Mechanistic Studies of Bismuth(V)-Mediated Thioglycoside Activation Reveal Differential Reactivity of Anomers

Article abstract of DOI:10.1021/acs.joc.6b00860

The mechanism of bismuth(V)-mediated thioglycoside activation was examined using reaction kinetics and quantum chemical reaction models. NMR experiments show an unusual nonlinear growth/decay curve for the glycosylation reaction. Further studies suggest an anomeric inversion of the β-glycoside donor to the α-donor during its activation, even in the presence of a neighboring 2-position acetate. Interestingly, in situ anomerization was not observed in the activation of an α-glycoside donor, and this anomer also showed faster reaction times and higher product diastereoselectivites. Density functional theory calculations identify the structure of the promoter triphenyl bismuth ditriflate, [Ph₃Bi(OTf)₂, 1], in solution and map out the energetics of its interactions with the two thioglycoside anomers. These calculations suggest that 1 must bind the thiopropyl arm to induce triflate loss. The computational analyses also show that, unlike most O-glycosides, the β- and α-donor S-glycosides are similar in energy. One energetically reasonable anomerization pathway of the donors is an S_N1-like mechanism promoted by forming a bismuth-sulfonium adduct with the Lewis acidic Bi(V) for the formation of an oxacarbenium intermediate. Finally, the computed energy compensations needed to form these α vs β Bi adducts is a possible explanation for the differential reactivity of these donors.

Full text of DOI:10.1021/acs.joc.6b00860

Subscriber access provided by West Virginia University | Libraries
Article
Mechanistic Studies of Bismuth(V)-Mediated Thioglycoside
Activation Reveal Differential Reactivity of Anomers
Manibarsha Goswami, Daniel C. Ashley, Mu-Hyun Baik, and Nicola L. B. Pohl
J. Org. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.joc.6b00860 • Publication Date (Web): 13 Jun 2016
Downloaded from http://pubs.acs.org on June 21, 2016
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
The Journal of Organic Chemistry is published by the American Chemical Society.
1155 Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 33
The Journal of Organic Chemistry
1
2
3
Introduction
4
5
6
7
8
9
Carbohydrates are vital to life; they form the building blocks of lipopolysaccharides,
glycoproteins or glycolipids,1ꢀ3 and constitute key components of natural products and
pharmaceuticals,4,5 such as the antimicrobials vancomycin6 and the cardiac glycosides.7 Thus, the ability
to manipulate the chemical composition of carbohydrates in a controlled and scalable fashion is
paramount to increased understanding of their roles.1 Despite significant efforts to predict the outcome
of the glycosylation reactions that are key to linking monosaccharides into more complex structures, the
dearth of mechanistic studies has prevented the development of reliable rules.8,9 Although an exciting
number of strategies have been developed, neighboring group participation (NGP) from the 2ꢀposition
remains one of the most reliable methods for setting the anomeric stereocenter.10,11,8,12,13 Nonetheless,
reactant concentrations, the nature of the leaving group and its mode of activation, substituents and their
effect on carbohydrate conformations, steric hindrance, metal coordination, temperature, and solvent
effects have also been noted as factors that affect reaction stereoselectivity.5 Clearly, nucleophilic
substitution of some type at the anomeric Cꢀ1 carbon is at play (Scheme 1) with some degree of
oxacarbenium ion formation,14,15 which are perceived to form by the dissociation of activated glycosyl
donorꢀpromoter complexes.16,17 Nuclear magnetic resonance (NMR)ꢀbased mechanistic studies of
chemical glycosylations carried out in the past two decades have been reviewed recently.18 However,
very few glycosylation reactions have been subjected to comprehensive mechanistic studies—detailed
experimental and analytical investigations along with inꢀdepth computational analyses.16,19,20
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 1. Chemical glycosylation pathways leading to 1,2 transꢀ and 1,2 cisꢀglycosides.
Recently, we developed a novel method for activating thiopropylglycosyl donors (2) and
coupling them to various acceptors that employed triphenylbismuth ditriflate (1) as the promoter
2
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 2 of 33
1
2
3
4
5
6
7
8
(Scheme 2).21 Bismuth is attractive for its high abundance and relative lack of toxicity; but, little is
known about its reactivity patterns, especially starting with Bi(V) species.22ꢀ25 The high valent bismuth
center could act as a soft Lewis acid that interacts with the Lewis basic site of the thioglycoside.
Bismuth can also easily access the common oxidation states of (III),26,27 (IV) and (V), which offers an
opportunity to access redoxꢀenhanced reaction mechanisms. Excellent yields were obtained in this
particular Bi(V)ꢀmediated glycosylation reaction without using additional coꢀpromoters or additives and
the strategy was extended to a wide range of carbohydrates with remarkable functional group tolerance,
including of alkenes and alkynes.28 Surprisingly, we observed that a promoter containing Bi(V) could
activate the thioglycosyl donors, whereas the more common Bi(III) analogues or triflic acid alone could
not. In addition to being sufficiently reactive even when substoichiometric amounts of the promoter
[Ph3Bi(OTf)2] were added, these reactions could also be performed at room temperature, which are
distinctively different, attractive features that are not commonly observed in previous methods for
thioglycoside activation.21
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 2. General scheme for thiopropylglycoside activation with pentavalent bismuth 1.
This newly discovered reactivity pattern for Bi(V), however, leaves many questions as to how
the reaction works, especially given the paucity of mechanistic studies of bismuthꢀmediated reactions.
Even the thioglycoside, one of the most commonly used building blocks in carbohydrate synthesis, has
only spawned mechanistic work focused primarily on the initial activation step, with the reactive
sulfonium intermediate having received special attention.29ꢀ31 Several key mechanistic questions remain
unanswered to date. What are the exact structures of the reactive intermediates? What are their
conformational preferences and which chemical features might govern the stereospecificity of the
reaction?
Results and Discussion
One of the most fundamental tasks for deciphering a reaction mechanism is determining the rate
law, which requires careful and sometimes challenging series of studies. Precise chemical kinetics
3
ACS Paragon Plus Environment

Page 3 of 33
The Journal of Organic Chemistry
1
2
3
4
5
6
7
8
studies on glycosylation are rare; Vernon et al.35ꢀ38 and Schroeder et al.39 were among the first who
attempted to study the kinetics of a KoenigsꢀKnorrꢀtype glycosylation by combining polarimetry and
Gas Liquid Chromatography (GLC) techniques. Following Wong and coworkers’ relative reactivity
work in thioglycoside building blocks,40 Huang and coworkers determined relative rates of reactions of
thioarylglycosides, which were activated by a mixture of Nꢀiodosuccinimide (NIS) and
trifluoromethanesulfonic acid (TfOH).41 Special attention was given to various substituent effects in this
study. To investigate the kinetics of activation and to identify the rateꢀdetermining step of the bismuthꢀ
mediated reaction, we carried out a series of variableꢀtemperature NMR studies, which can be
challenging due to signals that originate from different glycosyl reactants/products in solution that may
overlap with the characteristic signals that must be monitored.18 A benzylated βꢀthiopropyl galactoside
was selected to be the glycosyl donor (4ꢀβ) and methanol (CH3OH, 5) was chosen as the glycosyl
acceptor for its simplicity (5) along with the bismuth complex Ph3Bi(OTf)2 (1) as the promoter (Figure
1a). Although prior exploratory experiments were carried out in dichloromethane (CH2Cl2), these
experiments were conducted in chloroformꢀD (CDCl3, 1H δ=7.26 ppm), instead of CD2Cl2 (1H δ=5.32
ppm). This solvent was chosen to avoid overlapping (1H)ꢀpeaks with the sugarꢀring protons, thiopropyl,
or the Bi(V) compound. After some experimentation, we found that 313.15 K (40 °C) was the optimum
temperature for monitoring reaction kinetics, considering the required reaction time of a single
experiment on the NMR instrument. Figure 1b shows the 1HꢀNMR data collected for this glycosylation
reaction.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
4
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 4 of 33
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 1. (a) Model glycosylation with βꢀdonor for rateꢀorder studies; (b) Reaction progress monitored
by 1HꢀNMR spectra of model glycosylation reaction.
The progress of the reaction was monitored by measuring the disappearance and appearance of
the methoxy peak in the reactant (5, δ= 3.47ppm) and βꢀproduct (6ꢀβ, δ= 3.38 ppm), respectively, as
illustrated in Figure 2a. These peaks were selected as they do not overlap with the other proton peaks of
interest and could thereby be accurately integrated. The relative concentrations of these species derived
from the numerical integration of the NMRꢀsignals against reaction time is shown in Figure 2b. A
gradual decrease of the OCH3 signal, highlighted in blue on Figure 2b, indicates the depletion of 5 and
correlates with an increase of the OCH3 signal of the 6ꢀβ, shown in red. Interestingly, the concentration
traces show a relatively long incubation time characterized by a slow and steady decay of 5
accompanied by an even slower, but equally steady increase of 6ꢀβ. At ~70 minutes, there was a sudden
drop in the concentration of 5 paired with a similarly sharp increase in the 6ꢀβ concentration.
5
ACS Paragon Plus Environment

