEFPCM/Theory/6-311 C G(2d,p)//IEFPCM/
exp
corr
jυjT D
υi ꢀ υi
ꢀ3ꢁ
B3LYP/6-31G(d).
i
Copyright  2007 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2007; 45: 819–829
DOI: 10.1002/mrc

DFT methods for distinguishing stereoisomers via chemical shifts: penam test set 823
Table 2. Correlation data for methyl penams 1–4 using
IEFPCM/B3LYP/6-311CG(2d,p)//IEFPCM/B3LYP/6-31G(d) and
experimental 1H NMR chemical shiftsa
1corr
2corr
3corr
4corr
1exp
2exp
3exp
4exp
1.21
2.12
1.89
2.24
2.67
0.86
3.07
1.36
2.20
3.33
0.94
2.54
2.89
1.86
2.66
0.96
a
jυjT values reported in ppm.
Table 3. Correlation data for methyl penams 1–4 using
IEFPCM/PBE1/6-311 C G(2d,p)//IEFPCM/B3LYP/6-31G(d) and
experimental 1H NMR chemical shiftsa
Figure 3. Graphical representation of total absolute error
(jυjT) data for penam esters 1–4 using the indicated
theoretical method.
1corr
2corr
3corr
4corr
1exp
2exp
3exp
4exp
1.21
2.16
2.01
2.38
2.78
0.90
3.18
1.42
1.96
3.30
0.94
2.47
2.64
1.66
2.55
0.83
a
jυjT values reported in ppm.
Table 4. Correlation data for methyl penams 1–4 using
IEFPCM/WP04/6-311 C G(2d,p)//IEFPCM/B3LYP/6-31G(d)
and experimental 1H NMR chemical shiftsa
1corr
2corr
3corr
4corr
1exp
2exp
3exp
4exp
0.95
1.88
1.74
1.98
2.60
0.74
3.07
1.14
1.93
2.83
0.85
2.19
2.80
1.46
2.45
0.56
Figure 4. Graphical representation of R-ratios for penam esters
1–4 using the indicated theoretical method.
a
jυjT values reported in ppm.
(larger) for all other pairs (¾2 times larger for mismatched
pairs vs the correct pair).
Use of the relative configuration of penam
carboxylic acids to evaluate computed 1H NMR
chemical shifts
In general, the data in Tables 2–4 indicate that each of the
B3LYP, PBE1, and WP04 methods (with chloroform solva-
tion effects included) were able to distinguish convincingly
amongst all configurations. By considering the diagonal
sums of the jυjT values (boldface in tables) or the corre-
sponding MUE, WP04 computed chemical shifts are more
accurate than those of B3LYP and PBE1 for penams 1–4.
The jυjT sums for each method were 3.97 ppm for B3LYP,
3.88 ppm for PBE1, and 3.10 ppm for WP04 (¾20% lower than
PBE1). The diagonal jυjT data from Tables 2–4 are also com-
bined and displayed graphically in Fig. 3. For each method,
the MUE of the correct matches was 0.06 ppm for B3LYP,
0.06 ppm for PBE1, and a slightly better 0.05 ppm for WP04.
The relative ability of a theoretical method to distinguish
between configurations can be judged by comparing R-ratios,
the mismatch to match ratios of jυjT. The R-ratios for each
of the penams 1–4 using each of three DFT methods are dis-
played in Fig. 4. A larger value of R indicates a greater ability
to discriminate the matched from the mismatched diastere-
omers. For 1–4 WP04 has, on average, R-ratios that are 14 and
16% larger than those using PBE1 and B3LYP, respectively.
Tables 5–7 list the jυjT values from the various correlations
of experimental 1H shifts obtained for acids 6–8 to those
computed at the B3LYP, PBE1, and WP04 levels for penams
5–8. Again, each method was successfully able to make
stereochemical distinctions between all cases using the jυjT
criterion. However, this time the diagonal sums of the
jυjT values (emboldened matches) and the corresponding
MUEs indicate that PBE1 computed chemical shifts are more
accurate than those of B3LYP and WP04 for penams 6–8.
The jυjT sums for each method was 2.48 ppm for B3LYP,
2.22 ppm for WP04, and 1.84 ppm for PBE1 (¾17% lower
than WP04). The diagonal jυjT data of Tables 5–7 are also
displayed graphically in Fig. 5. For each method, the MUE
of the correct matches was 0.06 ppm for B3LYP, 0.06 ppm for
WP04, and 0.05 ppm for PBE1.
The R-ratios for the proton chemical shifts for penam
acids 6–8, again using each of three DFT methods, appear in
Fig. 6. On average, the PBE1 R-ratios are 16 and 25% larger
(better) than those of WP04 and B3LYP, respectively.
Copyright  2007 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2007; 45: 819–829
DOI: 10.1002/mrc

824
K. W. Wiitala, C. J. Cramer and T. R. Hoye
Table 5. Correlation data for penams 6–8 using
IEFPCM/B3LYP/6-311CG(2d,p)//IEFPCM/B3LYP/6-31G(d) and
experimental 1H NMR chemical shiftsa
5corr
6corr
7corr
8corr
6exp
7exp
8exp
1.80
1.16
1.92
0.63
2.49
1.10
2.86
0.80
2.04
1.64
2.16
1.05
a
jυjT values reported in ppm.
