Our glycal epoxide substrates are structurally and electronically distinct from the 1,2-dialkyl epoxide and spirodiepoxide substrates studied previously by the Jamison and Williams groups.9,10
In particular, the observed stereoselectivity for inversion of configuration at the anomeric carbon may be attributed to either of two possible reaction mechanisms (). An SN
2 reaction manifold (pathway A) would require sufficient hydrogen-bonding activation of the epoxide electrophile to induce nucleophilic substitution by the sidechain hydroxyl group without formation of a discrete oxocarbenium intermediate, which might lead to reduced stereoselectivity. Alternatively, an SN
1 mechanism (pathway B) might still afford high stereoselectivity if the C2-alkoxide could sterically and/or electronically block one face of the oxocarbenium ion-pair intermediate. While there is a wealth of literature devoted to describing the influence of nucleophiles, electrophiles, and reaction conditions in promoting SN
1 and SN
2 manifolds in glycosylation11,12
reactions, at benzylic positions14,15
and at acetals,16
the majority of these studies have involved inter
molecular reactions of relatively simple substrates. Moreover, attempts to distinguish between SN
1 and SN
2 manifolds are complicated by the fact that some of the reactions cited above fall into a gray area along the SN
2 continuum and, in some cases, are described as having concurrent, competing reaction manifolds.15,17,18
We envisioned that detailed kinetic studies of a series of electronically-tuned glycal epoxide substrates would allow us to distinguish between these two pathways in this intra
molecular reaction of these complex substrates.
Possible SN2 and SN1 mechanisms for MeOH-induced epoxide-opening spirocyclization of glycal epoxides with inversion of configuration at the anomeric carbon.
Selection of Reaction Probes.
We recently developed a systematic approach to the synthesis of benzannulated spiroketals from C1-aryl glycal substrates.6
To probe the mechanism of the epoxide-opening spirocyclization, we synthesized a series of these glycal substrates in which the electronic character of the anomeric carbon was modulated using various C1-aryl substituents (, 5–7
). These substrates were converted to the corresponding glycal epoxides in situ
and exposed to spontaneous thermal or MeOH-induced spirocyclization conditions.
Spontaneous Thermal Spirocyclizations of Substituted C1-Aryl Glycal Epoxides.a
Diastereomeric product ratios resulting from spontaneous thermal spirocyclizations (−78°C → rt) were first determined to assess the electronic influence of the various aryl substituents (). While most of these reactions favored spirocyclization with retention of configuration at the anomeric carbon to form thermodynamically-favored (bisanomeric) spiroketals (11–13
), increasing formation of the contrathermodynamic (mono-anomeric) inversion products (8–10
) was observed for more electron-deficient substituents (e.g.
). These results are as expected for formation of retention products via a presumed SN
1 mechanism, which is favored by electron-donating substituents that stabilize the discrete oxocarbenium intermediate and disfavored by electron-withdrawing substituents that destabilize this intermediate.19
Interestingly, larger ring systems showed enhanced preferences for spirocyclization with retention of configuration (5
), suggesting that the corresponding inversion products are formed, at least in part, via an SN
2 mechanism, which becomes less competitive with the SN
1 process as the rate of cyclization decreases, since the rate-limiting step in the latter reaction is oxocarbenium formation rather than cyclization.
Next, we investigated the corresponding MeOH-induced epoxide-opening spirocyclizations (). As expected, MeOH provided increased selectivity for inversion of configuration compared to the spontaneous thermal spirocyclizations (cf
. ), although inherent substrate biases could still be observed (e.g.
). As the methyl glycosides have previously been eliminated as intermediates in these reactions5
, double inversion, net retention), the retention products must form via an SN
1 mechanism. Consistent with this mechanism, electron-donating substituents again trended toward increased levels of retention products when considered across the entire substrate panel, presumably by stabilizing the oxocarbenium intermediate, while electron-withdrawing substituents conversely trended toward enhanced selectivity for inversion of configuration by destabilizing this intermediate. Larger ring systems likewise exhibited enhanced preferences for spirocyclization with retention of configuration (5
), as well as increased formation of methyl glycoside side products (16
) that result from competing intermolecular epoxide opening by MeOH, also with retention of configuration at the anomeric carbon. This again suggests that the inversion products are formed preferentially via an SN
2 mechanism that becomes less competitive with alternative SN
1 processes as the rate of cyclization decreases. Notably, the combination of electron-withdrawing substituents and MeOH-induced spirocyclization allowed selective access to 7-membered ring products with inversion of configuration (10d–f
) for the first time via this approach.6
MeOH-Induced Spirocyclizations of Substituted C1-Aryl Glycal Epoxides.a
Based on these results, we surmised that this panel of C1-aryl glycal substrates possessed sufficient electronic variability to influence product distribution and to allow us to probe the mechanism of the MeOH-induced epoxide-opening spirocyclization reaction through detailed kinetic studies.
