A minimalist NMR approach for the structural revision of mucoxin

Article abstract of DOI:10.1002/chem.201001619

In an attempt to revise the structural assignment of mucoxin, and faced with 64 diastereomeric possibilities, we resorted to the synthesis of truncated structures that contained the core stereochemical sites. Twelve stereochemical analogues were synthesiz

Full text of DOI:10.1002/chem.201001619

FULL PAPER
DOI: 10.1002/chem.201001619
A Minimalist NMR Approach for the Structural Revision of Mucoxin
Jun Yan, Atefeh Garzan, Radha S. Narayan, Chrysoula Vasileiou,* and Babak Borhan*[a]
Abstract: In an attempt to revise the
structural assignment of mucoxin, and
faced with 64 diastereomeric possibili-
ties, we resorted to the synthesis of
truncated structures that contained the
core stereochemical sites. Twelve ste-
reochemical analogues were synthe-
sized, their 1H and 13C NMR spectra
were analyzed and four recurring ste-
reochemical trends were distilled from
the data. Applying the observed trends
to the diastereomeric population pared
the possible choices for the correct
structure of mucoxin from 64 to 4. Syn-
thesis of these analogues led to the
identification of the correct structure
of mucoxin.
Keywords: acetogenins · mucoxin ·
natural products · NMR spectrosco-
py · structure elucidation
Introduction
great effect on the chemical environment of the resonating
nuclei and thus yield large changes in chemical shift. There-
fore, it is not uncommon that a complete set of possible dia-
stereomers must be synthesized in order to discover the real
structure of a misidentified natural product. Recently, we
faced this exact problem during the synthesis of mucoxin, a
reportedly bioactive Annonaceous acetogenin (Scheme 1).[1]
With the mismatch observed between the reported NMR
spectrum and that of our synthesized material we were
faced with the possibility of having to synthesize 64 different
diastereomers. To avoid this daunting task we resorted to
With the exception of X-ray crystallography, which yields
unequivocal assignment of structure for organic molecules
and new natural products, total synthesis is still the gold
standard for proof of structure. With a large number of nat-
ural product families that do not succumb to crystallization,
the initial structural assignment is often a non-trivial exer-
cise. Expectedly, errors in structure determination are
common (most often realized through synthesis of the pro-
posed structure), and the problem is exacerbated since an
error in stereochemical determination of one site of the
molecules can lead to significant changes in the chemical
shift of nearby nuclei, realizing that most natural products
are complex molecules containing multiple stereocenters.
Error in assignment of one or a combination of stereocen-
ters leads to observed mismatches between the NMR spec-
tra of the authentic natural compound and the synthetic ma-
terial. Often the magnitude of chemical shift discrepancies
for individual centers cannot be used reliably for identifica-
tion of misidentified stereocenters (particularly with com-
pounds that have contiguous stereocenters), since changes
in relative configuration can lead to a number of different
conformations, rotamers, and ring geometries. These have a
[a] Dr. J. Yan, A. Garzan, Dr. R. S. Narayan, Dr. C. Vasileiou,
Prof. B. Borhan
Scheme 1. General structure of Annonaceous acetogenins. Most but not
all have two disubstituted THF rings flanked by two hydrophobic chains.
On one end, a butenolide ring system is a common structural feature for
this class of molecules. Hydroxylation of the hydrophobic spacer is seen
but is not a common or necessary feature for activity. Dashed box: The
proposed structure of mucoxin (arbitrarily drawn with absolute configu-
ration based on the proposed relative configuration) is unique due to the
ring hydroxylation.
Department of Chemistry, Michigan State University
East Lansing, MI 48824 (USA)
E-mail: vassilio@chemistry.msu.edu
babak@chemistry.msu.edu
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201001619.
Chem. Eur. J. 2010, 16, 13749 – 13756
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
13749

Table 1. 1H NMR comparison of the natural and synthetic mucoxin
(NMR spectra recorded in CDCl3).[a]
synthesize a few selected diastereomers with the intention
of utilizing spectral trends obtained from the study of these
diastereomers to focus the field of possibilities to only a
handful of choices. Chemical shift correlations have been
used previously as a guide for structure determination for
acetogenins. Hoye and Suhadolnik et al., in an elegant set
of experiments, demonstrated the utility of chemical shift
correlations for relative configurational assignment of uvari-
cin.[2,3] Curran and co-workers performed an exhaustive
study of a stereoisomeric library of murisolin, illustrating
the use of Mosher ester derivatives for configurational as-
signment of single THF containing acetogenins.[4,5] Herein,
we take a complimentary approach to previous studies to
assign the relative configurational relationships in mucoxin.
A brief summary of our initial investigation, which led to
the latter strategy for the identification of the real structure
of mucoxin is provided below.
