Abstract

Graphical Abstract

Open in a separate window

Neighboring-group participation by C-2 acyloxy groups can control the 1,2-trans stereoselectivity of acetal substitution reactions with the employment of strong nucleophiles. 1,2-Trans selectivity increases with decresed electron-donating ability of the acyloxy groups at C-2.

Introduction

Neighboring-group participation involving acyloxy groups is an effective strategy to control stereoselectivity in substitution reactions of acetals and glycosylation reactions, allowing for the synthesis of a range of biologically active compounds.^[1] Upon activation of an acetal, a C-2-acyloxy group can form a stabilized cis-fused dioxolenium ion resembling 2 (scheme 1), which can be opened by a nucleophile to form the 1,2-trans product stereoselectively.^[1e,2] Although neighboring-group participation can be a powerful means to exert stereocontrol in acetal substitution reactions and has become one of the cornerstones of carbohydrate chemistry, the presence of a neighboring acyloxy group does not assure high selectivity.^[1a,1c,1d,3] Understanding the factors that contribute to the success or failure of an acyloxy group to control stereoselectivity in glycosylation reactions can be challenging considering that other, more remote substituents on glycosyl donors can also exert considerable influence on the stereochemical outcomes of acetal substitution reactions.^[4] It is therefore difficult to derive general principles to guide the use of neighboring groups to control the stereochemical courses of nucleophilic substitution reactions of acetals.

Open in a separate window

Scheme 1.

Oxocarbenium ion 1 converts to the stabilized cis-fused dioxolenium ion 2

By examining reactions of acetals carrying a single C-2-acyloxy group, we demonstrate how sensitive neighboring-group participation is to the nature of the acyloxy group at C-2 as well as to the reactivity of the incoming nucleophile. Substitutions of acetals bearing C-2 acyloxy groups with strong nucleophiles are generally highly 1,2-trans diastereoselective, but selectivity decreases as nucleophilicity decreases, in violation of the reactivity-selectivity principle.^[5] This trend is consistent with two reaction pathways involving two different reactive intermediates: the more stabilized cis dioxolenium ions can form the 1,2-trans products through a ring-opening substitution reaction by a strong nucleophile, while weak nucleophiles react with the more reactive oxocarbenium ions, leading to mixtures of products. The extent of stereochemical control depends on the electronic nature of the neighboring group. Reactions of weaker nucleophiles can be rendered highly 1,2-trans selective by use of a participating group that is less electron-donating,^[6] allowing the reaction to proceed through a less stabilized, more reactive cis-fused dioxolenium ion.

Results and Discussion

We started our investigation by studying the substitution of acetal 3, carrying a C-2-acetoxy group, using allyltrimethylsilane as a small nucleophile that reacts irreversibly with carbocations.^[7] In this reaction, the inability of the single acetoxy group to deliver the 1,2-trans selectivity immediately became apparent. Instead of formation of the expected products, the 1,2-cis products were observed.^[8] The magnitude of the selectivity, which was independent of the anomer ratio of the starting material, was consistent with the selectivity observed for 2-benzyloxy-substituted acetals (trans:cis = 17:83), where participation is unlikely.^[4a] The stereoselectivity was relatively insensitive to changes in activator, temperature, or solvent (Table 1). The fact that the presence of triflate ions did not alter the diastereomeric ratio reveals that pathways involving triflate counterions were not responsible for the observed stereoselectivity.^[9] Polar solvents such as acetonitrile and 1-nitropropane facilitated ionization of the anomeric acetal, as anticipated for a reaction involving charged intermediates, leading to higher conversions and enabling reactions at lower temperatures.^[10] The fact that the stereoselectivities were essentially the same for reactions in dichloromethane and acetonitrile indicates that the displacement of a nitrile from a nitrilium ion intermediate was not responsible for the stereoselectivity.^[11]

Table 1.

