Diastereoselective Ni-catalyzed Negishi cross-coupling approach to saturated, fully oxygenated C-alkyl and C-aryl glycosides

Article abstract of DOI:10.1021/ja8041564

A Ni-catalyzed Negishi cross-coupling approach to C-glycosides is described with an emphasis on C-aryl glycosides. The combination of NiCl ₂/PyBox in N,N′-dimethylimidazolidinone (DMI) enabled the synthesis of C-alkyl glycosides under mild reaction conditions. Moderate yields and β-selectivities were obtained for C-glucosides, and good yields and high α-selectivities were the norm for C-mannosides. For C-aryl glycosides, reactions employing Ni(COD)₂/^tBu-Terpy in N,N-dimethylformamide (DMF) were typically high yielding and provided C-glucosides with high β-selectivities (1:>10 α:β) and C-mannosides in moderate α-selectivities (3:1 α:β); α-C-aryl glycosides could be obtained by the combination of Ni(COD) ₂/PyBox in DMF (>20:1 α:β). The collective studies suggest that stereochemical control of the C-glycosides is dependent on the substrate and catalysts combination. The Negishi protocol displays excellent functional group tolerance, as demonstrated by its use in the first total synthesis of the natural product salmochelin SX.

Full text of DOI:10.1021/ja8041564

Published on Web 08/13/2008
Diastereoselective Ni-Catalyzed Negishi Cross-Coupling
Approach to Saturated, Fully Oxygenated C-Alkyl and C-Aryl
Glycosides
Hegui Gong and Michel R. Gagne´*
Department of Chemistry, UniVersity of North Carolina at Chapel Hill,
Chapel Hill, North Carolina 27599-3290
Received June 2, 2008; E-mail: mgagne@unc.edu
Abstract: A Ni-catalyzed Negishi cross-coupling approach to C-glycosides is described with an emphasis
on C-aryl glycosides. The combination of NiCl2/PyBox in N,N′-dimethylimidazolidinone (DMI) enabled the
synthesis of C-alkyl glycosides under mild reaction conditions. Moderate yields and ꢀ-selectivities were
obtained for C-glucosides, and good yields and high R-selectivities were the norm for C-mannosides. For
C-aryl glycosides, reactions employing Ni(COD)2/tBu-Terpy in N,N-dimethylformamide (DMF) were typically
high yielding and provided C-glucosides with high ꢀ-selectivities (1:>10 R:ꢀ) and C-mannosides in moderate
R-selectivities (3:1 R:ꢀ); R-C-aryl glycosides could be obtained by the combination of Ni(COD)2/PyBox in
DMF (>20:1 R:ꢀ). The collective studies suggest that stereochemical control of the C-glycosides is dependent
on the substrate and catalysts combination. The Negishi protocol displays excellent functional group
tolerance, as demonstrated by its use in the first total synthesis of the natural product salmochelin SX.
Scheme 1
1. Introduction
C-Glycosides are an important class of naturally occurring
and synthetic bioactive products.1-4 In contrast to their O-
analogues, C-glycosides are less vulnerable to metabolic
processing, which has led to significant interest in their viability
as drug candidates and inhibitors of carbohydrate processing
enzymes.1,2 These interests have advanced the synthesis of
C-glycosides, with the construction of the key anomeric C-C
bond centering on approaches that generate nucleophilic,
electrophilic, or radical character at the C1 center; concerted
reactions have also been examined.3-5 In general, these methods
rely on substrate control to achieve a stereoselective construction
of the anomeric center. Unfortunately, many reaction protocols
are prone to ꢀ-elimination of C2 substituents, leading to an
emphasis on C2-deoxy C-glycoside syntheses.5
over the anomeric center. Most transition metal-catalyzed
protocols to C-glycosides utilize sp2-hybridized anomeric
carbons (e.g., 1-halo glycals in Heck-type cross-couplings) or
π-allyl types, with both providing unsaturated products.6,7 These
constraints are the result of the sensitivity of putative, fully
saturated C1-organometallic intermediates to ꢀ-elimination
reactions (Scheme 1).
As part of an effort to develop cross-coupling approaches to
fully oxygenated C-glycosides, we became interested in catalysts
for the construction of C-C bonds at the C1 of sp3-hybridized
glycosyl halides. This strategy would provide a convergent route
to C-glycosides with the potential for catalyst control over the
stereochemistry of the anomeric center. As discussed above,
this approach is challenged by the susceptibility of C1-
organometallic intermediates to ꢀ-elimination of the C2 sub-
stituent (H or OR, Scheme 1).
Catalytic approaches to C-glycosides, on the other hand, are
relatively rare, though they promise to provide catalyst control
(1) For recent reviews in C-glycoside biology, see: (a) Hultin, P. G. Curr.
Top. Med. Chem. 2005, 5, 1299–1331. (b) Zou, W. Curr. Top. Med.
Chem. 2005, 5, 1363–1391. (c) Compain, P.; Martin, O. R. Bioorg.
Med. Chem. 2001, 9, 3077–3092. (d) Nicotra, F. Top. Curr. Chem.
1997, 187, 55–83.