Table 6. Correlation data for penams 6–8 using
IEFPCM/PBE1/6-311CG(2d,p)//IEFPCM/B3LYP/6-31G(d) and
experimental 1H NMR chemical shiftsa
Figure 6. Graphical representation of R-ratios for penam esters
6–8 using the indicated theoretical method.
5corr
6corr
7corr
8corr
6exp
7exp
8exp
1.81
1.25
2.07
0.49
2.32
1.03
2.73
0.64
1.91
1.31
1.92
0.71
while for the acids 6–8, PBE1 outperforms B3LYP and WP04.
Averaging the R-ratios across both sets of penams indicates
that WP04 and PBE1 are best and nearly equivalent.
a
jυjT values reported in ppm.
Use of the relative configuration of penam methyl
esters to evaluate computed 13C NMR chemical
shifts
Table 7. Correlation data for penams 6–8 using
IEFPCM/WP04/6-311CG(2d,p)//IEFPCM/B3LYP/6-31G(d) and
We also examined additional correlations for 1–4 by
comparing the B3LYP, PBE1, and WC04 predictions for
carbon chemical shifts with experiment using the same
statistical criterion (jυjT) described above. The data in
Table 8 indicate that the widely employed B3LYP model
fails to identify the proper isomer in two of four cases:
(jυjT) is minimized for the incorrect match of 2exp with 1corr
experimental 1H NMR chemical shiftsa
5corr
6corr
7corr
8corr
6exp
7exp
8exp
1.62
1.16
1.76
0.56
2.59
1.12
2.52
0.67
1.73
1.57
2.15
0.99
as well as 3exp with 1corr
.
a
jυjT values reported in ppm.
The 13C shift comparisons using the PBE1 and WC04
models are summarized in Tables 9 and 10. Like the B3LYP
model, neither PBE1 nor WC04 successfully distinguishes
1, 2, and 3 from one another. The failure of each of the
methods indicates that assigning stereochemistry using 13C
shifts computed by any of these methods is unwise.
For the B3LYP and PBE1 methods the largest contribu-
tions to the shift error are associated with the C-3 and C-5
atoms (i.e. two of the 17 carbons in each compound): 26.1%
of the error is due to C-3 and 12.5% to C-5. WC04 also had
Figure 5. Graphical representation of total absolute error
(jυjT) data for penam acids 6–8 using the indicated
theoretical method.
Overall evaluation of methods for computing 1H
NMR chemical shifts
The average R-ratios across the sets of penams 1–4, 6–8, and
the combined sets 1–4 and 6–8 are displayed in Fig. 7. For
both the penam esters (1–4) and acids (6–8), the popular
B3LYP method does not perform as well as PBE1 or WP04
in its ability to discriminate among stereoisomers. For the
penam esters 1–4, WP04 outperforms B3LYP and PBE1,
Figure 7. Comparison of average R-ratios for penam esters
1–4 and acids 6–8 using the indicated theoretical method.
Copyright  2007 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2007; 45: 819–829
DOI: 10.1002/mrc

DFT methods for distinguishing stereoisomers via chemical shifts: penam test set 825
Table 8. Correlation data for methyl penams 1–4 using
Table 11. Correlation data for penams 6–8 using
IEFPCM/B3LYP/6-311CG(2d,p)//IEFPCM/B3LYP/6-31G(d) and
IEFPCM/B3LYP/6-311CG(2d,p)//IEFPCM/B3LYP/6-31G(d) and
experimental 13C NMR chemical shiftsa
experimental 13C NMR chemical shiftsa
1corr
2corr
3corr
4corr
5corr
6corr
7corr
8corr
1exp
2exp
3exp
4exp
38.3
35.8
32.1
48.3
51.2
37.2
47.8
45.1
45.9
44.6
35.8
52.7
60.4
45.2
54.8
37.0
6exp
7exp
8exp
46.2
41.7
60.4
37.0
43.1
42.6
43.9
35.6
48.8
47.1
50.2
36.9
a
jυjT values reported in ppm.
a
jυjT values reported in ppm.
Table 12. Correlation data for penams 6–8 using
IEFPCM/PBE1/6-311CG(2d,p)//IEFPCM/B3LYP/6-31G(d) and
Table 9. Correlation data for methyl penams 1–4 using
experimental 13C NMR chemical shiftsa
IEFPCM/PBE1/6-311CG(2d,p)//IEFPCM/B3LYP/6-31G(d) and
experimental 13C NMR chemical shiftsa
5corr
6corr
7corr
8corr
1corr
2corr
3corr
4corr
6exp
7exp
8exp
42.2
37.6
58.1
35.0
40.8
39.1
38.6
30.1
44.8
44.8
46.7
35.8
1exp
2exp
3exp
4exp
32.6
31.8
27.6
46.5
47.8
34.4
44.3
40.6
40.7
41.8
33.0
48.4
55.2
41.7
49.6
34.5
a
jυjT values reported in ppm.
a
jυjT reported in ppm.
