Despite numerous advances in spectroscopic methods through the latter part of the 20th century, the unequivocal structure determination of natural products can remain challenging, and inevitably, incorrect structures appear in the literature. Computational methods that allow the accurate prediction of NMR chemical shifts have emerged as a powerful addition to the toolbox of methods available for the structure determination of small organic molecules. Herein, we report the structure determination of a small, stereochemically rich natural product from Laurencia majuscula using the powerful combination of computational methods and total synthesis, along with the structure confirmation of notoryne, using the same approach. Additionally, we synthesized three further diastereomers of the L. majuscula enyne and have demonstrated that computations are able to distinguish each of the four synthetic diastereomers from the 32 possible diastereomers of the natural product. Key to the success of this work is to analyze the computational data to provide the greatest distinction between each diastereomer, by identifying chemical shifts that are most sensitive to changes in relative stereochemistry. The success of the computational methods in the structure determination of stereochemically rich, flexible organic molecules will allow all involved in structure determination to use these methods with confidence.
Despite numerous advances in spectroscopic methods through the latter part of the 20th century, the unequivocal structure determination of natural products can remain challenging, and inevitably, incorrect structures appear in the literature. Computational methods that allow the accurate prediction of NMR chemical shifts have emerged as a powerful addition to the toolbox of methods available for the structure determination of small organic molecules. Herein, we report the structure determination of a small, stereochemically rich natural product from Laurencia majuscula using the powerful combination of computational methods and total synthesis, along with the structure confirmation of notoryne, using the same approach. Additionally, we synthesized three further diastereomers of the L. majusculaenyne and have demonstrated that computations are able to distinguish each of the four synthetic diastereomers from the 32 possible diastereomers of the natural product. Key to the success of this work is to analyze the computational data to provide the greatest distinction between each diastereomer, by identifying chemical shifts that are most sensitive to changes in relative stereochemistry. The success of the computational methods in the structure determination of stereochemically rich, flexible organic molecules will allow all involved in structure determination to use these methods with confidence.
Elucidating the structures
of natural products that are available
only in very small quantities is often exceptionally difficult, highlighted
by the high number of structure revisions reported every year in the
chemical literature.[1] Unequivocal establishment
of molecular structure may be possible using single-crystal X-ray
diffraction; however, for molecules that will not crystallize or that
form poorly diffracting crystals, alternative methods must be used.[2] Nuclear magnetic resonance (NMR) spectroscopy
remains one of the primary means of molecular structure determination.
Comparison of computationally predicted NMR chemical shifts, coupling
constants, and internuclear distances have also emerged as a powerful
and reliable way to assess the likelihood of a putative structure
being correct. NMR computations have thus been used to establish relative
stereochemistry,[3] to confirm or reassign
proposed natural product structures,[4] to
characterize the identity of a side product,[5] and in conformational assignment of cyclic peptides.[6] However, this approach can become particularly challenging
for molecules containing multiple stereogenic centers and with a high
degree of conformational flexibility. In such cases, weighted ensemble
averages must be considered in comparison against experimentally observed
NMR parameters. In the case of complex structures such as baulamycin,
hundreds and thousands of conformations may be relevant at room temperature.[7] Although elegant synthetic approaches have been
developed to access multiple diastereomers of a target molecule,[8] the ability to predict the most plausible relative
and absolute configuration of flexible natural product structures
is desirable. Flexible, halogenated natural products epitomize this
challenge and have been variously misassigned.[9]Comparison of experimental NMR parameters, such as chemical
shifts,
against calculated values can reveal obvious discrepancies and raise
doubts about a proposed structure.[4a] Furthermore,
several measures have been used to quantify the “best match”
from several putative structures with respect to an experimental spectrum.
In particular, statistical parameters developed in the Goodman group,
namely, CP3[10] and DP4,[11] use prior knowledge of the underlying empirical error distribution
of computational predictions to assign statistical confidence values
to a particular structural assignment. This has led to their widespread
adoption in computational natural product assignments.[12] These metrics emphasize the importance of ensuring
a narrow underlying prediction error as this allows greater significance
to be attached to the differences between incorrect structures and
experiment. More confident assignments can be made as a result. Empirical
linear scaling, aided greatly through contributions from the Tantillo
group and the CHESHIRE repository,[13] has
contributed to this as have modified DP4 models and the use of different
internal standards against which chemical shifts are referenced.[14] However, molecules with many accessible conformers
can give rise to larger prediction errors,[3] making the use of these now standard tools more precarious. Herein,
we report the full structure determination of a 2,2′-bifuranyl
chloroenyne natural product isolated from Laurencia
majuscula (2) using the powerful combination
of biosynthetic postulates, density functional theory (DFT) calculations
of NMR chemical shifts, and total synthesis. In addition, we report
the synthesis of all four biosynthetically relevant diastereomers
of 2 (2a–d) and show
that comparison of the experimental 1H and 13C NMR data for any of these four diastereomers with the computed 1H and 13C NMR data for all 32 diastereomers of
the natural product allows each specific biomimetic diastereomer to
be identified. Finally, we report the structure confirmation of (Z)-notoryne as (Z)-3a using the above combined approach of quantum chemical calculations
coupled with two distinct total syntheses.Recently, we assigned
the full structures of two 2,2′-bifuranyl
natural products using the powerful combination of biosynthetic postulates,
DFT calculations of NMR chemical shifts, and total synthesis. We used
this combined approach to fully elucidate the structure of elatenyne[15] and to reassign the stereostructure of laurefurenynes
A and B.[16] Previously, we demonstrated
that the gross structure of a chloroenyne isolated from Laurencia majuscula, originally assigned as a pyrano[3,2-b]pyran (1)[17] on
the basis of extensive NMR experiments and comparison with the NMR
data of the dactomelynes[18] and of the originally
assigned pyrano[3,2-b]pyran structure of elatenyne,[19] was actually a 2,2′-bifuranyl 2 (Chart ).[15a,15b] The 2,2′-bifuranyl 2 contains six stereocenters,
with the full structure of the natural product being one of 32 possible
diastereomers. We aimed to solve the complete stereostructure of this
natural product using the combined approach which we had previously
found successful, namely, biosynthetic postulates coupled with DFT
calculations of NMR chemical shifts and total synthesis.
Chart 1
Proposed
Structures of Chloroenyne from L. majuscula (Relative Configurations) along with the Structures of (Z)- and (E)-Notoryne (Absolute Configurations)
Clues from Biosynthesis
Algae of
the genus Laurencia produce a vast
array of structurally diverse
C15 halogenated natural products. Elatenyne,[19,20] notoryne (3),[21] laurendecumenyne
B,[20,22] laurefurenynes A and B,[23] and the above chloroenyne from L. majuscula2(17) are currently the only
2,2′-bifuranyl natural products that have been isolated from Laurencia spp. Notoryne 3 was the first
of these 2,2′-bifuranyls to be isolated, and its structure
was determined through extensive chemical degradation and by comparison
with chemical degradation products of laurencin.[21a] A plausible biosynthesis of notoryne 3a was
proposed by Suzuki and by Fukuzawa and Murai.[21a,24] For (3Z)-notoryne (Z)-3a, (3Z,12E,R,R)-laurediol 4a(25) is converted into (3Z)-deacetyllaurencin 6a via bromonium ion formation and cyclization (Figure a).[26]
Figure 1
(a)
Plausible biosynthesis of notoryne 3a as proposed
by Suzuki and by Fukuzawa and Murai along with the proposed biosynthesis
of diastereomer 2a of the chloroenyne from L. majuscula. (b) Proposed biosynthesis of diastereomer 2b of the chloroenyne from L. majuscula via the natural products bromofucin 8b and laurendecumeyne
B 3b. (c) Proposed stereostructures 2c and 2d of the chloroenyne from L. majuscula derived from the (12Z)-laurediols 4c and 4d.
(a)
Plausible biosynthesis of notoryne 3a as proposed
by Suzuki and by Fukuzawa and Murai along with the proposed biosynthesis
of diastereomer 2a of the chloroenyne from L. majuscula. (b) Proposed biosynthesis of diastereomer 2b of the chloroenyne from L. majuscula via the natural products bromofucin 8b and laurendecumeyne
B 3b. (c) Proposed stereostructures 2c and 2d of the chloroenyne from L. majuscula derived from the (12Z)-laurediols 4c and 4d.Further bromoetherification of (3Z)-deacetyllaurencin 6a gives the (3Z)-dibromide 8a, prelaurefucin, via 7a. Transannular displacement of
bromide gives (3Z)-tricyclic oxonium ion 9a, which on opening at C-7 with chloride gives (3Z)-notoryne (3Z)-3a. As we previously
proposed for the biosynthesis of laurefurenynes A and B,[16a] displacement of bromide by water (or a water
equivalent)[27] would give rise to 2a as a potential stereostructure for the chloroenyne from L. majuscula. Based on this proposed biosynthesis
of notoryne 3a, we recently proposed a biosynthesis of
elatenyne, laurendecumenyne B,[15d] and laurefurenynes
A and B,[16a] which begins from a different
diastereomer of the laurediols (4b)[25] and proceeds via the natural product bromofucin 8b.[28] Here, laurediol 4b undergoes
bromoetherification to give a diastereomer of deacetyllaurencin 6b, which undergoes further bromoetherification to give bromofucin 8b (Figure b). Transannular displacement of bromide gives the oxonium ion 9b, which on C-7 opening with chloride gives the notoryne
diastereomer 3b, (3Z)-laurendecumenyne
B. Displacement of bromide by water (or a water equivalent)[27] gives diastereomer 2b of the chloroenyne
from L. majuscula. The laurediols exist
naturally as unequal mixtures of (R,R), (S,S), (3E),
(3Z), (12E), and (12Z)-diastereomers.[25] The (3E) and (12Z)-laurediols 4c and 4d are therefore also plausible starting points for the biosynthesis
of the chloroenyne 2, leading to stereostructures 2c and 2d (Figure c).[29] Based on our biosynthetic
analysis, the structure of the chloroenyne from L.