(2) (a) Levy, D. E.; Tang, C. The Chemistry of C-Glycosides; Pergamon:
Tarrytown, NY, 1995. (b) Lin, C.-H.; Lin, H.-C.; Yang, W.-B. Curr.
Top. Med. Chem. 2005, 5, 1431–1457.
Fu, Knochel, and others have demonstrated that transition
metal-catalyzed cross-coupling of alkyl halides with ꢀ-hydrogens
(3) For reviews in C-glycoside synthesis, see: (a) Postema, M. H. D.
C-Glycoside Synthesis; CRC Press, Inc.: Boca Raton, FL, 1995. (b)
Somsak, L. Chem. ReV. 2001, 101, 81–135. (c) Meo, P.; Osborn,
H. M. I. Best Synthetic Methods: Carbohydrates; Elsevier Science
Ltd.: New York, 2003; pp 337-384. (d) Du, Y.; Linhardt, R. J.;
Vlahov, I. R. Tetrahedron 1998, 54, 9913–9959.
(6) For recent examples of C1-sp2 Heck, Suzuki, Hiyama, and Stille
couplings, see: (a) Chen, C.-L.; Martin, S. F. J. Org. Chem. 2006, 71,
4810–4817. (b) Potuzak, J. S.; Tan, D. S. Tetrahedron Lett. 2004, 45,
1797–1801. (c) Denmark, S. E.; Regens, C. S.; Kobayashi, T. J. Am.
Chem. Soc. 2007, 129, 2774–2776. (d) Jeanneret, V.; Meerpoel, L.;
Vogel, P. Tetrahedron Lett. 1997, 38, 543–546.
(4) For reviews in C-aryl glycoside biology and synthesis, see: (a) Bililign,
T.; Griffith, B. R.; Thorson, J. S. Nat. Prod. Rep. 2005, 22, 742–760.
(b) Lee, D. Y. W.; He, M. Curr. Top. Med. Chem. 2005, 5, 1333–
1350. (c) Wellington, K. W.; Benner, S. A. Nucleosides Nucleotides
Nucleic Acids 2006, 25, 1309–1333.
(7) For allyl type, see: (a) Daves, G. D., Jr. Acc. Chem. Res. 1990, 23,
201–206. (b) Brakta, M.; Lhoste, P.; Sinou, D. J. Org. Chem. 1989,
54, 1890–1896.
(5) Beau, J.-M.; Gallagher, T. Top. Curr. Chem. 1997, 187, 1–54.
9
10.1021/ja8041564 CCC: $40.75
2008 American Chemical Society
J. AM. CHEM. SOC. 2008, 130, 12177–12183 12177

A R T I C L E S
Gong and Gagne´
can be achieved under the guidance of suitable ligands.8 In
particular, Ni-catalyzed Negishi cross-coupling conditions have
been developed wherein pincer ligands effectively inhibit
detrimental ꢀ-elimination reactions that would serve to bypass
productive cross-coupling cycles.9 One presumes that the pincer
ligands operate by blocking the vacant site of the transition metal
cis to the putative metal-alkyl.10 If this reactivity pattern could
be transferred to elimination-prone glycosyl-type electrophiles,
Fu’s procedure might also enable stereoselective C-C bond
formation at the carbohydrate anomeric carbon.
Scheme 2
Table 1. Ligand Effects on the Coupling of Ph(CH2)3ZnBr and
Aceto-R-1-bromo-glucose
Recently, we demonstrated our first successful implementation
of this strategy to synthesize C-alkyl glycosides.11 Herein, we
present additional discussion of this development and extend
the methodology to include the stereoselective synthesis of
C-aryl glycosides. To the best of our knowledge, this latter work
represents the first Ni-catalyzed cross-coupling of an arylzinc
and a secondary sp3 carbon.12 The application of our C-aryl
glycoside methodology to an expedient synthesis of the enteric
bacterium natural product salmochelin SX is also described.13
Entrya
Catalyst/Ligand
Solventb
Product (R:ꢀ)
Glucal
1
2
3
4
5
6
7
8
NiCl2 ·glyme/Terpy
NiCl2 ·glyme/Terpy
NiCl2 ·glyme/Terpy
DMI
15% (ꢀ only) major
DMA
DMF
DMA
DMI
DMA
DMF
DMF
DMI
NDc
trace
major
major
NiCl2 ·glyme/EtO-Terpy
NiCl2 ·glyme/Pyr-Terpy
NiCl2 ·glyme/Pyr-Terpy
NiCl2 ·glyme/tBu-Terpy
Ni(COD)2/tBu-Terpy
NiCl2 ·glyme/BBP
15% (ꢀ only) major
10% (ꢀ only) 32%
trace
trace
trace
trace
30% (1:1)
10% (1:1)
trace
trace
ND
trace
NA
major
trace
19%
10%
trace
major
15%
2. Results and Discussion
9
10
11
12
13
14
NiBr2 ·diglyme/R-iPr-PyBox DMI
NiBr2 ·diglyme/S-iPr-PyBox DMI
After extensive experimentation, we discovered that a variant
of Fu’s Ni-catalyzed Negishi coupling conditions was compat-
ible with the dense functionality of a carbohydrate, thus enabling
an efficient synthesis of a variety of C-alkyl glycosides.11 The
conditions found by Fu to be optimum for unhindered cross-
couplings (iPr-PyBox) were not suitable to glycosyl electro-
philes, where elimination to glycal was particularly problematic.