Page 5 of 33
The Journal of Organic Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 2. (a) Expanded 1H NMR spectrum of model glycosylation showing the peaks of interest; (b)
Product and acceptor concentration vs. time. Reaction conditions: 4ꢀβ:1:5::1:1:1 equiv., 40 °C, CDCl3.
This nonꢀlinear behavior in the consumption of 5 and production of 6ꢀβ is unusual and not commonly
observed in standard organic reactions.42ꢀ44 A plausible speculation is that the initial, slow induction
phase is related to transforming the Bi(V) promoter to its active form. Once the promoter has passed
through the resting state, a rate escalation is observed and the glycosylation reaction proceeds to
completion. One obvious scenario is that the promoter activation step is unimolecular, i.e. it does not
involve 5. If so, simply incubating the promoter in the reaction solution that lacks 5 for ~80 minutes and
adding reactant 5 to the solution should show immediate reaction. Our experiments show, however, that
this is not the case, thereby indicating that nucleophile 5 is required for the activation of the promoter
6
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 6 of 33
1
2
3
4
and suggesting that its activation is not unimolecular. These interesting features require deeper
investigation into the influence of the substrates or reactants on the reaction kinetics.
5
6
1
The timeꢀresolved HꢀNMR spectra (Figure 1b) reveal interesting additional features. A
7
8
9
magnification of the relevant chemical shift ranges is shown in Figure 3. In the range of 5.5–5.6 ppm (d,
J = 4.8–5.2 Hz, 1H), there is a doublet signal that grew in strength during the course of the reaction and
disappeared as soon as all of the reactant was consumed. Similarly, multiplets at 2.7 and 1.6 ppm appear
to split into two sets as the reaction proceeded and this trend disappeared also at the end of the reaction.
These observations are consistent with the formation of a reactive intermediate, which is ultimately
consumed when the reaction goes to completion. This discovery is exciting; the direct detection of
reactive intermediates in glycosylation reactions is rare18,29,31,45,46 and particularly so in this case, given
that these experiments are not lowꢀtemperature studies and involve a metal, bismuth, as a promoter. As
illustrated in Scheme 1, these intermediates can form by a variety of pathways when the promoter acts
on a glycosyl donor.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 3. Expanded regions of the kinetics NMR spectra focusing on the transient peaks.
In our case, the growing doublet peak around 5.5 ppm can be assigned to the βꢀanomeric Hꢀ1
that is usually found in the range of 4.5ꢀ6.5 ppm. Interestingly, this doublet peak shifted about 0.02 ppm
during the course of the reaction, as illustrated in Figure 3. Moreover, the peak initially seemed to have a
coupling constant of 4.8 Hz, which later increased slightly to 5.5 Hz. This spectral signature is intriguing
and suggests the existence of a dynamic equilibrium between two closely related glycosyl species during
the activation. If formation of a βꢀglycosyl sulfonium intermediate30,31,45,46 is considered from the βꢀ
donor, then the anomeric Hꢀ1 proton peak is expected to be more deshielded than what a chemical shift
of 5.5 ppm indicates and the coupling constant should be in the range of 9ꢀ11 Hz.
7
ACS Paragon Plus Environment

Page 7 of 33
The Journal of Organic Chemistry
1
2
3
4
5
6
7
8
9
Next, we looked into the possibility of formation of a thiogalactoside radical cation. Previously,
thioglycosides have been activated by a singleꢀelectron transfer (SET) mechanism by methods such as
anodic oxidation or UVꢀ or visibleꢀlight activation in presence of photosensitizers.47,48 To probe the
possibility of this SET pathway, glycosylations were carried out (benchtop and NMRꢀbased) in the
presence of radical scavengers (Galvinoxyl, TEMPO), which can trap radical intermediates and
terminate a radicalꢀmediated activation. Interestingly, the addition of these scavengers to the reaction
mixture has no effect in the activation and still leads to the formation of the desired products (See
Supporting Information for details). Therefore, we concluded that the Biꢀpromoted activation does not
follow a radical pathway or form any in situ radical thioglycoside intermediates.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Another plausible intermediate that can be envisioned in the course of triaryl Bi(V)ꢀmediated
reactions is phenylation of the substrates, typically containing alcohols, amines or thiols.24 If the
thiopropylglycoside is phenylated to form a reactive species, we expected to observe phenylpropyl
sulfide (PhSPr) as one of the byꢀproducts during the activation. To analyze the presence of PhSPr, we
conducted Gas ChromatographyꢀMass Spectrometry (GCꢀMS) based studies. The model glycosylation
(Figure 1a, room temperature instead of 40 °C) was monitored at different time intervals [time (t) = 0
min, 25 min, 60 min (reaction end)]. We did not observe PhSPr (even in trace amounts) at any of these
time points of the thioglycoside activation, a fact that suggests that a phenyl transfer from the triaryl
Bi(V) promoter to the donor probably does not take place. We then synthesized PhSPr,49 added it to a
model glycosylation (Figure 1a) which has gone to completion, and probed the mixture by NMR. The
addition of PhSPr to the reaction did not increase the intensity of the existing peaks in the NMR
spectrum, but created new peaks in the phenyl and alkyl regions. This observation supports the GCMS
results that PhSPr is not formed in the reaction and that phenylation of the thioglycoside does not occur
during the glycosylation. Looking closely at the NMR peaks, we considered that the observed chemical
shifts and coupling constants may originate from an αꢀanomeric Hꢀ1 proton of a glycosyl compound,
suggesting that an anomerization reaction had taken place. In order to test this idea, a number of 1D and
2D correlation NMR studies were conducted on this intermediate. To maximize the population of this
intermediate in solution, the donor (4ꢀβ) and promoter (1) were mixed in absence of an acceptor (5)
under the same condition as employed before. As soon as the intermediate signal was observed, 1D and
2D NMR spectra were obtained and compared to the benzylated αꢀthiopropyl galactoside donor.
1
13
Correlating HꢀNMR (Figure 4), C as well as phaseꢀedit HSQC spectra (see Supporting Information),
we were able to establish that this intermediate was the αꢀdonor, thereby confirming the hypothesis that
the βꢀthioglycosyl donor anomerizes to the αꢀanomer during the reaction course. The discovery of this in
8
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 8 of 33
1
2
3
4
situ anomerization then opened the question of which anomer was actually reacting in the bismuthꢀ
mediated glycosylation.
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 4. 1HꢀNMR comparison of the reaction mixture and 4ꢀα donor.
The anomeric effect—a preference of the axial versus the equatorial glycosidic linkage—has
been well studied since its discovery in 1950s. This effect is generally found to be larger for Oꢀ
glycosides compared to Sꢀglycosides, owing to the lower electronegativity of sulfur compared to
oxygen.50 Nonetheless, few reports have demonstrated the utility of anomerization in thioglycosides to
generate αꢀglycosides preferentially. Interestingly, Boons et al. in their report51 mentioned that the
thioglycosides with small thioalkyl groups can be anomerized via an intermolecular exchange in the
presence of catalytic iodonium promoters. In another case, Murphy and coꢀworkers studied Tiꢀ and Snꢀ
catalyzed anomerization of uronic acids and acetylated thioglycosides and proposed a chelationꢀinduced
endocyclic ring cleavage mechanism for the anomerization.52 However, in both these cases, the
anomerization was achieved under thermodynamic conditions53 unlike our NMR and benchtop
experiments which are kineticallyꢀcontrolled. Therefore, a thermodynamic property like the anomeric
effect53 might not cause the β → α anomerization of donors during the Bi(V)ꢀmediated activation. Here,
we want to highlight that whereas the direct observation of carbohydrate anomerization is common, the
9
ACS Paragon Plus Environment