Annonaceous acetogenins are a series of natural products
isolated from the plant family Annonaceae, whose members
include edible fruits such as cherimoya and custard apple,
and are widely distributed in tropical and sub-tropical re-
gions.[1–9] They are found to have potent and diverse biologi-
cal effects such as cytotoxic, antitumor, antimalarial, pestici-
dal, and insecticidal activities.[6–13] The acetogenins share a
common skeleton characterized by an unbranched C32 or
C34 fatty acid ending in an a,b-unsaturated g-lactone
(Scheme 1).[7,12,14] Several oxygenated functional groups,
such as tetrahydrofuran (THF), tetrahydropyran (THP), ep-
oxide, hydroxy, and ketone could be present in the structure,
as well as double and triple bonds. There is a variety of ste-
reoisomers of the center THF cores, however, all isolated
compounds have an S configuration of the chiral center in
the g-lactone ring. Generally acetogenins are divided into
two broad groups. Classical acetogenins contain up to three
2,5-disubstituted THF units along with a terminal g-lactone.
Nonclassical acetogenins contain either a THP or a hydroxy-
lated THF ring, and in some examples in combination with
a 2,5-disubstituted THF ring. Although to date there are
more than 400 known acetogenins, their isolation and struc-
ture determination can be tedious since only small amounts
in complex mixtures are obtained from natural sources.[10,12]
Moreover, most acetogenins are waxes or gums, and cannot
be crystallized for X-ray crystallographic analysis in order to
determine the relative and absolute configuration of their
stereogenic centers.
Mucoxin, an example of nonclassical acetogenins, was iso-
lated from the bioactive leaf extracts of Rollinia mucosa by
McLaughlin and his co-workers in 1996.[15] It has exhibited
potent and selective cytotoxic properties against PACA-2
(pancreatic cancer) and MCF-7 (breast cancer) cell lines in
a panel of six human solid tumors (in vitro assays), however,
limited access to the material (1.8 mg was isolated from nat-
ural sources) has hampered further biological evaluations.[15]
We have recently disclosed the total synthesis of mucoxin
(the proposed structure) and after careful consideration con-
cluded that the stereochemical assignment was wrong.[1]
Table 1 summarizes the observed chemical shifts for both
À
H8
H9
H C
1[b]
Natural mucoxin
Dd (1-natural)
3.41
3.96
1.69, 2.02
2.02, 2.13
4.35
3.80
4.44
1.84, 2.02
4.09
3.41
73.6
84.1
79.3
83.3
74.7
39.0
81.6
74.3
3.42
3.95
1.91, 2.05
1.91, 2.05
4.31
3.71
4.32
1.91, 2.35
4.04
3.53
73.6
83.4
78.9
83.5
72.8
38.0
79.9
73.7
À0.01
+0.01
À0.22, À0.03
+0.11, +0.08
+0.04
H10
H11
H12
H13
H14
H15
H16
H17
C8
+0.09
+0.12
À0.07, À0.33
+0.05
À0.12
0.0
C9
+0.7
+0.4
C12
C13
C14
C15
C16
C17
À0.2
+1.9
+1.0
+1.7
+0.6
[a] Deviation greater than 0.1 ppm for proton and 1.0 ppm for carbon are
highlighted in bold. [b] The data in all Tables for all synthesized ana-
logues in this study have been referenced to 7.27 ppm (CDCl3) to be
comparable with the published data for mucoxin.
the naturally isolated compound and the synthetically pre-
pared material. A close inspection of the data reveals the
problem alluded to above with regards to identifying site(s)
that are stereochemically misassigned based on the gross
changes in chemical shift of each stereocenter; namely, most
of the chiral centers exhibit large differences that hamper
the formulation of a reasonable hypothesis as to the source
of the stereochemical misassignment. Excluding the chiral
center in the butenolide, mucoxin has seven closely situated
stereocenters, which translates to 64 possible diastereomers.
In the present study, we resorted to the synthesis of a small
1
subgroup of diastereomers such that trends in the H and
13C NMR spectra could be derived and subsequently used to
pare down the number of possible structures. The choice of
diastereomers synthesized is critical to provide a sufficient
13750
www.chemeurj.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 13749 – 13756

Structural Revision of Mucoxin
FULL PAPER
Table 2. Conformational analysis of the 13,14-cis-12,13-erythro struc-
grouping of various relative stereochemical relationships so
that conclusive trends could be obtained. Via the latter
strategy, we demonstrate how with 12 diastereomeric struc-
tures we were able to narrow the field of 64 possible diaste-
reomers to four candidates, leading to the identification of
two stereochemical mistakes and the eventual structural re-
vision of mucoxin.
ture.[a]
Results and Discussion
We have re-examined McLaughlinꢁs structural elucidation of
mucoxin[15] to delineate any ambiguities in their proposed
structure and possible sources of discrepancies between the
synthetic and the natural spectra. Based on the reported
COSY and HRMS analysis of the natural sample of mucox-
in, we find no reason to question the proposed constitution-
al structure of mucoxin. Therefore, we believe that the dif-
ferences in the spectra of the synthetic versus natural sam-
ples are most likely due to stereochemical mismatches. In
the case of natural mucoxin, the relative configuration of
[c]
[d]
Conformer Relative energy [kcalmolÀ1
]
OH–O[b] [ꢂ] q13–14 q12–13
1
2
3
4
5
6
7
8
9
10
11
12
13
14
0.00
0.31
0.62
0.69
0.73
0.94
0.95
1.13
1.22
1.30
1.33
1.43
1.54
1.64
2.44
2.44
2.43
2.37
1.84
1.84
2.35
1.84
2.44
1.85
1.85
1.85
2.44
1.85
À34.2 À173
À34.2 À173
À34.1 À173
À32.7 À171
À40.0 À65.4
À40.0 À65.7
À32.6 À173
À40.0 À65.6
À34.0 À173
À39.3 À64.4
À39.5 À64.5
À40.0 À64.6
À33.9 À173
À39.3 À64.5
À
À
the 2,5-substitution for both THF rings (C9 C12 and C13
C16) was assigned as trans based on the lack of NOESY cor-
relations. However, lack of NOE can be misleading in the
determination of relative configuration, especially with five-
membered ring systems.