As a general observation, slow catalysts led to products that
were the result of RZnX-mediated decomposition of the glycosyl
halide (glycal formation and hydrolysis/oligomerization).
NiCl2 ·glyme/R-Ph-PyBox
NiBr2 ·diglyme/terthiophene DMI
NiCl2 ·glyme/PMDETA DMI
DMI
a Reaction conditions: 1 (0.24 mmol, 0.19 M in solvent), Ni catalyst
(0.024 mmol), ligand (0.036 mmol), and RZnX (∼0.9 M, 0.72 mmol) at
room temperature for 12 h. b DMI, N,N′-dimethylimidazolidinone;
DMA, N,N-dimethylacetamide; DMF, N,N-dimethylformamide. c ND,
not detected by NMR.
oligomeric byproducts.14 The diastereoselectivity of the reactions
depended on the carbohydrate, with mannosyl halides providing
good to excellent yields and uniformly high R-selectivities,
whereas glucosyl halides gave moderate yields and a slight
preference for ꢀ-selectivities (Scheme 2).
Optimal was the combination of NiCl2 ·glyme (5 mol %) and
an unsubstituted PyBox in N,N′-dimethylimidazolidinone (DMI).
This catalyst afforded the desired C-glycosides in moderate to
good yields, along with glucal (up to 15%) and baseline
In the course of optimizing this reaction, it was noted that
the Terpy/NiCl2 · glyme catalyst formed the ꢀ-product only
with MeZnI in N,N-dimethylacetamide (DMA) (60% yield),15
in contrast to the PyBox result in Scheme 2 (1:2.2 R:ꢀ).
Unfortunately, this selectivity did not extend to primary
alkylzinc reagents, where poor yields were the norm (entries
1-8, Table 1). To improve this situation, the ligands in Chart
1 were screened with MeZnI, and those that were promising
were additionally examined with a primary alkylzinc reagent
(Table 1).
As mentioned above, increasing the size of the ligands and/
or the organozinc reagent generally decreased the yields, with
the mass balance being glycal or oligomeric products. This
general trend was rationalized as reflecting competitive back-
ground decomposition by the alkylzinc reagent, bulkier catalysts
being slower to consume the starting material before decom-
position ensued.
(8) For leading reviews, see: (a) Terao, J.; Kambe, N. Bull. Chem. Soc.
Jpn. 2006, 79, 663–672. (b) Knochel, P.; Calaza, M. I.; Hupe, E. Metal-
Catalyzed Cross-Coupling Reactions, 2nd ed.; Wiley-VCH Verlag
GmbH & Co. KGaA: Weinheim, Germany, 2004; Vol. 2, pp 619-
670. (c) Frisch, A. C.; Beller, M. Angew. Chem., Int. Ed. 2005, 44,
674–688. (d) Netherton, M. R.; Fu, G. C. AdV. Synth. Catal. 2004,
346, 1525–1532. (e) Ca´rdenas, D. J. Angew. Chem., Int. Ed. 2003,
42, 384–387.
(9) For pincer ligands used in the Ni-catalyzed Negishi approach to (sp3-
sp3) C-C bonds, see: (a) Arp, F. O.; Fu, G. C J. Am. Chem. Soc.
2005, 127, 10482–10483. (b) Jones, G. D.; Martin, J. L.; McFarland,
C.; Allen, O. R.; Hall, R. E.; Haley, A. D.; Brandon, R. J.; Kanovalova,
T.; Desrochers, P. J.; Pulay, P.; Vicic, D. A. J. Am. Chem. Soc. 2006,
128, 13175–13183. (c) Phapale, V. B.; Bun˜uel, E.; Garc´ıa-Iglesias,
M.; Ca´rdenas, D. J. Angew. Chem., Int. Ed. 2007, 46, 8790–8795.
(10) For other examples using pincer ligands, see: (a) Hahn, C.; Cucciolito,
M. E.; Vitagliano, A. J. Am. Chem. Soc. 2002, 124, 9038–9039. (b)
Cucciolito, M. E.; D’Amora, A.; Vitagliano, A. Organometallics 2005,
24, 3359–3361. (c) Feducia, J. A.; Campbell, A. N.; Doherty, M. Q.;
Gagne´, M. R. J. Am. Chem. Soc. 2006, 128, 13290–13297.
(11) Gong, H.; Sinisi, R.; Gagne´, M. R. J. Am. Chem. Soc. 2007, 129,
1908–1909.
With regard to anomer selectivity, the catalysts obtained from
i
the enantiomeric Pr-PyBox ligands were similarly selective,
(12) (a) For Ni-catalyzed Negishi coupling of arylzincs with primary halides,
see: Giovannini, R.; Knochel, P. J. Am. Chem. Soc. 1998, 120, 11186–
11187. (b) For Ni-catalyzed Kumada coupling of PhMgBr with primary
alkyl halides, see: Terao, J.; Watanabe, H.; Ikumi, A.; Kuniyasu, H.;
Kambe, N. J. Am. Chem. Soc. 2002, 124, 4222–4223.
though the R-antipode was higher yielding (entries 10 and 11,
(14) The baseline mixture was subject to ESI-MS studies. The peaks at
m/z 889.1 and 540.0 suggest the formation of probable O-linked
trisaccharide and disaccharide, respectively.