Temperature Dependence of MeOH-Catalyzed Epoxide-Opening Spirocyclization.
Next, in preparation for kinetic studies using low-temperature NMR, we sought to determine the temperature at which epoxide-opening spirocyclization occurs. To provide a direct comparison to our earlier mechanistic studies with the aliphatic glycal 17
we began by investigating the spirocyclization of this substrate (). The glycal was dissolved in CD3
OD (17.8 M solvent concentration) and cooled to −78 °C, followed by addition of DMDO in acetone to generate the corresponding glycal epoxide in situ
. The sample was then transferred to a precooled NMR probe, by which time the glycal had been converted completely to the corresponding glycal epoxide (<3 min). As observed by low-temperature NMR, the glycal epoxide intermediate was stable at −63 °C over a 2 h time frame (entry 1). Conversion to products within 2 h was observed beginning at −40 °C (not shown), with efficient formation of spiroketal 18
and methyl glycoside 20
occurring at −35 °C (entry 2).20
Optimal Temperatures for Stereoselective Epoxide-Opening Spirocyclization of Glycal Epoxide Substrates.a
In our initial report, we noted that decreased MeOH concentration led to marked decreases in both stereoselectivity (18
) and chemoselectivity (18
However, consistent with the enhanced selectivities observed with five-membered ring systems above (Tables and ), stereo- and chemoselective spirocyclization of the glycal epoxide derived from C1-aryl glycal 5c
could be achieved equally effectively at both the original concentration (17.8 M CD3
OD) and a somewhat lower concentration (11.9 M CD3
OD) (entries 3 and 4). Importantly, the glycal epoxide was unreactive at −35 °C in the absence of CD3
OD (entry 5). These findings were valuable as they allowed the convenient design of experiments in which the concentrations of all three solvents could be varied to assess their influences upon reaction outcome and in which the slower reaction rate at 11.9 M CD3
OD facilitated kinetic studies by low-temperature NMR (vide infra
The corresponding six- and seven-membered ring systems (6c, 7c) only began to spirocyclize at higher temperatures and did so with decreased stereo- and chemoselectivity (entries 6 and 7). Consistent with our mechanistic hypotheses above, we reasoned that, for these larger rings, epoxide-opening spirocyclization with inversion of configuration via an SN2 manifold requires temperatures at which SN1 reactions via oxocarbenium intermediates become competitive or even dominant.
Thus, having identified appropriate temperatures for NMR analysis of the epoxide-opening spirocyclization reactions, we were poised to carry out detailed kinetic studies of these reactions.
Hammett Analysis of Acid-Catalyzed Spiroketal Epimerization via an SN1 Mechanism.
Substrates having electron-withdrawing C1-aryl substituents exhibit increased preferences for epoxide-opening spirocyclization with inversion of configuration (Tables and ). Because oxocarbenium formation should be strongly disfavored in these substrates, we postulated that the inversion products are formed via an SN
2 pathway rather than an SN
1 pathway (). To explore this idea further, we first characterized reactions known to proceed by an SN
1 mechanism to serve as a benchmark against which to compare our proposed SN
2 MeOH-catalyzed epoxide-opening spirocyclization.15,18
Thus, contrathermodynamic (inversion) spiroketals 8a–f were treated with TsOH (10 mol% in CDCl3) at rt to induce epimerization to the corresponding thermodynamic (retention) spiroketals 11a–f via oxocarbenium intermediates 21a–f (). The rate of conversion via this SN1 mechanism was measured by NMR. Substrates having electron-donating C1-aryl substituents that stabilize the requisite oxocarbenium intermediate were expected to exhibit faster reaction rates, while substrates having electron-withdrawing substituents were expected to display slower rates. Accordingly, the p-methoxy-substituted spiroketal 8a reacted completely within <2.5 min (), while the p-nitro-substituted spiroketal 8f showed no conversion over 72 h. Spiroketals 8b–e, bearing substituents with intermediate electronic properties, displayed intermediate rates.
Figure 3 Hammett analysis of acid-catalyzed spiroketal epimerization via an SN1 mechanism. a) SN1 mechanism for acid-catalyzed epimerization of contrathermodynamic (inversion) spiroketals 8a–f to thermodynamic (retention) spiroketals 11a–f. R = (more ...)