[a] Non-stereogenic hydrogen atoms are omitted for clarity. [b] Distance
between C14-OH and C9 C12 THF oxygen. [c] H13–H14 dihedral angle.
À
We next turned to evaluate the assignment for the relative
configuration of the C12, C13, and C14 triad. In the NMR
spectrum of natural mucoxin, H13 appeared as a pseudo
triplet (J=3 Hz). Based on this and molecular models, the
[d] H12–H13 dihedral angle.
is 2.44 ꢂ. Thus, the 13,14-cis-12,13-erythro conformation
with strong intramolecular H-bonding results in the expect-
ed small coupling constant, and therefore, contrary to the ar-
gument used for structural assignment of mucoxin, it is pos-
sible for the cis–erythro compound to exhibit the observed
3 Hz coupling. As anticipated, the H12/H13 dihedral angle
does not change to a great extent in any of the conformers.
We further confirmed our modeling results by measuring
the NMR spectrum of 1 (proposed structure of mucoxin) in
methanol. H13 appears as a triplet with a 3 Hz coupling con-
stant in CDCl3, however, in CD3OD H13 appeared as a dou-
blet of a doublet (J=3.3 and 7.3 Hz). While the coupling
constant between H13 and H14 is expected to be more or
less independent of the solvent, the J value between H12
and H13 could vary with the nature of the solvent if intra-
molecular H-bonding is disturbed. Thus, as predicted by
À
authors proposed that the rotation of the C12 C13 bond
might be restricted possibly due to intramolecular hydrogen
À
bonding between the C14 hydroxyl and the C9 C12 THF
ring oxygen.[15] In such a rigid conformation, it was suggest-
ed that in order for H13 to maintain a 3 Hz coupling con-
stant with H12 and H14, the relative configuration of the
two THF rings must be threo and H14 and H13 must be cis.
This stereochemical assignment seems tenuous since the
only experimental evidence presented is the coupling con-
stant of H13 with H12 and H14. To gain further insight into
the conformations of the natural product, we carried out
Monte Carlo based conformational studies (MacroModel
version 7.0) of the proposed structure and its epimers at
C12 and C14. Modeling of the 13,14-cis-12,13-threo com-
pound led to structures (within 2 kcal) with dihedral angles
necessary for the small observed 3 Hz coupling (consistent
with the proposed structure). However, similar analysis of
the 13,14-cis-12,13-erythro diastereomer (2) provided low
energy conformers (within 2 kcal) that could also exhibit the
observed 3 Hz coupling. Conformational minimization of
13,14-cis-12,13-erythro isomer 2 leads to a series of conform-
ers with 50% population in agreement with conformation B
(Table 2). The ~658 dihedral angle between H12 and H13
would yield a small coupling constant (in order of 3 Hz),
consistent with the observed NMR spectrum of the natural
product. Notably, the OH–O distance between C14-OH and
À
modeling the non-H-bonded rotamer (longer OH O bond
length) leads to a >1708 dihedral angle and the larger ob-
served coupling constant (7.3 Hz).
In a similar fashion, we have examined the 13,14-trans-
12,13-threo and 13,14-trans-12,13-erythro systems, and find it
difficult to eliminate these possibilities upon further analysis
and in comparison to the reported spectroscopic data for
mucoxin. At this juncture it became clear that many alter-
nate isomers are possible, and since the synthesis of each
isomer would be unfeasible, we resorted to an alternate ap-
proach. The goal was to derive guides for NMR chemical
shifts that were unique to specific stereochemical relation-
À
C9 C12 THF oxygen in B is about 1.84 ꢂ whereas that in A
Chem. Eur. J. 2010, 16, 13749 – 13756
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
13751

B. Borhan, C. Vasileiou et al.
ships. These could be utilized not only to pare down the
number of possibilities for mucoxin, but also would add to
the library of chemical shift correlations that has been grow-
ing through mostly the efforts of Kishi and co-workers.[16–20]
Although tedious to accumulate, chemical shift correlations
are immensely useful for the structural determination of nat-
ural products and also for verification of structural assign-
ment in synthetic projects.
posed diastereomers, and therefore took advantage of inter-
jecting into the scheme at opportune locations that would
yield the necessary changes with simple chemistry (most
often by changing chiral auxiliaries for the Sharpless Asym-
metric Dihydroxylation (SAD) reaction or by selective de-
protection of hydroxyl groups followed by an oxidation-re-
duction sequence). Structures in bold in Figure 1 list the
chosen diastereomers for synthesis.