Table 13. Correlation data for penams 6–8 using
IEFPCM/WC04/6-311CG(2d,p)//IEFPCM/B3LYP/6-31G(d) and
experimental 13C NMR chemical shiftsa
Table 10. Correlation data for methyl penams 1–4 using
IEFPCM/WC04/6-311CG(2d,p)//IEFPCM/B3LYP/6-31G(d) and
5corr
6corr
7corr
8corr
experimental 13C NMR chemical shiftsa
6exp
7exp
8exp
56.0
55.1
68.2
50.1
61.0
60.5
51.8
52.7
62.6
46.0
58.8
51.1
1corr
2corr
3corr
4corr
1exp
2exp
3exp
4exp
60.8
59.7
60.1
73.1
64.6
57.3
68.6
67.7
60.4
61.7
59.8
75.1
65.3
54.3
67.9
59.3
a
jυjT values reported in ppm.
shift comparison. However, this success is mitigated by
the failure of these same methods to correctly predict the
stereochemistry of penam ester isomers 1–4 using 13C shifts.
This outcome significantly contrasts the use of H chemical
shifts and suggests that the latter are of greater utility for
drawing stereochemical distinctions.
a
jυjT reported in ppm.
1
somewhat larger average errors for these atoms (7.9% on
C-3 and 7.1% on C-5) but exhibited significant error on other
atoms as well, such as C-7 (11.2%), C-2 (14.6%), C-6 (9.1%),
and the ester carbonyl (8.2%). The proximity of C-3 and C-5
to the attached sulfur atom and the consistently too large
(PBE1, B3LYP) or too small (WC04) computed 13C shifts for
these atoms suggest that the modeling may be failing to
account for subtle features in the carbon sulfur bonds, even
though dimethyl sulfide was included in the WC04 training
set.32
Evaluating the use of only the global minimum energy
conformer
In Table 14 are listed the relative conformer energies and per-
centages of the families used to model methyl penams 1–4.
For each, the global minimum conformer comprises 37–68%
of the total conformational population. For the purpose of
possible simplification, we examined the merit of using only
those 1H chemical shifts computed for the global minimum
energy conformer to distinguish among the diastereomers.
Statistical data for the WP04 1H NMR chemical shifts
are provided in Table 15. All stereochemical distinctions
are made successfully using only the data of the global-
minimum conformer. Indeed, the accuracy is unchanged
considering the MUE value for stereochemical matches
(0.05 ppm, Table 15) compared to the MUE obtained using
the full complement of conformers (0.05 ppm, Table 4). The
R-ratio did degrade somewhat going from 8.52 for the full
conformer complement to 8.40.
Use of the relative configuration of penam
carboxylic acids to evaluate computed 13C NMR
chemical shifts
Correlation data for compounds 5–8 are presented in
Tables 11–13. The data in Tables 11 and 12 indicate that the
B3LYP and PBE1 methods are able to distinguish between
all available isomers for this set, although computed errors
continue to be dominated at the C-3 (26.6%) and C-5 (12.6%)
atoms. WC04, on the other hand, fails to distinguish 6exp
from 8corr (Table 13).
Within this set of penam acid isomers (5–8), the B3LYP
methods and PBE1 appeared to perform satisfactorily for the
prediction of relative configuration based on 13C chemical
An appealing idea is that the use of the global minimum
energy conformer may not be necessary, but that any single
Copyright  2007 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2007; 45: 819–829
DOI: 10.1002/mrc

826
K. W. Wiitala, C. J. Cramer and T. R. Hoye
Table 14. Relative energies and percentages of conformers
used to model the penam methyl esters 1–4a
CONCLUSIONS
The approach that provided the most reliable stereochemical
assignment of isomeric penam ˇ-lactam derivatives used
proton chemical shifts and a Boltzmann-averaged family
of conformations whose energies were determined at the
B3LYP/6-311CG(2d,p) level including chloroform solvation.
The less rigorous approach of using only the global minimum
energy geometry also provided correct answers and may be
considered for cases where molecular size dictates recourse
to this approximation. Among the DFT functionals that
were examined for computing the 1H shifts, each of B3LYP,
PBE1, and WP04 was capable of verifying all of the proper
diastereomeric correlations. Further evaluation using the
R-ratio criterion indicated that PBE1 and WP04 performed
comparably to one another and somewhat better than B3LYP
for stereochemical distinction.
1
2
3
4
Conf
G
%
G
%
G
%
G
%
a
b
c
d
e
f
0.0
0.9
1.1
1.7
2.2
68
16
11
4
1
0
0.0
0.2
0.3
0.6
7.6
37
27
21
14
0
0.0
0.4
0.9
1.9
2.5
8.0
55
29
12
2
1
0
0.0
0.3
1.1
1.2
7.8
9.6
55
30
8
8
0
10.9
0
a G values are relative energies (kcal molꢀ1) above the global
minimum (boldface) computed using IEFPCM/B3LYP/6-311C
G(2d,p).
1
By comparing H NMR chemical shifts with those com-
Table 15. Single conformer correlation data for methyl
penams 1–4 between IEFPCM/WP04/6-311CG(2d,p)//
IEFPCM/B3LYP/6-31G(d) and experimental 1H NMR chemical
shiftsa
puted by DFT, it is feasible to deduce the relative configura-
tion of an unknown compound having moderately complex
(both in size and functionality) constitution. Analogous anal-
yses of 13C data proved less successful; for purposes of
1
1corr
2corr
3corr
4corr
stereochemical distinction, the H chemical shifts are more
reliable than those of carbon because the shift differences
between related diastereomers are more likely to be larger
than the computational error.
1exp
2exp
3exp
4exp
1.17
1.79
1.98
2.20
2.87
0.64
3.34
1.14
1.88
2.69
0.80
2.22
3.02
1.47
2.55
0.71
EXPERIMENTAL
a
jυjT values reported in ppm.