majuscula is plausibly, therefore, one of the four
diastereomers 2a, 2b, 2c, or 2d, each of which could be produced from the above biogenetic
schemes.[30] We then used quantum chemistry
to predict the most likely stereostructure for 2, based
on computed 13C and 1H chemical shifts for each
of the 32 possible diastereomers.[31]
Results
and Discussion
Computational Prediction
The specific
challenges associated
with the computational structural assignment of chloroenyne 2 (and related molecules) influenced the computational methods
employed. First, each diastereomer is highly flexible. A Monte Carlo
conformational search with the Merck Molecular Force Field (MMFF)
was used to obtain the low-energy conformations within 10 kJ/mol of
the lowest-energy structure of each diastereomer.[32] Across all of the diastereomers, there are 1277 conformers
in this energy range that contribute to the predicted Boltzmann-weighted
chemical shifts! Furthermore, the level of theory used for geometry
optimization influences both the estimated populations and the computed
chemical shifts of each conformation. Although MMFF geometries have
been used in structure prediction,[31c,33] DFT optimization
allows for more confident structural assignment as there is a narrower
distribution of errors with respect to experimental chemical shifts.[16a,31c] As an illustration, for 82 molecules with 709 experimentally assigned 13C chemical shifts, we verified that DFT optimizations led
to a 2 ppm reduction in root-mean-square deviation (rmsd) compared
to MMFF geometries, even though both sets of calculations used the
same level of theory for shielding tensor calculation and empirical
scaling (Figure S1). In this work, we optimized
all structures with dispersion-corrected DFT, at the wB97XD/6-31G(d)
level of theory[34] with CPCM (Conductor-like
Polarizable Continuum Model) chloroform.[35] Conformer relative energies were checked against COSMO–DLPNO–CCSD(T)/cc-pVTZ
single-point energies[36] for one diastereomer
and showed a good level of agreement (R2 = 0.90, rmsd = 3.1 kJ/mol) against this high accuracy method (Figure S2). Manual data processing is prohibitively
difficult for so many conformations; a Python program was developed
to automate all analysis given a collection of output files and a
text file with experimental chemical shifts. Conformational Boltzmann
weighting, empirical scaling, symmetry averaging, consideration of
alternative assignments, and calculation of rmsd/MAD and DP4 for all
structures are fully automated (Supporting Information shows example usage). We note that the DP4 workflow has now been
automated (pyDP4) by Ermanis and Goodman.[12]Halogenated natural products pose a further challenge due
to the so-called heavy-atom light-atom (HALA) effect.[37] Relativistic spin–orbit contributions shift carbon
atoms bonded to Cl, Br, or I to lower parts per million. In this work,
we used the CHESHIRE database test set of molecules developed by Rablen,
Bally, and Tantillo[13] to derive new, optimal
scaling parameters for mPW1PW91/6-311G(d,p)//wB97XD/6-31G(d) GIAO
shielding tensors for 13C and 1H nuclei (Figure S3). Several chlorine-containing compounds
appear in this data set, from which we found an additive correction
of 7.6 ppm applied to the shielding tensors of C–Cl atoms results
in an rmsd no worse than if these atoms are excluded entirely. Such
a correction (which is level of theory dependent) has also been used
by Rzepa and Braddock.[38]We analyzed
the Boltzmann-weighted 13C and 1H chemical shifts
for all 32 diastereomers at every position except
the hydroxyl proton. The variability in computed shifts across the
entire set of diastereomers (Figure ) shows the extent to which each nucleus acts as a
reporter for stereochemical changes. Larger standard deviations are
obtained for positions which are inherently more sensitive to their
relative stereochemistry. Stereostructure assignment can be made more
confidently when these values are large in relation to the inherent
accuracy of the computed chemical shifts versus experiment (i.e.,
by ensuring a higher signal-to-noise ratio).
Figure 2
Standard deviations of
computed 13C and 1H chemical shifts at each
position over all 32 diastereomers of 2. These values,
illustrated by circle size, show the sensitivity
of each shift to stereochemical changes. Values in black were excluded
from experimental comparison.
Standard deviations of
computed 13C and 1H chemical shifts at each
position over all 32 diastereomers of 2. These values,
illustrated by circle size, show the sensitivity
of each shift to stereochemical changes. Values in black were excluded
from experimental comparison.Based on the work of Smith and Goodman,[11] standard deviations of 2.3 and 0.19 ppm for errors in 13C and 1H chemical shifts are representative of
the underlying
computational accuracy, although because previous calculations used
MM geometries, these values are likely to be pessimistic for the DFT
optimizations used here. For chloroenyne 2, a handful
of nuclei have low variability and are thus very poor reporters of
stereochemical information. Consistent with chemical intuition, these
are exocyclic positions remote from stereocenters. These nuclei were
excluded from further analysis as they contribute minimally to stereochemical
assignment. The inclusion of the halogenated carbon atom is important
as this is the best stereochemical reporter from all of the 13C shifts. This is important because halogenated carbons have been
omitted previously from structural prediction due to the aforementioned
challenges associated with the HALA.[15c] The ethyl 13CH2 also provides one of the more
diagnostic values. There are several protons which show a standard
deviation larger than 0.19 ppm. This is consistent with the idea that
structural differences in 1H chemical shifts are more diagnostic
than 13C in relation to the underlying theoretical accuracy.[39] We compared predicted chemical shifts for each
diastereomer against those of the natural product. The 13C spectrum was fully assigned, whereas for three pairs of protons,
the best possible (lowest rmsd) assignment was generated automatically
for each structure. The rmsd values and DP4 metrics were generated
using 10 13C and 14 1H chemical shifts (Figure ). If one assumes
that the underlying error distribution of n chemical
shifts is Gaussian (in its original formulation, the DP4 metric assumes
a t-distribution, although a Gaussian distribution
was also proposed), the sum of squared errors obeys a χ2-distribution with n degrees of freedom.
The rmsd values can therefore be interpreted in a probabilistic fashion
as is done for DP4: for example, incorrect structures can be rejected
at the p = 0.05 significance level, where the rmsd
falls above a critical value (Figure ). Importantly, the statistical significance of differences
between rmsd values is intrinsically linked to the number of chemical
shifts being compared. Although rmsd, MAE, and R2 measures have been used previously to compare the relative
likelihood of structures being correct, the statistical confidence
of these measures has not received much attention. Here, in addition
to DP4 values, we show the 95% confidence intervals for rmsd values,
above which structures are deemed to be unlikely candidates.
Figure 3
Comparison
of computed 13C and 1H chemical
shifts for 32 diastereomers of 2 against the natural
product. The smallest rmsd values and largest DP4 probabilities point
to biosynthetically predicted compound 2b.
Comparison
of computed 13C and 1H chemical
shifts for 32 diastereomers of 2 against the natural
product. The smallest rmsd values and largest DP4 probabilities point
to biosynthetically predicted compound 2b.We found that one of the four biogenetically plausible
structures
(2b) was the best match for 13C and 1H experimental data, giving the smallest rmsd and largest DP4* values.
The convergence of all four metrics, combined with biosynthetic arguments,
overwhelmingly suggests the identity of the chloroenyne from L. majuscula as 2b. In using the original t-distribution parameters (σ, υ) derived for
MMFF-optimized geometries, we are being deliberately conservative:
we expect a narrower error distribution using DFT optimizations that
would penalize the less likely structures more severely. DP4 relies
on individually scaling each structure against experiment, which can
lead to a fortuitous improvement of incorrect structures, particularly
for a small number of nuclei. Indeed, we found better predictive performance
without individually scaling the 13C shifts (indicated
as DP4*).All of our biosynthetic and computational analysis
provided compelling
evidence that the correct structure for the chloroenyne from L. majuscula was represented by 2b;
however, proof of this could only come through total synthesis.[1,2b] Additionally, we decided to synthesize all four biosynthetically
relevant diastereomers of the chloroenyne from L. majuscula, which would allow us to obtain NMR data for each diastereomer.
With these data in hand, we could further test the computational methods;
would it be possible computationally to correctly identify each of
the four biosynthetically predicted diastereomers (2a–d) from the computed data of the 32 possible
diastereomers of the chloroenyne from L. majuscula?