(13) (a) Hantke, K.; Nicholson, G.; Rabsch, W.; Winkelmann, G. Proc.
Natl. Acad. Sci. U.S.A. 2003, 100, 3677–3682. (b) Bister, B.; Bischoff,
D.; Nicholson, G. J.; Valdebenito, M.; Schneider, K.; Winkelmann,
G.; Hantke, K.; Suessmuth, R. D. BioMetals 2004, 17, 471–481.
t
(15) Although the Bu-Terpy/NiCl2 ·glyme catalyst gave only traces of
product in DMA (and none in DMI), it gave the ꢀ-methyl product in
40% yield in THF.
9
12178 J. AM. CHEM. SOC. VOL. 130, NO. 36, 2008

Negishi Cross-Coupling Approach to C-Glycosides
A R T I C L E S
Chart 1
Table 2. Optimum Conditions for the C-Alkylation of Glycosyl
Halides
Table 1).16 Other ligands such as terthiophene, PMDETA, and
tBu-Terpy collectively failed to provide more than trace amounts
of the coupling product. A number of Ni sources were
investigated, and NiBr2 provided results comparable to those
obtained with NiCl2, whereas Ni(COD)2, NiF2, NiI2,
Ni(PPh3)2Cl2, and Ni(dppp)Cl2 were all inferior (data not
shown).
Although the optimum catalyst (PyBox/NiCl2 in DMI) was
only slightly ꢀ-selective with glucosyl halides, it was very
R-selective with mannosyl halides. Examples of these transfor-
mations are collected in Table 2. It should be noted that benzyl-
protected carbohydrates required the use of chloride leaving
groups, as these more electron rich (armed) compounds were
too sensitive to hydrolysis and elimination as the bromide.
2.1. C-Aryl Glycoside. Extension of the Ni-catalyzed Negishi
strategy to C-aryl glycosides was compelling since many natural
products and drug candidates contain the C-aryl glycoside core.4
More importantly, the current synthesis of C-aryl glycosides is
often limited to the addition of electron-rich aromatic systems
to glycosyl electrophiles,4,17 reduction of 1-C-aryl-glucosides,18
or chemistry involving anomeric sp2 carbons.4,19 A direct
conversion of glycosyl halides to C-aryl glycosides employing
a transition metal-catalyzed cross-coupling protocol would
overcome many of these limitations.
Our first experiments to this end utilized the optimized C-alkyl
glycoside synthesis conditions (NiCl2/PyBox in DMI) to
investigate the cross-coupling of PhZnI ·LiCl20 and aceto-R-D-
bromoglucose (1); however, only the glucal elimination product
was observed (entry 1, Table 3). A trace amount of the desired
product was obtained using Ni(COD)2/PyBox, and changing the
solvent from DMI to DMA further increased the yield to 20%
with moderate R-selectivity (entry 3, Table 3). Screening various
a Typical reaction conditions: glycosyl bromide (0.24 mmol, 0.19 M
in DMI), NiCl2 ·DME (0.024 mmol), ligand (0.036 mmol), and RZnBr
(in DMI) at room temperature for 12 h. b No ꢀ-isomer detected by TLC
or NMR.
ligands (entries 4-11, Table 3, Chart 2) in DMA demonstrated
that terpyridine (entry 10, Table 3) was superior to PyBox,
providing the Ph-C-glucoside in 45% yield with moderate
ꢀ-selectivity. The ꢀ-selectivity was enhanced markedly when
tBu-Terpy was used (1:18 R:ꢀ, entry 11, Table 3). Finally, it
was found that, in DMF, the Ni(COD)2/tBu-Terpy combination
furnished the phenyl product in 71% yield, with a slightly
lowered diastereomer ratio (dr) of 1:12 R:ꢀ (entry 12, Table
3).21Further screening of ligands (Chart 2) and solvents did not
offer additional improvement (entries 13-15, Table 3).
(16) Similar results were observed with MeZnI, see ref 11
(17) For recent examples, see: (a) Yuan, Y.; Men, H.; Lee, C. J. Am. Chem.
Soc. 2004, 126, 14720–14721. (b) Rousseau, C.; Martin, O. R. Org.
Lett. 2003, 5, 3763–3766.
(18) Deshpande, P. P.; Ellsworth, B. A.; Buono, F. G.; Pullockaran, A.;
Singh, J.; Kissick, T. P.; Huang, M.-H.; Lobinger, H.; Denzel, T.;
Mueller, R. H J. Org. Chem. 2007, 72, 9746–9749.
(19) For lactone substrates, see: (a) Chen, C.-L.; Sparks, S. M.; Martin,
S. F. J. Am. Chem. Soc. 2006, 128, 13696–13697. (b) Martin, S. F.
Pure. Appl. Chem 2003, 75, 63–70. For examples using cross-coupling
approaches, see ref 6. For nucleophilic sp2 anomeric center, see ref 5
and references cited therein.