For NMR measurements the sample concentration was
approximately 0.7% by weight in CDCl3 for the 1D 1H NMR
spectra and 2.5% for all others. The 13C and 1H NMR spectra
were obtained at ambient temperature with chemical shifts
determined relative to CDCl3 (υ 77.23 ppm) for 13C and TMS
conformer of the family might be sufficient. This approach
will only be useful if the chemical shift differences between
each configuration for all permutations are always greater
than those caused by common conformational changes. It
is unlikely, though, that this will always/often be the case.
For example, Table 16 shows correlation data for 1–4 using
chemical shifts computed with the WP04 protocol and the
conformers c1, d2, e3, and e4 (Table 14). These conformers
produced the worse case scenario of using a single random
conformer representation for stereochemical distinction. As
is seen in Table 16 the computations now fail to correctly
distinguish the relative configurations of 3 and 4 (compare
to Tables 4 and 16). The R-ratio is also reduced considerably
from 8.52 (full complement) to a value of 4.79. The value
of Boltzmann averaging chemical shifts across a family of
conformers is apparent.
1
(υ 0.00 ppm) for H spectra. Proton spectra were recorded
with acquisition times of 2 s and a spectral width of 8000 Hz;
coupling constant values are significant to the nearest
0.25 Hz. A Varian VI-500 MHz NMR instrument or Varian
VXR-300 MHz NMR instrument was used throughout.
High resolution mass spectrometry (HRMS) was per-
formed using a BioTOF II ESI instrument. The source tem-
°
perature was set to 150 C, acceleration voltage was 8500 V,
and nitrogen was used as the carrier gas. A mass range of
100–1000 amu was used for the analysis; the resolution of
the measurement was 10 000 FWHM.
High pressure liquid chromatography (LC) analysis of
each compound reported was performed using an Agilent
1100 Series LC equipped with a 4.6 ð 150 mm Zorbax Eclipse
XDB-C18 ꢀ5 µmꢁ column using a programmed mobile phase
gradient of 10 mM aqueous ammonium acetate and methanol
(5–41% methanol over 10 min, then 41–98% over 6 min,
holding at 98% methanol for 6 min). Dual detection with
Agilent LC/MSD SL (G1978A) and diode array (254 nm,
G1315B) detectors was used. Samples were prepared in
acetonitrile (¾1 mg/ml) and the injection volume was 5 µl.
Table 16. Worse-case single conformer correlation data for
methyl penams 1–4 between
IEFPCM/WP04/6-311CG(2d,p)//IEFPCM/B3LYP/6-31G(d) and
experimental 1H NMR chemical shiftsa
1corr
2corr
3corr
4corr
1exp
2exp
3exp
4exp
1.07
2.53
0.91
1.78
2.36
1.26
2.57
0.87
1.77
2.88
1.75
2.10
2.49
1.78
2.61
1.25
(2S, 5S, 6S)-6-(1,3-dioxoisoindolin-2-yl)-3,3-
dimethyl-7-oxo-4-thia-1-aza-bicyclo[3.2.0]heptane-
2-carboxylic acid methyl ester (1)
1H NMR (CDCl3) υ 1.71 (s, 3H, ˇMe), 1.73 (s, 3H, ˛Me),
3.90 (s, 3H, CO2CH3), 4.00 (d, 1H, J D 1.0 Hz, H2), 5.29
a
jυjT values reported in ppm.
Copyright  2007 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2007; 45: 819–829
DOI: 10.1002/mrc

DFT methods for distinguishing stereoisomers via chemical shifts: penam test set 827
(d, 1H, J D 4.2 Hz, H-5), 5.63 (dd, 1H, J D 4.2, 0.9 Hz, H-
(2S, 5R, 6R)-6-(1,3-dioxoisoindolin-2-yl)-3,3-
dimethyl-7-oxo-4-thia-1-aza-bicyclo
[3.2.0]heptane-2-carboxylic acid (6)
6), 7.77 (m, 2H, Armeta), 7.89 (m, 2H, Arortho); 13C NMR υ
27.4 (qm, J D 129 Hz, ˛Me), 30.4 (qm, J D 128 Hz, ˇMe),
52.8 (q, J D 148 Hz, CO2CH3), 59.1 (dd, J D 149 and 4 Hz,
C-6), 63.6 (dd, J D 174 and 4 Hz, C-5), 65.1 (br m, C-3),
73.1 (dm, J D 143 Hz, C-2), 124.0 (dm, J D 166 Hz, Arortho),
131.7 (m, Aripso), υ 134.7 (dd, J D 165, 8 Hz, Armeta), 165.3
(ddd, J D 10, 7, and 7 Hz, C-7), 165.7 (dq, J D 8 and
4 Hz, CO2CH3), 166.8 (dd, J D 4 and 4 Hz, Ar C O);
HRMS Calcd for C17H16N2NaO5S ꢀM C Naꢁ 383.0672, found
383.0656 (ꢀ4.2 ppm error); LC (ꢂ D 254 nm) 19.503 min.