Synthesis
The proposed synthesis of the four biomimetically
plausible diastereomers of the chloroenyne form L.
majuscula (2a–d)
presented us with the opportunity to modify and improve our modular
route to 2,2′-bifuranyl natural products.[15d] The retrosynthetic synthetic route to diastereomers 2a and 2c is shown in Figure . Enynes 2a and 2c were to be readily derived from the 2,2′-bifuranyl 10, which itself was to be derived from 11 and
subsequently from diol 12 following a route analogous
to that described in our recent synthesis of elatenyne.[15d] Diol 12 was to be derived from
alkene 13, which was to be constructed by Julia–Kocienski
olefination of aldehyde 15 with sulfone 14.[40] The two coupling partners 14 and 15 were to be derived from protection of the known
enantiomeric epoxy alkenes (+)-16 and (−)-16. In the forward direction (Scheme ), the known benzyl-protected epoxy alkene 17, prepared according to the method of Crimmins,[41] was converted into alcohol 18 by
ozonolysis with in situ reduction (PPh3 then NaBH4). Alcohol 18 was then converted into the corresponding
tetrazole sulfone 14 by Mitsunobu reaction followed by
oxidation.[40] Aldehyde 15 was
readily prepared from alcohol (+)-16, which on PMB protection
gave ether 19.[15d] Copper-catalyzed
ring opening of epoxide 19 with methylmagnesium bromide[41] followed by silyl protection gave alkene 20. Ozonolysis of alkene 20 and reductive phosphine
workup gave the required aldehyde 15. Julia–Kocienski[40,42] coupling of sulfone tetrazole 14 with aldehyde 15 proceeded in good yield with >20:1 E/Z-selectivity to give alkene 13.[43] Sharpless asymmetric dihydroxylation[44] of 13 using super-AD-mix β[45] gave the corresponding diols in 91% yield as
a 6:1 mixture of syn-diastereomers; diol 12 could be isolated in pure form in 64% yield. Under acid catalysis,
diol 12 underwent cyclization to give THF 21. Formation of the second THF ring was achieved using a three-step
procedure. Exposure of diol 21 to mesyl chloride and
Hünig’s base gave dimesylate 23 which,
without purification, was treated with (±)-10-camphorsulfonic
acid to remove the silyl-protecting group. Exposure of the resulting
alcohol to potassium tert-butoxide gave 2,2′-bifuranyl 24 in 52% yield over three steps.
Figure 4
Retrosynthetic analysis
of diastereomers 2a and 2c.
Scheme 1
Synthesis of Chloride 28
Reagents and conditions: (a)
O3/O2, CH2Cl2, MeOH, −78
°C, then PPh3, 30 min, then NaBH4, −78
°C to rt, 2 h, 92%; (b) DIAD, PPh3, THF, 0 °C,
then 18, then 1-phenyl-1H-tetrazole-5-thiol,
rt, 16 h, 88%; (c) 3-chloroperbenzoic acid, CH2Cl2, rt, 72 h, 60%; (d) NaH, PMBBr, Bu4NI, THF, −78
°C to rt, 16 h; (e) MeMgBr, CuI, THF, −20 °C, 1 h,
88% (two steps); (f) TBSCl, imidazole, DMF, 60 °C, 16 h, 99%;
(g) O3/O2, CH2Cl2, −78
°C, then PPh3, −78 °C to rt, 16 h, 98%;
(h) 14, DME, NaHMDS, −78 °C, 15 min, then
add 15 in DME, −78 °C, 1 h, then warm to
rt, 2 h, 80%; (i) K2OsO4(OH)4, K3Fe(CN)6, (DHQD)2PHAL, K2CO3, MeSO2NH2, tBuOH,
water, rt, 24 h, 64% pure 12 (6:1 mixture of syn-diol diastereomers 91% total); (j) (±)-10-camphorsulfonic
acid, CH2Cl2, 0 °C, 2 h, 86%; (k) MsCl, i-Pr2NEt, CH2Cl2, 0 °C,
1 h; (l) (±)-10-camphorsulfonic acid, CH2Cl2, MeOH, rt, 24 h; (m) t-BuOK, tBuOH, 35 °C, 52% (three steps); (n) Bu4NI, toluene,
110 °C, 16 h, 80%; (o) CH2=CHMgBr, benzene,
THF, 40 °C, 3 h, 57%; (p) Li, 4,4′-di-tert-butyl-1,1′-biphenyl, THF, −78 °C, 77%; (q) CCl4, PPh3, CH2Cl2, rt, 3 h,
78%.
Retrosynthetic analysis
of diastereomers 2a and 2c.
Synthesis of Chloride 28
Reagents and conditions: (a)
O3/O2, CH2Cl2, MeOH, −78
°C, then PPh3, 30 min, then NaBH4, −78
°C to rt, 2 h, 92%; (b) DIAD, PPh3, THF, 0 °C,
then 18, then 1-phenyl-1H-tetrazole-5-thiol,
rt, 16 h, 88%; (c) 3-chloroperbenzoic acid, CH2Cl2, rt, 72 h, 60%; (d) NaH, PMBBr, Bu4NI, THF, −78
°C to rt, 16 h; (e) MeMgBr, CuI, THF, −20 °C, 1 h,
88% (two steps); (f) TBSCl, imidazole, DMF, 60 °C, 16 h, 99%;
(g) O3/O2, CH2Cl2, −78
°C, then PPh3, −78 °C to rt, 16 h, 98%;
(h) 14, DME, NaHMDS, −78 °C, 15 min, then
add 15 in DME, −78 °C, 1 h, then warm to
rt, 2 h, 80%; (i) K2OsO4(OH)4, K3Fe(CN)6, (DHQD)2PHAL, K2CO3, MeSO2NH2, tBuOH,
water, rt, 24 h, 64% pure 12 (6:1 mixture of syn-diol diastereomers 91% total); (j) (±)-10-camphorsulfonic
acid, CH2Cl2, 0 °C, 2 h, 86%; (k) MsCl, i-Pr2NEt, CH2Cl2, 0 °C,
1 h; (l) (±)-10-camphorsulfonic acid, CH2Cl2, MeOH, rt, 24 h; (m) t-BuOK, tBuOH, 35 °C, 52% (three steps); (n) Bu4NI, toluene,
110 °C, 16 h, 80%; (o) CH2=CHMgBr, benzene,
THF, 40 °C, 3 h, 57%; (p) Li, 4,4′-di-tert-butyl-1,1′-biphenyl, THF, −78 °C, 77%; (q) CCl4, PPh3, CH2Cl2, rt, 3 h,
78%.During optimization studies, the monomesylate 22 was
isolated and characterized. Mosher ester formation using 22 allowed the sense of the Sharpless asymmetric dihydroxylation reaction
to be confirmed.[46] As with the synthesis
of elatenyne,[15d] forcing conditions (Bu4NI, toluene, reflux) were required to convert the mesylate 24 into the corresponding iodide 25. Iodide 25 underwent displacement with vinylmagnesium bromide in a
mixed benzene/THF solvent system to give the allyl-substituted 2,2′-bifuranyl 11 in 57% yield.[47] Deprotection
of the benzyl group in 11 in the presence of the PMB
group was achieved by titrating LiDBB[48] into a THF solution of 11 to minimize formation of
diol 27. Attempted chlorination of alcohol 26 using triphenylphosphine with carbon tetrachloride as both reagent
and solvent led to very little conversion of 26 into
chloride 28.[49] Appel reported
a large solvent effect on the chlorination of alcohols using tertiary
phosphines and carbon tetrachloride with the use of dichloromethane
and acetonitrile, resulting in significant rate enhancements.[49] Conducting the chlorination of alcohol 26 using dichloromethane as solvent gave chloride 28 in 78% yield. Conversion of chloride 28 into the biosynthetically
plausible diastereomers 2a and 2c required
introduction of the (E)-enyne (Scheme ). Previously the Oxford group have used
a Wittig reaction with the ylide derived from (3-trimethylsilyl-2-propynyl)triphenylphosphonium
bromide[15a,15b,15d] for the stereoselective
synthesis of (E)-enynes from aldehydes; however,
the diastereocontrol in these reactions was never above 9:1 E/Z. We therefore elected to use the recently
developed methodology for (E)-enyne synthesis from
alkenes reported by the Seoul group, which gives very high E-selectivity.[15d,16a,50] Cross-metathesis of alkene 28 with crotonaldehyde and
Grubbs’ second generation catalyst 29 gave the
corresponding α,β-unsaturated aldehyde 30, which was immediately exposed to lithiated trimethylsilyl diazomethane
(Colvin–Ohira homologation) to give enyne 31.
Removal of the para-methoxy benzyl group[51] from 31 gave 2c, the
first of the enyne targets. Mitsunobu inversion[52] of the secondary alcohol in 2c followed by
ester methanolysis gave 2a, the second of the biomimetically
plausible diastereomers.
Scheme 2
Synthesis of Diastereomers 2a and 2c
Reagents and conditions:
(a)
crotonaldehyde, catalyst 29, CH2Cl2, 40 °C, 1 h then Me2SO, rt, 16 h, 88%; (b) (trimethylsilyl)diazomethane, n-BuLi, THF, add 30, −78 °C to
rt, 1 h 70%; (c) BCl3·SMe2, CH2Cl2, rt, 5 min, 84%; (d) DIAD, PPh3, THF, 0
°C, then 2c, then 4-nitrobenzoic acid, rt; (e) K2CO3, MeOH, rt, 20 min, 25% two steps).