Page 9 of 33
The Journal of Organic Chemistry
1
2
3
4
5
6
7
8
possible role of such anomerization processes in an ongoing glycosylation is largely ignored. Such an
epimerization process has only been reported54,55 by Poletti and coworkers; they found that βꢀ
trichloroacetimidate donors in ionic liquids—an uncommon glycosylation reaction medium—epimerize
to the αꢀdonor and also convert to the αꢀtriflate intermediate during their activation by strong acids.
Glycosylations are often conducted with the βꢀisomers of glycosyl donors such as thioglycosides; the 1,2
transꢀisomers, or Cꢀ1 βꢀdonors for most sugars are easier to prepare utilizing the NGP of the Cꢀ2 oxygen
protecting group, such as esters or phthalates. On the other hand, the 1,2 cisꢀisomers or the αꢀdonors are
more difficult to form in a stereoselective fashion and are often obtained by purification of an α:β
product mixture derived from a reaction where the Cꢀ2 position is decorated with a nonꢀparticipating
group, such as benzyl ethers or silyl ethers, if they are not just used as a mixture of anomers in the
glycosylation reaction.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 5. (a) Reaction scheme of 4ꢀα donor kinetics study; (b) 6ꢀβ product fraction of 4ꢀβ and 4ꢀα
activation vs. time. Reaction conditions: (Blue diamond) 4ꢀβ:1:5::1:1:1 equiv., (Red square) 4ꢀ
α:1:5::1:1:1 equiv., 40 °C, CDCl3.
To better understand this intriguing anomerization process and its role in the bismuthꢀmediated
reaction mechanism, we considered whether the αꢀthioglycoside would exhibit the same kinetic behavior
as the βꢀthioglycoside. To investigate this, the αꢀthioglycoside donor 4ꢀα was prepared and treated under
the same glycosylation conditions as described for 4ꢀβ (Figure 5a). Figure 5b compares the product
fractions of the αꢀdonor to those we had determined using the βꢀdonor activation. The notably more
rapid increase of the product fraction shown in red in Figure 5b indicates that the “more stable” αꢀdonor
1
reacts considerably faster than the βꢀdonor (76 min vs. 101 min). The HꢀNMR spectra of kinetics data
for 4ꢀα are presented in Figure 6a and show that the Hꢀ1 peak of 4ꢀα at δ 5.5 ppm displays several
10
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 10 of 33
1
2
3
4
5
6
7
8
characteristic features during the activation (Figure 6b). First, the area under this doublet peak
consistently decreased with reaction progress, which indicates that there was continuous consumption
and either no or much slower formation of the βꢀdonor during the activation. This observation is
consistent with the notion that the glycosylation is faster, because 4ꢀα does not need to undergo the
(β→α) anomerization that is required for 4ꢀβ. Secondly, the coupling constant of this doublet, or in other
words the shape of the peak changed from a fine doublet (J=5.4 Hz) to a broad singlet. This
characteristic was also seen with the peak at 5.5 ppm towards the end of 4ꢀβ donor activation.
Interestingly, the peaks at δ 2.5 and δ 1.6 ppm corresponding to the methylene protons on the thiopropyl
group (SCH2CH2–) also decayed with progress of the reaction. Unlike what is shown in Figure 3, the
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
multiplicity of these two peaks do not seem to change and a similar multiplet did not appear in the Hꢀ
NMR spectrum. A sharp triplet (δ 2.7 ppm) and a quartet (δ 1.7 ppm) peak were found to grow as the
reaction progressed.
Figure 6. (a) Reaction progress 1HꢀNMR spectrum of the αꢀdonor glycosylation (b) Expanded 1HꢀNMR
spectrum showing regions of δ 2.4ꢀ2.8 ppm and δ 5.3ꢀ5.8 ppm.
11
ACS Paragon Plus Environment

Page 11 of 33
The Journal of Organic Chemistry
1
2
3
4
5
6
7
8
These peaks may be assigned to methylene protons of the byꢀproduct containing the thiopropyl moiety.
Apart from the improved reaction time—an important discovery for the application of these
glycosylation reactions to automated protocols—the diastereoselectivity of the obtained products is
enhanced from 1:1 to a 1:1.8 ratio in favor of the βꢀanomer. This discovery leads to an improvement of
our current methodology as the previous glycosylation trials did not produce high stereoselectivites in
the products. Interestingly, we still observed the initial slow rise followed by the sudden rate escalation
in the kinetic spectrum of the αꢀdonor (Figure 5b) as seen with the βꢀdonor. This finding suggests that
the β → α anomerization process is not responsible for the nonꢀlinearity and a promoter/catalyst
activation step is involved during the activation.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
To gain a deeper understanding of how the bismuth promoter 1 interacts with the thioglycosides
in solution and construct a model that incorporates the experimental findings reported above, we carried
out a series of quantum chemical calculations using density functional theory (DFT). While theoretical
studies of bismuthꢀcontaining systems are less common than studies on completely organic or transition
metalꢀcontaining systems, a significant amount of work has shown that DFT can be effective for
modeling Bi(III) chemistry,56ꢀ64 although dispersion corrections are often needed. Fewer studies have
been performed on Bi(V) systems; however, there are some notable examples.65,66 Assuming that the
Bi(V) species acts as a Lewis acid, one plausible mode of interaction may be its coordination to the
glycosyl donor by binding to the thiopropyl arm. As mentioned earlier, another possible mechanism has
been proposed by Murphy and coworkers,52 where it was suggested that a Lewis acidic TiCl4 promoter
could chelate to the ring oxygen and a carbonyl group that had been installed on the Cꢀ6 carbon to form
a 5ꢀmembered ring, followed by cleavage of the Cꢀ1 carbon and the ring oxygen. This method of
activation was referred to as endocyclic cleavage. We considered endocyclic cleavage to be a less likely
mechanism for anomerization in our system for several reasons. Given that Ti(IV) is an unambiguously
hard acid, it is likely that it would prefer coordination to hard bases such as ether oxygens. On the other
hand, Bi(V) is a much softer acid, and will prefer coordination to a softer base such as a thioether. This
is also evident from the observation that the Bi(V) promoter selectively activates only Sꢀglycosides in
presence of Oꢀglycosides. In addition, the steric strain of trying to insert the much larger bismuth
complex 1 into a 5ꢀmembered ring will be much worse than for the small TiCl4 complex. Computational
results also suggest that endocyclic cleavage is higher in energy (vide infra). We therefore chose to
focus on the interaction of Bi(V) with the thioether group.
12
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 12 of 33
1
2
3
4
5
6
7
8
9
Scheme 3. Triflate loss from Ph3Bi(OTf)2. Energies are ꢁG(sol) in kcal/mol.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Because the strength of the Lewis acidity and specific mode of binding will depend on the
composition of the Bi(V) promoter, different forms of the promoter must be considered as detailed in
Scheme 3. The initial composition of 1 is as a fiveꢀcoordinate complex, which may lose a triflate ligand
to become a cationic fourꢀcoordinate species [Ph3Bi(OTf)]+ (1a). Loss of another triflate ligand may
afford the dicationic threeꢀcoordinate Bi(V)ꢀcomplex [Ph3Bi]2+ (1b). As summarized in Scheme 3,
dissociation of a triflate ion carries significant energetic penalties—21.4 and 43.1 kcal/mol,
respectively—strongly suggesting that the dominant species in solution is 1. The presence or release of
strong acids such as triflic acid (TfOH) in solution may cause anomerization of glycosides. However, in
this case, the high energetics suggests that if in situ generation of TfOH occurs, it is not mediated by the
bismuth promoter alone but by its interaction with the substrate(s). Consequently, the most likely
reaction scenario is that 1 initially binds to the thiopropyl arm of the sugar.
Given the large size and conformational complexity of the thioglycoside itself, we chose to first
study how a minimalistic model, Me2S, binds to bismuth complex 1. The thioether can either bind at the
equatorial or axial positions to afford the octahedral adducts M1eq or M1ax, respectively (Figure 7).
Calculations suggest that M1eq is unstable and spontaneously loses the thioether ligand, i.e. the sixꢀ
coordinate geometry is not a minimum on the potential energy surface. Binding thioether in the axial
position gives a structure that is a proper minimum, although its formation is uphill by 12.7 kcal/mol on
the solution phase free energy surface. This result is easy to understand; in M1ax the thioether is in trans
disposition to a triflate ligand, whereas in M1eq the thioether is trans to the much stronger σꢀdonor
phenyl ligand. The strong trans influence of the phenyl ligand weakens the Bi–S bond in M1eq.
13
ACS Paragon Plus Environment