Effect of conditions on the stereoselectivity of allylations of acetal 3.^[a],[b]

Open in a separate window
Entry	Solvent	Lewis acid	trans:cis	Conversion [%][c]
1[d]	CH2Cl2	SnCl4	14:86	22
2	CH2Cl2	SnCl4	23:77	100
3	CH2Cl2	Me3SiOTf	22:78	24
4	CH2Cl2	BF3•OEt2	21:79	35
5	PhMe	BF3•OEt2	20:80	19
6[e]	TCE	BF3•OEt2	21:79	7
7[f]	THF	BF3•OEt2	–	0
8	C3H7NO2	BF3•OEt2	22:78	100
9[g]	MeCN	BF3•OEt2	22:78	100

Open in a separate window

^[a]Diastereomeric ratios were determined by ¹³C NMR spectroscopy analysis of the unpurified reaction mixtures.^[12]

^[b]Acetal 3 was used as mixtures of diastereomers. The anomer ratio of the starting material should not exert influence on the diastereomeric ratios.

^[c]Conversion determined based on recovered starting material.

^[d]Reaction conducted at –78 °C.

^[e]TCE = Trichloroethylene.

^[f]No reaction.

^[g]Reaction conducted at –40 °C.

Changing the nucleophilicity of the acceptor, however, had an effect on the extent of stereocontrol exerted by the single C-2 acyloxy group (Table 2). The two weakest nucleophiles, allylchlorodimethylsilane and allyltrimethylsilane, reacted with the same diastereoselectivity (Table 2, entry 1 and 2), which is similar to that for allylation of the C-2 benzyloxy-substituted acetal.^[4a] The similar diastereoselectivities likely reflects the intrinsic selectivity of reactions involving an oxocarbenium ion. With increasing nucleophilicity, as measured by the nucleophilicity parameter N,^[7c,13] the relative amount of the 1,2-trans product increased. The 1,2-trans product trans-7 was formed as a single diastereomer in the substitution with sterically hindered silyl ketene acetal 13 (entry 6). A dioxolane product 14 was also isolated in 15% yield (scheme 2), providing evidence for the formation of the cis dioxolenium ion (i.e., 2).^[14] Steric interactions may contribute to the high diastereoselectivity in the formation of products trans-7 and 14 considering the steric bulk of the nucleophile 13.

Open in a separate window

Scheme 2.

Substitutions of acetal 3 with silyl ketene acetal 13.

Table 2.

Nucleophilic substitution reactions of acetal 3 with C-nucleophiles.^[a],[b],[c]

Open in a separate window
Entry	Product	Nu–[M]	N Number[d]	trans:cis
1	4	8	−0.6	23:77
2	4	9	1.7	23:77
3	5	10	4.4	48:52
4[e]	4	11	5.5	55:45
5	6	12	6.2	83:17
6[f]	7	13	9.0	≥ 99:1

Open in a separate window

^[a]Diastereomeric ratios were determined by ¹³C NMR spectroscopy analysis of the unpurified reaction mixtures.^[12]

^[b]Acetal 3 was used as mixtures of diastereomers. The anomer ratio of the starting material should not exert influence on the diastereomeric ratios.

^[c]Attempts to use stronger silyl ketene acetal nucleophiles failed due to decomposition of the nucleophile under various reaction conditions.

^[d]Higher N numbers correlate to higher nucleophilicities.^[13]

^[e]BF₃•OEt₂ used as the activator.^[15]

^[f]N number estimated with the corresponding Me₃Si-protected silyl ketene acetal.^[13]

The trend observed with carbon nucleophiles also holds for oxygen nucleophiles (Table 3). Substitutions of acetal 15 and 16, which possess acyl groups that are most frequently used to provide neighboring-group assistance in O-glycosylation reactions,^[1a,16] revealed a similar relationship between nucleophilicity and stereoselectivity. Reactions with a weak nucleophile, 2,2,2-trifluoroethanol, occurred with low selectivity (entry 1), and with increasing nucleophilicity of the alcohols, as measured by decreasing field inductive parameters (F),^[17] the preference for the formation of the 1,2-trans products increased. Formation of the acetal products was accompanied by the generation of the corresponding orthoester side products, which arose from nucleophilic addition to the cis-fused dioxolenium ions.^[14] In all cases, control experiments indicate that these orthoester products are not reactive intermediates and that their presence does not influence the diastereomeric ratios of the acetal products.^[18]

Table 3.