1H NMR (CDCl3) υ 1.62 (s, 3H, ˛Me), 1.85 (s, 3H, ˇMe),
4.71 (s, 1H, H-2), 5.59 (d, 1H, J D 3.9 Hz, H-5), 5.70 (d, 1H,
J D 4.2 Hz, H-6), 7.78 (m, 2H, Armeta), 7.90 (m, 2H, Arortho);
13C NMR υ 28.2 (qm, J D 129 Hz, ˛Me), 30.5 (qm, J D 129 Hz,
ˇMe), 58.4 (dd, J D 151, 3 Hz, C-6), 65.9 (br m, C-3), 66.8
(ddd, J D 176, 3, 3 Hz, C-5), 71.1 (dm, J D 146 Hz, C-2), 124.1
(dm, J D 166 Hz, Arortho), 131.6 (m, Aripso), 134.8 (dd, J D 164,
7 Hz, Armeta), 166.8 (dd, J D 4, 4 Hz, Ar C O), 169.1 (ddd,
J D 9, 7, 5 Hz, C-7), 172.6 (d, J D 5 Hz, CO2H); HRMS Calcd
C16H13N2O5S for (M ꢀ H) 345.0551, found 345.0544 (ꢀ2 ppm
error); LC (ꢂ D 254 nm) 16.934 min.
(2S, 5R, 6R)-6-(1,3-dioxoisoindolin-2-yl)-3,3-
dimethyl-7-oxo-4-thia-1-aza-bicyclo
[3.2.0]heptane-2-carboxylic acid methyl ester (2)
1H NMR (CDCl3) υ 1.51 (s, 3H, ˛Me), 1.83 (s, 3H, ˇMe), 3.81
(s, 3H, CO2CH3), 4.68 (s, 1H, H-2), 5.61 (d, 1H, J D 4.2 Hz,
H-5), 5.68 (d, 1H, J D 3.9 Hz, H-6), 7.77 (m, 2H, Armeta), 7.89
(m, 2H, Arortho); 13C NMR υ 28.2 (qm, J D 129 Hz, ˛Me),
31.1 (qm, J D 129 Hz, ˇMe), 52.7 (q, J D 148 Hz, CO2CH3),
58.6 (dd, J D 150, 3 Hz, C-6), 66.2 (br m, C-3), 67.1 (ddd,
J D 176, 4, 3 Hz, C-5), 71.1 (dm, J D 146 Hz, C-2), 124 (dm,
J D 166 Hz, Arortho), 131.7 (m, Aripso), 134.7 (dd, J D 164, 7 Hz,
Armeta), 166.8 (dd, J D 4, 4 Hz, Ar C O), 168.6 (m, CO2CH3),
168.6 (m, C-7); HRMS Calcd C17H16N2NaO5S for (M C Na)
383.0672, found 383.0666 (ꢀ1.6 ppm error); LC (ꢂ D 254 nm)
19.640 min.
(2S, 5S, 6R)-6-(1,3-dioxoisoindolin-2-yl)-3,3-
dimethyl-7-oxo-4-thia-1-aza-bicyclo
[3.2.0]heptane-2-carboxylic acid (7)
1H NMR (CDCl3) υ 1.57 (s, 3H, ˛Me), 1.73 (s, 3H, ˇMe),
4.03 (s, 1H, H-2), 5.39 (d, 1H, J D 2.1 Hz, H-5), 5.51 (d,
1H, J D 1.8 Hz, H-6), 7.78 (m, 2H, Armeta), 7.91 (m, 2H,
Arortho); 13C NMR υ 26.9 (qm, J D 129 Hz, ˛Me), 29.7 (qm,
J D 129 Hz, ˇMe), 61.1 (d, J D 153 Hz, C-6), 65.2 (br m, C-
3), 66.1 (ddd, J D 177, 5, 3 Hz, C-5), 72.2 (dm, J D 152 Hz,
C-2), 124.2 (dm, J D 167 Hz, Arortho), 131.6 (m, Aripso), 135.0
(dm, J D 165 Hz, Armeta), 166.7 (dd, J D 4, 4 Hz, Ar C O),
167.6 (d, J D 5 Hz, CO2H), 168.8 (ddd, J D 7, 4, 4 Hz, C-7);
HRMS Calcd for C16H14N2NaO5S ꢀM C Naꢁ 369.0516, found
369.0511 (ꢀ1.4 ppm error); LC (ꢂ D 254 nm) 16.796 min.
(2S, 5S, 6R)-6-(1,3-dioxoisoindolin-2-yl)-3,3-
dimethyl-7-oxo-4-thia-1-aza-bicyclo
[3.2.0]heptane-2-carboxylic acid methyl ester (3)
1H NMR (CDCl3) υ 1.49 (s, 3H, ˛Me), 1.69 (s, 3H, ˇMe), 3.84
(s, 3H, CO2CH3), 3.90 (s, 1H, H-2), 5.44 (d, 1H, J D 2.0 Hz,
H-5), 5.56 (d, 1H, J D 2.5 Hz, H-6), 7.77 (m, 2H, Armeta),
7.90 (m, 2H, Arortho); 13C NMR υ 25.0 (qm, J D 129 Hz,
˛Me), 31.4 (qm, J D 129 Hz, ˇMe), 52.8 (q, J D 148 Hz,
CO2CH3), 60.1 (d, J D 152 Hz, C-6), 66.0 (br m, C-3), 66.6
(ddd, J D 177, 4, 4 Hz, C-5), 70.5 (dm, J D 151 Hz, C-2),
124.1 (dm, J D 168 Hz, Arortho), 131.8 (m, Aripso), 134.8 (dm,
J D 164 Hz, Armeta), 165.9 (ddd, J D 7, 3, 2 Hz, C-7), 166.8
(dd, J D 4, 4 Hz, Ar C O), 167.3 (dq, J D 4, 4 Hz, CO2CH3);
HRMS Calcd for C17H16N2NaO5S ꢀM C Naꢁ 383.0672, found
383.0674 (ꢀ0.5 ppm error); LC (ꢂ D 254 nm) 19.176 min.
