Rhodium-phosphoramidite catalyzed alkene hydroacylation: Mechanism and octaketide natural product synthesis

Article abstract of DOI:10.1021/ja305593y

We describe a method that allows salicylaldehyde derivatives to be coupled with a wide range of unactivated alkenes at catalyst loadings as low as 2 mol %. A chiral phosphoramidite ligand and the precise stoichiometry of heterogeneous base are key for high catalytic activity and linear regioselectivity. This protocol was applied in the atom- and step-economical synthesis of eight biologically active octaketide natural products, including anticancer drug candidate cytosporone B. Mechanistic studies provide insight on parameters affecting decarbonylation, a side reaction that limits the turnover number for catalytic hydroacylation. Deuterium labeling studies show that branched hydride insertion is fully reversible, whereas linear hydride insertion is largely irreversible and turnover-limiting. We propose that ligand (R _a,R,R)-SIPHOS-PE effectively suppresses decarbonylation, and helps favor a turnover-limiting insertion, by lowering the barrier for reductive elimination in the linear-selective pathway. Together, these factors enable high reactivity and regioselectivity.

Full text of DOI:10.1021/ja305593y

Article
pubs.acs.org/JACS
Rhodium-Phosphoramidite Catalyzed Alkene Hydroacylation:
Mechanism and Octaketide Natural Product Synthesis
Max von Delius,† Christine M. Le,† and Vy M. Dong*,†,‡
†Department of Chemistry, University of Toronto, 80 St. George Street, Toronto, Ontario M5S 3H6, Canada
‡Department of Chemistry, University of California, Irvine, California 92697-2025, United States
S
* Supporting Information
ABSTRACT: We describe a method that allows salicylalde-
hyde derivatives to be coupled with a wide range of
unactivated alkenes at catalyst loadings as low as 2 mol %. A
chiral phosphoramidite ligand and the precise stoichiometry of
heterogeneous base are key for high catalytic activity and linear
regioselectivity. This protocol was applied in the atom- and
step-economical synthesis of eight biologically active octake-
tide natural products, including anticancer drug candidate
cytosporone B. Mechanistic studies provide insight on
parameters affecting decarbonylation, a side reaction that limits the turnover number for catalytic hydroacylation. Deuterium
labeling studies show that branched hydride insertion is fully reversible, whereas linear hydride insertion is largely irreversible and
turnover-limiting. We propose that ligand (Ra,R,R)-SIPHOS-PE effectively suppresses decarbonylation, and helps favor a
turnover-limiting insertion, by lowering the barrier for reductive elimination in the linear-selective pathway. Together, these
factors enable high reactivity and regioselectivity.
Scheme 1. Rh-Phosphoramidite Catalyzed Hydroacylation
INTRODUCTION
■
Although alkene hydroacylation1 is a conceptually ideal way to
prepare ketones, this atom-economical2 strategy remains
limited in application due to a lack of catalysts that can
transform a broad range of simple olefins with high
regiocontrol.3 Willis and Weller have recently reported a highly
efficient rhodium/bisphosphine catalyst for the coupling of β-S-
substituted aldehydes4 with a wide variety of unactivated
alkenes.5 Despite the frequent use of salicylaldehydes in
intermolecular hydroacylation,6 however, a method with
comparable efficiency, alkene scope, and practicality still
remains elusive for this type of aldehyde substrate.7,8
Developing such a method would open up a general route to
a range of arylketones that could be applied to the synthesis of
biologically active polyketide natural products.9
Our group has recently communicated the enantioselective
hydroacylation of sulfur-containing alkenes with salicylalde-
hydes based on a double-chelation approach.6h,10 Using chiral
phosphoramidite ligand (Ra,R,R)-SIPHOS-PE,11 homoallylic
sulfides such as 2A could be transformed into the branched
arylketone products (3A) with high regio- and enantiocontrol
(Scheme 1a). This study represented the first example of
asymmetric hydroacylation of unactivated olefins and high-
lighted the unique ability of phosphoramidites in hydro-
acylation.12 Given these initial results, we aimed to better
understand this catalyst and apply our system to the
functionalization of simple olefins, without the need for a
directing sulfide or other heteroatom functionality.6j,k In this
article, we show that our previously reported rhodium-
phosphoramidite catalyst can be used to achieve linear-selective
hydroacylation of a wide range of monosubstituted olefins
(Scheme 1b). Mechanistic studies show that the chiral ligand
(Ra,R,R)-SIPHOS-PE, although initially used for enantiocontrol
in the branched-selective protocol (Scheme 1a), possesses
unique stereoelectronic properties that enhance reactivity13a−c
and promote regioselectivity13d−f in the challenging linear-
selective hydroacylation of simple alkenes (Scheme 1b). In
addition to fundamental insights on hydroacylation, our study
contributes to the unexpected importance of using chiral
ligands to enhance both reactivity and regiocontrol in
nonasymmetric transformations.13
Received: June 9, 2012
© XXXX American Chemical Society
A
dx.doi.org/10.1021/ja305593y | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Moreover, our catalytic protocol could be applied to the step-
and atom-economical total synthesis of eight structurally related
octaketides that have been isolated from different endophytic
fungi within the past decade (Scheme 2).9,14 The majority of
Table 1. Effect of Phosphorus-Based Ligands
Scheme 2. Natural Products Targeted in This Study
these compounds have not been previously synthesized and
limited data is currently available on their biological activity due
to the low quantities typically isolated.14e Nevertheless, recent
reports have identified octaketide natural product cytosporone
B as a promising anticancer therapeutic based on its ability to
regulate cell proliferation, differentiation, and apoptosis up-
stream in a biochemical pathway.15 Simple structural analogues
of cytosporone B have been shown to decrease xenograft tumor
growth in vivo,16 indicating the potential benefit of a synthetic
method that allowed the convenient preparation of more
complex derivatives.
In principle, all members of this natural product family could
be prepared through the coupling of simple 2-hydroxybenzal-
dehydes with unactivated terminal alkenes (Scheme 2).
However, the previously reported hydroacylations by Jun7d
and Suemune6f were not applicable to the total synthesis of
cytosporone B.17 Applying Jun’s protocol, based on activation
of the aldehyde via reversible imine formation, gave no
appreciable amount of product in our hands. Suemune’s
protocol, requiring the use of 40 mol % rhodium catalyst,
resulted in formation of the natural product, albeit in 5% yield.
These experiments highlight the need for more general and
efficient hydroacylation methods.
a
b
The ratio of rhodium to ligand was 1:1 in each case. NMR yield
(internal standard: 1,3,5-trimethoxybenzene); isolated yield in
c
parentheses. Regioisomeric ratio determined by integration of crude
1H NMR. See Supporting Information, Table S1, for all ligand
structures and results obtained with further ligands.
RESULTS AND DISCUSSION
and linear selectivity at both 20 and 10 mol % catalyst loading
(entries 1 and 2). Neither less bulky ligand (R)-SIPHOS
(entries 3 and 4) nor phosphoramidites, derived from a BINOL
or biphenyl backbone (entries 5 and 6, respectively), resulted in
comparable reactivity or selectivity. Commercially available
diastereomer, (Sa,R,R)-SIPHOS-PE (entry 7),11 furnished the
product in low yield and selectivity, indicating a mismatch
between the stereogenic elements in this ligand.19,20 These
results suggest that a synergistic effect between the three
stereogenic elements in (Ra,R,R)-SIPHOS-PE is responsible for
the excellent reactivity and regioselectivity observed. Structur-
ally related phosphine or phosphite ligands led to poor
reactivity (entries 8 and 9).