Synthesis of Diastereomers 2a and 2c
Reagents and conditions:
(a)
crotonaldehyde, catalyst 29, CH2Cl2, 40 °C, 1 h then Me2SO, rt, 16 h, 88%; (b) (trimethylsilyl)diazomethane, n-BuLi, THF, add 30, −78 °C to
rt, 1 h 70%; (c) BCl3·SMe2, CH2Cl2, rt, 5 min, 84%; (d) DIAD, PPh3, THF, 0
°C, then 2c, then 4-nitrobenzoic acid, rt; (e) K2CO3, MeOH, rt, 20 min, 25% two steps).The synthesis of the final two biosynthetically plausible
diastereomers
of the chloroenyne from L. majuscula (enynes 2b and 2d) began from the known
2,2′-bifuranyl 32, an intermediate in our recent
synthesis of elatenyne (Scheme ).[15d] Selective removal of the para-bromobenzyl group in the presence of the more electron-rich para-methoxy benzyl group was achieved using LiDBB[48] in the presence of a proton donor,[53] which gave the requisite alcohol 34 in 66% yield along with 18% of the benzyl ether 33.[54] Chlorination of 34 as before[49] provided chloride 35 in 96% yield,
which was readily deprotected under Lewis acidic conditions to give
alcohol 36.[51] Mitsunobu inversion
of 36(52) to give 37, followed by enyne introduction, as before,[15d,50] gave the third biomimetic diastereomer 2b. A further
Mitsunobu reaction gave the fourth and final biomimetic diastereomer 2d.[55]
Scheme 3
Synthesis of Diastereomers 2b and 2d
Reagents
and conditions: (a)
Li, 4,4′-di-tert-butyl-1,1′-biphenyl,
bis(2-methoxyethyl)amine, THF, −78 °C, 18% of 33, 66% of 34; (b) CCl4, PPh3, CH2Cl2, rt, 3 h, 96%; (c) BCl3·SMe2, CH2Cl2, rt, 5 min; 94%; (d) DIAD,
PPh3, THF, 0 °C, then 36, then 4-nitrobenzoic
acid, rt; (e) K2CO3, MeOH, rt, 30 min, 79% (two
steps); (f) crotonaldehyde, catalyst 29, CH2Cl2, 40 °C, 90 min; (g) (trimethylsilyl)diazomethane, n-BuLi, THF, add 37, −78 to 0 °C;
(h) Bu4NF, THF, 0 °C, 5 min, 39% (three steps); (i)
DIAD, PPh3, THF, 0 °C, then 2b, then
4-nitrobenzoic acid, rt; (j) K2CO3, MeOH, rt,
30 min.
Synthesis of Diastereomers 2b and 2d
Reagents
and conditions: (a)
Li, 4,4′-di-tert-butyl-1,1′-biphenyl,
bis(2-methoxyethyl)amine, THF, −78 °C, 18% of 33, 66% of 34; (b) CCl4, PPh3, CH2Cl2, rt, 3 h, 96%; (c) BCl3·SMe2, CH2Cl2, rt, 5 min; 94%; (d) DIAD,
PPh3, THF, 0 °C, then 36, then 4-nitrobenzoic
acid, rt; (e) K2CO3, MeOH, rt, 30 min, 79% (two
steps); (f) crotonaldehyde, catalyst 29, CH2Cl2, 40 °C, 90 min; (g) (trimethylsilyl)diazomethane, n-BuLi, THF, add 37, −78 to 0 °C;
(h) Bu4NF, THF, 0 °C, 5 min, 39% (three steps); (i)
DIAD, PPh3, THF, 0 °C, then 2b, then
4-nitrobenzoic acid, rt; (j) K2CO3, MeOH, rt,
30 min.
Analysis
Having completed the total
synthesis of all
four of the biosynthetically relevant diastereomers of the chloroenyne
from L. majuscula (2a–d), we were delighted to find that the 1H and 13C NMR data for diastereomer 2b were in excellent
agreement with that reported for the natural product.[17] Our synthesis of the chloroenyne from L.
majuscula2b proceeds in 16 steps (longest
linear sequence) from (+)-16. The optical rotation for
our synthetic 2b was in good agreement in terms of both
sign and magnitude with that recorded for the natural product,[17] demonstrating that the absolute configuration
of the chloroenyne from L. majuscula is as represented by 2b. The chloroenyne from L. majuscula2b thus sits on the same
proposed biosynthetic pathway as elatenyne,[15d] laurendecumenyne B,[15d] and laurefurenynes
A and B,[16] proceeding from (3E/Z)-laurediols 4b via the bromofucins 8b.The identification of the full stereostructure of
the chloroenyne from L. majuscula as 2b on the basis of DFT methods demonstrates the power and
utility of these methods to aid in the structure determination of
stereochemically rich organic molecules. The synthesis of the remaining
three biosynthetically relevant diastereomers 2a, 2c, and 2d provided us with the opportunity to
further test the computational methods. We had acquired 1H and 13C NMR data for the chloroenynes 2a, 2c, and 2d, which allowed us to determine
whether it would be possible, computationally and correctly, to identify
each of these diastereomers from the computed data of the 32 possible
diastereomers of the chloroenyne from L. majuscula. We found the correct stereostructures identified among structures
with the smallest rmsd values: the two structures with the lowest
rmsd values contain the correct structure in all but one case (1H of 2d), where it is in the lowest four (Figure ). 1H
DP4 values were more diagnostic, with the correct structure predicted
with 60–98% confidence (compared to 40–55% for 13C). The product of the DP4/DP4* values for both nuclei provide
an unequivocal and, importantly, correct stereostructure prediction
for all four diastereomers studied, including the natural product.
Figure 5
Comparison
of computed 13C and 1H chemical
shifts for 32 diastereomers of 2 against experimental
spectra for 2a, 2c, and 2d.
Comparison
of computed 13C and 1H chemical
shifts for 32 diastereomers of 2 against experimental
spectra for 2a, 2c, and 2d.
Notoryne
Notoryne(Z)-3a is the first halogenated 2,2′-bifuranly
natural product isolated
from Laurencia spp.[56] The structure of (Z)-notoryne (Z)-3a was originally assigned by careful chemical
degradation and comparison with chemical degradation products from
laurefucin and laurencin, whose structures and absolute configurations
were securely established through single-crystal X-ray analysis.[21] As noted above, a biosynthesis of notoryne 3 was first proposed by Suzuki and by Fukuzawa and Murai (Figure a).[21,24] Previously, the Oxford group had prepared the 2,2′-bifuranyl 28 (Scheme ), with the necessary stereochemical arrangement for ready conversion
into notoryne (Z)-3a. Additionally,
the Seoul group had demonstrated a number of biomimetic syntheses
of halogenated natural products from Laurencia spp., including the synthesis of the 2,2′-bifuranyl natural
products (Z)- and (E)-elatenyne[15d] and laurendecumenyne B.[15d] In their synthesis of laurefucin,[50a] the Seoul group had prepared the bromooxocene 39, which
on treatment with N-phenylselenophthalimide (N-PSP) under aqueous acidic conditions gave rise to the
[5.2.1]dioxabicyclic bromide 44 in quantitative yield
(Scheme ); the bromide
was readily converted into the natural product laurefucin 45. The formation of the [5.2.1]dioxabicyclic bromide 44 most likely follows the mechanism indicated. Here, seleniranium
ion formation occurs from the oxocene 39, giving 40, which undergoes ether formation to yield the selenide 41. Activation of the selenide group in 41 by
reaction with further N-PSP gives prelaurefucin surrogate 42. Transannular C–O bond formation then occurs, giving
the key oxonium ion 43. Attack at C-10 by water with
loss of a proton leads to the laurefucin precursor 44 that was readily transformed into the natural product 45. Opening of the oxonium ion 43 at C-7 by chloride with
inversion of configuration would yield the notoryne precursor 46 with the correct absolute configuration for synthesis of
the natural product (Z)-3a. Given the
previous preparation of the 2,2′-bifuranyl 28 and
the oxocene 39, we reasoned that synthesis of the natural
product notoryne could be readily achieved by two independent routes
(Scheme ). The Oxford
group began their synthesis from the previously prepared chloroalcohol 28, which was readily converted into chloroalcohol 10 through deprotection, Mitsunobu inversion, and saponification. Bromination
of alcohol 10 using the Hooz procedure[57] gave bromide 47. The Seoul team prepared the
same bromide beginning with their previously prepared oxocene 39. After extensive experimentation, the Seoul team found
that exposure of the oxocene alcohol 39 to phenylselenyl
chloride in the presence of activated silica gel followed by treatment
of the crude mixture with water in acetonitrile gave the 2,2′-bifuranyl
chloride 46 along with alcohol 44.[58] It is of note that in the absence of silica
gel, the [5.2.1]-bicyclic chloride 49 was formed.[50a] In both the Oxford and Seoul syntheses, the
chlorine-bearing carbon atoms could be identified using 13C NMR chlorine-induced isotopic shift.[59] The Seoul group converted the benzyl-protected alcohol 46 into the corresponding alkene 47 using standard procedures.
Comparison of the 13C NMR chemical shifts of the 2,2′-bifuranyls 47 synthesized in Oxford and Seoul, with the corresponding
chemical shifts for notoryne, indicated that the synthesized material
had the same stereostructure as that of the natural product (Scheme ). Completion of
the synthesis of the natural products was accomplished by two independent
routes. In Oxford, the terminal alkene in 47 was subject
to ozonolysis followed by reductive workup to give the corresponding
aldehyde (uncharacterized) that was subject to a Yamamoto–Petersen
reaction[60] to give the (Z)-enyne 48 with high diastereoselectivity. Removal of
the terminal silyl group was readily achieved on brief exposure of 48 to fluoride to give notoryne (Z)-3a. In Seoul, the (Z)-enyne was introduced
directly from the terminal alkene 47 via a relay cross-metathesis
using enyne 50 and catalyst 51,[15d,50b,61] which gave the desired enyne 52 as a 3:1 mixture of Z/E-enynes in 82% combined yield.