The scope of arylzinc reagents was next investigated using
the Ni(COD)2/tBu-Terpy/DMF conditions. Gratifyingly, the
(20) Krasovskiy, A; Malakhov, V; Gavryushin, A; Knochel, P Angew.
Chem., Int. Ed. 2006, 45, 6040–6044.
(21) The Ph-C-glucoside was obtained in 68% yield (1:10 R:ꢀ) when 5
t
mol % Ni(COD)2 and 7.5 mol % Bu-Terpy were employed.
9
J. AM. CHEM. SOC. VOL. 130, NO. 36, 2008 12179

A R T I C L E S
Gong and Gagne´
Table 4. Scope of Organozinc Reagents in the Coupling with 1
Table 3. Reaction Optimization for the Coupling of PhZnI ·LiCl with
Acetobromo-R-D-glucose
Entrya
Precatalyst/Ligand
Solventb Product (R:ꢀ)
Glucal
1
2
3
4
NiCl2 ·glyme/PyBox
Ni(COD)2/PyBox
Ni(COD)2/PyBox
DMI
DMI
DMA 20% (2:1)
DMA trace
NDc
trace
70%
substantial
substantial
substantial
Ni(COD)2/ThiopheneBox
5
Ni(COD)2/bathophenanthroline DMA 13% (1:1.5) 37%
6
7
Ni(COD)2/phenanthroline
Ni(COD)2/2,2′-dibromoTerpy DMA trace
DMA 21% (2:1)
35%
substantial
8
9
Ni(COD)2/EtO-Terpy
Ni(COD)2/Cl-Terpy
Ni(COD)2/Terpy
DMA 22% (1:4)
6%
DMA 20% (1:2.5) substantial
DMA 45% (1: 3.5) 10%
DMA 45% (1:18) 15%
DMF 71% (1:12) 7%
DMF 46% (1:4.7) 13%
DMF 69% (1:2.8) trace
10
11
12
13
14
15
16
Ni(COD)2/tBu-Terpy
Ni(COD)2/tBu-Terpy
Ni(COD)2/4-MePh-Terpy
Ni(COD)2/Terpy
Ni(COD)2/tBu-Terpy
Ni(COD)2/tBu-Terpy
NMP 10%
THF 10%
10%
10%
a Reaction conditions: 1 (0.24 mmol, 0.19 M in DMF), Ni(COD)2
(0.024 mmol), ligand (0.036 mmol), PhZnI ·LiCl (∼0.5 M in DMF) at rt
for 12 h. b DMI, N,N′-dimethylimidazolidinone; DMA, N,N-dimethylacet-
amide; DMF, N,N-dimethylformamide; NMP, N-methylpyrrolidone; THF,
tetrahydrofuran. c ND ) not detected by NMR.
Chart 2
reaction was tolerant of a diverse array of functionalities such
as halides, nitriles, acetals, ethers, and esters. Both electron-
rich and electron-deficient arylzinc reagents provided good to
excellent yields of the desired products (entries 1-3, Table 4),
and excellent ꢀ-selectivities (1:>10 R:ꢀ) were observed through-
out. While meta and para substituents were tolerated (entries
1-4, Table 4), ortho substituents generally produced poor results
(entry 5, Table 4).
Extending the arylation method to heteroaromatic systems
was also pursued. Furan and both 2- and 3-thiophene derivatives
were good partners (entries 6-9, Table 4), the latter two
derivatives being electronically quite different.22 Unfortunately,
pyridine-based zinc reagents failed to give the desired products.
Finally, formation of the trans-phenyl vinyl derivate (entry 11,
Table 4) was obtained in good yield, but with a low dr (1:1
R:ꢀ). A literature search did not uncover other Ni-catalyzed
Negishi coupling of vinylzinc reagents with sp3 alkyl halides.23
The versatility of the Ni-catalyzed Negishi protocol for other
glycosyl halides was studied using PhZnI ·LiCl as the coupling
a Reaction conditions: 1 (0.24 mmol, 0.19 M in DMF), Ni(COD)2
(0.024 mmol), tBu-Terpy (0.036 mmol), and ArZnI·LiCl (∼0.5 M in
DMF, 0.75 mL) at room temperature for 12 h. b ND ) not detected by
NMR.
partner. Whereas the aceto-R-D-bromo mannose provided phenyl
C-mannoside as a 2.9:1 mixture of R:ꢀ anomers, the aceto-R-
D-bromo galactose was considerably more ꢀ-selective, giving
the phenyl C-galactoside as a 1:10 mixture of R:ꢀ anomers
(entries 1 and 2, Table 5). The 5-dealkylated glucose analogue,
tri-O-acetyl-ꢀ-D-arabinosyl bromide, gave the desired product
with a 1:2.5 R:ꢀ ratio (entry 3, Table 5). Perhaps mechanistically
revealing was the observation that both the isomerically pure
R-Br and the 1:1 mixture of R- and ꢀ-2-phthalimido glucosyl
bromides converge to the same yield of ꢀ-Ph product (entry 4,
Table 5). Ironically, the 2-deoxyglucose failed to give product
due to decomposition of the glycosyl bromide and chloride.