Prior to the synthesis of the proposed diastereomers we
demonstrated that a truncated structure containing the bis-
THF core that includes all stereogenic centers in doubt is
sufficient for spectral comparisons, thus eliminating the
need to synthesize the full-length molecules. This is illustrat-
Strategy for selection of representative structures
Figure 1 depicts the list of all possible diastereomeric struc-
tures of mucoxin without the inclusion of the C36 stereogen-
ic center (assumed to be S by inference from all known
structures of acetogenins). Scheme 1 illustrates the short
hand codes used to identify each acetogenin. The first three
lower case letters refer to the stereochemical relationships
that are either cis (c) or trans (t). Placing the butenolide on
the right hand side, the cis/trans relationships are read from
left to right. Thus, the proposed structure of mucoxin (1,
1
ed in Table 3, in which the H NMR spectrum of the trun-
Table 3. 1H NMR comparison of
a truncated structure with the full
length acetogenin (proposed structure of mucoxin) (NMR spectra record-
ed in CDCl3).
À
À
À
Scheme 1) is tct (C16 C13 trans; C14 C13 cis; C12 C9
trans). Similarly, erythro (E) and threo (T) relationships that
would fully describe the stereochemical picture of each dia-
stereomer are given as the last three letters in capitals
À
À
À
(C17 C16 threo; C13 C12 threo; C9 C8 threo, thus TTT for
structure 1).
À
H C
tctTTT Model
Proposed mucoxin
Dd Value
ACHTUNGTRENNUNG[ppm]
H
C
H
C
H
H
H
H
H
C
H
C
H
C
H
H
C
H
C
H
C
3.41
73.4
3.95
83.9
1.71
2.03
2.03
2.13
3.41
73.6
3.96
84.1
1.69
2.02
2.02
2.13
4.35
0.00
8
À0.2
À0.01
À0.2
+0.02
+0.01
+0.01
0.00
9
10
11
12
13
14
4.34
79.0
À0.01
À0.3
0.00
À0.2
À0.01
À0.3
À0.01
+0.01
À0.3
0.00
79.3
3.80 (t, J=3.4 Hz)
3.80 (t, J=3 Hz)
83.3
83.1
4.43
74.4
1.83
2.03
38.7
4.09
81.3
3.41
74.1
4.44
74.7
1.84
2.02
39.0
4.09
81.6
3.41
Figure 1. List of 64 possible diastereomers considering only the stereo-
genic centers between C8 and C17. Isomers in bold were chosen as initial
synthetic targets. Isomers highlighted in italics satisfy the restrictions con-
cluded after analysis of the data in Tables 5–8.
15
16
17
À0.3
0.00
As described above, the goal was to select 12 diastereo-
mers for synthesis and complete NMR assignment. Depend-
ence of chemical shift changes in both 1H and 13C NMR
spectra as a function of relative stereochemical relationships
were sought in order to narrow down the number of diaste-
reomeric possibilities. The selection of the 12 diastereomers
was governed by two major factors. First and foremost, it
was necessary to have a list of structures that would yield re-
dundant stereochemical relationships at each site except for
the location that was changed. This would ensure that if a
trend is observed from the chemical shift analysis for each
74.3
À0.2
cated tctTTT molecule A is compared to the full-length
tctTTT acetogenin 1 (proposed structure of mucoxin). There
was less than 0.02 ppm deviation between protons on C8 to
C17 of the truncated and full length molecule. This allowed
the use of simpler truncated structures in an effort to deter-
mine the structure of mucoxin.
Synthesis of truncated diastereomers of mucoxin
À
diastereomeric relationship (such as C9 C8 erythro vs
threo), then it must have been immune to stereochemical
changes in other parts of the molecule. Second, we relied on
the synthetic scheme developed for 1 to synthesize the pro-
Scheme 2 summarizes the strategies utilized to establish the
À
stereogenic centers in 1. The stereochemistries of C17 C16
À
and C9 C8 were established via Sharpless asymmetric dihy-
13752
www.chemeurj.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 13749 – 13756

Structural Revision of Mucoxin
FULL PAPER
Scheme 2. Retrosynthetic strategy used to synthesize the proposed structure of mucoxin. Most of the stereogenic centers were established via the use of
either Sharpless Asymmetric Dihydroxylation (SAD) or Sharpless Asymmetric Epoxidation (SAE), thus enabling the use of appropriate ligands to
effect stereochemical inversion.
1
À
droxylation, while the stereochemistry of C14 C13 origi-
nates from Sharpless asymmetric epoxidation.[1] The stereo-
chemistry at C12 was arrived at via chelation controlled ad-
dition of an organometallic reagent with an incipient alde-
hyde. The trisubstituted tetrahydrofuran ring in 5 was closed
regioselectively with the aid of the thiophenyl directed ep-
oxydiol cyclization,[21] while the disubstituted ring in 3 was
furnished through the 1,2,n-triol cyclization.[22] Inspection of
the synthetic strategy indicates the ease by which stereogen-
ic changes to the molecule can be accomplished. These
could be either through changes in the ligands used for
asymmetric induction or by inversion of stereochemistry of
the three hydroxyl groups present in the target molecule.
The synthetic details by which the 12 diastereomers depicted
in Scheme 3 were synthesized are relegated to the Support-
ing Information. It should be noted, however, that the ste-
reochemistry of each diastereomer was rigorously confirmed
by a battery of 1D and 2D NMR experiments.