Rate constants were determined using the initial rates for these SN
1 processes, and the logarithms of these observed rate constants were plotted against reported σ
values to define the Hammett correlation ().21
Generally, Hammett ρ
values range from +5 to −5,22
with greater absolute values associated with greater buildup of charge at the reactive center. We anticipated that a positive charge would develop at the anomeric center in these reactions, narrowing the range of ρ
values to between 0 and approximately −5.23
As expected, a linear relationship with a ρ
value (slope) of −5.1 indicated a distinctly positive SN
1 transition state.
Hammett Analysis of MeOH-Catalyzed Epoxide-Opening Spirocyclization via a Proposed SN2 Mechanism.
We next used the analogous analysis to probe the transition state for the MeOH-catalyzed epoxide-opening spirocyclization with inversion of configuration (). If an SN2 mechanism is operative (, pathway A), the ρ value should be closer to zero than that observed for the TsOH epimerization above. In contrast, if the epoxide-opening spirocyclization occurs via an SN1 mechanism involving a discrete oxocarbenium intermediate (, pathway B), the ρ value should be similar to that of the TsOH epimerization.
Figure 4 Hammett analysis of methanol-catalyzed epoxide-opening spirocyclization via a proposed SN2 mechanism. a) Methanol-catalyzed epoxide-opening spirocyclization of glycal epoxides 22a–f with inversion of configuration to afford spiroketals 8a–f (more ...)
OD-catalyzed spirocyclization with inversion of configuration of the glycal epoxides 22a–f
generated in situ
(−78 °C, 10 min) was monitored in 11.9 M CD3
OD at −35 °C by NMR (). Rate constants were again determined using the method of initial rates, and the logarithms of the observed rate constants were plotted against reported σ values ().21
The resulting Hammett correlation exhibited a negative slope with a ρ
value of −1.3, consistent with an SN
2 (or SN
2-like) transition state.14f,15c,d,18,24
The slope was significantly shallower than that observed for the SN
1 process above (ρ
= −5.1), indicating the smaller electronic influence of the aryl substituents upon this SN
Kinetic Analysis of MeOH-Catalyzed Epoxide-Opening Spirocyclization.
To assess the role of MeOH in hydrogen-bonding catalysis of the epoxide-opening spirocyclization with inversion of configuration, we determined the kinetic order of MeOH in the transition state. Thus, the initial rates of spirocyclization of the glycal epoxide 22c
(R = H) derived in situ
from glycal 5c
(R = H) were measured in the presence of varying CD3
OD concentrations at −35 °C by NMR.25
A polynomial curve was obtained, suggesting more than one equivalent of MeOH in the transition state (). Non-linear least squares fit of the data to the equation f
) = c
gave values of c
= 2.1, consistent with second-order catalysis by MeOH. Indeed, plotting the observed rate against [CD3
yielded a linear fit with r2
= 0.96 ().
Figure 5 Second-order catalysis by methanol in the epoxide-opening spirocyclization of glycal epoxide 22c (R = H). a) Plot of kobs (min−1) for inversion product formation at varying [CD3OD] yields a polynomial curve. Mean values over two replicate experiments (more ...)
Based on these data, we envision three possible transition states (). In transition state 23
, one molecule of MeOH interacts with the epoxide oxygen and another with the attacking nucleophilic alcohol by hydrogen-bonding interactions. In transition state 24
, one MeOH again interacts with the epoxide oxygen, while the second MeOH engages in two hydrogen bonds to both the alcohol nucleophile and the tetrahydropyran ring oxygen. In this model, the second MeOH may direct the nucleophile to anti
-attack and also enhance reaction selectivity by disfavoring oxocarbenium formation in competing SN
1 pathways. In transition state 25
, both MeOH molecules participate in hydrogen-bonding activation of the epoxide leaving group. This transition state mimics the active site mechanism proposed in epoxide hydrolase enzymes.26,27
Figure 6 Three possible SN2 transition states for MeOH-catalyzed epoxide-opening spirocyclization with inversion of configuration under MeOH hydrogen-bonding catalysis. In transition state 23, both the epoxide leaving group and alcohol nucleophile are activated (more ...)
Strikingly, Jamison and coworkers have also reported second-order catalysis by water in their studies of regioselective epoxide-opening cascades to generate ladder polyethers, and have proposed related hydrogen-bonding interactions in their systems.9a,b
Role of the Glycal Ring Oxygen in MeOH-Catalyzed Epoxide-Opening Spirocyclization.