Several furanose derivatives were also investigated. Whereas
the benzoyl-protected 2,3,5-tri-O-benzoyl-R-D-arabinofuranosyl
bromide produced largely the hydrolysis product, 2,3,5-tri-O-
aceto-D-ribofuranosyl chloride gave the desired product in low
yield with high ꢀ-selectivity (entries 6 and 7, Table 5). We
continue to seek improvements in furanoside reactivities.
(22) While 2-thiophene-zinc reagent can be prepared from the bromo
precursor, 3-thiophene-zinc reagent can only be prepared from
3-iodothiophene (see Supporting Information for organozinc prepara-
tion).
(23) Only trace amount of Kumada coupling products was detected in the
Ni-catalyzed coupling of vinyl Grignard reagents with primary alkyl
halides, see: Terao, J.; Watabe, H.; Kambe, N. J. Am. Chem. Soc.
2005, 127, 3656–3657.
The impact of protecting groups on the reactivity of the sugar
derivatives was also studied. For the benzyl-protected sugars,
9
12180 J. AM. CHEM. SOC. VOL. 130, NO. 36, 2008

Negishi Cross-Coupling Approach to C-Glycosides
A R T I C L E S
Table 5. Investigation of the Scope of Sugars under the
Ni-Catalyzed Negishi Cross-Coupling Conditions
Table 6. Reaction Optimization for PhZnI ·LiCl and
Aceto-R-Br-D-mannose
Entrya
Conditions
Product (R:ꢀ)
Glucal
1
2
3
4
5
6
7
NiCl2 ·glyme, PyBox, DMI
Ni(COD)2, PyBox, DMA
Ni(COD)2, Terpy, DMA
Ni(COD)2, Bu-Terpy, DMA
Ni(COD)2, PyBox, DMF
40% (R)
30%
15%
trace
15%
7%
∼30% (R)
56% (6.6:1)
65% (1.6:1)
80% (20:1)
79% (10:1)
76% (2.9:1)
t
Ni(COD)2, Terpy, DMF
10%
10%
t
Ni(COD)2, Bu-Terpy, DMF
a Reaction conditions: mannosyl bromide (0.24 mmol, 0.19 M in
DMF), Ni(COD)2 (0.024 mmol), ligand (0.036 mmol), and PhZnI ·LiCl
(∼0.5 M in DMF) at room temperature for 12 h.
The ability to at least partially overcome the strong bias of a
mannoside for R-products suggested that a proper ligand might
eventually provide the desired catalyst control.
2.2. Catalyst or Substrate Control? Our experiments, aimed
at developing efficient and selective catalysts for the C-alkylation
of glycosyl halides, showed that the combination of PyBox/
NiCl2 in DMI was effective for the stereoselective generation
of C-glycosides of biased sugars. For example, the C2 axial
substituent of mannosides sterically (and perhaps electronically)
favored axial alkylation and high R-selectivities. In contrast,
when the carbohydrate had a less defined anomer preference
(glucose, galactose), the R:ꢀ ratios were significantly eroded.
Like the high R-selectivities observed in mannosyl bromide
alkylation reactions with PyBox/NiCl2 in DMI, the PyBox/
Ni(COD)2 catalyst in DMF was similarly R-aryl selective and
high yielding. Despite the different optimum solvent and nickel
sources, the PyBox-based catalysts appear to be inherently
unselective and prone to strong substrate control (mannosides).
Experiments using Terpy/NiCl2 catalysts in DMA showed
that this catalyst had a significant preference for the ꢀ-methy-
lation of 1 (Vide supra), and although primary alkyl zinc reagents
were poor yielding, they also were ꢀ-selective (entries 1, 4, and
t
5, Table 1). In a seemingly parallel fashion, the Bu-terpy/
Ni(COD)2 arylation catalysts provided excellent ꢀ-selectivities
for glucosides and galactosides in DMF.
a Reaction conditions: glycosyl bromide (0.24 mmol, 0.19 M in
DMF), Ni(COD)2 (0.024 mmol), tBu-Terpy (0.036 mmol), and
ArZnI·LiCl (∼0.5 M in DMF) at room temperature for 12 h.
Taken together, these data suggest that the PyBox-nickel and
Terpy- or tBu-terpy-Ni catalysts functioned by different mech-
anisms, with oxidative addition to the former occurring with
little inherent stereochemical bias, while the latter preferred an
invertive pathway (assuming retentive transmetalation and
reductive elimination elementary steps).
the arming effect24 required the use of R-glucosyl and mannosyl
chlorides (for C-alkylation), since the bromo analogues undergo
rapid hydrolysis in air. Nevertheless, the use of benzyl-protected
R-chloro substrates gave low yields of the C-aryl glycoside
(entries 8 and 9, Table 5), accompanied by substantial amounts
of hydrolysis. As anticipated, the benzoyl-protected glucose gave
results comparable to those obtained with the acetyl-protected
glucose (entry 10, Table 5).
The optimal reaction conditions for C-Ph-mannoside were
examined more carefully, as the standard conditions with
tBuTerpy provided the product as a 1.6:1 ratio of R:ꢀ isomers
(entry 4, Table 6), in contrast to the ꢀ-selective gluco series.
The effect of ligand was considerable, as good to excellent
R-selectivities were achieved using Terpy (10:1, entry 6, Table
6) and PyBox (>20:1, entry 5, Table 6), the latter in 80% yield.