Upon completion of the syntheses and inspection of the H
and 13C NMR data, none of the spectra from 12 isomers A–
L (Scheme 3) fully matched the spectra obtained for natural
mucoxin. In order to efficiently compare the data for each
1
of the isomers with natural mucoxin, the deviations of H
and 13C NMR chemical shift between the truncated struc-
tures and the reported natural mucoxin are tabulated in
Table 4. The data of protons exhibiting differences greater
than 0.1 ppm and carbons exhibiting differences greater
than 1.0 ppm are shown in bold. A careful analysis of the
data enabled the extraction of four reoccurring trends.
These are listed below:
À
1) C8 C9 relationship: Table 5 lists the chemical shifts ob-
À
served for H8 C8 for analogues A–L. The chemical shift
of H8 in the 8,9-erythro-isomers is shifted downfield rela-
tive to 8,9-threo-isomers. Generally, erythro isomers ex-
hibit a chemical shift for H8 that is larger than 3.8 ppm.
On the contrary, threo isomers have chemical shifts
smaller than 3.5 ppm for H8. A similar but opposite
trend is observed for the 13C NMR data. The C8 chemi-
cal shift of erythro isomers
Comparison of spectral data of diastereomers A–L with mu-
coxin and structural revision
falls below 72 ppm, while
the threo isomers have
chemical
shifts
above
73 ppm. This trend has been
previously observed and uti-
lized for predicting the rela-
tive configuration of 2,5-dis-
ubstituted THF rings that
have a hydroxyl group on
the carbon adjacent to the
ring (Bornꢁs rule),[23] and is
further verified by the pres-
ent data for bis-THF sys-
À
tems. The reported H8 C8
chemical shifts for mucoxin
are 3.42 and 73.6 ppm, re-
spectively, suggesting a threo
À
relationship for C8 C9.
À
2) C16 C17 relationship: Sim-
ilar to the observations
made above for the 2,5-dis-
Scheme 3. Structure of synthesized truncated diastereomers of mucoxin.
Chem. Eur. J. 2010, 16, 13749 – 13756
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
13753

B. Borhan, C. Vasileiou et al.
Table 4. Chemical shift differences between the truncated analogues and the reported chemical shifts for mu-
tems. As listed in Table 6,
the chemical shift of H17 in
coxin (NMR spectra recorded in CDCl3). Dd=d (ppm) truncated analogue Àd [ppm] reported mucoxin[a]
À
the C16 C17 erythro iso-
mers
is
larger
than
3.90 ppm, while the C17
chemical shifts are less than
72.0 ppm. On the other
À
H C
A
B
C
D
E
F
G
H
I
J
K
L
H
À0.01 +0.40 À0.01 À0.02 À0.03 À0.05 +0.44 À0.05 À0.01 À0.02 +0.03 À0.02
À0.2 À2.1 0.0 +0.2 0.0 +0.3 À2.3 +0.3 +0.3 +0.1
0.00 +0.05 À0.03 À0.10 À0.11 À0.12 +0.02 À0.18 À0.09 À0.12 À0.01 À0.16
+0.5 +0.4 +0.5 0.0 À0.3 À0.2 À0.1 +0.1 +0.2
À0.2
À0.20 À0.10 À0.27 À0.2
8
À
hand, C16 C17 threo iso-
C
H
C
–
–
mers have H17 chemical
shifts below 3.55 ppm and
C17 chemical shifts above
73.5 ppm. The latter obser-
vations suggest that Bornꢁs
rule could also be applied
for the relative stereochemi-
cal assignment of trisubsti-
tuted THF systems. None-
theless, care must be applied
since only the trend and not
the absolute chemical shifts
follow Bornꢁs rule. This ne-
cessitates the investigation
of both threo and erythro
analogues in order to obtain
the chemical shift character-
istics for different ring sys-
9
–
–
H
H
H
H
H
C
H
C
H
C
À0.20 À0.25 À0.04 À0.27 À0.28 À0.21 À0.20 À0.17
10
11
12
13
14
À0.05 À0.15 À0.04 À0.02 À0.06 À0.06 À0.18 À0.09 À0.11 À0.04 À0.02 À0.12
+0.10 À0.01 +0.02 À0.07 +0.08 À0.09 À0.09 +0.05 +0.05 À0.01 À0.05 +0.02
+0.07 +0.08 +0.07 +0.16 À0.06 +0.03 +0.15 À0.09 +0.14 +0.12 +0.16 +0.08
+0.02 +0.09 À0.02 À0.09 À0.24 À0.48 À0.02 À0.36 À0.19 À0.08 À0.05 À0.13
+0.1
+0.09 +0.06 +0.16
À0.4 À0.2 +0.6
+0.11 +0.09 +0.12 +0.20 +0.13 À0.03 À0.13 À0.09 À0.10 +0.03 À0.08 +0.18
+1.6 +1.6 +1.4 +0.3 +1.2 +1.0 +1.3
+0.2
+0.2
À1.1
0.00 +0.05 +0.02 À0.05 +0.15 À0.14 À0.12
+0.5 +3.5 +4.6 +2.3 +5.7 +2.6
+0.3
+1.3
+0.3
+0.8
À0.2
–
À0.2
0.00 À0.02
+2.1
–
–
–
À0.5
À1.2
–
À0.3
–
H
H
C
À0.08 À0.10 À0.07 À0.08 +0.04 À0.09 À0.04 À0.12 À0.25 À0.04 +0.01 +0.02
À0.35 À0.33 À0.23 À0.32 À0.25 +0.06 +0.02 +0.13 À0.14 +0.01 +0.06 À0.33
15
+0.7
+0.7
À3.3
À0.5
À0.1
À0.9
+0.8
+0.1
À3.7
–
+0.8
–
H
C
H
C
+0.05 +0.03 +0.19 +0.08 +0.08 À0.04 +0.02 +0.02 +0.06 À0.06 +0.06 +0.09
+1.4 +1.5 +1.3 +1.7 +1.5 +1.2 +0.3 +1.6 +1.3 +0.3
À0.12 À0.12 +0.39 À0.13 À0.14 À0.01 À0.04 À0.03 +0.38 À0.01 0.00 À0.13
+0.4 +0.5 +0.4 +0.6 +0.7 +0.4 +0.8 À2.0 +0.4
À2.1
16
17
–
–
–
–
[a] Deviations greater or equal to 0.10 ppm for proton and 1.0 ppm for carbon are highlighted in bold.