Page 13 of 33
The Journal of Organic Chemistry
1
2
3
4
OTf
Ph
OTf
Ph
Ph
R2S
Bi
Ph
Ph
TfO
Bi
Ph
5
OTf
SR2
6
7
M1eq
M1ax
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
OTf
Bi
OTf
Bi
Ph
Ph
Ph
Ph
Ph
S
Ph
OTf
OTf
S
R
R
R
R
Figure 7. Different relevant binding geometries of a SR2 ligand to 1.
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Whereas Me2S is an adequate general model for the Bi–S bonding, it may not be sufficient for
understanding thioglycoside binding to the promoter; the much larger thioglycoside will have significant
attractive van der Waals interactions that will assist adduct formation. Indeed, sixꢀcoordinate
thioglycoside adducts for both axial and equatorial binding were located. As expected, the Bi–S
distances in the equatorially bound structures were about 0.5 Å longer than in the axial bound structures
(Table 1). Thus, the Bi–S bond in 9ꢀβeq and 9ꢀαeq (see Scheme 4 for compound numbering) is weak, if at
all existent and is not responsible for binding the thioglycosides; instead, van der Waals interactions are
key to maintaining the structural integrity. Note that unless stated otherwise, all calculations use the
B3LYPꢀD367ꢀ72 functional, which accounts for dispersion interactions. When the geometries of 9ꢀβeq and
9ꢀαeq were optimized using the B3LYP functional without Grimme’s dispersion corrections, both
adducts dissociated into the respective fragments, as seen in the small model. Our decision to use this
method is discussed in detail in the Supporting Information.
Compound
Bi–S
Length (Å)
2.97
M1ax
9ꢀβax
9ꢀαax
9ꢀβeq
9ꢀαeq
10ꢀβ
10ꢀα
2.81
2.89
3.21
3.42
2.90
2.94
Table 1. Calculated Bi–S bond lengths for various intermediates. Compound numbering is from Scheme
4.
14
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 14 of 33
1
2
3
4
5
6
7
8
OBn
O
BnO
BnO
(12.1)
BnO
S
Pr
9-αax
+ 9.21
– 0.22
– OTf–
+ 8.43
+
OBn
O
Ph
BnO
BnO
Ph Bi Ph
OBn
O
BnO
OTf
OTf
BnO
BnO
S
9
Pr
+ Ph3Bi(OTf)2
BnO
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
SPr
Ph
Ph
Bi Ph
OTf
(2.86)
4-α
OBn
O
BnO
BnO
+ 6.06
(11.9)
(3.42)
+ 0.56
BnO
10-α
9-αeq
S
Pr
TfO
Ph Bi Ph
OBn
BnO
BnO
O
OTf
Ph
(17.9)
+ 2.86
OBn
11
+ Ph3Bi(OTf)(SPr)
+ 8.30
Ph
OTf
Ph
OBn
BnO
BnO
Bi
TfO
S
(8.23)
Ph
O
9-βax
+ 8.23
+ 1.39
– OTf–
– 2.81
Pr
OBn
OBn
BnO
BnO
+
Ph
O
OTf
Ph
OBn
SPr
BnO
BnO
Bi
OBn
O
+ Ph3Bi(OTf)2
S
Ph
Pr
(0.0)
4-β
OBn
(9.62)
Ph
Ph
OTf
Ph
OBn
O
BnO
BnO
Bi
(12.4)
10-β
TfO
S
+ 12.4
9-βeq
Pr
OBn
Scheme 4. Calculated energetics of various intermediates. Energies are ꢁG(sol) in kcal/mol. Energies in
parentheses are relative to 4ꢀβ.
The energetics of the most relevant intermediates is illustrated in Scheme 4. These calculated
energies are valuable information in that they allow us to comment on the thermodynamic viability of
proposed reaction pathways. It is important to note that in the absence of calculated barriers for these
steps, we cannot completely validate a proposed reaction step even if it is thermodynamically viable.
We do model the barrier heights for oxacarbenium formation, which is discussed in more detail below.
For the free thioglycoside the αꢀanomer 4ꢀα was higher in energy than the βꢀanalogue 4ꢀβ by 2.9
kcal/mol. Electronically the two species were essentially isoenergetic, but entropy corrections gives
preference to the βꢀanomer. This result was intriguing as we expected the αꢀthioglycoside to be more
stable than the βꢀdonor, owing to the anomeric effect. But it needs to be noted here that although the
anomeric effect has been observed in some Sꢀcontaining glycosyl donors, extensive highꢀlevel
calculations have not been done to compute energy differences of α or β donors, especially with the
15
ACS Paragon Plus Environment

Page 15 of 33
The Journal of Organic Chemistry
1
2
3
4
5
6
7
8
thioglycosides we have chosen to study. While these calculated anomer energies did not predict the αꢀ
thioglycoside to be lower in energy, they do indicate that the two anomers are very close in energy,
likely within the error of our computational methodology. This implies that the observed interconversion
of donors is possible, and that formation of the higher energy αꢀdonor during the reaction is not
unreasonable.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Further on, binding the Biꢀcomplex 1 to 4ꢀβ in either the axial or equatorial fashion was found to
be uphill by 8.2 and 12.4 kcal/mol, respectively. Once bound, the triflate loss was found to be
substantially easier, possibly as the Bi center is more electronꢀrich, and was associated with 1.4 and ꢀ2.8
kcal/mol for the axial and equatorial species respectively. More interestingly, the binding energetics
were notably different for 4ꢀα. Axial binding of 1 was uphill by 9.2 kcal/mol, similar to what was seen
for 4ꢀβ, but binding it in the equatorial position was much easier, costing only 0.6 kcal/mol. In other
words, formation of the 9ꢀαeq intermediate via 4ꢀα or the αꢀdonor requires much less energy than the
formation of the other possible intermediates. We suspect that these energy differences may be crucial
for determining the reactivities of the donors. At this stage, the triflate loss from the Biꢀglycoside
complexes are much more favorable. From 9ꢀα, the release is easy when bound axially (similar to 9ꢀβ)
but is more difficult for the equatorially bound structure. In all cases, these energies demonstrate a
considerable activation of the Biꢀcomplex for triflate loss when compared to free Bi promoter 1 (Scheme
3). Note that additional triflate loss from 10 is difficult (25ꢀ30 kcal/mol), and is not considered relevant
(see Supporting Information for reaction energies). Altogether, the computed energies for different
reaction pathways indicate that formation of a thioglycosylꢀbismuthonium type intermediate is more
feasible in an equatorial fashion by the αꢀdonor compared to the βꢀdonor, which explains the essential β
→ α anomerization. Mechanistically, we have not yet determined how this isomerization of the donors
takes place. It is likely however, that the anomerization is caused by favorable interactions of the Sꢀ
glycoside with Bi complex and not simply by trace amounts of triflic acid, particularly early on in the
mechanism.
As stated earlier, an SN1ꢀlike mechanism involving an oxacarbeniumꢀlike intermediate is the
accepted mechanism for glycosylations. These glycosyl cations have always been elusive but recently,
Blériot et al. were able to stabilize these cations in superꢀacid medium and study them by NMR in
condensed phase.73 In our system, the Lewis acidic Bi(V) species can bind to the thiopropyl group,
which may make it a better leaving group, and facilitate the oxacarbenium formation. Energetically, it is
most reasonable to dissociate the triflate from 9 and abstract the thiolate group, which is associated with
modest energies at 8.3 and 6.1 kcal/mol for 10ꢀβ and 10ꢀα, respectively (Scheme 4). The thiolate
abstraction prior to triflate loss was found to be much more demanding thermodynamically and will not
16
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 16 of 33
1
2
3
4
5
6
7
8
be considered further (see Supporting Information for all reaction energetics). Overall, the dissociated
fiveꢀcoordinate Ph3Bi(OTf)SPr and oxacarbenium intermediate 11 are 17.9 kcal/mol higher in energy
than the initial reactants. This is a reasonable energy for a fleeting highꢀenergy intermediate. Our
calculations predict that triflate loss from the Ph3Bi(OTf)SPr species is only uphill by 1.3 kcal/mol.
For 10ꢀβ an interesting alternative structure labeled as 10ꢀβch was found where the benzyl
oxygen on Cꢀ6 also coordinates to Bi, essentially allowing the thioglycoside to act as a chelating ligand.
At 3.00 Å the Bi–S bond is lengthened compared to 10ꢀβ and the Bi–O distance is now much shorter
(3.12 Å reduced from 4.62 Å). In addition, the ring oxygen is also 2.85 Å from Bi, suggesting it may be
involved in this coordination as well. These structures are shown in Figure 8. This structure was ~5
kcal/mol higher in energy than 10ꢀβ, suggesting that this alternative structure may be accessible during
the course of the reaction. While it is not yet clear what, if any, relevance this binding mode might have
in the actual chemistry, this unique type of binding was interesting in that it can only occur for the β
form, where the benzyl and sulfur group are on the same side of the ring. It should also be noted that this
structure is somewhat reminiscent of the chelate structure proposed by Murphy discussed earlier.
Notably however the primary interaction is still with the sulfur atom, and not constrained to form a fiveꢀ
membered ring with the two oxygen bases. This suggests that activation at the sulfur arm will still be
possible, if not preferred. While we did not extensively probe the endocyclic cleavage mechanism, we
did use DFT to probe the stability of the critical 6ꢀcoordinate intermediate (10En). As shown in Scheme
5 this structure is 9.29 kcal/mol higher in solvated free energy than Ph3Bi(OTf)SPr and 11. This suggests
that even if endocyclic cleavage did occur, forming an oxacarbenium ion will still be the most likely
energetic path.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 8. Optimized structures for (a) 10ꢀβ and (b) 10ꢀβch. Hydrogens omitted for clarity.
Assuming that thiolate abstraction from 10 is the most relevant for oxacarbenium formation, we
modeled the barrier for this process by performing a relaxed scan (Figure 10) along the C−S distance for
17
ACS Paragon Plus Environment