■
It occurred to us that the Rh/(Ra,R,R)-SIPHOS-PE catalyst
(Scheme 1a)6h could be effective for the intermolecular
hydroacylation of a broader range of alkenes and ultimately
be suitable for the total synthesis of octaketide natural products
(Scheme 2). We hypothesized that phosphoramidite ligands
such as (Ra,R,R)-SIPHOS-PE that are good π-acceptors and
relatively bulky should promote this reaction by decreasing the
barrier to reductive elimination, which has been implicated as
the turnover-limiting step for related hydroacylations.8c,18 To
test our hypothesis, we used Rh/(Ra,R,R)-SIPHOS-PE for the
model reaction of salicylaldehyde (1) with unactivated alkene 2
and obtained the linear ketone 4 in 90% yield and excellent
selectivity (entry 1, Table 1).
Phosphorus-Based Ligands. In light of this initial finding,
we evaluated 30 other commercially available phosphorus-based
ligands for the transformation of salicylaldehyde (1) with 1-
octene (2) (Table 1 and Table S1 in the Supporting
Information (SI)). Phosphoramidite ligands (entries 1−7)
were among the few ligands that effectively promoted the
transformation. (Ra,R,R)-SIPHOS-PE gave excellent reactivity
The solid state structure of precatalyst complex [Rh(COD)-
Cl((Ra,R,R)-SIPHOS-PE)] (5, Figure 1) illustrates the large
steric bulk of (Ra,R,R)-SIPHOS-PE, as well as the intricate
spatial arrangement of its stereochemical components (see SI
for a crystal structure of mismatched diastereomeric complex
[Rh(COD)Cl((Sa,R,R)-SIPHOS-PE)]).21
Finally, monodentate phosphines PPh3 and DavePhos22 were
effective at 20 mol % catalyst loading, but a significant drop in
B
dx.doi.org/10.1021/ja305593y | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
deprotonates salicylaldehyde to make the strongly coordinating
phenolate anion,6a which can then undergo oxidative addition
onto Rh. Accordingly, we observed no reactivity in the absence
of base (entry 1).
However, we observed an increase in yield when the base/
[Rh] ratio was increased from 1:4 to 1:1 (entries 2−4). A
drastic drop in reactivity was observed when a base/[Rh] ratio
of 2:1 or 8:1 was used (entries 5 and 6).
When we submitted the potassium and sodium salts (6) of
salicylaldehyde to the reaction conditions, less than 10% yield
was obtained in both cases (eq 3). This finding indicates that an
excess of 6a relative to rhodium inhibits catalytic turnover (see
also Table 2, entry 6). However, when salicylaldehyde (1) was
treated with 1-octene (2) in the presence of catalytic amounts
of salts 6a or 6b in lieu of K3PO4, excellent reactivity was
observed (eq 4). When we investigated a wide range of
heterogeneous and homogeneous bases for the reaction of 1
with 2, we found that all soluble bases were completely
Figure 1. Solid-state structure of catalyst complex 5: [Rh(COD)Cl-
((Ra,R,R)-SIPHOS-PE)]. Single crystals were obtained by slow
diffusion of hexanes into a CHCl3 solution. P, purple; O, red; N,
blue; Rh, orange; Cl, green; C, black. Key bond lengths (Å) and angles
(deg): Rh−P, 2.2614(6); Rh−Cl, 2.3617(5); P−N, 1.6534(19); Cl−
Rh−P, 89.70(2); O−P−O, 101.41(8); Rh−P−N, 121.80(7).
reactivity was observed at 10 mol % loading, presumably due to
premature catalyst deactivation via decarbonylation (entries 4
and 5).23 All other Buchwald-type ligands, as well as all
surveyed bisphosphine ligands, resulted in poor reactivity (see
SI).
Base Stoichiometry. A study on the base/[Rh] ratio
revealed that the relative amount of heterogeneous base
critically impacts reactivity (Table 2).24 Presumably, the base
Table 2. Variation of K3PO4/[Rh] Ratio
ineffective for the transformation (see Table S5 in the SI for
details). Mechanistic studies support the notion that excess
soluble base promotes catalyst deactivation via decarbonylation
processes (vide infra).
Optimizing Solvent and Concentration. A brief study of
various rhodium sources revealed [Rh(COD)Cl]2 was ideal
(see SI, Table S6). Among the chlorinated solvents examined,
DCE gave the best reactivity (see SI Table S7 for details). The
reaction also proceeded well in tetrahydrofuran, 1,4-dioxane
and the renewable solvent,25 2-methyltetrahydrofuran.26 No
significant reactivity was observed in more polar solvents, such
as propylene carbonate, or strongly coordinating solvents, such
as acetonitrile. The catalytic protocol is also effective in the
absence of added solvent (i.e., in neat octene, see SI). The
a
NMR yield (internal standard: 1,3,5-trimethoxybenzene); isolated
b
yield in parentheses. Regioisomeric ratio determined by integration of
c
crude 1H NMR. Based on analysis by GC−MS. n.d.: not determined.
Table 3. Optimization of Rhodium Catalyst Loading
a
b
NMR yield (internal standard: 1,3,5-trimethoxybenzene); isolated yield in parentheses. Regioisomeric ratio determined by integration of crude 1H
c
NMR. The remainder of the mass balance constituted salicylaldehyde (1).
C
dx.doi.org/10.1021/ja305593y | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
product was isolated in 97% yield when biphasic conditions
(DCE/water, 1:1) were used (eq 5). To the best of our
knowledge, this represents the first report of a biphasic
aqueous/organic hydroacylation reaction. Separating base and
catalyst into two phases minimizes the detrimental effect of
excess base on catalytic activity and results in a more robust
protocol. We observed 90% yield of product 4 when a 10-fold
excess of K3PO4 with respect to catalyst was used (in DCE
alone this would result in poor yield; refer to Table 2, entry 6).
Finally, we were pleased to find that the reaction proceeded
smoothly with only 5 mol % [Rh], provided that an equimolar
amount of K3PO4 base was used (Table 3, entry 2). A further
decrease in catalyst loading, however, necessitated an increase
in reaction concentration.27 An isolated yield of 99% was
obtained when 2 mol % of rhodium, ligand and base were used
at a concentration of 1.0 M (entry 3). When the catalyst
loading was decreased to 1 mol % [Rh], 75% yield of product 4
was observed due to incomplete conversion of aldehyde
starting material (entry 4).
methoxybenzaldehyde (1h, entry 8) confirmed that the reaction
indeed requires the 2-hydroxy group to proceed. Our method
was also effective, however, for 2-(methylthio)benzaldehyde
(1i), a substrate previously only transformed with cationic
rhodium catalysts.1,4f,5
Alkene Scope. Next, we investigated a wide range of
monosubstituted alkenes (Table 5). Electron-rich, electron-
Table 5. Alkene Scope
Aldehyde Scope. With these optimized conditions in hand
(1:1:1 ratio of rhodium, ligand and base), we studied the
reactivity of other salicylaldehyde derivatives at 5 mol % catalyst
loading and 0.4 molar concentration. As shown in Table 4, both
Table 4. Aldehyde Scope
a
b
Combined isolated yield. Regioisomeric ratio determined by
c
1
a
b
integration of crude H NMR spectrum. Four equivalents of alkene
and 10 mol % [Rh] used. Five equivalents of alkene (neat) and 2.5
mol % base used. Ten mole percent [Rh] used. The branched isomer
Isolated yield. Regioisomeric ratio was >95:5 in all cases
d
c
(determined by integration of crude 1H NMR spectrum). Four
e
f
d
equivalents of alkene used. Solvent: 2-Me-THF; 2.5 mol % base used.
e
was isolated and an ee of <5% was determined by chiral SFC (see SI).