Potentially conflicting evidence for this simple notion are the
notable differences in the alkylation and arylation protocol’s
optimum nickel source and solvent. For both C-alkylation-based
methods, Ni(II) was optimum, while the arylation procedure
required a Ni(0) source. Similarly, all alkylation reactions
worked best in DMI, while the arylations worked best in DMF,
regardless of the specific catalyst in question. Unanswerable as
of yet is why ligand-dependent mechanisms, as implied above,
would have optimum metal and solvent coupled to whether the
reaction was alkylative or arylative rather than the overall
mechanism. Of course, without knowing the mechanisms, it is
difficult to reconcile these observations.
The propensity for equatorial arylation by tBuTerpy/Ni(COD)2
catalysts in DMF was consistent with the convergence of R-
(24) Mootoo, D. R.; Konradsson, P.; Udodong, U.; Fraser-Reid, B. J. Am.
Chem. Soc. 1988, 110, 5583–5584.
9
J. AM. CHEM. SOC. VOL. 130, NO. 36, 2008 12181

A R T I C L E S
Gong and Gagne´
Scheme 3
Scheme 5. Free Radical Reactivity of Glucosyl and Mannosyl
Radicals and Their Respective EPR-Derived Structures
Scheme 4. Mechanistic Proposals for Ni-Catalyzed Cross-Coupling
Scheme 6. Stereochemical Outcome of the Reaction of Glycosyl
Radicals with Reducing Metals
If such radicals were generated in the present case, they would
be subject to anomeric effects. Giese has shown that both
glucosyl and mannosyl radicals selectively react with alkenes
to form R-addition products (Scheme 5).26 In conjunction with
these synthetic studies, the structure of these radicals was
obtained by EPR.27 If a stereochemistry-determining Ni-C bond
formation were to occur in analogy to the addition of acrylates
or acrylonitrile to these radicals, then one would expect that all
catalysts would generate the R-organometallic intermediate and
R-products, again assuming retentive transmetalation and reduc-
tive elimination elementary steps (Scheme 3). The change in
selectivity for alkylation and arylation reactions, coupled with
the ligand effects, argues against a simple scenario.
Perhaps reflective of the complexity of radical-metal addition
reactions are the three reactions in Scheme 6.28 Each of these
results reasonably invokes the reaction of a glycosyl radical and
a reducing metal (SmI2 or NiI2). In these cases, the stereochem-
ical data follow exactly the general trends observed herein:
glucosyls can be made ꢀ-selective, and mannosyls can be made
R-selective. The divergence from the Giese free radical cases
and 1:1 R:ꢀ-glucosaminyl bromide substrates to the ꢀ-Ph
product (entry 4, Table 5), and also with arylation trends in the
mannosyl bromide substrate. In this manno series, the catalyst
most prone to ꢀ-arylation of glucosides (tBuTerpy) performed
most poorly, with the R-product only being favored by 2.9:1
(entry 7, Table 6). However, as the catalysts became progres-
sively less ꢀ-gluco-selective, the reaction became progressively
more R-manno-selective (10:1 for Terpy/Ni(COD)2 and 20:1
for PyBox/Ni(COD)2). This trend was indicative of a stereo-
chemical mismatch between the catalyst and substrate, the
tBuTerpy/Ni(COD)2 catalyst preferring to ꢀ-arylate while the
mannosyl framework favored R-products. Catalysts without a
predilection (PyBox/Ni(COD)2) provided good R-arylation
selectivities by substrate control, while ꢀ-selective catalysts led
to compromised results (Scheme 3).
Recent mechanistic studies of Ni-catalyzed Negishi cross-
coupling reactions have provided evidence for stepwise oxidative
addition mechanisms involving alkyl halide-derived free radicals
prior to Ni-C bond formation.9b,c Two variants on this
mechanism, a halogen abstraction (inner sphere) and a single
electron transfer (outer sphere), are outlined in Scheme 4. They
differ in the details of how the C-radical is generated (single
electron transfer versus halogen abstraction) but are similar in
that the oxidative addition to a Ni(I) alkyl is stepwise,
proceeding via a sequence of one-electron elementary steps and
not via two-electron processes.25
(26) (a) Giese, B.; Dupuis, J. Angew. Chem., Int. Ed. Engl. 1983, 22, 622–
623. (b) Dupuis, J.; Giese, B.; Hartung, J.; Leising, M.; Korth, H.-G.;
Sustmann, R. J. Am. Chem. Soc. 1985, 107, 4332–4333. (c) Adlington,
R. M.; Baldwin, J. E.; Basak, A.; Kozyrod, R. P. J. Chem. Soc., Chem.
Commun. 1983, 944–945.
(27) (a) Korth, H.-G.; Sustmann, R.; Dupuis, J.; Giese, B. J. Chem. Soc.,
Perkin Trans. 2 1986, 1453–1459. (b) Dupuis, J.; Giese, B.; Ru¨egge,
D.; Fishcer, H.; Korth, H.-G.; Sustmann, R. Angew. Chem., Int. Ed.
Engl. 1984, 23, 896–897.