Page 17 of 33
The Journal of Organic Chemistry
1
2
3
4
10ꢀβ, and 10ꢀα. These scans did not yield genuine SN1 transition states but they provided an upperꢀlimit
estimate for the barrier height for these pathways. The points along the trajectories (Figure 9) are located
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 5. Depiction of endocyclic cleavage from 10ꢀβ. Energies are ꢁG(sol) in kcal/mol. Energies in
parentheses are relative to 4ꢀβ.
by constrained optimizations and hence entropic and zero point energy (ZPE) corrections are not
meaningful. These scans show a sharp initial increase in energy corresponding to breaking the C–S
bond. Instead of decreasing at larger C–S distances, the energy levels off and or gently increases. Thus,
electronically, there is no saddle point that would correspond to a transition state. At long C–S distances
the α and β trajectories have similar energies, especially when solvation corrections are added, which
considerably lower the energy at these distances due to growing charge separation in the dissociating
fragments. These energies at long distances can be used to estimate the barriers to be ~18 and ~14
kcal/mol for the 10ꢀβ and 10ꢀα, respectively. Since 10ꢀβ and 10ꢀα are 9.6 and 11.8 kcal/mol higher in
energy than the reactants, as illustrated in Scheme 4, the overall reaction barriers can be estimated to be
~28 and ~26 kcal/mol for 10ꢀβ and 10ꢀα, respectively. These energies are referenced to 4ꢀβ and since 4ꢀ
α is 2.9 kcal/mol higher in energy than 4ꢀβ, the reaction may be more facile when starting directly from
4ꢀα. The estimated barrier heights are reasonable for a reaction that proceeds slowly at 40 °C and these
calculations suggest that oxacarbenium formation is consistent with the time scale of the reaction. In
addition, given that the reverse reaction(s) will be substantially faster, this also offers a reasonable
pathway for anomerization to either 4ꢀα or 9ꢀα when it is equatorially bound.
18
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 18 of 33
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 9. (a) Electronic energies of 10 as a function of C–S distance. (b) E(sol) (Electronic energies
with solvation corrections) of 10 as a function of C−S distance. The β anomer is represented with blue
diamonds, and the α is represented with red squares.
A final comment is necessary on the quality of our computational results and methodology.
Given the challenges in both properly searching the conformational space, and in accurately modeling
the delicate nonꢀbonding energetics (both of which are discussed in the Supporting Information), the
results here should be interpreted appropriately. It is not our intent to propose that we have necessarily
located the absolute lowest energy structures (global minima) for every possible species, nor that we
have definitively found the only possible mechanisms of anomerization. Rather, we present a thorough
DFT study that demonstrates the kinetic and thermodynamic viability of our proposed mechanism and
complex speciation within the level of accuracy and precision afforded to us by presentꢀday
computational methodology.
With the computational evidence for the favorable formation of a Biꢀglycosyl sulfonium species,
we chose to investigate these ions further with NMR spectroscopy. Glycosyl sulfonium species have
been extensively studied previously75 and although notoriously unstable, they have been isolated and
19
ACS Paragon Plus Environment

Page 19 of 33
The Journal of Organic Chemistry
1
2
3
4
5
6
7
8
studied by NMR.31,76 Most of these studies are on βꢀglycosyl sulfonium species,75 except a report by
Yoshida et al. on the generation and reactivity of αꢀ, βꢀglucosyl sulfonium ions.45 Curiously, they found
that the αꢀsulfonium species reacted faster than their βꢀcounterparts and attributed this to the distortion
of the tetrahydropyran ring of the activated species. This outcome complements our observations that
the βꢀdonor converts to the αꢀdonor as it presumably forms a much more reactive sulfonium. As
observed in the 1H NMRꢀkinetic studies of the 4ꢀα donor, the Hꢀ1 doublet peak shape and chemical shift
changes (Figure 6b) are indicative of a reactive intermediate, and to analyze it, we performed a variety
of 1D and 2D NMR experiments. These experiments were done at various stages of both donor 4ꢀα/β
activation with the Bi(V) promoter without any acceptor in CDCl3 at different temperatures.
Unfortunately, our efforts to obtain structural information of the intermediate by through bondꢀbased
(HSQC, HMBC, COSY, TOCSY) and diffusionꢀbased (DOSY) NMR approaches were not successful.
Correlations of the reactant 4ꢀβ protons with sugar ring carbons and the promoter protons with phenyl
ring carbons could be seen throughout the reaction, but correlations between the donor and promoter
were not detected. It needs to be pointed out here that although sulfonium species have been well
studied, glycosyl sulfonium species connected to Bi or any other heavyꢀmetal atoms have not been
reported or studied by NMR earlier.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The other strategy that we attempted to gain evidence for formation of a Biꢀglycosyl sulfonium
species was application of through spaceꢀbased (NOESY, ROESY) NMR spectroscopy. Fortunately, the
1Dꢀselective NOE studies, where the peaks of interest can be irradiated to show correlations to other
peaks, proved to be informative, although the 2D NOE experiments failed to show interesting peaks. As
shown in Figure 10; (b) when CH3 (1.0 ppm) on thiopropyl group of the thiopropyl group was irradiated,
it correlated with the Ph protons on 1; (c) CH2 on the thiopropyl group (2.6 ppm) on irradiating
surprisingly showed strong correlation signals with the phenyl protons, but weak signals with the rest of
the thiopropyl protons; (d) when transient Hꢀ1 peak on the thioglycoside was irradiated (5.5 ppm), it
correlated with the Ph protons on 1 as well as other sugar ring protons; (e) when the in situ generated
peak (9.2 ppm) belonging to an activated PhBiꢀ species was selectively irradiated, it correlated with
peaks on the glycoside, specifically the methyl (CH3). Based on these interactions, the Biꢀsulfonium
intermediate generated by the computational studies appears to be reasonable [Figure 10(a)]. For
example, the thiopropyl arm of the glycoside is more flexible and hence the terminal CH3 (1.0 ppm) can
have more spatial interactions with the phenyl protons on 1. Similarly, the new Ph proton peak (9.2
ppm) on the Biꢀspecies correlated with this CH3 peak. Interestingly, when the original Ph peaks on 1 (δ
7.7 – 8.2 ppm) were selected, they failed to correlate to any peaks on the glycoside except each other.
20
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 20 of 33
1
2
3
4
5
6
This suggests when 1 binds or is modified, new peaks corresponding to some of these protons were
created, which can be seen 7ꢀ9.5 ppm region. These interactions showcases some structural assignments,
but more studies are required to definitively confirm the Biꢀsulfonium species.
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 10: a) Structure of probable intermediate; 1D selective NOESY of peaks at chemical
shifts δ (b) 1.0 ppm (c) 2.6 ppm (d) 5.5 ppm (e) 9.2 ppm showing correlation with other peaks, (f)
1HNMR of the activation shown in figure 4a.
To gain more insight into the anomerization of donors, we were curious if the presence of a
coordinating group like anacetate at the Cꢀ2 position of the thioglycosyl donor would hinder this
isomerization. As mentioned above, certain functional groups at the Cꢀ2 position of the glycosyl donor
can direct the stereochemistry and afford 1,2 trans-products exclusively. It has been shown that the
carboxyl oxygen of the ester on the Cꢀ2 group can form a transient, bridged dioxalenium ion with the
anomeric carbon.77ꢀ79 In this putative intermediate, the attack of a nucleophile from the “bottom side of
the molecule”, i.e. the αꢀface of the glycosyl donor, becomes difficult. Hence, the incoming nucleophile
attacks preferentially from the equatorial or the βꢀface resulting in the 1,2 transꢀ diastereomers. In our
experiments so far, we observed the β→α isomerization when a nonꢀparticipating group or a benzyl
ether was present at the neighboring Cꢀ2 position of the βꢀthioglycoside. To test this hypothesis, the αꢀ
and βꢀ3,4,6ꢀOꢀbenzyl 2ꢀOꢀacetate thiopropyl galactoside (7ꢀα/β) donors were prepared. This
thioglycoside donor was chosen as it is a galactoside donor like 4ꢀα/β and the functional groups present
21
ACS Paragon Plus Environment