No base added, 6.0 equiv of alkene used (neat).
g
Biphasic reaction conditions (DCE/H2O, 1:1) were applied.
sterically hindered (entry 1), as well as electron-deficient
(entries 2 and 3) and electron-rich (entries 5 and 6) 2-
hydroxybenzaldehydes gave good to excellent yields of the
corresponding linear products. 5-Iodosalicylaldehyde (1d, entry
4) reacted cleanlyno products indicative of competing
protodeiodination or Mizoroki-Heck coupling were observed.
Dihydroxybenzaldehyde (1f), which is not soluble in DCE,
reacted smoothly in 2-Me-THF (entry 6), and served as a
model substrate for the total synthesis of octaketide natural
products (vide infra). 2-Hydroxy-1-naphthaldehyde (1g)
furnished the desired product in excellent yield and
regioselectivity (entry 7). A control experiment with 2-
neutral, as well as electron-deficient alkenes (entries 1−13)
displayed good to excellent reactivity under the optimized
reaction conditions or slight modifications thereof. Electron-
rich vinylsilane 2B (entry 1), for example, underwent
hydroacylation in good yield and with excellent linear
regioselectivity. Acrolein diethyl acetal (2E, entry 4), not
previously reported to undergo hydroacylation, gave 89% yield
and 89:11 selectivity. Upon acetal deprotection, this allows for
the convenient synthesis of hard-to-access 1,4-ketoaldehydes.
Essential to our proposed natural product syntheses, polar
functional groups, as present in alcohol 2H (entry 7) and
D
dx.doi.org/10.1021/ja305593y | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
protected amine 2I (entry 8), were tolerated under the reaction
conditions. Styrene 2J and two representative derivatives 2K
and 2L (entries 9−11) underwent hydroacylation in good
yields and with synthetically useful selectivities. The conven-
tional route toward the corresponding linear reaction products
involves a two step aldol condensation/hydrogenation
sequence.28 Intermolecular hydroacylation provides a one
step protocol to these compounds from inexpensive starting
materials. Under biphasic reaction conditions, electron-deficient
acrylamide 2M underwent hydroacylation in 64% yield and
excellent linear selectivity (entry 12). Biphasic reaction
conditions tend to be superior to standard conditions for
strongly chelating alkene substrates (such as 2M). Acrylate 2N
(entry 13) gave excellent reactivity, yet only modest selectivity.
Terminal alkyne 2O (entry 14) gave good reactivity, but only
poor selectivity. No products of hydroacylation were obtained
when conjugated terminal alkyne 2P (entry 15) was subjected
to standard reaction conditions. We observe a deep red color
upon addition of base and broad peaks in the 1H NMR
spectrum; substrate 2P may undergo deprotonation and
subsequent polymerization under these reaction conditions.
Symmetrical, internal alkyne 2Q (entry 16) was moderately
reactive and resulted in the exclusive formation of the (E)
product.6a Diene 2R (entry 17) illustrates two limitations of our
method: (i) disubstituted double bonds, as present in 2-octene
or compound 2R, are generally not reactive; (ii) alkenes that
can chelate to the Rh-center due to a neighboring double bond
or heteroatom can significantly reduce the linear selectivity (in
such cases, the regioselectivity is partially controlled by the
substrate, not the catalyst; see Scheme 1a).29
of both the triflate and ketone functionalities present in 7a to
furnish 3-ylidenephthalide 7d via cyclocarbonylation.32
Natural Product Synthesis. Having explored the scope of
our new hydroacylation method, we turned our attention
toward preparing natural products 10a−h (Table 6). Aldehyde
coupling partners 8 were prepared in two steps from
commercial materials via Vilsmeier−Haack formylation,33
followed by AlCl3 deprotection of the aryl methyl ethers34
(Scheme 4a). Alkene building blocks 9 were either
commercially available or could be prepared in one step by
copper catalyzed addition of simple Grignard reagents to
enantiopure epoxides (Scheme 4b).35
Using our hydroacylation protocol, we prepared eight
octaketides (10a−h) in good isolated yields (Table 6).
Presumably due to a chelation effect, dothiorelone B was
obtained in only moderate yield under standard hydroacylation
conditions (10c, entry 3). Pleasingly, we were able to obtain
86% yield for this product under biphasic reaction conditions
(DCE/H2O). In cases where the reactions proceeded with less
than 95:5 linear selectivities, we were able to separate the two
regioisomers by chromatography. The reduced linear selectivity
in these substrates presumably stems from chelation of the
hydroxyl groups on the alkene building blocks to the rhodium
catalyst. A ninth natural product, cytosporone A (10i, R1−5
:
H),14a was synthesized from cytosporone B (10a) by LiOH-
mediated saponification of the ethyl ester (see SI).
This study represents the first total synthesis of compounds
10c, 10d, 10f, 10g, 10h, and 10i.36 While we generally observed
good agreement between our spectroscopic data and the data
presented in the isolation reports, we found significant
Removing or Functionalizing the 2-Hydroxy Group in
Hydroacylation Product 4. Encouraged by the relatively
broad alkene scope (Table 5), we examined methods to use the
2-hydroxy moiety as a functional handle. As shown in Scheme
3, hydroacylation product 4 can be smoothly transformed into
1
deviations in the H and 13C NMR data for compound 10f
(see SI for a comparison table). We propose that the material
isolated by Huang et al. in 200914d is not identical to
compound 10f, which we have synthesized and fully
characterized as part of this study. Octaketides 10b, 10c, 10f,
and 10h were prepared in enantiopure form, allowing the
assignment of the absolute configuration as (R) in natural
products 10b and 10h, based on comparison of [α]D values
(see Table 6). For natural product dothiorelone B (10c), the
stereochemical assignment was confirmed by an X-ray crystal
structure (Figure 2).
a
Scheme 3. Derivatization of Hydroacylation Product 4
The advantages of our synthetic approach include easy access
to functionalized terminal alkenes and high functional group
tolerance for the C−C bond forming reaction.37 Dothiorelone
A, for example, is the only natural product with a hydroxy-
substituent on the acyl side chain that has been previously
synthesized,36c but the reported synthetic route required 12
linear steps from commercial starting materials. We were able
to prepare this compound in three steps in the longest linear
sequence (and four steps in total) using an atom-economic
intermolecular hydroacylation reaction as the key step.
a
Reaction conditions: (i) Tf2O (1.5 equiv), pyridine (1.5 equiv),
DMAP (0.4 equiv), DCM, 0 °C → RT, 16 h, 85%; (ii) BH3·Me2NH
(1.05 equiv), K2CO3 (1.0 equiv), Pd(PPh3)4 (5 mol %), MeCN, 40
°C, 16 h, 90%; (iii) Ar−B(OH)2 (1.5 equiv), K2CO3 (2.5 equiv),
Pd(PPh3)4 (5 mol %), toluene, reflux, 6 h, 98% (Ar = 4-Et-Ph); (iv)
CO (1 atm), Et3N (2.0 equiv), Pd(OAc)2 (10 mol %), dppp (10 mol
%), DMF, 60 °C, 1 h, 90%.
Deuterium Labeling: Insights into Regioselectivity
and the Turnover-Limiting Step. To gain insight into the
reaction mechanism, we subjected deuterium-labeled salicylal-
dehyde (d-1 or h/d-1)38 and selected alkenes to standard
reaction conditions for 16−24 h (Table 7). After typical
the corresponding triflate 7a. Using a method reported by
Lipshutz and co-workers, we were able to remove the triflate
functionality to afford 7b in 90% yield without observable
reduction of the reactive aromatic ketone.30 The 2-hydroxy
functionality in aldehydes 1a−g can thus be considered a
removable directing group.31
Triflate 7a can engage in Pd-catalyzed cross-coupling
reactions, as exemplified by the highly efficient Suzuki−Miyaura
coupling shown in Scheme 3 (product 7c). We took advantage
1
workup, H NMR analysis revealed the deuterium content in
residual salicylaldehyde h/d-1. Both regioisomeric products h/
d/d1/d2/d3/d4-3 and h/d/d1/d2/d3/d4-4 were subsequently
1
2
isolated, and the purified material was analyzed by H and H
NMR spectroscopy, as well as by high-resolution mass
spectrometry.