Substitutions of thioacetal 15 and 16 with O-nucleophiles.^{[a],[b],[c],[d]}

Open in a separate window
Entry	Product	Nucleophile	F Value[e]	trans:cis (R1 = Me)	trans:cis (R1 = Ph)
1[c]	17a/19a	21	0.38	54:46	65:35
2	17b/19b	22	0.29	64:36	73:27
3	17c/19c	23	0.15	70:30	78:22
4	17d/19d	24	0.00	79:21	86:14

Open in a separate window

^[a]Diastereomeric ratios were determined by ¹³C NMR spectroscopy analysis of the unpurified reaction mixtures.^[12]

^[b]Thioacetal 15 and 16 were used as mixtures of diastereomers. The anomer ratio of the starting material should not exert influence on the diastereomeric ratios.

^[c]Diastereomeric ratios were obtained under kinetically controlled conditions.

^[d]Product yields reported as a mixture of 1,2-disubstituted products and orthoester side products (entries 1, 2, and 4), and others as a mixture of 1,2-disubstituted products.

^[e]Field inductive effect parameters. Decreasing F values corresponding to increasing nucleophilicities.^[17]

Similar to the stereoselectivities observed in the reactions of the carbon nucleophiles, the stereoselectivities of the substitutions with the O-nucleophiles do not depend significantly on reaction parameters such as temperature or activation method (Table 4). These variables were investigated using 2,2,2-trifluoroethanol, which reacted with low stereoselectivity, to ensure that any change in stereoselectivity as a result of reaction conditions would be readily apparent. The differences in selectivity were generally modest enough so that it is difficult to draw a conclusion about how changes in reaction conditions alter the stereochemical outcomes of the nucleophilic substitution reactions. The fact that substitution reactions with or without triflate ions occurred with similar selectivities suggests that anomeric triflates were not involved in the stereochemistry-determining step (entries 3–5). Activation of the SPh group with both N-bromo- and N-iodosuccinimide afforded similar results, indicating that glycosyl halides were not involved.^[19] The preference for the formation of the 1,2-trans products decreased with increasing polarity of the solvent, with acetonitrile providing more of the cis-products. The difference in stereoselectivity noted upon changing from the non-polar solvent CH₂Cl₂ to the polar solvent CH₃CN (Table 4, entries 1 and 11) could result from differences in the importance of contact ion pairs between the two solvents. The difference in selectivity, however, was not enough to justify a different mechanism involving nitrilium ions as reactive intermediates in these cases.^[11]

Table 4.

Effect of conditions on the substitutions of thioacetal 16 with 2,2,2-trifluoroethanol.^[a],[b]

Open in a separate window
Entry	Solvent	Temperature (°C)	Activator	trans:cis [c]
1	CH2Cl2	−45 to 25	NIS	63:37
2	CH2Cl2	−45	NIS	65:35
3	CH2Cl2	−45	NIS/Me3SiOTf	57:43
4	CH2Cl2	−45	NIS/TfOH	53:47
5	CH2Cl2	−45	NIS/AgOTf	50:50
6	CH2Cl2	−45	NBS	48:52
7[d]	CH2Cl2	−45	NIS	40:60
8[e]	CH2Cl2	−45	NIS	–
9[f]	PhMe	−45	NIS	54:46
10	MeCN	−40	NIS	28:72
11	MeCN	−45 to 25	NIS	27:73

Open in a separate window

^[a]Diastereomeric ratios were determined by ¹³C NMR spectroscopy analysis of the unpurified reaction mixtures.^[12]

^[b]Thioacetal 16 was used as mixtures of diastereomers. The anomer ratio of the starting material should not exert influence on the diastereomeric ratios.

^[c]Reaction conducted with 1 M concentration in CH₂Cl₂.

^[d]Reaction conducted with 0.01 M concentration in CH₂Cl₂.

^[e]No reaction.