(25) Phillips has recently reported a detailed DFT study on the cross-
coupling process and has shown inner-sphere I• abstraction from I-iPr
by (Terpy)NiI-CH3 to be energetically feasible, see: Lin, X.; Phillips,
D. L. J. Org. Chem. 2008, 73, 3680–3688.
(28) (a) Hung, S.-C.; Wong, C.-H. Angew. Chem., Int. Ed. Engl. 1996, 35,
2671–2674. (b) Miquel, N.; Doisneau, G.; Beau, J.-M. Angew. Chem.,
Int. Ed. 2000, 39, 4111–4114.
9
12182 J. AM. CHEM. SOC. VOL. 130, NO. 36, 2008

Negishi Cross-Coupling Approach to C-Glycosides
A R T I C L E S
Scheme 7. Retrosynthesis of Salmochelin SX
Scheme 8. Total Synthesis of Salmochelin SXa
(pure R-selectivity) may reflect the more sterically encumbered
reducing metals choosing the ꢀ-face of a free boat-like glucosyl
radical (Scheme 6), or the unique stereoselectivity of putative
radical anion intermediates. The ligand effects observed herein
also suggest the possibility of a gradient of mechanisms that
respond to changes in the catalyst structure, once again raising
the intriguing possibility of eventual catalyst control over this
important C-C bond-forming reaction.
2.3. Total Synthesis of Salmochelin SX. The applicability of
this methodology to natural product synthesis was examined
t
a Reaction conditions: (a) Ni(COD)2 (10 mol %), Bu-Terpy (15 mol
%), ArZnI·LiCl (150 mol %), DMF, room temperature, 12 h, R:ꢀ ) 1:20,
55%. (b) Na2CO3 (500 mol %), MeOH. (c) BnBr (600 mol %), Bu4NI (10
mol %), NaH (480 mol %), DMF, 75% over two steps. (d) NaOH (1000
mol %), THF:MeOH (3:1), 82%. (e) SOCl2 (400 mol %), DMF (catalytic),
DCM. (f) S-(BnOOC)CH(CH2OCbz)CHNH2-CF3COOH salt (150 mol %),
Et3N (150 mol %), DCM, 0 °C then room temperature, 30 min, 88% over
two steps. (g) H2, Pd(OH)2 (20 wt % on carbon), MeOH:EtOAc 1:1, 18 h,
94%.
by synthesizing the ꢀ-C-aryl-glucoside, salmochelin SX,13
a
metabolite of the ferric-binding siderophores produced by
Escherichia coli and Salmonella enterica.29 A retrosynthesis
provides the aryl-C-glucoside 2 and the doubly protected serine
3 (Scheme 7). The key step in the synthesis of 2 was the cross-
coupling of arylzinc 430 and acetobromo-R-D-glucose (1) using
the standard conditions to afford 5 in moderate yield (55%) and
with high ꢀ-selectivity (1:20 R:ꢀ, Scheme 8). The methyl-
protected catechol zinc reagent was slightly more well behaved,
and better yields and selectivities were obtained.
Unfortunately, the Bn-protected glucosyl chloride was prone
to hydrolysis and decomposition by the zinc reagent, requiring
a change of protecting groups. Deacetylation of 5, followed by
perbenzylation, yielded the aryl-C-glucoside 6 in 75% yield over
two steps. Saponification, conversion to the acid chloride, and
immediate addition to the trifluoroacetate salt of 3, obtained by
TFA deprotection of Boc-Ser(Cbz)-OBn,31 in the presence of
NEt3 provided the perbenzylated salmochelin precursor 8 after
30 min at room temperature (88%); debenzylation gave salmo-
chelin SX in 94% yield.
selective synthesis of C-alkyl and C-aryl glycosides. The
approach furnishes C-aryl glycosides in high yields and with a
tolerance for a variety of para and meta substituents, heteroaro-
matics, and vinylzinc nucleophiles. High ꢀ-selectivities for
aceto-R-D-glucose and excellent R-selectivities for aceto-R-D-
mannose could be achieved by suitable choice of the N3 ligand.
The former selectivity enabled the efficient synthesis of salmo-
chelin SX. We continue to search for catalysts capable of broad
control over product stereoselectivity.
Acknowledgment. Andrew Blanchard and Riccardo Sinisi are
thanked for help. We additionally thank the Department of Energy,
Office of the Basic Energy Sources, for generous support (DE-
FG02-05ER15630).
3. Conclusions
In summary, we have demonstrated that existing Ni-catalyzed
Negishi coupling methodologies can be adapted to the stereo-
Supporting Information Available: Full experimental details
and characterization data for new compounds. This information
is available free of charge via the Internet at http://pubs.acs.org.
(29) Fischbach, M. A.; Lin, H.; Liu, D. R.; Walsh, C. T. Proc. Natl. Acad.
Sci. U.S.A. 2005, 102, 571–576.
JA8041564
(30) The zinc reagent was prepared according to a standard procedure from
the iodo precursor that was synthesized from commercially available
2-hydroxy-3-methoxybenzoic acid methyl ester in three steps. See
Supporting Information.
(31) Ramesh, R.; De, K.; Chandrasekaran, S. Tetrahedron 2007, 63, 10534–
10542.
9
J. AM. CHEM. SOC. VOL. 130, NO. 36, 2008 12183