E
dx.doi.org/10.1021/ja305593y | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Table 6. Total Synthesis of Natural Products via Alkene Hydroacylation
a
b
c
d
1
Nonspecified R substituents are H. Combined isolated yield. Regioisomeric ratio determined by integration of crude H NMR spectrum. An
e
[α]D of +3.4 was reported for synthetic (S)-dothiorelone A (ref 36c). Biphasic conditions used (1:1 DCE:H2O). General reaction conditions:
[Rh(COD)Cl]2 (5 mol %), (Ra,R,R)-SIPHOS-PE (5−10 mol %), K3PO4 (5−10 mol %), DCE, 70 °C, 18−72 h. See SI for details.
Scheme 4. Representative Synthesis of Aldehyde and Alkene
Building Blocks 8a and 9a
substrates 2J and 2N and largely irreversible40 for unactivated
alkenes such as 2. In conjunction with our observation of
catalytic resting state [Rh(COD)Cl((Ra,R,R)-SIPHOS-PE)] (5,
vide infra), these findings indicate that hydride insertion is
turnover-limiting in the linear pathway. These mechanistic
studies provide deeper insight into the high regioselectivity
observed in our catalytic hydroacylation using Rh/(Ra,R,R)-
SIPHOS-PE.
a
When 1-octene was subjected to standard conditions (Table
7, entry 1), a deuterium content of ca. 25% was found in
residual aldehyde 1 and both the α- and β-positions in product
4. This observation indicates that the branched hydride
insertion (iii-a) is readily occurring and fully reversible under
the reaction conditions (see Scheme 5). Reversible insertion to
generate the branched intermediate IIIa would lead to
incorporation of deuterium into the terminal position of olefin
2. Subsequent linear deuterioacylation of this olefin would
generate ketone 4 with deuterium incorporated at both the α-
and β-positions. Because no appreciable quantities of branched
product 3 were formed, we conclude that reductive elimination
to form ketone 3 (iv-a, Scheme 5) is turnover-limiting in the
branched-pathway and relatively slow compared to the
turnover-limiting step in the linear pathway (>98:2 linear
selectivity observed).
a
Reaction conditions: (i) DMF (5.0 equiv), POCl3 (1.2 equiv), RT, 19
h, 84%; (ii) AlCl3 (10 equiv), DCM, 0 °C → 40 °C, 87%; (iii) CuCN
(10 mol %), THF, −78 °C → RT, 14 h, 72%.
Treating the same substrates with less selective catalyst
system Rh/(R)-SIPHOS allowed characterization of both
regioisomers 3 and 4 (entry 2). Incorporation of deuterium
into both the α- and β-position of product 4 occurred in
approximately equal amounts, which provides further support
that the branched insertion is reversible for this substrate (iii-a,
Scheme 5). In branched product 3, a small but measurable
quantity of deuterium was found in the α-position, indicating
that linear hydride insertion (iii-b, Scheme 5) is largely but not
fully irreversible.41 If linear insertion was fully reversible, a
deuterium content of 15−20% (statistical scrambling of all
exchangeable hydrogen/deuterium atoms) would be expected
in the α-position of product 3.
Further support for largely irreversible linear hydride
insertion into unactivated olefins is provided by an experiment
using deuterium-labeled alkene d-2S (Table 7, entry 3). When
nonlabeled aldehyde 1 was treated with d-2S,42 a nonstatistical
deuterium distribution was observed in product 4S (ca. 50% D
found in β, which is higher than the 30% D incorporation
expected for statistical scrambling). However, a relatively small
Figure 2. Solid-state structure of octaketide natural product
dothiorelone B (10c). Single crystals were obtained by slow diffusion
of hexanes into a DCE solution. O, red; C, black; H, white.
The mechanistic conclusions derived from these studies are
summarized in the catalytic cycle presented in Scheme 5. In
agreement with literature reports,8c,18 we propose that oxidative
addition (step i) and alkene coordination (step ii) are rapid and
reversible under the reaction conditions. The magnitude of the
kinetic barriers for elemental steps iii-a,b and iv-a,b (Scheme 5)
can vary significantly for different types of alkenes (unactivated,
activated or chelating), but a general trend is that branched
hydride insertion (iii-a) is fast and fully reversible. In contrast to
previous reports on olefin hydroacylation,8c,18 we propose that
linear hydride insertion (iii-b) is fully irreversible for activated
F
dx.doi.org/10.1021/ja305593y | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Table 7. Deuterium-Labeling Studies Employing Deuterated Aldehyde or Alkene
a
b
1
Deuterioenriched salicylaldehyde h/d-1 (62% D) was conveniently synthesized as shown in Scheme 6b. Determined by integration of crude H
c
NMR spectrum. Deuterium content of purified 3 and 4 determined by 1H and 2H NMR spectroscopy, and high-resolution DART mass
spectrometry (estimated experimental error: 2−3%). Reaction conditions: [Rh(COD)Cl]2 (5 mol %), ligand (10 mol %), K3PO4 (10 mol %),
d
e
DCE, 70 °C, 16−24 h. Three equivalents of alkene used. DCM was used as solvent; reaction temperature: 30 °C. See SI for further details.
of substrates.43 Higher levels of deuterium were observed in the
ab
Scheme 5. Proposed Catalytic Cycle
α-position than in the β-position in linear products 4J and 4N.
This result can be explained by a mechanistic scenario in which
the branched hydride insertion, iii-a, is fast and steps i and ii
(Scheme 5) are fully reversible. At the beginning of this
process, alkene substrates with a relatively high terminal
deuterium content would be generated, along with aldehyde
h/d-1 with a reduced deuterium content. Once the slower and
irreversible linear insertion occurs, lower levels of deuterium are
incorporated into the β-position, while the high levels of
deuterium in the α-position reflect the high terminal deuterium
content of the initially formed alkene.
Homoallylic sulfide 2A was transformed to branched product
3A with no measurable deuterium incorporation in the α-
position (entry 6). This observation can be explained by slow
and irreversible linear hydride insertion, which is particularly
plausible for this substrate where chelation of the sulfur atom to
the rhodium catalyst is expected to favor branched over linear
hydride insertion through a five-membered rhodacycle.6h
When deuterium-labeled aldehyde d-1 was subjected to
standard reaction conditions in the absence of alkene (Scheme
6a), a decrease of the aldehyde deuterium content to about
a
57% was found after 40 h. This result indicates that the
phenolic hydrogen atoms can slowly exchange with the
aldehyde deuterium via rhodium catalysis.44 This H/D
exchange at the aldehyde does not affect our mechanistic
proposal, however, because we derive our conclusions from the
relative, not absolute, deuterium content in products 3 and 4.
As shown in Scheme 6b, we used this process to achieve a one-
step synthesis of deuterium-enriched aldehyde h/d-1 (62% D)
from 1 and deuterated methanol.45
In addition, we confirmed that product 4 does not
incorporate deuterium at the α-position through a deprotona-
tion-reprotonation mechanism involving exchangeable phenolic
D atoms. No measurable deuterium incorporation was found in
product 4 after submitting 4 and h/d-1 to standard reaction
conditions (Scheme 6c), indicating that the α-deuterium
incorporation presented in Table 7 stems indeed from an
organometallic mechanism.