Page 21 of 33
The Journal of Organic Chemistry
1
2
3
4
5
6
7
8
are identical to 4ꢀα/β except at the Cꢀ2 position. These type of donors containing electronꢀrich benzyl
groups (Bn) on the Cꢀ3,4,6 positions and a participating group at Cꢀ2 position (like acetates, benzoates,
etc.) have previously been found to be highly reactive substrates and are generally labeled as superarmed
donors.80
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
Figure 11. (a) Scheme of superarmed βꢀdonor activation (7ꢀβ); HꢀNMR comparison of the activation;
(b) 1HꢀNMR βꢀdonor (7ꢀβ); (c) reaction mixture 1HꢀNMR of activation of 7ꢀβ after 10 min; (d) 1HꢀNMR
of the αꢀdonor (7ꢀα)
The glycosylation of donor 7ꢀβ was carried out with methanol (5) in the presence of promoter 1
(Figure 11a). The reaction was relatively fast, reaching completion within 20 minutes, in contrast to the
activation of donor 4ꢀβ, which required 101 minutes for complete reaction. The 1HꢀNMR spectra of the
22
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 22 of 33
1
2
3
4
5
6
7
8
1
pure anomers (Figure 12b, d) were compared to the HꢀNMR of a reaction mixture (Figure 7c).
Correlating the peaks in the region of 2.2 ꢀ 6.0 ppm, where most carbohydrate signals appear, some
common characteristic peaks were observed as shown by dotted lines in Figure 11. Figure 11c shows a
doublet at δ=4.32 ppm (J=9.9 Hz) and a multiplet at δ=2.7 ppm, which correspond to the Hꢀ1 and the
CH2 of the beta donor 7ꢀβ. Apart from these peaks, another doublet at δ=5.69 ppm (J=5.6 Hz) can be
seen in the reaction mixture, which corresponds to the Hꢀ1 peak of the alphaꢀdonor 7ꢀα and therefore
indicates that this donor is present in the reaction mixture. In addition, another multiplet at δ=2.4 ppm
assigned to the CH2 protons of the αꢀdonor was seen to appear as the reaction progressed. These
observations support the conclusion that the β anomer 7ꢀβ isomerizes to the α anomer 7ꢀα during the
activation. In addition to NMR evidence, thin layer chromatography (TLC) taken during the activation
also showed the appearance of the αꢀdonor 7ꢀα along with the βꢀdonor 7ꢀβ spot (both have distinct Rfs)
as the reaction progressed. Taken together, this evidence suggests that the presence of a coordinating
acetate group at the Cꢀ2 position does not preclude the β→α anomerization of the superarmed
thioglycosyl donor during its activation by a Bi(V) promoter. Interestingly, the Oꢀmethyl glycoside
obtained at the end of the 7ꢀβ activation was found to be exclusively the 1,2ꢀtrans isomer (α/β=1:20),
which is in stark contrast to no stereoselectivity (α/β=1:1) obtained with 4ꢀβ activation. This result
suggests that although the acetate group at Cꢀ2 does not hinder the in situ anomerization, it plays a
crucial role in the stereochemical outcome of the activation.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Conclusions
In summary, the first mechanistic studies of a Ph3Bi(OTf)2ꢀmediated thiopropylglycoside
activation have revealed an unprecedented anomerization of the βꢀdonor to the αꢀanalog as an essential
step during the glycosylation. As a consequence, use of the αꢀdonor alone gave rise to better product
stereoselectivity and increased reaction speeds, thereby addressing some of the practical goals for these
mechanistic studies. Interestingly, both donors followed a sigmoidalꢀtype kinetics pathway, consisting of
an initial slow promoter activation step. Computational analysis delineated the most likely intermediates
involved in activating the thioglycoside, including showing that a pentaꢀ rather than tetraꢀcoordinate
bismuth species is most likely to coordinate to the sulfur of the thioglycoside initially, and that
oxacarbenium formation mediated by Ph3Bi(OTf)2 was viable and in the energy range expected based on
experimental conditions. In addition, we found that the αꢀthioglycoside was effectively equal in energy
to the βꢀform when entropy terms were ignored. As the calculations did not completely explain the
necessary isomerization of the β to αꢀdonor during activation or the observed rate acceleration with the
αꢀanomer, it suggests that this effect may be related to the formation of favorable αꢀ versus βꢀglycosyl
23
ACS Paragon Plus Environment

Page 23 of 33
The Journal of Organic Chemistry
1
2
3
4
5
6
7
8
sulfoniumꢀbismuth intermediates. The oxacarbenium formation mechanism was also investigated, but
did not provide us with information about the observed reactivity of the donors. Some preliminary
insights into the bismuthꢀsulfonium complex were obtained with NOE NMR studies and this data
supports the existence of the computationally generated reactive intermediates. Moreover, the βꢀ to αꢀ
anomerization reaction could still be seen with the superarmed donor, i.e., in the presence of NGP by a
Cꢀ2 acetate group. Although these studies obviously cannot settle every question as to how
thioglycoside reactions work and how Bi(V) species operate, they serve as a good foundation for future
experiments and computational work in this area.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Experimental Section
General Experimental
All reactions were carried out in ovenꢀdried or flameꢀdried glassware, septumꢀcapped under atmospheric
pressure of argon using anhydrous solvents, unless otherwise stated. Airꢀ and moistureꢀsensitive liquids
and solutions were transferred via syringe or stainless steel cannula. Solvents used in the reactions were
obtained from a solvent purification system having a alumina packed column. For the kinetic
experiments, methanol, d1ꢀchloroform were distilled and kept over 3Å molecular sieves. All saccharides
were preꢀdried by azeotropic removal of water using toluene. Prior to use and to ensure dryness,
Ph3Bi(OTf)2 was kept in a vacuum desiccator containing P2O5 overnight. Analytical thinꢀlayer
chromatography (TLC) plates were detected under UV light or by spraying the plates with a 0.02 M
solution of resorcinol in 20% ethanolic H2SO4 solution followed by heating. Proton (1H) NMR, carbon
(13C) NMR and 2D NMR were recorded using the residual signals from d1ꢀchloroform (CDCl3), δ 7.26
ppm and 77.0 ppm, as internal references for 1H and 13C chemical shifts (δ) respectively.
Syntheses of compounds
n-Propylꢀ2,3,4,6ꢀtetraꢀOꢀbenzylꢀ1ꢀthioꢀβꢀDꢀgalactopyranoside (4ꢀβ)
Glycosyl donor was converted into the thiopropylgalactoside (4ꢀβ) following reported methods. The 1H
and 13C NMR spectral data matched with literature.21
n-Propylꢀ2,3,4,6ꢀtetraꢀOꢀbenzylꢀ1ꢀthioꢀαꢀDꢀgalactopyranoside (4ꢀα)
24
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 24 of 33
1
2
3
4
5
6
7
8
9
During the preparation of beta galactoside donor (4ꢀβ), the alpha galactoside donor (4ꢀα) was obtained
as a minor product. Compound is a pale yellow oil; Rf 0.62 (Ethyl acetate:hexanes 1:5) [α]D +83.4 cm3 gꢀ
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
o
1dmꢀ1 (c 0.14 g cmꢀ3, CHCl3); H NMR (400 MHz, [Dꢀ1] CDCl3, 25 C, TMS) δ 7.42 – 7.27 (m, 20H;
PhCH2O), 5.46 (d, J = 5.5 Hz, 1H; Hꢀ1), 4.95 (d, J = 11.4 Hz), 4.84 (d, J = 11.8 Hz), 4.75 (d, J = 11.7
Hz), 4.72 – 4.66 (m), 4.59 – 4.55 (m), 4.49 – 4.37 (m) [8H, PhCH2O], 4.33 – 4.29 (m), 4.29 – 4.25 (m),
3.93 (d, J = 2.4 Hz), 3.81 (dd, J = 9.9, 2.9 Hz) [4H, Hꢀ2,3,4,5], 3.54 (d, J = 6.4 Hz, 2H, Hꢀ6ab), 2.61 –
2.39 (m, 2H, SCH2CH2CH3), 1.65 – 1.58 (m, 2H, SCH2CH2CH3), 0.97 (t, J = 7.3 Hz, 3H,
o
SCH2CH2CH3); 13C NMR (151 MHz, [Dꢀ1] CDCl3, 25 C, TMS) δ 13C NMR (151 MHz, CDCl3) δ
138.91, 138.75, 138.39, 138.18 (4C, 4×Cꢀ1' CPh), 128.4, 128.4, 128.4, 128.3, 128.3, 128.0, 127.7, 127.7,
127.6, 127.6, 127.5 (20C, CPh), 83.8 (1C; Cꢀ1), 79.7, 76.4, 75.3 (3C, Cꢀ2,3,4), 74.9, 73.5, 73.5, 72.6 (4C,
PhCH2O), 69.8 (1C, Cꢀ5), 69.2 (1C,Cꢀ6), 31.6 (1C, SCH2CH2CH3), 22.9 (1C, SCH2CH2CH3), 13.7 (1C,
SCH2CH2CH3); HRMS (ESIꢀQTOF) m/z calc. for C37H42O5 SEt3NH+ 700.4035 found 700.4031, m/z
calc. for C37H42O5SNa+ 621.2645 found 621.2641.
Triphenyl bismuth ditriflate (1)
Promoter (1) was prepared by following previous procedure developed in our lab.81
n-Propylꢀ2ꢀOꢀacetylꢀ3,4,6ꢀtetraꢀOꢀbenzylꢀ1ꢀthioꢀDꢀgalactopyranoside (7ꢀβ, 7ꢀα)
A 0.50 M solution of glycosyl acetate donor82 (1.50 g, 2.81 mmol) and propanethiol (PrSH, 0.256 g,
3.37 mmol) was stirred for 0.5 h in anhydrous dichloromethane at 0 °C. Then, boron trifluoride
dietherate (BF3⋅OEt2, 1.19 g, 8.42 mmol) was added dropwise to the reaction mixture and it was stirred
over an ice bath (0–5 °C) until consumption of the starting donor was seen by TLC. The reaction was
quenched with excess triethylamine, diluted with CH2Cl2, filtered, and washed sequentially with 2 M
25
ACS Paragon Plus Environment