Scheme 4
Key Step in the Seoul Synthesis of
Laurefucin, along with the Proposed
Route to Notoryne
Scheme 5
Oxford and Seoul Syntheses of Notoryne
Reagents
and conditions: (a)
BCl3·SMe2, CH2Cl2, rt, 5 min, 95%; (b) DIAD, PPh3, THF, 0 °C, then
4-nitrobenzoic acid, rt, 74%; (c) K2CO3, MeOH,
rt, 20 min, 91%; (d) CBr4, PPh3, toluene, 80
°C, 75 min, 75%; (e) O3, CH2Cl2, −78 °C then PPh3, −78 °C to
rt, 15 h; (f) TMSC≡CCH2TBS, tBuLi, THF, −78
°C, 1 h, then Ti(OiPr)4, 10 min, then add 47, −78 °C, 30 min, rt, 30 min, 32% (two steps); (g) TBAF,
THF, −20 °C, 5 min, quant.; (h) PhSeCl, n-hexane; (i) PhSeCl (3 equiv), activated silica gel, n-hexane, rt, 72 h; (j) CH3CN/H2O (9:1), rt,
24 h, 80% of 46, 20% of 44; (k) H2, Pd(OH)2/C, EtOH, 1 h, 95%; (l) o-nitrophenylselenocyanide,
(Oct)3P, THF, rt, 10 min, then H2O2, 0 °C to rt, 24 h, 85%; (m) 50, catalyst 51, benzene, 70 °C, 1.5 h, then additional 50 and 51, 82% 3:1 Z/E; (n) TBAF, THF, 0 °C, 1 h, 95%; (o) crotonaldehyde, catalyst 29, CH2Cl2, 40 °C, 1.5 h then Me2SO, rt, 12 h; (p) (trimethylsilyl)diazomethane, LDA, THF,
−78 to 0 °C, 2 h, 88% (two steps). Note: The difference
in the13C NMR chemical shifts between
the synthetic compoundsprepared by the Oxford and Seoul groups and natural notoryne (Z)-are shown adjacent to the
relevant carbon atoms in structures47.
Oxford and Seoul Syntheses of Notoryne
Reagents
and conditions: (a)
BCl3·SMe2, CH2Cl2, rt, 5 min, 95%; (b) DIAD, PPh3, THF, 0 °C, then
4-nitrobenzoic acid, rt, 74%; (c) K2CO3, MeOH,
rt, 20 min, 91%; (d) CBr4, PPh3, toluene, 80
°C, 75 min, 75%; (e) O3, CH2Cl2, −78 °C then PPh3, −78 °C to
rt, 15 h; (f) TMSC≡CCH2TBS, tBuLi, THF, −78
°C, 1 h, then Ti(OiPr)4, 10 min, then add 47, −78 °C, 30 min, rt, 30 min, 32% (two steps); (g) TBAF,
THF, −20 °C, 5 min, quant.; (h) PhSeCl, n-hexane; (i) PhSeCl (3 equiv), activated silica gel, n-hexane, rt, 72 h; (j) CH3CN/H2O (9:1), rt,
24 h, 80% of 46, 20% of 44; (k) H2, Pd(OH)2/C, EtOH, 1 h, 95%; (l) o-nitrophenylselenocyanide,
(Oct)3P, THF, rt, 10 min, then H2O2, 0 °C to rt, 24 h, 85%; (m) 50, catalyst 51, benzene, 70 °C, 1.5 h, then additional 50 and 51, 82% 3:1 Z/E; (n) TBAF, THF, 0 °C, 1 h, 95%; (o) crotonaldehyde, catalyst 29, CH2Cl2, 40 °C, 1.5 h then Me2SO, rt, 12 h; (p) (trimethylsilyl)diazomethane, LDA, THF,
−78 to 0 °C, 2 h, 88% (two steps). Note: The difference
in the13C NMR chemical shifts between
the synthetic compoundsprepared by the Oxford and Seoul groups and natural notoryne (Z)-are shown adjacent to the
relevant carbon atoms in structures47.Fluoride treatment of 52 gave notoryne(Z)-3a. The Oxford and Seoul 1H and 13C NMR data were in excellent agreement with each
other and with the
data reported by Suzuki.[21a,62] Additionally, the optical
rotations of the synthetic materials confirm that the absolute configuration
of the natural product is as represented by (Z)-3a as originally assigned by Suzuki. Additionally, the Seoul
team synthesized (E)-notoryne (E)-3a[21b] from alkene 47. Cross-metathesis of alkene in 47 using crotonaldehyde
and the Grubbs–Hoveyda catalyst 29 gave the corresponding
α,β-unsaturated aldehyde as a single E-isomer,[15d,50] which on Colvin–Ohira
homologation gave (E)-notoryne (E)-3a in 88% overall yield from alkene 47.[21b]Having synthesized notoryne
by two independent routes, we elected
to further test the computational methods for prediction/confirmation
of structure of these halogenated 2,2′-natural products. As
with the previous 2,2′-bifuranyls from Laurencia spp. we have studied, notoryne (Z)-3a contains six stereocenters, resulting in 32 diastereomeric notorynes.
We decided to challenge the computational method to see if it could
predict the correct structure of notoryne from the pool of 32 diastereomeric
notorynes.Upon computing the Boltzmann-weighted chemical shifts
for all 32
diastereomers, we found maximum variance to stereochemical changes
occurs for carbon atoms directly attached to halogen atoms and the
attached protons (Figure ). As with earlier studies, the exocyclic positions offer
little for predictive power and were excluded from further analysis.
The stereostructure for notoryne, (Z)-3a, is clearly favored in the analysis of 13C predictions,
whereas from 1H chemical shifts, it ranks in the top two
structures. Again, the cumulative 1H/13C DP4
metric gives (Z)-3a as the single most
likely stereostructure. The total synthesis and structure confirmation
of notoryne (Z)-3a further demonstrates
the utility of computational methods to not only predict but also
confirm the structures of stereochemically rich, functionalized, and
flexible organic molecules and natural products.
Figure 6
Comparison of computed 13C and 1H chemical
shifts for 32 diastereomers against experimental spectra of notoryne.
Data for (Z)-3a are shown as blue circles
in rmsd and blue area in DP4*.
Comparison of computed 13C and 1H chemical
shifts for 32 diastereomers against experimental spectra of notoryne.
Data for (Z)-3a are shown as blue circles
in rmsd and blue area in DP4*.
Conclusions
We have demonstrated that computational
methods are able to predict
the structure of a highly flexible chloroenyne natural product from L. majuscula containing six stereocenters from the
32 possible diastereomeric structures of the natural product. Moreover,
we have synthesized three further “biomimetic” diastereomers
of the natural product. Using the NMR data of these diastereomers,
we have shown that the same computational methods can identify each
diastereomer out of the set of 32 possible diastereomers. Key to these
computational methods was to use the computed NMR chemical shift data
only for those atoms that are good reporters of stereochemical information
across all 32 diastereomers, that is, those atoms that show a large
standard deviation in computed NMR chemical shift among the diastereomers.
Furthermore, we applied both computational methods and synthesis to
confirm the structure of notoryne, a further halogenated 2,2′-bifuranyl
natural product isolated from Laurencia spp.
Experimental Section
General Procedures
Proton (1H), carbon (13C), and fluorine (19F) NMR spectra were recorded
on a Bruker AV 500 (500/125 MHz), Bruker AV 400 (400/100 MHz), or
Bruker DPX 300 (300/75 MHz) spectrometer. Proton and carbon chemical
shifts (δ) are quoted in parts per million and referenced to
tetramethylsilane with residual protonated solvent as internal standard.
Resonances are described as s (singlet), d (doublet), t (triplet),
q (quartet), m (multiplet), br (broad), dd (double doublet), and so
on. Coupling constants (J) are given in hertz and
are rounded to the nearest 0.1 Hz. H and H′ refer to diastereotopic
protons attached to the same carbon and imply no particular stereochemistry.
All assignments are confimed by 1H–1H
COSY and 1H–13C HSQC experiments. Low-resolution
mass spectra were recorded on a Fisons Platform spectrometer (ES).