(i) Oxidative addition; (ii) alkene coordination; (iii) hydride
b
insertion; (iv) reductive elimination. L: (Ra,R,R)-SIPHOS-PE is
likely bound to Rh throughout; the second ligand in intermediates I
and III is likely alkene or solvent.39
but measurable quantity of deuterium was found in both
residual aldehyde h/d-1 and in the α-position of product 4S.
This result indicates that once intermediate IIIb is formed, only
a small fraction of molecules can undergo β-hydride elimination
back to intermediate II, explaining the deuterium incorporation
into 1 and into the α-position of 4S (Scheme 5).
Subjecting styrene (2J) and tert-butyl acrylate (2N)
separately to deuterium-enriched aldehyde h/d-1 under Rh/
(Ra,R,R)-SIPHOS-PE catalysis gave almost identical deuterium
distributions for the two alkenes in both the corresponding
branched and linear products (entries 4 and 5). In the branched
products 3J and 3N, no measurable deuterium incorporation
was observed at the α-position. This result suggests that linear
hydride insertion (iii-b, Scheme 5) is irreversible for these types
While our observation of turnover-limiting hydride insertion
is unprecedented in the olefin hydroacylation literature, Casey
G
dx.doi.org/10.1021/ja305593y | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
of the aldehyde starting material (Figure 3a).47 Toward the end
of this transformation, 31P NMR data indicated the formation
of a significant amount of complex 11, which is tentatively
assigned to [Rh(CO)Cl((Ra,R,R)-SIPHOS-PE)]48 on the basis
of high-resolution ESI mass spectrometry49 and comparison of
the obtained 31P NMR (162 MHz, DCE, 298 K; δ = 126.4
ppm, JRh−P = 205 Hz;) and IR (neat; ν(CO) = 1994 cm−1)
spectroscopic data with that of a closely related, known
rhodium complex.50
Scheme 6. Control Experiments and Synthesis of Deuterium-
Enriched Salicylaldehyde h/d-1
a
When an 8-fold excess of alkene 2 was used, the aldehyde
was consumed in only 20 h, despite the lack of stirring (Figure
3b).47 A linear kinetic profile was found up to approximately
60% conversion under these conditions. This finding suggests
that the potassium salt of salicylaldehyde (6a), not
salicylaldehyde itself, is the reactive species and that the
concentration of 6a is likely limited by its low solubility in DCE
and thus constant for a significant period.51 After 22 h, analysis
of the reaction mixture by 31P NMR indicated that a relatively
small amount of CO complex 11 was formed (observation of
this complex implies reductive decarbonylation of salicylalde-
hyde). Instead, the major species observed was resting state
complex 5 ([Rh(COD)Cl((Ra,R,R)-SIPHOS-PE)]).52
In contrast, when an 8-fold excess of aldehyde 1 was used,
the transformation proceeded with less than 10% conversion of
the starting alkene (Figure 3c). The 31P NMR spectrum
recorded after 24 h showed that complex 5, the observed
resting state for hydroacylation, had disappeared entirely and
CO complex 11 represented the predominant species.53 When
monitoring this experiment by 1H NMR spectroscopy, we
observed significant broadening of the phenolic singlets in 1
and 4 at the time when the reaction progress began to slow
down. Because this reaction mixture, unlike all others described
in this section, eventually turned into a homogeneous mixture,
we propose that the observed peak broadening is a result of the
gradual solubilization of base 6a under the reaction conditions.
From these observations, we conclude: (i) excess alkene
generally increases the rate of hydroacylation and disfavors
decarbonylation; (ii) the precise stoichiometry of base is critical
because deprotonated aldehyde 6 is necessary to guarantee high
reaction rates, but an excess amount of 6 in solution leads to
irreversible catalyst decomposition via reductive decarbon-
ylation. This finding explains the sensitivity of the reaction to
the amount and nature of base (Tables 2 and S4), as well as to
a
Reaction conditions: 10 mol % [Rh], 10 mol % ligand and 10 mol %
1
base. Deuterium content of 1 determined by H NMR spectroscopy.
Deuterium content of 4 determined by 1H and 2H NMR spectroscopy,
as well as high-resolution DART mass spectrometry (estimated
experimental error: 2−3%).
et al. have made similar mechanistic findings when studying a
highly linear-selective rhodium-catalyst for the hydroformyla-
tion of 1-hexene.46
In Situ NMR Monitoring: Insights into Decarbon-
ylation and the Role of Ligand. Having established a
catalytic proposal for the hydroacylation pathway, we sought
insight into the decarbonylation side reaction that can
drastically limit catalyst turnover. Because of the heterogeneous
base required in our catalytic protocol, it was difficult to
conduct kinetic studies and establish a precise rate law for
hydroacylation or decarbonylation. Therefore, we obtained
qualitative kinetic data and aimed to identify catalytic
intermediates. In three separate experiments, we varied the
relative concentration of substrates 1 and 2 and monitored the
reaction progress by 1H and 31P NMR spectroscopy (Figure 3).
Under standard hydroacylation conditions, using 2 equiv of
alkene, the reaction followed an exponential time course with
no observable induction period, leading to consumption of 96%
1
Figure 3. In situ H NMR monitoring (400 MHz, DCE, 343 K) of the reaction of 1 with 2 at different starting material equivalents.
H
dx.doi.org/10.1021/ja305593y | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
the solvent (Table S6), and provides a rationale for why all
studied soluble bases led to low yields.
We can only speculate at this point about how excess base
triggers catalyst decomposition via reductive decarbonylation.
In reactions that resulted in rapid decarbonylation, we generally
observed small amounts of phenol (12, Chart 1), as well as
complex alkene building blocks, our method is ideally suited for
the preparation of analogues of anticancer drug candidate
cytosporone B. Mechanistic studies revealed that for
unactivated alkenes, branched hydride insertion is rapid and
fully reversible, while linear hydride insertion is largely
irreversible. Comparison with a number of related, less effective
ligands showed that bulky phosphoramidite (Ra,R,R)-SIPHOS-
PE is uniquely suited to promote reactivity and regioselectivity
in this transformation. Our findings provide insights that will be
useful for future catalyst development and ligand design, which
will be key to solving the remaining challenges in hydro-
acylation.
Chart 1. Observed Products of Reductive Decarbonylation
and Proposed Relative Rate-Dependence
ASSOCIATED CONTENT
■
S
* Supporting Information
bis(2-hydroxyphenyl)methanone (13). We propose that the
organometallic pathway leading to this side product is higher
than first order in deprotonated salicylaldehyde (6), while
hydroacylation is likely first order in both 6 and alkene (2).
Such rate dependences would explain the relatively high
propensity for decarbonylation at increased concentrations of 6,
as well as the beneficial effect of excess alkene (Chart 1, eq 12).
Iridium complexes are known to catalyze aldehyde decarbon-
ylation in preference to olefin hydroacylation.8a However, when
we replaced precatalyst [Rh(COD)Cl]2 with [Ir(COD)Cl]2 for
the reaction of 1 with 2, we observed 35% isolated yield for
linear hydroacylation product 4, corresponding to a turnover
number of seven. This result indicates that ligand (Ra,R,R)-
SIPHOS-PE is effective at suppressing competing reductive
decarbonylation even when bound to an iridium center. Our
results obtained with structurally related ligands (Table 1)
indicate that three key factors are responsible for the unique
reactivity and selectivity observed with (Ra,R,R)-SIPHOS-PE:
(i) large steric bulk, (ii) a synergistic effect between the three
stereogenic elements, and (iii) the electronic properties of
phosphoramidite ligands.12a
Full synthetic, crystallographic, and characterization details.
Crystallographic data have been deposited with the Cambridge
Crystallographic Data Centre under CCDC nos. 893320−
893322. This material is available free of charge via the Internet
at http://pubs.acs.org.