Page 25 of 33
The Journal of Organic Chemistry
1
2
3
4
5
6
aqueous HCl, saturated aqueous NaHCO3, and water. The organic layer was dried over MgSO4,
concentrated under reduced pressure at 40 °C, and the resulting residue was purified by silica gel
column chromatography by a solvent system (Ethyl acetate:hexanes, 1:7), which yielded:
7
8
9
Compound (7ꢀβ) as a white amorphous solid (0.99 g, 65%); Rf 0.65 (Ethyl acetate:hexanes 1:4); [α]D +
67.3 cm3 gꢀ1dmꢀ1 (c 0.12 g cmꢀ3, CHCl3; 1H NMR (600 MHz, [Dꢀ1] CDCl3, 25 oC, TMS) 7.31 (m, 15 H,
PhCH2O) 5.41 (t, J = 9.7 Hz, 1H, Hꢀ2), [4.95 (d, J = 11.7 Hz), 4.68 (d, J = 12.2 Hz), 4.58ꢀ4.55 (2 d, J =
11.9 Hz), 4.48 – 4.39 (m)] (6H, 6×PhCH2O), 4.32 (d, J = 9.9 Hz, 1H, Hꢀ1), 3.99 (dd, J = 2.8, 0.9 Hz,
1H, Hꢀ4), 3.62 – 3.57 (m, 3H, Hꢀ5,Hꢀ6ab), 3.54 (dd, J = 9.7, 2.8 Hz, 1H, Hꢀ3), 2.64 (dddd, J = 12.4, 8.3,
6.5 Hz, 2H, SCH2CH2CH3), 2.04 (s, 3H, CH3C=O) 1.64 – 1.56 (m, 2H, SCH2CH2CH3), 0.94 (t, J = 7.4
Hz, 3H, SCH2CH2CH3); 13C NMR (151 MHz, CDCl3) δ 169.8, 138.7, 138.1, 137.9, 128.5, 128.5, 128.3,
128.1, 128.0, 127.9, 127.8, 127.6, 127.5, 83.9, 81.6, 77.6, 77.4, 77.2, 76.9, 74.5, 73.7, 73.0, 72.1, 69.8,
68.7, 31.7, 23.2, 21.2, 13.6; HRMS (ESIꢀQTOF) m/z calc. for C32H38O6SNa+ 573.2286 found 573.2291.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Compound (7ꢀα) as a clear syrup (0.23 g, 15%); Rf 0.72 (Ethyl acetate:hexanes 1:4); [α]D + 67.3 cm3 gꢀ
1
o
1dmꢀ1 (c 0.12 g cmꢀ3, CHCl3; H NMR (600 MHz, [Dꢀ1] CDCl3, 25 C, TMS) δ 7.36 – 7.27 (m, 15H,
PhCH2O), 5.69 (d, J = 5.6 Hz, 1H, Hꢀ1), 5.41 (dd, J = 10.3, 5.7 Hz, 1H, Hꢀ2), [4.92 (d, J = 11.5 Hz),
4.68 (s), 4.56 (d, J = 11.5 Hz), 4.48 (d, J = 11.8 Hz), 4.42 (d, J = 11.7 Hz), (6H, 6×PhCH2O)], 4.30 (t, J
= 6.5 Hz, 1H, Hꢀ4), 3.97 (d, J = 3.0 Hz, 1H, Hꢀ4), 3.81 (dd, J = 10.3, 2.9 Hz, 1H, Hꢀ3), 3.62 – 3.53 (m,
2H, H, Hꢀ6ab), 2.48 (dddd, J = 12.8, 8.1, 6.6 Hz, 2H SCH2CH2CH3), 2.06 (s, 3H, CH3C=O), 1.63 – 1.56
(m, 2H, SCH2CH2CH3), 0.93 (t, J = 7.3 Hz, 3H, SCH2CH2CH3); 13C NMR (151 MHz, CDCl3) δ 170.3,
138.6, 138.4, 138.1, 128.5, 128.5, 128.4, 128.3, 127.8, 127.7, 127.5, 82.5, 77.7, 77.2, 74.9, 74.7, 73.6,
73.0, 71.4, 69.7, 68.9, 32.1, 23.0, 21.2, 13.6; HRMS (ESIꢀQTOF) m/z calc. for C32H38O6SNa+ 573.2287
found 573.2284.
Kinetics experimental data
Various temperatures were tried for monitoring the reaction kinetics and out of these 40 °C (313 K) was
found to be the optimum temperature, considering the total reaction time. NMR tubes were flameꢀdried
before use.
Typical procedure: The glycosyl donor (1 equiv., 0.025 mmol) and promoter (1 equiv., 0.025 mmol)
were weighed in a septumꢀcapped oven dried 1ꢀdram vial, purged with argon, and then taken to the
NMR instrument. The thermostat of the NMR instrument was allowed to stabilize at 313 K. Then, the
instrument was locked and shimmed using an NMR tube containing only CDCl3. A stock solution of the
acceptor, MeOH in deuterated CDCl3 was prepared and then the required amount (1 equiv., 0.025
26
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 26 of 33
1
2
3
4
5
6
7
8
mmol) was added to the vial via a syringe. The total volume of the solvent was kept constant for each
experiment (0.6 mL). The vial was shaken until reaction mixture reaches homogeneity (generally ∼5ꢀ7
sec), transferred to a new flameꢀdried NMR tube and immediately inserted into NMR machine for data
acquisition. This was taken as time zero and automatic spectra, of 4 scans with a 24 s acquisition delay
(D1) between scans, were registered automatically every 120 seconds for the first 69 FIDs, after which
scans were taken at the intervals of 300 seconds (Table S1). The spectra obtained were processed
manually to obtain the concentration of reactant and product species during the experiment.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Table S1. Time points of data acquisition for reaction kinetics experiments
FID Time FID # Time FID # Time
(min) (min)
65
FID Time FID Time
(min)
216
#
(min)
33
35
37
39
41
43
45
47
49
51
53
55
57
59
61
63
#
(min)
136
141
146
151
156
161
166
171
176
181
186
191
196
201
206
211
#
1
2
1
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
3
5
67
69
221
226
231
236
241
246
251
256
261
266
271
276
281
286
290
3
4
7
71
5
9
76
6
11
13
15
17
19
21
23
25
27
29
31
81
7
86
8
91
9
96
10
11
12
13
14
15
16
101
106
111
116
121
126
131
Radical mechanism studies with scavengers
27
ACS Paragon Plus Environment