^[f]34% Conversion, determined by ¹³C NMR spectroscopy analysis of the unpurified reaction mixture.^[12]

The stereoselectivity trends observed for substitutions with C- and O-nucleophiles suggest that both cis-fused dioxolenium ions and oxocarbenium ions can be involved in the stereochemistry-determining step (Figure 1). Formation of dioxolane and orthoester products 14, 18, and 20 provided evidence for the involvement of dioxolenium ions (i.e., 26, Figure 1).^[14] Strong nucleophiles can attack these more stabilized and therefore less electrophilic dioxolenium ions on the face opposite to the acyloxy group, leading to the formation of 1,2-trans products.^[20] In contrast, the formation of the 1,2-cis products originates from attack of the oxocarbenium ions by the weaker nucleophiles (i.e., 25, Figure 1).^[4a,4b,21] Reactions that provide significant quantities of both products could arise from a combination of these two pathways, or from reactions of the oxocarbenium ion 25 with a nucleophile at rates approaching the rate of diffusion, which would lead to both stereoisomers of product.^[22] The consistent behavior for both carbon and oxygen nucleophiles argues against the importance of hydrogen bonding to define diastereoselectivity.^[23]

Open in a separate window

Figure 1.

Nucleophilic additions to oxocarbenium ion 25 and cis dioxolenium ion 26.

We reasoned that the relative importance of the two pathways depicted in Figure 1 could depend upon the electron-donating ability of the neighboring group^[24] so we investigated reactions of acetals with different C-2-benzoyloxy groups (Table 5). The reactions of benzoates 28–30 with nucleophiles 9–11 followed a general trend that was unanticipated: the amount of 1,2-trans product increased with decreasing electron-donating capacity along the series 4-MeOC₆H₄ ≈ Ph < 4-NO₂C₆H₄. Considering that the 1,2-cis product must arise from the oxocarbenium ion, the trends show that the dioxolenium ions formed from the weaker participating groups are more readily displaced by the nucleophile. This trend suggests that the transition states for ring-opening of the dioxolenium ion involve considerable bond breakage (vide infra).^[1e,3a]

Table 5.

Substitutions of benzoates 28–30 with C-nucleophiles.^[a],[b]

Open in a separate window
Entry	Nu–[M]	N Number[c]	trans:cis (X = OMe)	trans:cis (X = H)	trans:cis (X = NO2)
1	9	1.7	33:67	34:66	41:59
2	10	4.4	52:48	55:45	84:16
3[d]	11	5.5	54:46	62:38	68:32

Open in a separate window

^[a]Diastereomeric ratios were determined by ¹³C NMR spectroscopy analysis of the unpurified reaction mixtures.^[12]

^[b]Acetal 28, 29 and 30 were used as mixtures of diastereomers. The anomer ratio of the starting material should not exert influence on the diastereomeric ratios.

^[c]Higher N numbers correlate to higher nucleophilicities.^[13]

^[d]BF₃•OEt₂ used as the activator.^[15] Reactions warmed to 25 °C for efficient activation of the acetal leaving group.

The use of a nitrobenzoyloxy group as a participating group to achieve higher stereoselectivity appears to be general. Substitution reactions of a furan acetal bearing a nitrobenzoyloxy group (36) with an oxygen nucleophile occurred with higher stereoselectivity than the corresponding reaction of benzoate 34 (Scheme 3).

Open in a separate window

Scheme 3.

Substitutions of thioacetal 34 and 36 with 2,2,2-trifluoroethanol.