High-resolution mass spectra were recorded by the mass spectrometry
staff at the Chemistry Research Laboratory, University of Oxford,
using a Bruker Daltronics microTOF spectrometer (ES) or a Micromass
GCT (FI). The m/z values are reported
in Daltons with their percentage abundances and, where known, the
relevant fragment ions in parentheses. High-resolution values are
calculated to four decimal places from the molecular formula, with
all found values being within a tolerance of 5 ppm. Infrared spectra
were recorded on a Bruker Tensor 27 Fourier transform spectrometer,
as a thin film on diamond ATR. Absorption maxima (νmax) are quoted in wavenumbers (cm–1). Optical rotations
were measured using a PerkinElmer 241 polarimeter in a cell of 1 dm
path length (l). TLC was performed on MerckDC-Alufolien
60F254 0.2 mm precoated plates and visualized using an acidic vanillin
or basic potassium permanganate dip. Retention factors (R) are reported with the solvent system
used in parentheses. Flash column chromatography was performed on
Merck 60 silica (particle size 40–63 μm, pore diameter
60 Å), and the solvent system used is recorded in parentheses.All nonaqueous reactions were carried out in oven-dried glassware
under an inert atmosphere of nitrogen and employing standard techniques
for handling air-sensitive materials. Solvents and commercially available
reagents were dried and purified before use, as appropriate. In particular,
DCM and THF were distilled from CaH2 and stored over 3
Å molecular sieves. “Petrol” refers to the fraction
of light petroleum ether boiling in the range of 40–60 °C
unless otherwise stated. All water used experimentally was distilled,
and the term “brine” refers to a saturated solution
of sodium chloride in water.
Alkene 17 (3 g, 14.7 mmol)
was dissolved in DCM/MeOH (1:1, 200 mL) and the stirred solution cooled
to −78 °C. O2 was sparged through the solution
for 5 min followed by O3/O2 until a faint blue
hue appeared. Then the reaction was sparged with O2 for
5 min and PPh3 was added (11.6 g, 44 mmol) and the reaction
stirred at −78 °C for 30 min. To the reaction mix was
added NaBH4 (1.6 g, 44 mmol), and the reaction was allowed
to warm to rt over 2 h. The reaction was quenched with H2O (100 mL) and then diluted with DCM (100 mL). The aqueous phase
was separated and extracted with DCM (3 × 50 mL). The combined
organic phases were dried (MgSO4), filtered, and concentrated
in vacuo. The crude mixture was dry-loaded onto silica and purified
by rapid flash column chromatography (2:1→ 1:1 petrol bp 30–40
°C/diethyl ether 1% NEt3) to give the title compound
as a colorless oil (2.81 g, 13.5 mmol, 92%): R 0.40 (1:1 petrol/diethyl ether); νmax/cm–1 (thin film) 3434s br, 2875m; 1H NMR (400 MHz C6D6) δ 7.06–7.23
(m, 5H, ArH), 4.46 (d, J = 11.6
Hz, 1H, CHH′Ar), 4.23 (d, J = 11.6 Hz, 1H, CHH′Ar), 3.48–3.73
(m, 2H, CH2OH), 3.24 (dt, J = 6.8, 5.6 Hz, 1H, CHOBn), 2.61 (ddd, J = 5.6, 3.8, 2.8 Hz, 1H, CHOCH2), 2.35
(dd, J = 5.3, 2.6 Hz, 1H, CHOCHH′), 2.29 (dd, J = 5.4, 3.8 Hz, 1H, CHOCHH′), 1.75–1.60 (m, 3H, COH, CH2CH2OH); 13C{1H} NMR (100 MHz, C6D6) δ
139.0 (Ar), 128.2 (Ar), 128.0 (Ar), 127.8 (Ar), 76.9 (CHOBn), 72.4 (CH2Ar), 59.5 (CH2OH), 53.1 (CHOCH2) 45.3
(CHOCH2), 35.6 (CH2CH2OH); HRMS (ESI-TOF) m/z [M + Na]+ calcd for C12H16O3Na 231.0992; found 231.0992; [α]D25 +32.0 (c =
1.0 in CHCl3).
To a stirred solution of diol 21 (900 mg, 1.57 mmol) in DCM (30 mL) at 0 °C were added ethyldiisopropylamine
(2.72 mL, 15.6 mmol) and MsCl (1 mL, 12.5 mmol) and stirred for 1
h before being quenched with saturated aqueous NH4Cl (30
mL) and diluted with H2O (30 mL) and DCM (30 mL). The aqueous
layer was separated and extracted with DCM (3 × 30 mL). The combined
organic layers were dried (Na2SO4), filtered,
and concentrated in vacuo to give a crude oil (23). The
crude oil was dissolved in DCM/MeOH (1:1, 30 mL); then CSA (363 mg,
1.57 mmol) was added, and the reaction was stirred for 24 h before
being quenched by the addition of saturated aqueous NaHCO3 (40 mL). The aqueous layer was separated and extracted with DCM
(3 × 30 mL). The combined organic layers were dried (Na2SO4), filtered, and concentrated in vacuo to give a crude
oil. The crude oil was then dissolved in BuOH (20 mL) and warmed to 35 °C; BuOK (527 mg, 4.71 mmol) was added and the reaction stirred for 1
h. The reaction was quenched with saturated aqueous NH4Cl (30 mL). The aqueous layer was separated and extracted with EtOAc
(3 × 30 mL). The combined organic layers were dried (MgSO4), filtered, and concentrated in vacuo. Purification via flash
column chromatography (2:1 petrol/ethyl acetate) gave the title compound
as a colorless oil (427 mg, 0.82 mmol, 52%): R 0.56 (2:1 petrol/ethyl acetate); νmax/cm–1 (thin film) 2935m, 1356s, 1175s; 1H NMR (400 MHz CDCl3) δ 7.22–7.38
(m, 7H, ArH), 6.88 (d, J = 8.6 Hz,
2H, ArH), 4.61 (d, J = 11.8 Hz,
1H, CHH′Ar), 4.36 (d, J =
11.9 Hz, 1H, CHH′Ar), 4.32–4.52 (m,
5H, 2 × CHH′Ar, CHORCH2OMs), 4.12–422 (m, 2H, CHOBn, CHOPMB), 4.00–4.09 (m, 2H, CHORCHOR), 3.85–3.92 (m, 1H, EtCHOR), 3.81 (s, 3H, OMe), 3.00 (s, 3H, SMe), 2.21–2.30 (m, 2H, 2 × CHH′CHOR), 2.16 (dt, J = 13.1, 4.4 Hz, 1H,
CHH′CHOBn), 2.00 (dt, J =
12.6, 4.4 Hz, 1H, CHOPMBCHH′), 1.48 (qn, J = 7.6 Hz, 2H, CH3CH2), 0.93 (t, J = 7.6 Hz, 3H, CH3CH2); 13C{1H} NMR (100 MHz,
CDCl3) δ 159.2 (Ar), 137.5 (Ar), 130.2 (Ar), 129.3
(Ar), 128.5 (Ar), 127.9 (Ar), 127.8 (Ar), 113.8 (Ar), 84.9 (EtCHOR), 82.4 (CHOPMB), 80.9 (CHOBn), 79.8 (CHORCHOR), 79.3 (CHORCHOR), 78.7 (CH2OMs), 77.3 (CHORCH2OMs), 71.1 (CH2Ar), 69.8 (CH2Ar), 55.3 (OMe), 37.8 (SCH3), 34.6 (CH2), 34.1 (CH2), 26.7 (CH3CH2), 10.1 (CH3); MS (ESI-TOF) m/z 543 [M + Na]+; HRMS (ESI-TOF) m/z [M + Na]+ calcd for C27H36O8SNa 543.2023; found 543.2022; [α]D20 +4.0 (c = 1.0 in CHCl3).
(2R,2′S,4R,4′R,5R,5′S)-5′-allyl-4′-chloro-5-ethyloctahydro-[2,2′-bifuran]-4-yl
4-nitrobenzoate (20 mg, 49 μmol) was dissolved in MeOH (2 mL)
and cooled to 0 °C, and K2CO3 (40 mg, 290
μmol) was added. The reaction was stirred for 30 min and then
diluted with H2O (5 mL) and extracted with EtOAc (3 ×
10 mL). The combined organic layers were then dried (MgSO4), filtered, and concentrated in vacuo. Purification by flash column
chromatography (4:1 petrol/ethyl acetate) gave the title compound
as a colorless oil (13 mg 48 μmol, 98%): R 0.32 (5:1 petrol/ethyl acetate); νmax/cm–1 (thin film) 3461br (OH), 2968m,
1430m, 1050s; 1H NMR (400 MHz CDCl3) δ
5.82 (ddt, J = 17.0, 10.0, 6.8 Hz, 1H, CH=CH2), 5.17 (dd, J = 17.0, 1.5
Hz, 1H, CH=CHH′), 5.15 (dd, J = 10.0, 1.5 Hz, 1H, CH=CHH′),
4.44 (ddd, J = 9.1, 6.8, 2.6 Hz, 1H, CHORCHOR), 4.16 (dt, J = 9.8, 2.6 Hz, 1H, CHORCHOR), 4.11–4.14 (m, 1H, CHORCHCl),
3.99–4.04 (m, 2H, CHOH, CHCl), 3.54 (tdfz, J = 6.9, 2.5 Hz, 1H, EtCHOR), 3.24 (br s, 1H, OH), 2.33–2.47
(m, 2H, CH2CH=CH2),
2.25 (ddd, J = 14.0, 9.8, 5.5 Hz, 1H, CHH′CHOH), 2.15 (ddd, J = 13.8, 6.8, 3.7 Hz,
1H, CHH′CHCl), 2.00 (ddd, J = 13.8, 9.6, 7.8 Hz, 1H, CHH′CHCl), 1.81
(dd, J = 14.0, 3.3 Hz, 1H, CHH′CHOH),
1.65–1.74 (m, 2H, CH3CH2), 0.97 (t, J = 7.6 Hz, 3H, CH3); 13C{1H} NMR (100 MHz, CDCl3) δ 132.7 (CH=CH2), 118.6
(CH=CH2), 86.6 (CHORCHCl), 85.7 (EtCHOR), 79.0 (CHORCHOR), 77.9 (CHORCHOR), 70.9 (CHOH),
58.4 (CHCl), 38.2 (CH2CHCl), 37.6 (CH2CH=CH2), 34.7 (CH2CHOH), 21.8 (CH3CH2), 10.5 (CH3); MS (ESI-TOF) m/z 283 [35M + Na+], 285 [37M + Na+]; HRMS
(ESI-TOF) m/z [M + Na]+ calcd for C13H21O335ClNa 283.1071; found 283.1072; [α]D20 −40.2 (c =
1.0 in CHCl3).