À
tems. The reported H17
C17 chemical shifts for mu-
coxin are 3.53 and 73.7 ppm,
À
Table 5. Proton and carbon chemical shifts of H8 C8 for analogues A–L (NMR spectra recorded in CDCl3).
À
À
C8 C9 erythro
C8 C9 threo
À
H C
B
G
A
C
D
E
F
H
I
J
K
L[a]
respectively, suggesting
a
H8
C8
3.82
71.5
3.86
71.3
3.41
73.4
3.41
73.6
3.40
73.8
3.39
73.6
3.37
73.9
3.37
73.9
3.41
73.9
3.40
74.0
3.45
73.7
3.40
–
À
threo relationship for C16
C17.
[a] Overlapping chemical shifts did not allow for an absolute assignment of C8 in analogue L.
À
3) C14 C16
relationship:
Careful analysis of the data
in Table 4 revealed a unique
trend in the chemical shift
difference of the two C15
À
Table 6. Proton and carbon chemical shifts of H17 C17 for analogues A–L (NMR spectra recorded in
CDCl3).
À
À
C16 C17 erythro
C16 C17 threo
À
H C
C
I
A
B
D
E
F
G
H
J
K
L[a]
methylene protons as
a
À
function of the C14 C16
stereochemical relationship.
Table 7 illustrates this more
succinctly by listing the
chemical shifts and the Dd
values for each isomer. In
H17
C17
3.92
71.6
3.91
71.7
3.41
74.1
3.41
74.2
3.40
74.1
3.39
74.3
3.52
74.4
3.49
74.1
3.50
74.5
3.52
73.8
3.53
74.1
3.40
–
[a] Overlapping chemical shifts did not allow for an absolute assignment of C8 in analogue L.
Table 7. Difference in chemical shift of H15a and H15b for analogues A–L (NMR spectra recorded in
CDCl3).
À
all examples, C14 C16 trans
À
À
C14 C16 trans
C14 C16 cis
H
isomers exhibit Dd smaller
than 0.30 ppm for H15a and
H15b. On the other hand,
the isomers that have cis
A
B
C
D
E
L
F
G
H
I
J
K
H15a
H15b
Dd
2.00
1.83
0.17
2.02
1.81
0.21
2.12
1.84
0.28
2.03
1.83
0.20
2.10
1.95
0.15
2.02
1.93
0.09
2.41
1.82
0.59
2.37
1.87
0.50
2.48
1.79
0.69
2.18
1.63
0.55
2.36
1.87
0.49
2.41
1.92
0.49
À
C14 C16 relationship pro-
duce spectra with larger
ubstituted THF ring, the trisubstituted THF ring exhibits
the same dependence of chemical shifts for erythro
(downfield proton and upfield carbon) and threo (upfield
proton and downfield carbon) isomers. The only differ-
ence is in the magnitude of chemical shifts for each
isomer as compared to the disubstituted THF ring sys-
chemical shift disparity for H15a and H15b (Dd is larger
than 0.45 ppm). The reported Dd for mucoxin is
0.44 ppm, thus suggesting a cis stereochemical relation-
À
ship of C14 C16 in the natural product. This observation
can be accounted for by considering the different aniso-
tropic environment experienced by H15a and H15b in
13754
www.chemeurj.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 13749 – 13756

Structural Revision of Mucoxin
FULL PAPER
segregated arrangement, whereby H15a enjoys larger de-
shielding as the result of the diamagnetic anisotropical
effect of the substituents on the THF ring.
À
4) C12 C13 relationship: The final discernable trend was
observed for the splitting of H13 as a function of the rel-
À
ative configuration of C12 C13. As illustrated in Table 8,
12,13-threo isomers yield a dd splitting pattern (or a trip-
let with identical J1 and J2) for H13 with two similar cou-
pling constants in the range of 2.4–3.9 Hz. On the other
hand, the erythro isomers exhibit dd signals with a large
J1 (greater than 6.5 Hz) and a small J2 (less than 3.5 Hz).