To provide insight into the general trends in diastereoselectivity as a function of the nature of the acyloxy group and the reactivity of the nucleophlile, we studied the substiutution reactions computationally. All computations were performed with ORCA5.03^[25] at the SMD(dichloromethane)-revDSD-PBEP86-D4-def2TZVPP//PCM(dichloromethane)-B3LYP-D3BJ-def2TZVPP^[26] level of theory, which should provide an accurate assessment of the energy of oxocarbenium ions.^[27] The first step involved establishing the conformational energy landscape^{[4c, 28]} of the dioxolenium (R_Diox) and oxocarbenium ions (R_3H4 and R_4H3) for the four aryl-substituted pyrans 38 (2-O-acetyl), 39 (2-O-p-methoxybenzoyl), 39 (2-O-benzoyl) and 40 (2-O-p-nitrobenzoyl) (Supplementary Figure S1). In all cases, the dioxolenium ions, adopting a ¹S₃/B_3,O conformation, were significantly favored over their oxocarbenium ion counterparts, which preferentially adopted a ³H₄ structure.^[1e] As the electron-donating nature of the acyloxy group decreased,^[6] the dioxolenium ion became less favored. The O₁–C₁ bond length of the dioxolenium ion reflected the stabilization by the acyloxy group: the p-nitrobenzoyloxy group coordinated less tightly to the cationic anomeric center, as reflected in its shorter O₁–C₁ and longer O₂–C₁ bonds compared to those for the other benzoyloxy groups (Table 6).

Table 6.

Ground-state conformations of cis-fused dioxolenium ions 39–41.^[a]

Open in a separate window
		Bond length (Å)
Cation	Relative Energy, R3H4 / RDiox (G, kcal mol–1)	O1–C1	O2–C1
39	−13.4	1.344	1.526
40	−11.2	1.338	1.545
41	−8.8	1.331	1.569

Open in a separate window

^[a]Computed at energy level SMD(dichloromethane)-revDSD-PBEP86-D4-def2TZVPP//PCM(dichloromethane)-B3LYP-D3BJ-def2TZVPP.

Next, we explored the nucleophilicity-stereoselectivity trends. To get a clear picture how the nucleophilicity influences the diastereoselectivity of the reactions, calculations were performed with the carbon nucleophiles presented in Table 2 (for computational feasibility, some groups were replaced with electronically similar but smaller groups). The C-nucleophile dataset is best suited for these investigations because we previously found good agreement with experimental results.^[27] The proton transfer involved in the reactions with the O-nuclfeophiles represents a major challenge in computing reaction trajectories because methods that do not employ explicit solvents generally overestimate hydrogen bonds in transitions states.^[29]

The computed reaction profiles for the 2-acetoxy substrate 38 (the reaction profiles for cations 39–41 can be found in Supplementary Figure S2–S4) show that two possible pathways are available for reactions with nucleophiles: direct substitution of the dioxolenium ions (right pathway in Figure 2) or opening of the dioxolenium ion to give a mixture of oxocarbenium conformers and subsequent attack on this ion (left pathway in Figure 2). The highest barrier on the latter pathway is the transition state for the ring opening to give the oxocarbenium ion intermediates (38–R_3H4 and 38–R_4H3). The S_E2’ additions to the oxocarbenium ion were determined to be barrierless for all nucleophiles (Supplementary Figure S5–S10),^[18] The activation energy for direct ring-opening of the dioxolenium ion (38–R_Diox) by the incoming nucleophiles decreases with increasing nucleophilicity (N number, Table 2) of the C-nucleophiles. For the weaker nucleophiles (nucleophiles 8 and 9), the reaction barrier for ring-opening of the dioxolenium ion (respectively 38–TS_A and 38–TS_B) was computed to be higher than the barrier for formation of the oxocarbenium ion (38–TS_F), while the barriers for the opening of the dioxolenium ion by the stronger nucleophiles 10 and 11 (38–TS_C and 38–TS_D) were similar to the barrier for formation of the oxocarbenium ion. The computations show that substitution of the dioxolenium ion with nucleophile 12 (38–TS_E) and 13 (Supplementary Figure S11).^[18] were more favorable than the generation of the oxocarbenium ion.^[18] With increasing reactivity of the nucleophiles, the geometries of the transition states change, with the Nuc–C₁ distance becoming longer and the O₂–C₁ distance shortening, suggesting that the transition states are becoming earlier (atom numbering is shown in Table 6). Overall, the computed reaction profiles follow the observed reactivity-stereoselectivity trends and support the presence of two intermediates leading to products.

Open in a separate window

Figure 2.