One of the proposed transition states for the MeOH-catalyzed epoxide-opening spirocyclization invokes hydrogen bonding between the glycal ring oxygen and one molecule of MeOH (24, ). This proposal is attractive in that this interaction might also disfavor competing oxocarbenium formation leading to alternative SN1 pathways. Thus, to investigate the possible role of this glycal ring oxygen, we explored an analogous carbocyclic system lacking that ring oxygen (). As expected, the cyclohexene substrate 26 was less reactive to DMDO oxidation, requiring much higher temperature and longer reaction time than the analogous glycals. Moreover, the resulting cyclohexene oxide intermediate 27 could be isolated at rt, facilitating the introduction of other solvents prior to the subsequent spirocyclization step.
Figure 7 Epoxide-opening spirocyclizations of a cyclohexene oxide substrate lacking the glycal ring oxygen. The thermal spirocyclization in toluene-d8 proceeds to only 85% conversion even after 72 h. Stereochemical assignments were determined by 1H-NMR and NOESY (more ...)
Spirocyclization of 27
in neat MeOH (24.6 M) required warming to 60 °C and 24 h reaction time, affording the spiroether 28
with inversion of configuration. Reaction in toluene-d8
required even higher temperature and longer reaction time (110 °C, 72 h, 85% conversion), establishing the catalytic activity of MeOH in this reaction. Interestingly, treatment of epoxide 27
with TsOH led to rapid conversion to spiroether 28
, again with inversion of configuration, despite the presumably contrathermodynamic nature of this product (axial aryl group observed by NMR).25
That epoxide opening occurs with inversion of configuration even with TsOH further emphasizes the differences in reactivity between alkyl epoxides such as 27
and those studied previously by Jamison and coworkers,9
and the corresponding glycal epoxides that are the focus of this study (cf
. ), for which stereoselectivity becomes a key consideration due to the electronic influence of the glycal ring oxygen.
We next determined the kinetic order of MeOH in the transition state of the spirocyclization of cyclohexene oxide 27
to spiroether 28
(), for comparison to our results with the analogous glycal epoxide 22c
. Initial rates of spirocyclization were measured in the presence of varying CD3
OD concentrations at 60 °C by NMR (). There was no reaction in neat toluene ([CD3
OD] = 0) at 60 °C over the timescale of these experiments. Intriguingly, fitting the data to f
) = c
gave values of c
= 0.59, indicative of fractional-order kinetics. Indeed, plotting the observed rate against [CD3
yielded a linear fit with r2
= 0.99 (). Fractional orders often suggest a complex kinetic scenario.28
Figure 8 Fractional-order catalysis by methanol in the epoxide-opening spirocyclization a substrate lacking the glycal ring oxygen. a) Methanol-catalyzed epoxide-opening spirocyclization of cyclohexene oxide 27 with inversion of configuration to afford spiroether (more ...)
Importantly, this result clearly demonstrates that the glycal ring oxygen, while not required for MeOH catalysis in general, is required for second-order catalysis of the epoxide-opening spirocyclization reaction. This may also be considered additional circumstantial support for proposed transition state 24 (), which is the only one of the three structures that invokes a specific interaction with the ring oxygen, although removal of the ring oxygen also clearly results in a drastic decrease in the inherent reactivity of the epoxide electrophile, which could certainly cause the change in mechanism. In addition, it remains a formal possibility that subtle steric or conformational changes resulting from replacement of the ring oxygen with a methylene group could also disable either of the other proposed transition states 23 or 25.
Inhibition of MeOH-Catalyzed Epoxide-Opening Spirocyclization by Acetone.
The in situ
DMDO epoxidations of the glycal substrates are carried out at −78 °C, a temperature at which spirocyclization does not occur (), making this step insignificant in our kinetic analyses. However, this step necessarily introduces acetone into the reaction milieu, which remains present during the epoxide-opening spirocyclization reaction.29
Under our typical reaction conditions, acetone is maximally one-seventh of the total reaction volume (1.9 M), a relatively minor component. Nonetheless, since acetone is both a hydrogen-bond acceptor and an electrophile that may react with MeOH, it is possible that this cosolvent could affect the spirocyclization reaction. Indeed, initial studies indicated that increased relative acetone concentrations led to increased methyl glycoside formation and decreased stereoselectivity.5,30
Thus, to determine the potential role of acetone in the spirocyclization reaction, initial rates of spirocyclization of the glycal epoxide 22c
(R = H) derived in situ
from glycal 5c
(R = H) were measured in the presence of varying acetone concentrations and constant CD3
OD concentration (11.9 M) at −35 °C by NMR.31
An inhibitory curve was obtained when the rate of spirocyclization was plotted as a function of acetone concentration (). Fitting the data to f
) = c
gave values of c
= −1.0, indicative of first-order inhibition. Indeed, plotting the observed rate against [acetone]−1
yielded a linear fit with r2
= 0.99 ().