(2R,2′S,4R,4′R,5R,5′S)-4′-Chloro-5-ethyl-5′-((E)-pent-2-en-4-yn-1-yl)octahydro-[2,2′-bifuran]-4-ol, L. majuscula Enyne (2b)
To
a stirred solution of alkene 37 (12 mg, 46 μmol)
(62 mg, 0.16 mmol) in dry degassed DCM (1.4 mL) were added crotonaldehyde
(38 μL, 0.46 mmol) and Grubbs’ second generation catalyst
(3.9 mg, 4.6 μmol). The reaction mixture was stirred for 1.5
h at 40 °C and then cooled to rt and quenched with the addition
of DMSO (30 μL) and stirred for 16 h. The mixture was concentrated
in vacuo and purified by flash column chromatography (5:1 petrol ethyl
acetate) to give the intermediate enal (38) (12.6 mg),
which was used immediately in the next reaction: MS (ESI-TOF) m/z 311 [35M + Na+], 313 [37M + Na+]; HRMS (ESI-TOF) m/z [M + Na]+ calcd for C14H21O435ClNa 311.1021; found
311.1026.To a solution of (diazomethyl)trimethylsilane (230
μL, 2 M in ether, 0.46 mmol) in THF (1 mL) at −78 °C
was added BuLi (288 μL, 1.6 M in hexanes, 0.46 mmol) dropwise.
The reaction mixture was stirred for 30 min before a solution of the
intermediate enal 38 was added as a solution in THF (0.5
+ 0.5 mL) at −78 °C. The reaction was stirred at −78
°C for 1 h before being warmed to 0 °C for a further 1 h.
The reaction was quenched with acetic acid (0.1 mL) and then saturated
aqueous NaHCO3 (10 mL). The aqueous layer was separated
and extracted with EtOAc (3 × 10 mL). The combined organic layers
were dried (MgSO4), filtered, and concentrated in vacuo.
Purification by flash column chromatography (20:1 petrol/ethyl acetate)
gave the OTMS-protected version title compound (m/z 379 M + Na+, 100), which was then
dissolved in THF (1 mL) and cooled to 0 °C, and TBAF (0.13 mL,
1 M in THF, 0.13 mmol) was added and the reaction stirred for 5 min.
The reaction was quenched with saturated aqueous NH4Cl
(1 mL). The aqueous layer was separated and extracted with EtOAc (3
× 8 mL). Purification by flash column chromatography (5:1 petrol/ethyl
acetate) gave the title compound as a colorless oil (5 mg, 18 μmol,
39%): R 0.40 (5:1: petrol/ethyl
acetate); νmax/cm–1 (thin film)
3461br (OH), 3296m (CH), 2968m, 1442m, 1066s; 1H NMR (500
MHz CDCl3) δ 6.22 (dt, J = 15.9,
7.4 Hz, 1H, CH2CH=CH), 5.59 (dddd, J = 15.9, 2.2, 1.6, 1.5 Hz, 1H, CH2CH=CH), 4.41 (ddd, J = 9.1, 6.8, 2.6 Hz, 1H,
CHORCHOR), 4.11 (ddd, J = 9.8, 3.3,
2.6 Hz, 1H, CHORCHOR), 4.08 (ddd, J = 6.5, 5.6, 5.4 Hz, 1H, CHORCHCl), 4.05 (br m,
1H, CHOH) 3.96 (ddd, J = 7.8, 5.4,
4.2 Hz, 1H, CHCl), 3.54 (dt, J =
6.9, 2.5 Hz, 1H, EtCHOR), 3.00 (br s, 1H, OH), 2.83 (d, J = 2.2 Hz, 1H), 2.40–2.53
(m, 2H, CH2CH=CH), 2.25 (ddd, 1H, J = 14.0, 9.8, 5.5 Hz, CHH′CHOH),
2.18 (ddd, J = 13.8, 6.8, 4.2 Hz, 1H, CHH′CHCl), 2.05 (ddd, J = 13.8, 9.1, 7.8 Hz,
1H, CHH′CHCl), 1.78 (dd, J = 14.0, 3.3 Hz, 1H, CHH′CHOH), 1.70 (qn, J = 7.5 Hz, 2H, CH3CH2), 0.98 (t, J = 7.5 Hz, 1H, CH3); 13C{1H} NMR (75 MHz, CDCl3) δ 139.9 (CH2CH=CH), 112.4
(CH=CH2), 86.0 (CHORCHCl), 85.7 (EtCHOR), 81.7 (CCH), 79.2 (CHORCHOR), 77.9 (CHORCHOR),
77.2 (CCH), 71.0 (CHOH), 58.2 (CHCl), 38.1 (CH2CHCl), 36.5
(CH2CH=CH2), 35.1(CH2CHOH), 21.7 (CH3CH2), 10.5 (CH3); MS (ESI-TOF) m/z 307 [35M + Na+], 309 [37M + Na+]; HRMS (ESI-TOF) m/z [M + Na]+ calcd for C15H2135ClO3Na 307.1071; found
307.1066; [α]D20 −51.9 (c = 0.45 in CHCl3), lit. [α]D22 −67.8 (c = 0.09 in CHCl3).
A mixture of ozone and oxygen was gently
bubbled through a stirred solution of 47-Oxford (5 mg,
15 μmol) in DCM (3 mL) at −78 °C until the solution
became pale blue (approximately 2 min). The excess ozone was purged
from the solution by bubbling oxygen through for a further 5 min.
Triphenylphosphine (20 mg, 75 μmol) was added, and the reaction
mixture was allowed to warm to rt over 15 h. The reaction mixture
was dry-loaded onto silica and rapidly purified by flash column chromatography
(5:1 petrol bp 30–40 °C/diethyl ether) to give the corresponding
aldehyde as a colorless oil (4.4 mg, 13.6 μmol, 91%) that was
used in the subsequent transformation without characterization.To a solution of 3-(tert-butyldimethylsilyl)-1-trimethylsilanyl-propyne
(135 mg, 0.63 mmol) in dry THF (1 mL) at −78 °C was added
dropwise tert-butyllithium (0.37 mL, 1.7 M in pentane,
0.63 mmol), and this solution was stirred at −78 °C for
1 h. A solution of titanium(IV) isopropoxide (0.19 mL, 180 mg, 0.63
mmol) in dry THF (0.5 mL) was added dropwise to the reaction mixture;
the resulting solution was stirred for 10 min, and then 1.94 mL was
removed and discarded. To the remaining 0.12 mL (36 μmol) was
added dropwise a solution of the aldehyde prepared above in dry THF
(0.5 mL + 0.5 mL rinse). The reaction mixture was stirred at −78
°C for 30 min, and then the cooling bath was removed and the
reaction mixture stirred at rt for 30 min. The reaction mixture was
then poured into a separating funnel containing 0.1 M aqueous HCl
(2 mL), and the aqueous layer was extracted with diethyl ether (3
× 5 mL). The combined organic layers were dried (MgSO4), filtered, and concentrated in vacuo. Purification by flash column
chromatography (50:1 petrol bp 30–40 °C/diethyl ether)
gave the title compound as a colorless oil (2 mg, 4.8 μmol,
32% from 47-Oxford, >30:1 (Z):(E) from crude 1H NMR analysis and characterized
as such a mixture): νmax/cm–1 (thin
film) 2963; 1H NMR (500 MHz, CDCl3) δ
6.02 (dt, J = 10.9, 7.5 Hz, 1HCH2CH=CH), 5.63 (dt, J = 10.9, 1.3 Hz,
1H, CH2CH=CH), 4.25 (td, J = 7.3, 5.6 Hz, 1H, CHClCH2CHO), 4.13–4.08 (m, 2H, CHCH2CH=CH,
CHCl), 3.97 (ddd, J = 8.3, 6.7,
5.5 Hz, 1H, CHBrCH2CHO), 3.91 (td, J = 7.5, 4.0 Hz, 1H, CHEt), 3.86 (dt, J = 8.7, 7.3 Hz, 1H, CHBr), 2.66 (dt, J = 13.2, 6.7 Hz, 1H, CHBrCHH), 2.69–2.63
(m, 1H, CHHCH=CH), 2.61–2.55 (m, 1H,
CHHCH=CH), 2.29–2.22 (m, 2H, CHClCH2), 2.18 (dt, J = 13.2, 8.3
Hz, 1H, CHBrCHH), 1.75 (dqd, J =
13.9, 7.4, 4.0 Hz, 1H, CHHCH3), 1.49 (dqn, J = 14.0, 7.4 Hz, 1H, CHHCH3), 1.00 (t, J = 7.4 Hz, 3H, CH2CH3), 0.20 (s, 9H, Si(CH3)3); 13C{1H} NMR (125 MHz,
CDCl3) δ 139.3 (CH2CH=CH),
112.4 (CH2CH=CH), 101.6 (C≡C), 100.0 (C≡C), 87.3 (CHEt), 86.5 (CHCH2CH=CH),
80.3 (CHClCH2CH), 79.1 (CHBrCH2CH), 59.4 (CHCl), 47.5 (CHBr), 39.5 (CHBrCH2), 38.3
(CHClCH2), 34.6 (CH2CH=CH), 25.6 (CH2CH3), 10.2 (CH2CH3), 0.11
(Si(CH3)3); [α]D25 +37.5 (c = 0.16 in CHCl3).