Page 27 of 33
The Journal of Organic Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
(a) For the NMR reaction monitoring, we followed the same protocol as described above. The amounts
of the reactants were: donor 4β (1 equiv., 15.0 mg, 0.025 mmol), promoter 1 (1 equiv., 18.0 mg,
0.025 mmol), scavenger (1 equiv., 0.025 mmol), MeOH (1 equiv., 0.001 mL, 0.025 mmol). The
scavenger was added along with MeOH in CDCl3. For these experiments, special care was taken to
degas the solvents before use, which was done by three freezeꢀpumpꢀthaw cycles. The reaction was
then monitored by NMR.
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
(b) For benchtop reaction monitoring, the donor 4β (1 equiv., 15.0 mg, 0.025 mmol) and promoter 1 (1
equiv., 18.0 mg, 0.025 mmol) were taken in a dry 10 mL roundꢀbottomed flask under Ar. To the
flask, a solution of MeOH (1 equiv., 0.001 mL, 0.025 mmol) and scavenger (1 equiv., 0.025 mmol)
in CDCl3 was added. The progress of the reaction or consumption of the donor/formation of products
was then monitored by TLC.
Both the NMR and benchtop reaction monitoring were done in absence of promoter as control studies to
rule out any interfering effects by the scavengers on the reactants.
GC-MS Studies
Materials
PrSH was commercially bought and distilled before use. PrSSPr and PhSPr49 were synthesized
following literature methods and distilled before use.
Procedure for separation of analytes: To find the retention times of the analytes, a stock solution was
prepared by mixing 1 ꢀL of each of the analyte; PrSH, PrSSPr, PhSPr and an internal standard
(dodecane) in 100 mL CH2Cl2 and analyzed on the GCꢀMS instrument. The retention times were
recorded and the relative response factors (RRF) were then calculated. All the analytes were separated
on the chromatogram and could be monitored.
28
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 28 of 33
1
2
3
4
5
6
7
RT
Peak
Peak
Stock
Conc.
Analyte
Formula MW
Density
RRF
(min) height Area
8
9
PrSH
C3H8S
76
2.124 870
8.166 1680
9.092 271
9.575 2449
21.54 0.84
34.37 0.96
5.69 0.75
54.06 0.99
42
48
3.375
4.471
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
PrSSPr
C6H14S2 150
Dodecane C12H26 170
PhSPr C9H12S 152
37.45 1.000
50 7.116
General procedure of reaction monitoring by GC-MS
The glycosyl donor (1 equiv., 0.033 mmol, 20.0 mg) and promoter (1 equiv., 0.033 mmol, 25.0 mg)
were taken in a tightly sealed oven dried flask under Ar. To it a solution of MeOH (1 equiv., 0.033
mmol, 1.35 ꢀL) in CH2Cl2 was added and reaction left to stir under argon. Aliquots (1 ꢀL) from the
reaction mixture at various times were then taken out, mixed with dodecane to make 50 mL stock
solutions in CH2Cl2 and then immediately analyzed by GCꢀMS. The donor consumption was also
monitored by TLC.
Note: Only trace amounts of PrSSPr was observed at the end of the glycosylation (see GCꢀMS spectra
T= 1 h in supporting information)
Corresponding Authors
npohl@indiana.edu, mbaik2805@kaist.ac.kr
Notes
The authors declare no competing financial interests.
Acknowledgment
This material is based in part upon work supported by the National Science Foundation under CHEꢀ
1362213. Purchase of the AVIIIꢀ600 NMR spectrometer used to obtain results included herein was
29
ACS Paragon Plus Environment

Page 29 of 33
The Journal of Organic Chemistry
1
2
3
4
5
6
supported by the National Science Foundation under Grant No. MRI 1040098. We thank the Institute for
Basic Science in Korea for partial support of this work (IBSꢀR10ꢀD1).
Supporting Information
7
8
9
The Supporting Information is available free of charge on the ACS Publications website, which contains
1D, 2D NMR spectra of compounds 4a, 7a, 7b, reaction kinetics NMR and GC spectra, computational
methodologies and calculations (PDF).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
References
(1)
(2)
Bertozzi, C. R.; Kiessling, L. L. Science 2001, 291, 2357.
Varki, A. Essentials of Glycobiology; Second Edition Essentials of Glycobiology. 2nd
edition, 2009.
(3)
(4)
(5)
DeMarco, M. L.; Woods, R. J. Glycobiology 2008, 18, 426.
Werz, D. B.; Seeberger, P. H. Chem. Eur. J. 2005, 11, 3194.
Dwek, R. A.; Butters, T. D.; Platt, F. M.; Zitzmann, N. Nat. Rev. Drug Discov. 2002, 1,
65.
(6)
(7)
(8)
(9)
Kahne, D.; Leimkuhler, C.; Lu, W.; Walsh, C. Chem. Rev. 2005, 105, 425.
Prassas, I.; Diamandis, E. P. Nat. Rev. Drug Discov. 2008, 7, 926.
Nigudkar, S. S.; Demchenko, A. V. Chem. Sci. 2015, 6, 2687.
Boltje, T. J.; Buskas, T.; Boons, G.ꢀJ. Nat. Chem. 2009, 1, 611.
(10) Demchenko, A. V. In Handbook of Chemical Glycosylation; WileyꢀVCH Verlag GmbH
& Co. KGaA: 2008, p 1.
(11) Nicolaou, K. C.; Mitchell, H. J. Angew. Chem. Int. Ed. 2001, 40, 1576.
(12) Kim, J.ꢀH.; Yang, H.; Park, J.; Boons, G.ꢀJ. J. Am. Chem. Soc. 2005, 127, 12090.
(13) Zhu, X.; Schmidt, R. R. Angew. Chem. Int. Ed. 2009, 48, 1900.
(14) Whitfield, D.; Whitfield Adv. Carbohydr. Chem. Biochem. 2009, 62, 83.
(15) Hosoya, T.; Takano, T.; Kosma, P.; Rosenau, T. J. Org. Chem. 2014, 79, 7889.
(16) Mydock, L. K.; Demchenko, A. V. Org. Biomol. Chem. 2010, 8, 497.
(17) Ranade, S. C.; Demchenko, A. V. J. Carbohydr. Chem. 2013, 32, 1.
(18) Frihed, T. G.; Bols, M.; Pedersen, C. M. Chem. Rev. 2015, 115, 4963.
(19) Whitfield, D. Adv. Carbohydr. Chem. Biochem. 2009, 62, 83.
(20) Crich, D. Acc. Chem. Res. 2010, 43, 1144.
(21) Goswami, M.; Ellern, A.; Pohl, N. L. B. Angew. Chem. Int. Ed. 2013, 52, 8441.
(22) Barton, D.; Barton, D. H. R.; Barton, N.; Bhatnagar, J.ꢀP.; Finet, W.; Motherwell
Tetrahedron 1986, 42, 3111.
(23) Barton, D. H. R.; Kitchin, J. P.; Lester, D. J.; Motherwell, W. B.; Papoula, M. T. B.
Tetrahedron 1981, 37, 73.
(24) Matano, Y. In Bismuth-Mediated Organic Reactions; Ollevier, T., Ed.; Springer Berlin
Heidelberg: 2012; Vol. 311, p 19.
(25) Matano, Y.; Ikegami, T. In Organobismuth Chemistry; Suzuki, H., Matano, Y., Eds.;
Elsevier Science: Amsterdam, 2001, p 247.
(26) Ollevier, T. Org. Biomol. Chem. 2013, 11, 2740.
(27) Bothwell, J. M.; Krabbe, S. W.; Mohan, R. S. Chem. Soc. Rev. 2011, 40, 4649.
(28) Adams, J. R.; Goswami, M.; Pohl, N. L. B.; Mallapragada, S. K. RSC Adv. 2014, 4,
15655.
(29) Park, J.; Kawatkar, S.; Kim, J.ꢀH.; Boons, G.ꢀJ. Org. Lett. 2007, 9, 1959.
(30) Fang, T.; Mo, K.ꢀF.; Boons, G.ꢀJ. J. Am. Chem. Soc. 2012, 134, 7545.
(31) Mydock, L. K.; Kamat, M. N.; Demchenko, A. V. Org. Lett. 2011, 13, 2928.
30
ACS Paragon Plus Environment