AUTHOR INFORMATION
■
Corresponding Author
dongv@uci.edu
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We thank the University of Toronto, Canada Foundation for
Innovation, Ontario Research Fund, and NSERC for funding.
V.M.D. is grateful for an Alfred P. Sloan Fellowship. M.v.D. is
grateful for a postdoctoral fellowship provided by the German
National Academy of Sciences Leopoldina (LPDS 2009-43).
C.M.L. is grateful for an Ontario Graduate Scholarship. The
authors wish to acknowledge the Canadian Foundation for
Innovation, project number 19119, and the Ontario MRI for
funding of the Centre for Spectroscopic Investigation of
Complex Organic Molecules and Polymers. We thank Drs.
Burrow and Burns of the CSICOMP NMR facility for
assistance with NMR spectroscopy and Dr. Alan Lough for
assistance with X-ray crystallography.
Collectively, our mechanistic findings suggest that ligand
(Ra,R,R)-SIPHOS-PE increases the rate of hydroacylation by
decreasing the barrier for reductive elimination toward linear
product 4, while suppressing catalyst decomposition via
reductive decarbonylation.
REFERENCES
■
(1) For a recent review article, see: Willis, M. C. Chem. Rev. 2010,
110, 725−748.
(2) (a) Trost, B. M. Science 1991, 254, 1471−1477. (b) Trost, B. M.
Angew. Chem., Int. Ed. 1995, 34, 259−281.
CONCLUSION
■
We developed a highly efficient method for the linear-selective
intermolecular hydroacylation of a wide range of monosub-
stituted alkenes with salicylaldehyde derivatives. Compared to
the previously reported method that allows hydroacylation of 2
with 1,6f our method stands out due to the broader alkene
scope and a 20-fold decrease in rhodium loading. In contrast to
most hydroacylation methods reported to date, the catalytic
protocol is compatible with environmentally benign solvent, as
well as neat or biphasic reaction conditions. We have
demonstrated that through a triflate intermediate, the hydro-
acylation products can be transformed into deoxygenated or
more complex building blocks. This catalytic strategy allowed
us to prepare eight natural products from the dothiorelone,
cytosporone, and phomopsin families, five of which were
prepared for the first time. Because of the ease of access to
(3) To the best of our knowledge, there is only one report of an
intermolecular alkene hydroacylation reaction as part of a natural
product synthesis: Kim, G.; Lee, E.-j. Tetrahedron: Asymmetry 2001,
12, 2073−2076.
(4) For selected studies of alkene hydroacylation relying on S-
chelating aldehydes: Willis, M. C.; McNally, S. J.; Beswick, P. J. Angew.
Chem., Int. Ed. 2004, 43, 340−343. (b) Willis, M. C.; Randell-Sly, H.
E.; Woodward, R. L.; Currie, G. S. Org. Lett. 2005, 7, 2249−2251.
(c) Willis, M. C.; Woodward, R. L. J. Am. Chem. Soc. 2005, 127,
18012−18013. (d) Willis, M. C.; Randell-Sly, H. E.; Woodward, R. L.;
McNally, S. J.; Currie, G. S. J. Org. Chem. 2006, 71, 5291−5297.
(e) Moxham, G. L.; Randell-Sly, H. E.; Brayshaw, S. K.; Woodward, R.
L.; Weller, A. S.; Willis, M. C. Angew. Chem., Int. Ed. 2006, 45, 7618−
7622. (f) Moxham, G. L.; Randell-Sly, H. E.; Brayshaw, S. K.; Weller,
A. S.; Willis, M. C. Chem.Eur. J. 2008, 14, 8383−8397. (g) Osborne,
J. D.; Willis, M. C. Chem. Commun. 2008, 5025−5027. (h) Randell-Sly,
I
dx.doi.org/10.1021/ja305593y | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
H. E.; Osborne, J. D.; Woodward, R. L.; Currie, G. S.; Willis, M. C.
Tetrahedron 2009, 65, 5110−5117.
hydroacylation using a chiral catalyst system, see: (f) Tanaka, K.; Fu,
G. C. J. Am. Chem. Soc. 2003, 125, 8078−8079.
(5) Chaplin, A. B.; Hooper, J. F.; Weller, A. S.; Willis, M. C. J. Am.
Chem. Soc. 2012, 134, 4885−4897.
(14) (a) Brady, S. F.; Wagenaar, M. M.; Singh, M. P.; Janso, J. E.;
Clardy, J. Org. Lett. 2000, 2, 4043−4046. (b) Xu, Q.; Wang, J.; Huang,
Y.; Zheng, Z.; Song, S.; Zhang, Y.; Su, W. Acta Oceanol. Sin. 2004, 23,
541−547. (c) Huang, Z.; Cai, X.; Shao, C.; She, Z.; Xia, X.; Chen, Y.;
Yang, J.; Zhou, S.; Lin, Y. Phytochemistry 2008, 69, 1604−1608.
(d) Huang, Z.; Guo, Z.; Yang, R.; Yin, X.; Li, X.; Luo, W.; She, Z.; Lin,
Y. Chem. Nat. Prod. 2009, 45, 625−628. (e) Xu, J.; Kjer, J.; Sendker, J.;
(6) For selected studies of alkene hydroacylation relying on O-
chelating aldehydes: (a) Kokobu, K.; Matsumasa, K.; Nishinaka, Y.;
Miura, M.; Nomura, M. Bull. Chem. Soc. Jpn. 1999, 72, 303−311.
(b) Tanaka, M.; Imai, M.; Yamamoto, Y.; Tanaka, K.; Shimowatari,
M.; Nagumo, S.; Kawahara, N.; Suemune, H. Org. Lett. 2003, 5, 1365−
1367. (c) Imai, M.; Tanaka, M.; Tanaka, K.; Yamamoto, Y.; Imai-
Ogata, N.; Shimowatari, M.; Nagumo, S.; Kawahara, N.; Suemune, H.
J. Org. Chem. 2004, 69, 1144−1150. (d) Tanaka, K.; Tanaka, M.;
Suemune, H. Tetrahedron Lett. 2005, 46, 6053−6056. (e) Stemmler, R.
T.; Bolm, C. Adv. Synth. Catal. 2007, 349, 1185−1198. (f) Imai, M.;
Tanaka, M.; Nagumo, S.; Kawahara, N.; Suemune, H. J. Org. Chem.
2007, 72, 2543−2546. (g) Phan, D. H. T.; Kou, K. G. M.; Dong, V. M.
J. Am. Chem. Soc. 2010, 132, 16354−16355. (h) Coulter, M. M.; Kou,
K. G. M.; Galligan, B.; Dong, V. M. J. Am. Chem. Soc. 2010, 132,
16330−16333. (i) Zhang, H.-J.; Bolm, C. Org. Lett. 2011, 13, 3900−
3903. (j) Murphy, S. M.; Petrone, D. A.; Coulter, M. C.; Dong, V. M.
Org. Lett. 2011, 13, 6216−6219. (k) Murphy, S. M.; Coulter, M. C.;
Dong, V. M. Chem. Sci. 2012, 3, 355−358.
(7) For selected studies of alkene hydroacylation relying on
stoichiometric or catalytic transformation of the aldehyde component
into an imine, see: (a) Suggs, J. W. J. Am. Chem. Soc. 1979, 101, 489.
(b) Suggs, J. W.; Wovkulich, M. J.; Cox, S. D. Organometallics 1985, 4,
1101−1107. (c) Jun, C.-H.; Hong, J.-B.; Lee, D.-Y. Synlett 1999, 1−12.
(d) Jun, C.-H.; Lee, D.-Y.; Lee, H.; Hong, J.-B. Angew. Chem., Int. Ed.
2000, 39, 3070−3072. (e) Willis, M. C.; Sapmaz, S. Chem. Commun.