The only exception to this rule was observed with ana-
logue I, which exhibits a dd with a large and small cou-
pling for H13. Considering the molecular modeling anal-
ysis presented in Table 2 on a similar system, it is not sur-
prising that exceptions to the rule are observed, since the
coupling constant is highly dependent on the dihedral
angle. A number of mitigating factors control the overall
favored population of an ensemble of rotameric possibili-
ties, and as such, analogue I seems to defy the general
trend that would dictate two small J values. Nevertheless,
since most of the analogues fit the classification as de-
picted in Table 8, we proceeded to include H13 coupling
as a defining characteristic for elucidation of mucoxinꢁs
structure, keeping in mind that the relative configuration
of H13 might have to be
À
Scheme 4. Energy minimized structures of C14 C16 cis and trans model
compounds (Spartan 2008, PM3). The hydrogen atoms are omitted for
clarity except for H15a and H15b. The larger environmental discrepancy
in the cis-isomer leads to the larger chemical shift difference between
H15a and H15b.
both the cis and trans ring systems. As depicted in
Scheme 4, the C14 C16-trans isomer leads to a more bal-
anced distribution of co-parallel C C bonds (major
factor that leads to diamagnetic anisotropical deshielding
in saturated ring systems) with respect to the pseudo ax-
ially and pseudo equatorially disposed hydrogen atoms
on C15. On the other hand, the C14 C16-cis isomer
places the two hydrogen atoms in a more structurally
À
À
À
probed further if a structural
Table 8. The splitting pattern and the coupling constant of H13 for analogues A–L (NMR spectra recorded in
match to the available spec-
tral data for mucoxin was
not found. The observed
coupling of H13 in mucoxin
CDCl3).
threo
erythro
H13
A
B
C
E
G
H
I[a]
K
D
F
L
J
splitting
t
dd
3.4
3.9
t
t
dd
2.3
2.9
dd
2.4
3.9
dd
2.7
6.9
dd
3.4
4.8
dd
3.1
7.2
dd
3.4
6.8
dd
2.9
7.3
dd
3.3
7.2
is a triplet with J1 ꢀ J2
ꢀ
J [Hz]
3.3
3.9
3.5
3.0 Hz, thus suggesting
a
À
threo relationship for C12
C13.
[a] The data for analogue I does not fit the trend observed for the rest of the threo isomers. The source of this
discrepancy is discussed in the text.
Considering the four conclu-
sions derived above through
analysis of the truncated ana-
Table 9. 1H and 13C NMR comparison of ttcTTT and cccTTT truncated analogues and cccTTT full length ace-
togenin with naturally isolated mucoxin (NMR spectra recorded in CDCl3).
À
logues of mucoxin: C8 C9
À
À
threo, C16 C17 threo, C14 C16
À
cis, C12 C13 threo, the field of
isomeric possibilities is nar-
rowed from 64 total diastereo-
meric choices to only 4. These
ttcTTT M (truncated)
cccTTT N (truncated)
cccTTT (10)
À
C H
Dd H
Dd C
Dd H
Dd C
Dd H
Dd C
isomers
are
italicized
in
8
9
À0.05
+0.9
À0.8
–
0.00
À0.1
+0.1
–
–
0.0
+0.2
+0.1
+0.8[a]
+0.2
+0.1
0.00
0.00
+0.02, +0.01
+0.02, +0.01
0.00
+0.04
0.00
+0.02, 0.00
+0.01
+0.01
+0.1
+0.1
–
–
0.0
+0.2
+0.1
+0.8[a]
+0.2
À0.2
À0.09
À0.02
Figure 1. As luck would have it,
two of the four possibilities (H
and K) were synthesized as part
of the 12 diastereomers chosen
for spectral analysis, thus fur-
ther narrowing the field to the
remaining 2 isomers (ttcTTT M
and cccTTT N, see Figure 1,
bold and italicized). Both of the
latter truncated isomers were
synthesized (see Supporting In-
10
11
12
13
14
15
16
17
À0.07, À0.10
0.04, À0.10
À0.31
0.00, 0.00
–
0.00, 0.00
0.00
+0.03
À0.01
0.00, 0.00
+0.8
+5.1
+1.2
À0.1
+1.7
+1.0
+0.17
À0.05
À0.07, +0.09
+0.04
0.00
0.00
À0.01
[a] The reported deviation is based on tabulated data from the original isolation and characterization report,
however, close inspection of the original NMR spectrum suggests that the listed data for C15 was transcribed
incorrectly by ~1 ppm, and in actuality the deviation between the NMR spectrum of cccTTT and the original
data is less than 0.2 ppm.
Chem. Eur. J. 2010, 16, 13749 – 13756
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
13755

B. Borhan, C. Vasileiou et al.
formation for synthetic scheme), and gratifyingly, the
cccTTT analogue N was found to be a perfect match
(Table 9). Further verification of this was obtained via the
synthesis of the full-length cccTTT molecule (10), which
Acknowledgements
Generous support was provided by the NIH (No. R01-GM082961).
1
produced identical H and 13C NMR spectra as compared to
[1] R. S. Narayan, B. Borhan, J. Org. Chem. 2006, 71, 1416–1429.
[2] T. R. Hoye, Z. P. Zhuang, J. Org. Chem. 1988, 53, 5578–5580.