To a stirred solution of 48 (2 mg, 4.8 μmol) in
THF (1 mL) at −20 °C was added TBAF (20 μL of a
2.0 M solution in THF, 40 μmol), and stirring was continued
for 5 min. The reaction mixture was quenched by the addition of aqueous
ammonium chloride (3 mL), and the aqueous phase was extracted with
ether (3 × 5 mL). The combined organic extracts were dried (Na2SO4). Purification by flash chromatography (30:1
petrol bp 30–40 °C/diethyl ether) gave the title compound
as a colorless oil (1.7 mg, 4.8 μmol, quant.): νmax/cm–1 (thin film) 3294, 1728, 1460, 1289, 1071,
966; 1H NMR (500 MHz, CDCl3) δ 6.08 (dtd, J = 10.9, 7.5, 0.9 Hz, 1H, CH2CH=CH), 5.60 (ddt, J = 10.9, 2.7, 1.4 Hz, 1H,
CH2CH=CH), 4.25 (td, J = 7.3, 5.6 Hz, 1H, CHClCH2CHO), 4.11–4.07
(m, 2H, CHCH2CH=CH, CHCl), 3.98 (ddd, J = 8.3, 6.8, 5.6 Hz, 1H, CHBrCH2CHO), 3.91 (td, J = 7.5,
3.9 Hz, 1H, CHEt), 3.86 (dt, J =
8.8, 7.3 Hz, 1H, CHBr), 3.12 (dd, J = 2.2, 0.9 Hz, 1H, CCH), 2.66 (dt, J = 13.1, 7.0 Hz, 1H, CHBrCHH), 2.69–2.64
(m, 1H, CHHCH=CH), 2.60 (dddd, J = 14.6, 7.4, 6.0, 1.4 Hz, 1H, CHHCH=CH),
2.29–2.24 (m, 2H, CHClCH2), 2.17
(dt, J = 13.2, 8.5 Hz, 1H, CHBrCHH), 1.75 (dqd, J = 14.0, 7.5, 3.9 Hz, 1H, CHHCH3), 1.49 (dqn, J = 14.0,
7.4 Hz, 1H, CHHCH3), 1.00 (t, J = 7.4 Hz, 3H, CH2CH3); 13C{1H} NMR (125 MHz, CDCl3)
δ 139.9 (CH2CH=CH), 111.1
(CH2CH=CH), 87.2 (CHEt), 86.1 (CHCH2CH=CH), 82.3
(HC≡C), 80.1 (CHClCH2CH), 79.9 (HC≡C), 78.9 (CHBrCH2CH), 59.3 (CHCl), 47.3 (CHBr), 39.3 (CHBrCH2), 38.1
(CHClCH2), 34.4 (CH2CH=CH), 25.4 (CH2CH3), 10.2 (CH2CH3); HRMS
(ESI-TOF) m/z [M + Na]+ calcd for C15H2079Br35ClNaO2 369.0227; found 369.0231, calcd for C15H2079/81Br37/35ClNaO2 371.0206; found 371.0203; [α]D25 +19.2 (c = 0.125 in CHCl3) {lit.[21a] [α]D = +40.3 (c 1.03 CHCl3)}.
Notoryne:
Seoul Route
General Procedures
Proton (1H) and carbon
(13C) NMR spectra were obtained on a JEOL JNM-LA300 (300/75
MHz), Bruker AV 400 (400/100 MHz), Bruker AMX 500 (500/125 MHz), Bruker
Avance 600 (600/150 MHz), or Bruker Avance 900 (900/225 MHz) spectrometer.
Chemical shifts are reported in parts per million units with Me4Si or CHCl3 as the internal standard. All reactions
were routinely carried out under an inert atmosphere of dry nitrogen
or argon. Reactions were checked by thin layer chromatography (Kieselgel
60 F254, Merck). Spots were detected by viewing under a UV light and
by colorizing with charring after immersion in a p-anisaldehyde solution or phosphomolybdic acid solution. In an aqueous
workup, all organic solutions were dried over anhydrous sodium sulfate
and filtered prior to rotary evaporation at water pump pressure. The
crude compounds were purified by column chromatography on a silica
gel (Kieselgel 60, 70-230 mesh, Merck). Unless otherwise noted, materials
were obtained from commercial suppliers and were used without purification.
All solvents were purified and dried by standard techniques just before
use. THF and Et2O were freshly distilled from sodium and
benzophenone. Methylene chloride, toluene, and benzene were purified
by refluxing with CaH2. Hexanes and ethyl acetate were
purified by simple distillation.
To a solution of terminal olefin 47-Seoul (7.7 mg, 0.024
mmol) in dry CH2Cl2 (0.47 mL, 0.05 M) were added
crotonaldehyde (0.015 mL, 0.19 mmol) and Grubbs’ catalyst 29 (2.04 mg, 0.0024 mmol) at room temperature. After the mixture
was stirred at 40 °C for 1.5 h, the reaction was quenched with
DMSO (0.02 mL). The resulting mixture was stirred at room temperature
for 12 h and concentrated in vacuo. The residue was purified by column
chromatography (silica gel, n-hexane/ethyl acetate,
6/1) to afford (E)-4-((2R,2′S,4S,4′S,5R,5′R)-4′-bromo-4-chloro-5′-ethyloctahydro-[2,2′-bifuran]-5-yl)but-2-enal
(7.3 mg, 87%) as a colorless oil: R 0.37 (n-hexane/ethyl acetate, 4/1). To a
cooled (−78 °C) solution of LDA (0.42 mL, 0.5 M solution
in THF, 0.21 mmol) was added dropwise TMSCH2N2 (0.105 mL, 2.0 M solution in Et2O, 0.021 mmol) under
argon atmosphere. After the mixture was stirred at −78 °C
for 30 min, (E)-4-((2R,2′S,4S,4′S,5R,5′R)-4′-bromo-4-chloro-5′-ethyloctahydro-[2,2′-bifuran]-5-yl)but-2-enal
(7.3 mg, 0.021 mmol) in THF (0.42 mL, 0.05 M) was added dropwise at
−78 °C. After the reaction mixture was stirred at −78
°C for 1 h, and then at 0 °C for 1 h, the reaction was quenched
with saturated aqueous NH4Cl. The layers were separated,
and the aqueous layer was extracted with diethyl ether (2 × 2
mL). The organic layers were washed with brine, dried over anhydrous
Na2SO4, and concentrated in vacuo. The residue
was purified by column chromatography (silica gel, n-hexane/ethyl acetate, 4/1) to afford (E)-notoryne
(E)-3a (6.4 mg, 88%) as a colorless
oil: R 0.69 (n-hexane/ethyl acetate, 4/1); [α]D25 = +47.6 (c 0.225,
CHCl3); 1H NMR (400 MHz, CDCl3) δ
6.24 (ddd, J = 7.2, 7.2, 16.0 Hz, 1H), 5.57 (dd, J = 1.8, 16.0 Hz, 1H), 4.22 (ddd, J = 5.6,
7.0. 7.0 Hz, 1 H), 4.03–3.94 (m, 3H), 3.93–3.84 (m,
2H), 2.83 (d, J = 2 Hz, 1H), 2.66 (ddd, J = 13.2, 7.0, 7.0 Hz, 1H), 2.46 (dddd, J = 12.9,
6.5, 5.4, 1.6 Hz, 1H), 2.40–2.33 (m, 1H), 2.32–2.20
(m, 2H), 2.14 (dt, J = 13.3, 8.4 Hz, 1H), 1.75 (ddddd, J = 14.1, 7.5, 7.5, 7.5, 3.8 Hz, 1H), 1.55–1.44 (m,
1H), 1.00 (t, J = 7.4 Hz, 3H); 13C{1H} NMR (100 MHz, CDCl3) δ 140.7, 111.8, 87.3,
85.7, 81.9, 80.0, 78.8, one peak buried under CDCl3 peak
by HSQC, 58.7, 47.3, 39.4, 37.9, 36.6, 25.5, 10.0; IR (neat) 3294,
2963, 1731, 1712, 1295, 1072, 959, 925 cm–1; HRMS
(EI) m/z [M – C2H5]+ calcd for C10H1579Br35ClO2 280.9938; found 280.9944.
Authors: Frédéric H Vaillancourt; Ellen Yeh; David A Vosburg; Sylvie Garneau-Tsodikova; Christopher T Walsh Journal: Chem Rev Date: 2006-08 Impact factor: 60.622
Chart 1
Figure 1
Figure 2
Figure 3
Figure 4
Scheme 1
Scheme 2
Scheme 3
Figure 5
Scheme 4
Scheme 5
Figure 6