Reaction profiles of nucleophilic substitution reactions of 2-O-acetyl pyran cation 38 with C-nucleophiles 8 (A), 9 (B), 10 (C), 11 (D),^[b] 12 (E).^[b] Computed at SMD(dichloromethane)-revDSD-PBEP86-D4-def2TZVPP//PCM(dichloromethane)-B3LYP-D3BJ-def2TZVPP and expressed as relative Gibbs free energy (T=228.15 K) in kcal mol^–1. [a] Electronically barrierless, determined by constrained PES analysis.^[18] [b] The t-Bu groups of nucleophiles 11, 12 were replaced with electronically similar but smaller methyl groups for computational feasibility.

Computational studies also explain the correlation between how electron-donating the C–2 acyloxy group is and the stereoselectivities of substitution reactions. The reaction profiles for the substitution of the aryl dioxolenium ions 39–41 with nucleophile 10 are depicted in Figure 3. The reaction profiles show minimal differences in barrier height between the ring-opening of the dioxolenium ion (right pathway, 39 and 40–TS_C) and formation of the oxocarbenium ion (left pathway, 39 and 40–TS_F) for both the p-methoxybenzoyloxy and benzoyloxy ions. By contrast, the transition state for the substitution reaction of the p-nitrobenzoyloxy group (41–TS_C) is lower in energy than the transition state leading to the oxocarbenium ion (41–TS_F). The net result is that the p-nitrophenyl dioxolenium ion 41 would form more product by the route involving the dioxolenium ion (the right pathway), leading to higher 1,2-trans stereoselectivity compared to the reactions of dioxolenium ions 39 and 40. Examination of the transition state geometries for the substitution reactions show that the barrier height correlates with the O₂•••C₁ stretch (the elongation of the bond in the TS) of 1.056 Å, 1.034 Å and 0.970 Å and a Nuc–C₁ distance of 2.378 Å, 2.417 Å and 2.470 Å for 39, 40 and 41–TS_C respectively. This trend indicates a significantly earlier transition state for the nucleophilic attack on the p-nitrophenyl ion 41.

Open in a separate window

Figure 3.

Reaction profiles of nucleophilic substitution reactions of 2-O-aceloxy pyran cations (39, 40 and 41) with 2-methallyltrimehtylsilane (10, C). Computed at SMD(dichloromethane)-revDSD-PBEP86-D4-def2TZVPP//PCM(dichloromethane)-B3LYP-D3BJ-def2TZVPP and expressed as relative Gibbs free energy (T=228.15 K) in kcal mol^–1. [a] Electronically barrierless, determined by constrained PES analysis.^[18]

When the related profiles for the weaker nucleophile 9 were generated (Supplementary Figure S2–S4), the computations showed the substitution of the dioxolenium ions 39–41 by nucleophile 9 to be significantly higher (G^‡ ~+2–3 kcal mol^–1) than the barrier to form the oxocarbenium ions, explaining why changing the substituent on the benzoyl group has little effect on the stereoselectivity of these reactions because substitution on the dioxolenium ion is too unfavourable. Overall, the balance between the stabilization and reactivity of the dioxolenium ions and the reactivity of the incoming nucleophile determines the stereoselectivity of the acetal substitution reactions.

Conclusion

In conclusion, the ability of acyloxy groups at C-2 of an acetal to guide the stereoselectivity of acetal substitution reactions follows two general trends: 1) 1,2-trans selectivity increases with increased reactivity of the nucleophiles; and 2) 1,2-trans selectivity increases with decreased electron-donating ability of participating groups. Nucleophilic substitutions of acetals bearing common participating groups such as an acetate or a benzoate can be 1,2-trans diastereoselective for highly reactive nucleophiles, whereas reactions with weaker nucleophiles are generally unselective. To achieve 1,2-trans selectivity with less reactive nucleophiles, the installation of a less electron-donating acyloxy group, a group that might be regarded as a less powerful participating group, can lead to higher stereoselectivity because the resulting dioxolenium ion intermediate is less stabilized and therefore reacts more readily with weaker nucleophiles.

Experimental Section

Supplementary Material

Acknowledgements

Footnotes

References