2001, 2558−2559. (f) Jun, C.-H.; Lee, J. H. Pure Appl. Chem. 2004, 76,
577−587. (g) Park, Y. J.; Park, J.-W.; Jun, C.-H. Acc. Chem. Res. 2008,
41, 222−234. (h) Vautravers, N. R.; Regent, D. D.; Breit, B. Chem.
Commun. 2011, 47, 6635−6637.
Wray, V.; Guan, H.; Edrada, R.; Muller, W. E. G.; Bayer, M.; Lin, W.;
̈
Wu, J.; Proksch, P. Bioorg. Med. Chem. 2009, 17, 7362−7367.
(15) Zhan, Y.; Du, X.; Chen, H.; Liu, J.; Zhao, B.; Huang, D.; Li, G.;
Xu, Q.; Zhang, M.; Weimer, B. C.; Chen, D.; Cheng, Z.; Zhang, L.; Li,
Q.; Li, S.; Zheng, Z.; Song, S.; Huang, Y.; Ye, Z.; Su, W.; Lin, S.-C.;
Shen, Y.; Wu, Q. Nat. Chem. Biol. 2008, 4, 548−556.
(16) Liu, J.; Zeng, H.; Zhang, L.; Zhan, Y.; Chen, Y.; Wang, Y.;
Wang, J.; Xiang, S.; Liu, W.; Wang, W.; Chen, H.; Shen, Y.; Su, W.;
Huang, P.; Zhang, H.; Wu, Q. Cancer Res. 2010, 70, 3628−3637.
(17) The reactions were carried out according to the standard
procedures, as specified in the original reports by Jun (reaction time: 2
h; ref 7d) and Suemune (reaction time: 24 h; ref 6f).
(18) Fairlie, D. P.; Bosnich, B. Organometallics 1988, 7, 946−954.
(19) For a study in which a similar reactivity difference was observed
when using two diastereomeric BINOL-derived phosphoramidite
ligands, see: Leitner, A.; Shekhar, S.; Pouy, M. J.; Hartwig, J. F. J.
Am. Chem. Soc. 2005, 127, 15506−15514. We did not find evidence
for a cyclometalation reaction between the rhodium centre and the
ligand, as reported by Hartwig for his iridium catalyst.
(20) For a seminal report on the use of ligand (Ra,R,R)-SIPHOS-PE
in transition metal catalysis and the observation of a mismatch in
(Ra,S,S)-SIPHOS-PE, see: Zhou, H.; Wang, W.-H.; Fu, Y.; Xie, J.-H.;
Shi, W.-J.; Wang, L.-X.; Zhou, Q.-L. J. Org. Chem. 2003, 68, 1582−
1584.
(8) Krische has recently reviewed alkene hydroacylation methods
that do not require chelation on part of the aldehyde/imine substrate
(alkene scope is narrow in each case): (a) Leung, J. C.; Krische, M. J.
Chem. Sci. 2012, 3, 2202−2209. For important original reports, see:
(b) Lenges, C. P.; Brookhart, M. J. Am. Chem. Soc. 1997, 119, 3165−
3166. (c) Roy, A. H.; Lenges, C. P.; Brookhart, M. J. Am. Chem. Soc.
2007, 129, 2082−2093. (d) Tanaka, K.; Shibata, Y.; Suda, T.;
Hagiwara, Y.; Hirano, M. Org. Lett. 2007, 9, 1215−1218. (e) Shibahara,
F.; Bower, J. F.; Krische, M. J. J. Am. Chem. Soc. 2008, 130, 14120−
14122. (f) Taniguchi, H.; Ohmura, T.; Suginome, M. J. Am. Chem. Soc.
2009, 131, 11298−11299. (g) Shibata, Y.; Tanaka, K. J. Am. Chem. Soc.
2009, 131, 12552−12553.
(21) (a) Tomori, H.; Fox, J. M.; Buchwald, S. L. J. Org. Chem. 2000,
65, 5334−5341. (b) Mauger, C. C.; Mignani, G. A. Aldrichimica Acta
2006, 39, 17−24.
(22) For solid-state structures of other transition metal complexes
including SIPHOS-type ligands, see: (a) Hu, A.-G.; Fu, Y.; Xie, J.-H.;
Zhou, H.; Wang, L.-X.; Zhou, Q.-L. Angew. Chem., Int. Ed. 2002, 41,
2348−2350. (b) Shi, W.-J.; Wang, L.-X.; Fu, Y.; Zhu, S.-F.; Zhou, Q.-L.
Tetrahedron: Asymmetry 2003, 14, 3867−3872. (c) Shi, W.-J.; Xie, J.-
H.; Zhou, Q.-L. Tetrahedron: Asymmetry 2005, 16, 705−710. (d) Yan,
P.-C.; Xie, J.-H.; Hou, G.-H.; Wang, L.-X.; Zhou, Q.-L. Adv. Synth.
Catal. 2009, 351, 3243−3250.
(23) An attempt to enhance reactivity with ligands PPh3 and
DavePhos by varying the [Rh]/ligand ratio led to only minor
improvements (see SI Table S2 for details).
(9) Gunatilaka, A. A. L. J. Nat. Prod. 2006, 69, 509−526.
(10) In a double-chelation assisted olefin hydroacylation, both the
aldehyde and alkene substrates bind to the Rh catalyst.
(24) A [Rh]/ligand ratio of 2:1 was used for part of the optimization
studies, because we had found that at this catalyst loading, the [Rh]/
ligand ratio was not a critical parameter (see SI, Table S4).
(25) Anastas, P. T.; Warner, J. C. Green ChemistryTheory and
Practice; Oxford University Press: New York, 1998.
(26) Aycock, D. Org. Process Res. Dev. 2007, 11, 156−159.
(27) In a recent study, Weller and Willis have found a similar
concentration dependency of reactivity (see ref 5).
(11) Diastereomeric ligands (Ra,R,R)-SIPHOS-PE and (Sa,R,R)-
SIPHOS-PE are commercially available under the misleading names
(R)-SIPHOS-PE and (S)-SIPHOS-PE.
(12) (a) Teichert, J. F.; Feringa, B. L. Angew. Chem., Int. Ed. 2010, 49,
2486−2528. (b) de Vries, A. H. M.; Meetsma, A.; Feringa, B. L. Angew.
Chem., Int. Ed. 1996, 35, 2374−2376. (c) Hoffman, T. J.; Carreira, E.
M. Angew. Chem., Int. Ed. 2011, 50, 10670−10674. (d) Trost, B. M.;
Silverman, S. M. J. Am. Chem. Soc. 2012, 134, 4941−4954. (e) Xie, J.-
H.; Yan, P.-C.; Zhang, Q.-Q.; Yuan, K.-X.; Zhou, Q.-L. ACS Catal.
2012, 2, 561−564.
(13) For the use of chiral ligands to enhance reactivity in other
nonasymmetric transformations, see for example: (a) Wolfe, J. P.;
Wagaw, S.; Buchwald, S. L. J. Am. Chem. Soc. 1996, 118, 7215−7216.
(b) Roy, A. H.; Hartwig, J. F. J. Am. Chem. Soc. 2003, 125, 8704−8705.
(c) Shen, Q.; Shekhar, S.; Stambuli, J. P.; Hartwig, J. F. Angew. Chem.,
Int. Ed. 2005, 44, 1371−1375. For an example of site-selective
transformation using a chiral catalyst, see: (d) Lewis, C. A.; Miller, S. J.
Angew. Chem., Int. Ed. 2006, 45, 5616−5619. For a review of
regioselective reactions on chiral substrates employing chiral catalysts,
see: (e) Kumar, R. R.; Kagan, H. B. Adv. Synth. Catal. 2010, 352, 231−
242. For an example of the parallel kinetic resolution of 4-alkynals via
(28) Murphy, W. S.; Sompong, W. J. Chem. Soc., Perkin Trans. 1
1980, 1567−1577.