(2S, 5R, 6S)-6-(1,3-dioxoisoindolin-2-yl)-3,3-
dimethyl-7-oxo-4-thia-1-aza-bicyclo
[3.2.0]heptane-2-carboxylic acid (8)
1H NMR (CDCl3) υ 1.60 (s, 3H, ˛Me), 1.68 (s, 3H, ˇMe), 4.64 (s,
1H, H-2), 5.42 (d, 1H, J D 1.8 Hz, H-6), 5.56 (d, 1H, J D 1.8 Hz,
H-5), 7.78 (m, 2H, Armeta), 7.90 (m, 2H, Arortho); 13C NMR υ
25.8 (qm, J D 129 Hz, ˛Me), 34.3 (qm, J D 129 Hz, ˇMe),
64.2 (d, J D 153 Hz, C-6), 64.4 (br m, J D 179 Hz, C-3), 69.0
(dm, J D 180 Hz, C-5), 69.3 (dm, J D 145 Hz, C-2), 124.2 (dm,
J D 166 Hz, Arortho), 131.7 (m, Aripso), 134.9 (dm, J D 165, 7 Hz,
Armeta), 166.7 (dd, J D 4, 4 Hz, Ar C O), 167.6 (ddd, J D 7,
4, 4 Hz, C-7), 172.0 (d, J D 5 Hz, CO2H); HRMS Calcd for
C16H13N2O5S ꢀM ꢀ Hꢁ 345.0551, found 345.0536 (ꢀ4.3 ppm
error); LC (ꢂ D 254 nm) 16.841 min.
(2S, 5R, 6S)-6-(1,3-dioxoisoindolin-2-yl)-3,3-
dimethyl-7-oxo-4-thia-1-aza-bicyclo[3.2.0]heptane-
2-carboxylic acid methyl ester (4)
1H NMR (CDCl3) υ 1.49 (s, 3H, ˛Me), 1.66 (s, 3H, ˇMe), 3.80
(s, 3H, CO2CH3), 4.64 (s, 1H, H-2), 5.41 (d, 1H, J D 1.8 Hz,
H-6), 5.58 (d, 1H, J D 1.8 Hz, H-5), 7.79 (m, 2H, Armeta),
7.88 (m, 2H, Arortho); 13C NMR υ 25.6 (qm, J D 129 Hz,
˛Me), 34.7 (qm, J D 128 Hz, ˇMe), 52.6 (q, J D 148 Hz,
CO2CH3), 64.3 (br m, C-3), 64.5 (d, J D 152 Hz, C-6), 69.2
(ddd, J D 178, 7, 4 Hz, C-5), 69.4 (dm, 145 Hz, C-2), 124.1
(dm, J D 166 Hz, Arortho), 131.7 (m, Aripso), 134.9 (dd, J D 165,
7 Hz, Armeta), 166.6 (dd, J D 4, 4 Hz, Ar C O), 167.4 (ddd,
J D 6, 4, 4 Hz, C-7), 167.8 (dq, J D 5, 4 Hz, CO2CH3);
HRMS Calcd for C17H16N2NaO5S ꢀM C Naꢁ 383.0672, found
383.0667 (ꢀ1.3 ppm error); LC (ꢂ D 254 nm) 19.659 min.
2-(5,5-dimethyl-2,5-dihydrothiazol-2-yl)-2-(1,3-
dioxoisoindolin-2-yl)acetic acid (9)
1H NMR (CDCl3) υ 1.54 (s, 3H, Me), 1.57 (s, 3H, Me), 4.83
(d, 1H, J D 9.9 Hz, H-2), 6.26 (dd, 1H, J D 9.6, 2.3 Hz,
H-20), 7.21 (d, 1H, J D 2.1 Hz, H-40), 7.71 (m, 2H, Armeta),
7.83 (m, 2H, Arortho); 13C NMR υ 29.0 (Me), 30.0 (Me), 57.9
(C-2), 64.4 (C-50), 79.0 (C-20), 123.9 (Arortho), 131.8 (Aripso), 134.4
(Armeta), 167.1 (ArC O), 170.8 (COOH), 173.4 (C-40); HRMS
Calcd for C15H13N2O4S ꢀM ꢀ Hꢁ 317.0602, found 317.0599
(ꢀ0.9 ppm error); and Calcd for C15H14N2O4SNa ꢀM C Naꢁ
341.0566, found 341.0560 (ꢀ1.8 ppm error); LC (ꢂ D 254 nm)
15.240 min.
Copyright  2007 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2007; 45: 819–829
DOI: 10.1002/mrc

828
K. W. Wiitala, C. J. Cramer and T. R. Hoye
32. Wiitala KW, Hoye TR, Cramer CJ. J. Chem. Theory Comput. 2006;
2: 1085.
33. Migda W, Rys B. Magn. Reson. Chem. 2004; 42: 459.
34. Wang B, Hinton JF, Pulay P. J. Comput. Chem. 2002; 23: 492.
35. Wang B, Fleischer U, Hinton JF, Pulay P. J. Comput. Chem. 2001;
22: 1887.
Supplementary material
Supplementary electronic material for this paper is available
in Wiley InterScience at: http://www.interscience.wiley.
com/jpages/0749-1581/suppmat/
36. Giesen DJ, Zumbulyadis N. Phys. Chem. Chem. Phys. 2002; 4:
5498.
37. Sebag AB, Forsyth DA, Plante MA. J. Org. Chem. 2001; 66: 7967.
38. Rablen PR, Pearlman SA, Finkbiner J. J. Phys. Chem. A 1999; 103:
7357.
Acknowledgements
This work was supported in part by the National Science Foundation
(CJC, CHE-0203346) and the National Cancer Institute, National
Institutes of Health (TRH, CA-76497).
39. Bifulco G, Dambruoso P, Gomez-Paloma L, Riccio R. Chem. Rev.
2007 (in press).
40. Pretsch E, Bu¨hlmann P, Affolter C. Structure Determination of
Organic Compounds: Tables of Spectral Data. Springer: Berlin, 2004.
41. Claridge T. High-Resolution NMR Techniques in Organic
Chemistry. Elsevier: London, 2004.
42. Becke AD. Phys. Rev. A 1988; 38: 3098.
43. Lee C, Yang W, Parr RG. Phys. Rev. B 1988; 37: 785.
44. Becke AD. J. Chem. Phys. 1993; 98: 5648.
REFERENCES
1. Bagno A, Rastrelli F, Saielli G. Chem. Eur. J. 2006; 12: 5514.
2. Perez M, Peakman TM, Alex A, Higginson PD, Mitchell JC,
Snowden MJ, Morao I. J. Org. Chem. 2006; 71: 3103.
3. Subramaniam G, Karimi S, Phillips D. Magn. Reson. Chem. 2006;
44: 1118.
4. Taubert S, Konschin H, Sundholm D. Phys. Chem. Chem. Phys.
2005; 7: 2561.
5. Baladina A, Mamedov V, Franck X, Figadere B, Latypov S.
Tetrahedron Lett. 2004; 45: 4003.
6. Facelli JC. Concepts Magn. Reson.Part A 2004; 20A: 42.
7. Boggioni L, Bertini F, Zannoni G, Tritto I, Carbone P, Ragazzi M,
Ferrro DR. Macromolecules 2003; 36: 882.
8. Tahtinen P, Bagno A, Klika KD, Pihlaja K. J. Am. Chem. Soc. 2003;
125: 4609.
9. Price DR, Stanton JF. Org. Lett. 2002; 4: 2809.
10. Barone G, Gomez-Paloma L, Duca D, Silvestri A, Riccio R,
Bifulco G. Chem. Eur. J. 2002; 8: 3233.
11. Ragazzi M, Carbone P, Ferro DR. Int. J. Quantum Chem. 2002; 88:
663.
45. Stephens PJ, Devlin FJ, Chabalowski CF, Frisch MJ. J. Phys.
Chem. 1994; 98: 11623.
46. Perdew JP. In Electronic Structure of Solids ‘91. Ziesche P,
Eschrig H (eds). Akademie Verlag: Berlin, 1991; 11.
47. Perdew JP, Wang Y. Phys. Rev. B 1992; 45: 13244.
48. Kukolja S, Jones ND, Chaney MO, Elzey TK, Gleissner MR,
Paschal JW, Dorman DE. J. Org. Chem. 1975; 40: 2388.
49. Fekner T, Baldwin JE, Adlington RM, Jones TW, Prout CK,
Schofield CJ. Tetrahedron 2000; 56: 6053.
50. Kukolja S. J. Am. Chem. Soc. 1971; 93: 6267.
51. Fisher JW, Trinkle KL. Tetrahedron Lett. 1994; 35: 2505.
52. Chang CJ, Hem SL. J. Pharm. Sci. 1979; 68: 64.
53. Bodurow CC, Levy JN, Wiitala KW. Tetrahedron Lett. 1991; 33:
4283.
54. Schutz A, Ugi I. J. Chem. Res. (S) 1979; 157.
55. Baldwin JE, Chan RY, Sutherland JD. Tetrahedron Lett. 1994; 35:
5519.
12. Bagno A. Chem. Eur. J. 2001; 7: 1652.
13. Rich JE, Manalo MN, de Dios AC. J. Phys. Chem. A 2000; 104:
5837.
14. Kolehmainen E,
Koivisto J,
Nikiforov V,
Perakyla M,
56. Diaz N, Suarez D, Sordo TL. J. Comp. Chem. 2003; 24: 1864.
57. Polak E, Ribiere G. Rev. Franc. Informat. Rech. Oper. 1969; 16: 35.
58. Chang G, Guides WC, Still WC. J. Am. Chem. Soc. 1989; 111: 4379.
59. Weiner SJ, Kollman PA, Case DA, Singh UC, Ghio C,
Alagona G, Profeta S, Weiner P. J. Am. Chem. Soc. 1984; 106:
765.
60. Weiner SJ, Kollman PA, Nguyen DT, Case DA. J. Comp. Chem.
1986; 7: 230.
61. Still WC, Tempczyk A, Hawley RC, Hendrickson T. J. Am. Chem.
Soc. 1990; 112: 6127.
62. Hehre WJ, Radom L, Schleyer PvR, Pople JA. Ab Initio Molecular
Orbital Theory. Wiley: New York, 1986.
63. Mennucci B, Tomasi J. J. Chem. Phys. 1997; 106: 5151.
64. Mennucci B, Cance`s E, Tomasi J. J. Phys. Chem. B 1997; 101: 10506.
65. Tomasi J, Mennucci B, Cance`s E. J. Mol. Struct. (Theochem) 1999;
464: 211.
66. Barone V, Cossi M, Tomasi J. J. Chem. Phys. 1997; 107: 3210.
67. London F. J. Phys. Radium (Paris) 1937; 8: 397.
68. Wolinski K, Hinton JF, Pulay P. J. Am. Chem. Soc. 1990; 112: 8251.
69. Gauss J. J. Chem. Phys. 1993; 99: 3629.
Tuppurainen K, Laihia K, Kauppinen R, Miltsov SA, Kara-
van VS. Magn. Reson. Chem. 1999; 37: 743.
15. Ortiz PJ, Evleth EM, Montero LA. J. Mol. Struct. (Theochem) 1998;
432: 121.
16. Vogt FG, Spoor GP, Su Q, Andemichael YW, Wang H,
Potter TC, Minick DJ. J. Mol. Struct. 2006; 797: 5.
17. Wipf P, Kerekes AD. J. Nat. Prod. 2003; 66: 716.
18. Harper JK, Barich DH, Hu JZ, Strobel GA, Grant DM. J. Org.
Chem. 2003; 68: 4609.
19. Backes AC, Schatz J, Siehl H. J. Chem. Soc., Perkin Trans. 2 2002;
3: 484.
20. Barone G, Duca D, Silvestri A, Gomez-Paloma L, Riccio R,
Bifulco G. Chem. Eur. J. 2002; 8: 3240.
21. Colombo D, Ferraboschi P, Ronchetti F, Toma L. Magn. Reson.
Chem. 2002; 40: 581.
22. Monica CD, Randazzo A, Bifulco G, Cimino P, Aquino M, Izzo I,
De Riccardis F, Gomez-Paloma L. Tetrahedron Lett. 2002; 43: 5707.
23. Ochsenfeld C, Brown SP, Schnell I, Gauss J, Spiess HW. J. Am.
Chem. Soc. 2001; 123: 2597.
24. Brown SP, Schaller T, Seelbach UP, Koziol F, Ochsenfeld C,
Klarner F, Spiess HW. Angew. Chem. Int. Ed. 2001; 40: 717.
25. Malkina OL, Hricovin M, Bizik F, Malkin VG. J. Phys. Chem. A
2001; 105: 9188.
70. Cammi R. J. Chem. Phys. 1998; 109: 3185.
71. Cammi R, Mennucci B, Tomasi J. J. Chem. Phys. 1999; 110: 7627.
72. Bondi A. J. Phys. Chem. 1964; 68: 441.
73. Frisch MJ, Trucks GW, Schlegel HB, Scuseria GE, Robb MA,
Cheeseman JR, Montgomery JA Jr, Vreven T, Kudin KN,
Burant JC, Millam JM, Iyengar SS, Tomasi J, Barone V,
Mennucci B, Cossi M, Scalmani G, Rega N, Petersson GA,
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 JE, Hratchian HP, Cross JB,
Bakken V, Adamo C, Jaramillo J, Gomperts R, Stratmann RE,
Yazyev O, Austin AJ, Cammi R, Pomelli C, Ochterski JW,
Ayala PY, Morokuma K, Voth GA, Salvador P, Dannenberg JJ,
Zakrzewski VG, Dapprich S, Daniels AD, Strain MC, Farkas O,
26. Harper JK, Mulgrew AE, Li JY, Barich DH, Strobel GA,
Grant DM. J. Am. Chem. Soc. 2001; 123: 9837.
27. Sebag AB, Friel CJ, Hansen RN, Forsyth DA. J. Org. Chem. 2000;
65: 7902.
28. Clark AJ, Ellard JM. Tetrahedron Lett. 1998; 39: 6033.
29. Stahl M, Schopfer U, Frenking G, Hoffman RW. J. Org. Chem.
1996; 61: 8083.
30. Cheeseman JR, Trucks GW, Keith TA, Frisch MJ. J. Chem. Phys.
1996; 104: 5497.
31. Cramer CJ. Essentials of Computational Chemistry: Theories and
Models. John Wiley & Sons: Chichester, 2004.
Copyright  2007 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2007; 45: 819–829
DOI: 10.1002/mrc

DFT methods for distinguishing stereoisomers via chemical shifts: penam test set 829
Malick DK, Rabuck AD, Raghavachari K, Foresman JB, Ortiz JV,
Gonzalez C, Pople JA. Gaussian 03, Revision B.05, Gaussian:
Wallingford, 2004.
74. Wiitala KW, Al-Rashid ZF, Dvornikovs V, Hoye TR, Cramer CJ.
J. Phys. Org. Chem. 2007; 20: 345.
Cui Q, Baboul AG, Clifford S, Cioslowski J, Stefanov BB,
Liu G, Liashenko A, Piskorz P, Komaromi I, Martin RL,
Fox DJ, Keith T, Al-Laham MA, Peng CY, Nanayakkara A,
Challacombe M, Gill PMW, Johnson B, Chen W, Wong MW,
Copyright  2007 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2007; 45: 819–829
DOI: 10.1002/mrc