[3] T. R. Hoye, J. C. Suhadolnik, J. Am. Chem. Soc. 1987, 109, 4402–
4403.
[4] D. P. Curran, Q. S. Zhang, H. J. Lu, V. Gudipati, J. Am. Chem. Soc.
2006, 128, 9943–9956.
[5] D. P. Curran, Q. S. Zhang, C. Richard, H. J. Lu, V. Gudipati, C. S.
Wilcox, J. Am. Chem. Soc. 2006, 128, 9561–9573.
[6] M. Carmen Zafra-Polo, M. C. Gonzalez, E. Estornell, S. Sahpaz, D.
Cortes, Phytochemistry 1996, 42, 253–271.
reported mucoxin. It is noteworthy that upon spectroscopic
analysis of the additional two truncated structures ttcTTT
(M) and cccTTT (N), the four trends derived from the origi-
nal 12 diastereomers also fit these structures.
In the final analysis, our results indicate that the original
À
trans stereochemical relationship for both rings (C9 C12
À
and C13 C16) is the source of error in the structure deter-
mination (trans stereochemistry of the ring substituents was
assumed from lack of NOE). This clearly demonstrates the
fallibility of using negative NOE results to assign relative
stereochemistries, especially in five-membered rings (pseudo
axial and pseudo equatorial disposition of substituents) if
both cis and trans isomers are not available for comparative
analysis. These problems are not isolated in structure deter-
mination, but also pose a challenge to synthetic organic
chemists in assigning stereochemistry of synthetic products.
The data also point to the fact that empirical results can be
of great value in stereochemical assignments since they can
be used as a guide and/or secondary verification for structur-
al elucidations.
[7] M. C. Zafra-Polo, B. Figadere, T. Gallardo, J. Tormo, D. Cortes, Phy-
tochemistry 1998, 48, 1087–1117.
[8] L. Zeng, Q. Ye, N. H. Oberlies, G. Shi, Z.-M. Gu, K. He, J. L.
McLaughlin, Nat. Prod. Rep. 1996, 13, 275–306.
[9] A. Cavꢃ, B. Figadꢄre, A. Laurens, D. Cortes, Prog. Chem. Org. Nat.
Prod. 1997, 70, 81–288.
[10] F. Q. Alali, X. X. Liu, J. L. McLaughlin, J. Nat. Prod. 1999, 62, 504–
540.
[11] J. R. Tormo, T. Gallardo, M. C. Gonzꢅlez, A. Bermejo, N. Cabedo, I.
Andreu, E. Estornell, Curr. Top. Phytochem. 1999, 2, 69–90.
[12] A. Bermejo, B. Figadere, M.-C. Zafra-Polo, I. Barrachina, E. Estor-
nell, D. Cortes, Nat. Prod. Rep. 2005, 22, 269–303.
[13] J. K. Rupprecht, Y.-H. Hui, J. L. McLaughlin, J. Nat. Prod. 1990, 53,
237–278.
[14] N. Maezaki, N. Kojima, T. Tanaka, Synlett 2006, 993–1003.
[15] G. Shi, J. F. Kozlowski, J. T. Schwedler, K. V. Wood, J. M. MacDou-
gal, J. L. McLaughlin, J. Org. Chem. 1996, 61, 7988–7989.
[16] S. Higashibayashi, W. Czechtizky, Y. Kobayashi, Y. Kishi, J. Am.
Chem. Soc. 2003, 125, 14379–14393.
Conclusion
[17] Y. Kobayashi, C. H. Tan, Y. Kishi, J. Am. Chem. Soc. 2001, 123,
2076–2078.
In summary, we have demonstrated a practical approach to-
wards solving stereochemical misassignments for molecules
that contain multiple chiral centers. In this “minimalist” ap-
proach, the initial goal is to generate spectroscopic data that
can correlate structural relationships with observable trends.
The choice of test structures is critical so that erroneous
trends are avoided as a result of having a compendium of
different stereochemical relationships besides the pair that
is being probed. In this manner, we were able to use 12 of
64 possible diastereomers to derive four trends, which re-
duced the field of possibilities to four isomers. The spectro-
scopic data for one of the four isomers matched the report-
ed data for mucoxin.
[18] Y. Kobayashi, C. H. Tan, Y. Kishi, Angew. Chem. 2000, 112, 4449–
4451; Angew. Chem. Int. Ed. 2000, 39, 4279–4281.
[19] Y. Kobayashi, J. Lee, K. Tezuka, Y. Kishi, Org. Lett. 1999, 1, 2177–
2180.
[20] J. Lee, Y. Kobayashi, K. Tezuka, Y. Kishi, Org. Lett. 1999, 1, 2181–
2184.
[21] R. S. Narayan, M. Sivakumar, E. Bouhlel, B. Borhan, Org. Lett.
2001, 3, 2489–2492.
[22] T. Zheng, R. S. Narayan, J. M. Schomaker, B. Borhan, J. Am. Chem.
Soc. 2005, 127, 6946–6947.
[23] L. Born, F. J. Lieb, P. Lorentzen, H. Moeschler, M. Nonfon, R. Soll-
ner, D. Wendisch, Planta Medica 1990, 56, 312–316.
Received: June 8, 2010
Published online: November 18, 2010
13756
www.chemeurj.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 13749 – 13756