Figure 9 First-order inhibition by acetone in the epoxide-opening spirocyclization of glycal epoxide 22c (R = H). a) Plot of kobs (min−1) for inversion product formation at varying [acetone] yields an inhibitory curve. Mean values over two replicate experiments (more ...)
The inhibitory activity of acetone suggests that it may sequester MeOH catalyst molecules, lowering the effective concentration of MeOH, leading to a lower rate of spirocyclization and decreased stereoselectivity for inversion of configuration. However, simple subtraction of [acetone] from [MeOH] is insufficient to explain the rate decreases observed in . This suggests that complex solvent interactions are involved, in which acetone alters the catalytic activity of MeOH in a non-linear fashion. Indeed, previous experimental and computational studies have shown that the hydrogen-bonded network of MeOH species in neat MeOH is both complex and significantly reorganized by the addition of a hydrogen-bond acceptor such as acetone.32
Within the range of acetone concentrations examined herein, reorganizations have been described that decrease higher-order MeOH branching off of hydrogen-bonded MeOH chains. Thus, higher concentrations of acetone may disfavor related higher-order MeOH hydrogen-bonding interactions required in the transition state necessary for selective epoxide opening (cf
Additionally, the previously observed increase in competing methyl glycoside formation5,30
in the presence of increased acetone concentrations suggests that this side reaction becomes more competitive at lower effective concentrations of MeOH (vide infra
Competing Intermolecular Methyl Glycoside Formation.
Intermolecular addition of MeOH is a side reaction that can be competitive with the desired epoxide-opening spirocyclization. In our initial studies, we found that, paradoxically, decreased MeOH concentration led to a profound increase in this competing side reaction.5
As noted above, increased acetone concentrations also lead to increased methyl glycoside formation.5,30
To understand these effects, we carried out kinetic analysis of the methyl glycoside formation reaction using a substrate that cannot undergo spirocyclization.
Thus, the primary alcohol functionality of glycal 5c (R = H) was protected as a TBS ether in 29 (). This glycal was then epoxidized with DMDO (−78 °C, 10 min) and initial rates of methyl glycoside formation were measured in the presence of varying CD3OD concentrations at 15 °C by NMR. Notably, this intermolecular reaction requires a much higher temperature than the corresponding intramolecular spirocyclization, which occurs at −35 °C (cf. ). Fitting the data to f(x) = c[x]n gave values of c = 1.3×10−3 and n = 1.0, indicative of first-order dependence of the methyl glycoside formation reaction upon methanol, and plotting the observed rate against [CD3OD] yielded a linear fit with r2 = 0.94.
Figure 10 First-order dependence on methanol in the intermolecular methyl glycoside formation side reaction. a) Methanolysis of the glycal epoxide generated in situ from protected glycal 29 to afford methyl glycoside 30. b) Plot of kobs (min−1) versus [CD (more ...)
We also observed that the methyl glycoside 30
is formed with retention of configuration at the anomeric carbon.33
At this elevated temperature of 15 °C, we posit that the reaction occurs via an SN
1 mechanism, involving initial methanol-catalyzed epoxide opening to an oxocarbenium intermediate 32
in the rate-limiting step,34
followed by stereoelectronically favored axial attack of methanol ().
Proposed SN1 mechanism for competing intermolecular methyl glycoside-forming side reaction.
Based on these results, it is evident that decreased MeOH concentrations lead to increased methyl glycoside formation by decreasing the relative rate of the desired spirocyclization, as the intermolecular reaction is first-order dependent upon MeOH while the intramolecular reaction is second-order dependent upon MeOH. This is, perhaps, not surprising, as an intermolecular reaction via any of the proposed SN2 transition states () would require three molecules of methanol, with third-order dependence likely rendering that reaction manifold kinetically inaccessible. These results are also consistent with the increased methyl glycoside formation observed in the presence of increased acetone concentrations above, whereby acetone lowers the effective concentration of MeOH, increasing the relative rate of the intermolecular reaction compared to the intramolecular reaction.
Interestingly, in their studies of regioselective epoxide-opening cascades, Jamison and coworkers have reported that differential kinetic orders of dependence upon water similarly influence the relative rates of competing endo
reaction manifolds, playing a critical role in the regioselectivity of those reactions.9b
Finally, the fact that a single equivalent of MeOH is sufficient to catalyze epoxide opening in the intermolecular SN1 manifold suggests that only one such hydrogen-bonding interaction is likewise required in the SN2 epoxide-opening spirocyclization reaction, in contrast to alternative bidentate activation mechanisms precedented in epoxide hydrolase enzymes (23,24 vs. 25, ).