(29) For a report on hexadienes as chelating hydroacylation
substrates, see ref 6b.
(30) Lipshutz, B. H.; Buzard, D. J.; Vivian, R. W. Tetrahedron Lett.
1999, 40, 6871−6874.
(31) Rousseau, G.; Breit, B. Angew. Chem., Int. Ed. 2011, 50, 2450−
2494.
(32) Ciattini, P. G.; Mastropietro, G.; Morera, E.; Ortar, G.
Tetrahedron Lett. 1993, 34, 3763−3766.
(33) Cannon, J. G.; Hamer, L. R.; Ilhan, M.; Bhatnagar, R. K.; Longi,
J. P. J. Med. Chem. 1984, 27, 190−195.
(34) Wei, W.-G.; Yao, Z.-J. J. Org. Chem. 2005, 70, 4585−4590.
J
dx.doi.org/10.1021/ja305593y | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(35) Jastrzebska, I.; Scaglione, J. B.; DeKoster, G. T.; Rath, N. P.;
Covey, D. F. J. Org. Chem. 2007, 72, 4837−4843.
(50) 31P NMR and IR spectroscopic data of complexes 5 (162 MHz,
DCE, 298 K; δ = 116.1 ppm, JRh‑P = 246 Hz) and 11 (162 MHz, DCE,
298 K; δ = 126.4 ppm, JRh‑P = 205 Hz; IR (neat): ν(CO) = 1994
cm−1) is in excellent agreement with data reported for closely related
complexes [Rh(COD)Cl(L2)] (121.5 MHz, CDCl3, RT; δ = 112.9
ppm, JRh‑P = 243 Hz) and [Rh(CO)Cl(L2)2] (121.5 MHz, CDCl3,
RT; δ = 122.3 ppm, JRh‑P = 194 Hz; IR (neat): ν(CO) = 1983 cm−1).
L2: bis(4-methoxyphenyl)dibenzylphosphoramidite. Shejwalkar, P.;
Sedinkin, S. L.; Bauer, E. B. Inorg. Chim. Acta 2011, 366, 209−218.
(51) The observed saturation kinetics could also be explained by the
formation of a rhodium−salicylaldehyde complex as a catalytic resting
state. Our 31P and ESI−MS data do not support such a hypothesis,
however. Once the complex [Rh(COD)1−((Ra,R,R)-SIPHOS-PE)] is
formed, rapid oxidative addition to the corresponding Rh(III)-hydride
species seems to occur.
(36) Previous total syntheses: (a) Huang, H.; Zhang, L.; Zhang, X.;
Ji, X.; Ding, X.; Shen, X.; Jiang, H.; Liu, H. Chin. J. Chem. 2010, 28,
1041−1043 (cytosporone B). (b) Yoshida, H.; Morishita, T.; Ohshita,
J. Chem. Lett. 2010, 39, 508−509 (cytosporone B and phomopsin C).
(c) Izuchi, Y.; Koshino, H.; Hongo, Y.; Kanomata, N.; Takahashi, S.
Org. Lett. 2011, 13, 3360−3363 (dothiorelone A).
(37) Cytosporone B, and simple derivatives thereof, has been
prepared via Friedel−Crafts acylation (ref 16). However, this route is
only viable for ketones with nonfunctionalized side chains.
(38) Deuterated salicylaldehyde d-1 was synthesized according to ref
6c. Deuteroenriched aldehyde h/d-1 was prepared as shown in Scheme
6b.
(39) When monitoring our reactions by ESI mass spectrometry, we
did not observe fragments corresponding to two (Ra,R,R)-SIPHOS-PE
ligands bound to rhodium. For less bulky ligand (R)-SIPHOS, we have
in contrast observed such fragments. We thus propose that one ligand
(R)-SIPHOS-PE is bound to rhodium throughout the catalytic cycle,
while any vacant sites are occupied by alkene or solvent.
(52) Complex 5 is likely an off-cycle resting state in equilibrium with
the corresponding product-bound complex (not observed in large
quantity). See also ref 50.
(53) Analysis of the reaction mixture by GCMS confirmed that
reductive decarbonylation occurred (i.e., formation of side products 12
and 13, see Chart 1).
(40) We use the term ‘largely irreversible’ hydride insertion to
describe a partially reversible process that within the continuum
between a fully reversible and a fully irreversible reaction is relatively
close to the irreversible extreme case. In terms of kinetic barriers, our
findings imply that from intermediate IIIb (Scheme 5) the barrier for
β-hydride elimination is approximately 1−2 kcal/mol higher than the
barrier for reductive elimination, leading to product 4. See also ref 46
and: (a) Hartwig, J., Ed.; Organotransition Metal Chemistry, University
Science Books: Sausalito, CA, 2010; pp 760−762. (b) Casey, C. P.;
Martins, S. C.; Fagan, M. A. J. Am. Chem. Soc. 2004, 126, 5585−5592.
(41) Ligands (R)-SIPHOS and (Ra,R,R)-SIPHOS-PE differ most
significantly in regioselectivity, not reactivity. The results obtained in
the deuterium labeling studies using (R)-SIPHOS are thus likely
applicable to (Ra,R,R)-SIPHOS-PE.
(42) 2-Deuteroundec-1-ene (d-2S) was prepared according to:
Anderson, B. J.; Keith, J. A.; Sigman, M. S. J. Am. Chem. Soc. 2010,
132, 11872−11874.
(43) An alternative explanation for this result could be that the
microscopic reverse of hydride insertion, β-hydride elimination, occurs
with stereospecifity. In other words, if deuterium inserts into the
double bond, the deuterium atom, and not the geminal hydrogen
atom, will undergo the β hydride elimination. Such a scenario could
arise when the substrate strongly chelates to the rhodium atom,
effectively preventing free rotation around the σ bond. Our
experiments are not suitable to rule out such a mechanistic scheme
for substrates 2N and 2A (alkenes 2 and 2S cannot chelate to
rhodium; 2J should not be able to bind strongly).
(44) Suemune has reported an analogous loss of deuterium when d-1
was subjected to similar conditions (ref 6c). For literature precedence
of deuterium/hydrogen scrambling between rhodium hydride
complexes and ethanol and a mechanistic proposal for this process
see: (a) Detellier, C.; Gelbard, G.; Kagan, H. G. J. Am. Chem. Soc.
1978, 100, 7556−7561. (b) Strathdee, G.; Given, R. Can. J. Chem.
1974, 52, 2216−2225.
(45) For previous (multistep) syntheses of deuterated salicylaldehyde
(d-1), see ref 6c and: (a) Manetsch, R.; Zheng, L.; Reymond, M. T.;
Woggon, W.-D.; Reymond, J.-L. Chem.Eur. J. 2004, 10, 2487−2506.
(b) Moore, J. L.; Silvestri, A. P.; De Alaniz, J. R.; Dirocco, D. A.; Rovis,
T. Org. Lett. 2011, 13, 1742−1745.
(46) Casey, C. P.; Petrovich, L. M. J. Am. Chem. Soc. 1995, 117,
6007−6014.
(47) Because magnetic stirring was not possible in the NMR
spectrometer, the described experiments proceeded significantly
slower compared to reactions that were stirred.
(48) Complex 11 could also include an additional ligand that readily
dissociates under ESI−MS conditions (e.g., solvent or alkene).
(49) Fragment [Rh(CO)((Ra,R,R)-SIPHOS-PE)]+ was observed by
high-resolution ESI+−MS (found, 636.1169; calculated, 636.1186).
K
dx.doi.org/10.1021/ja305593y | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX