Driving Force of the Pyranoside-into-Furanoside Rearrangement.

Ab initio calculations of fully O-sulfated model monosaccharides, including common hexoses (glucose, galactose, fucose, and mannose) and pentoses (arabinose and xylose), were performed to study the energetic properties of the recently discovered pyranoside-into-furanoside (PIF) rearrangement. It was shown that the per-O-sulfated derivatives of furanoside isomers generally had lower energies than the corresponding per-O-sulfated pyranosides, while nonsulfated furanosides were always less favored than nonsulfated pyranosides. Mannose, which is known to be unreactive in PIF rearrangement, was the only exception. The results of the theoretical calculations were confirmed by experimental studies of monosaccharide models and explained the driving force of such unusual ring contraction process as PIF rearrangement. The conclusions of performed investigation can be used for prediction of new substrates applicability for PIF rearrangement.

Introduction

It is well known that furanosides are generally less thermodynamically stable than the corresponding pyranosides.[1,2] In an acid-promoted Fischer reaction, furanosides are formed as the kinetic products, whereas thermodynamic equilibrium predominantly results in the formation of pyranosides.[3] The conversion between pyranosides and furanosides containing a substituent at the anomeric position is complicated; however, it is possible under enzymatic conditions. For example, mutase enzymes catalyze the transformation of uridine diphosphate galactopyranose (UDP-Galp, 1) into the corresponding galactofuranose (UDP-Galf, 2) (Scheme A). However, the equilibrium concentration in the latter process is only 5%.[4]

Scheme 1

Examples of PIF Rearrangements: (A) Conversion of UDP-Galp into UDP-Galf by UDP-Galactopyranose Mutase;[4] (B) Equilibrium between Pyranoside and Furanoside in N-Acetylgalactosamine;[5] and (C) PIF Rearrangement under Acid-Promoted Per-O-Sulfation Conditions[6]

Certain substituents at C-2 also facilitate the acid-promoted conversion between the pyranoside and furanoside forms. For example, N-acetylgalactosamine (3) under mild acidic conditions gives 5% of isomeric furanoside 4 (Scheme B). The same ratio of monosaccharides 3 and 4 can be reached by acid treatment of furanoside 4. The participation of the acetyl group was proposed as a key step in the mechanism facilitating the endocyclic cleavage of the O(5)–C(1) bond.[5]

The analogous participation of the 2-O-sulfonato group plays a key role in pyranoside-into-furanoside (PIF) rearrangement (Scheme C) under the acid-promoted per-O-sulfation conditions[23,24] that were discovered in 2014.[6,7] Unlike the previous examples, the PIF rearrangement gives predominantly furanoside products. This process was already successfully used in the syntheses of oligosaccharides related to Aspergillus galactomannan,[8−10]Enterococcus diheteroglycan,[11] galactan I from Klebsiella pneumoniae,[12] and some others.[13]

In our previous investigations, the kinetic aspects of PIF rearrangements were studied, including ab initio calculations of the activation energies of monosaccharides with gluco, galacto, fuco, and manno configurations.[6,7] Additionally, there was a study by Satoh et al. reporting the importance of the conformational strain caused by protecting groups for the endocyclic cleavage reaction.[14] In this work, we studied the energetic aspects of PIF rearrangements, which revealed for the first time the driving force of the nontrivial chemical process of ring contraction. Ab initio calculations were used to explore the relative stability of the corresponding pyranoside and furanoside forms of pentoses and hexoses depending on the presence or absence of sulfate groups on the structures.

Results and Discussion

Experimentally, it was previously demonstrated[7] that in the case of β-galactosides and β-fucosides, the PIF rearrangement proceeds irreversibly until the initially formed fully O-sulfated pyranosides are completely consumed. Thus, allyl galactoside 5 and allyl fucoside 8 under PIF rearrangement conditions give corresponding furanosides 7 and 9, and no traces of the initially formed per-O-sulfated allyl pyranosides 6 and 10 were detected in the reaction mixture after 24 h (Table , entries 1–2).[7] On the other hand, propyl α- and β-mannopyranosides 11 and 14 under the same conditions gave only per-O-sulfated pyranosides 12 and 15, and no traces of proposed furanoside products 13 and 16 were found. The behavior of these mannosides is probably determined by the axial orientation of the OH group at C-2.

Table 1

Acid-Promoted Sulfationa of Galacto-, Fuco-, Manno-, Gluco-, Arabino- and Xylo-Pyranosides

Standard conditions for PIF rearrangement were used for all experiments: (1) Py·SO3 (5 equiv/OH-group), HSO3Cl (2 equiv/OH-group), and dimethylformamide (DMF), 25 °C; and (2) NaHCO3 (aq).

Glucose represented a rather complicated case. Previous experiments on PIF rearrangements of allyl β-d-glucopyranoside resulted in the formation of a mixture of the pyranoside and furanoside, and complete transformation of the glucopyranoside into the furanoside required a long reaction time[7] that caused the formation of undesirable degradation products. To determine if this was due to thermodynamic or kinetical reasons, in the present work, a more reactive propyl β-d-glucopyranoside (17) was used, as the relative rate of the PIF rearrangement of propyl glycosides was shown to be generally higher.[7] Indeed, during 24 h, the initially formed fully O-sulfated propyl β-d-glucopyranoside 19 was consumed to give furanoside 18 in 75% NMR yield.

The presence of CH2OR or CH3 groups at C(5) of the hexoses anchors the conformation and may significantly influence the relative stability of the monosaccharide forms. Therefore, along with hexosides, the PIF rearrangements of pentosides 20 and 23 were also included in this study. The reactivity of pentosides was higher than that of hexosides and can be accompanied by aglycon cleavage as a side reaction. Thus, methyl α-l-arabinopyranoside 20 was completely transformed into α-l-arabinofuranoside 21 as the key product in 2 h while no pyranoside 22 was observed in the NMR spectra (Figure ). In the case of methyl β-d-xylopyranoside 23, the transformation into methyl β-d-xylofuranoside 24 was completed in 5 h.

Figure 1

Part of the 1H NMR spectra of the reaction mixtures: (A) per-O-sulfation of methyl α-l-arabinofuranoside 20 (10 min); (B) acid-promoted PIF rearrangement is almost complete after 2 h, and the signals of initially formed pyranoside 22 have almost disappeared.

For the α-l-arabinoside model structure (entries 1 and 2), the lowest energy conformers[15] were considered for the pyranoside form (4C1 for the nonsulfated form and 1C4 for the per-O-sulfated form) and EO for the furanoside. The complete sulfation causes the furanoside form to be 2.5 kcal/mol more preferable than α-arabinopyranoside. The same situation was observed for the β-d-xyloside model (entries 3 and 4): the pyranoside form tended to adopt an inverted chair 1C4 (Figure ) conformation upon the introduction of sulfates, and the furanoside form became dominant over the pyranoside. This, in our opinion, is the basis for the driving force of the isomerization as the final step of the studied PIF rearrangement process.

Figure 2

Chair (4C1 and 1C4) and skewed (0S2, 3S1 and 1S5) conformations of per-O-sulfated β-glucosides and β-xylosides and their relative energies (kcal/mol). The energies of the lowest energy conformations (1C4 for xylose and 0S2 for glucose) were taken as zero.

A similar preference for the furanoside forms in the case of per-O-sulfated monosaccharides was revealed for the three studied hexosides with β-d-galacto (entries 5 and 6), β-l-fuco (entries 7 and 8), and β-d-gluco configurations (entries 9 and 10). The previous NMR studies[15] reveal that manno- and galacto-pyranosides exist mostly in normal chair configuration 4C1, whereas for the per-O-sulfated β-glucosides, the skewed conformer 0S2 is dominant (Figure ). The conformers observed in the NMR spectra had the lowest energies in the ab initio calculations and were used to estimate the relative stability of the corresponding conformers.

These data clearly indicate that in the unsubstituted forms of all the studied hexoses, the pyranoside isomers are preferable by several kcal/mol. However, upon the introduction of sulfate substituents, the situation is reversed, and the furanoside isomers become more energetically stable, which provides the driving force for the isomerization of per-O-sulfated pyranosides into corresponding furanosides.

The only exceptions are the cases of α- and β-d-mannosides (entries 11–14). In these examples, the pyranoside form is the dominant form for both free and per-O-sulfated compounds, but in the β-isomers, this favorability is less pronounced. This might also account for the fact that in the course of our studies of the PIF rearrangement, we failed to find conditions under which either α- or β-mannosides could be transformed into the furanoside form.

The origin of such a change in the furanoside/pyranoside preference clearly lies in the repulsive interactions between the bulky and charged sulfate groups (Figure ). The presence of these interactions in highly sulfated carbohydrates and their ability to influence the conformation of carbohydrate rings was confirmed previously.[15−17] However, in the furanoside form, one of the sulfates is expelled from the ring and moves to the side chain, which gives it more degrees of freedom and allows it to avoid unfavorable interactions with the ring sulfates. In mannosides, these repulsions are decreased because the O-2 and O-3 sulfates are ax/eq oriented. In the α-mannosides, additionally, the ax/ax orientation of the aglycon and O-2 sulfate reduces the repulsion; thus, these sugars exhibit a stronger preference for the pyranoside form, especially their α-anomers. To prove this hypothesis, NCE energies were computed for all the sulfated molecules in this work using NBO 6.0 software.[18] The results given in Table indicate that for the sulfated pentoses, the NCE energies of the furanosides are lower than those of the pyranosides. For the hexoses with galacto or fuco configurations, the situation is the same, whereas for the sulfated glucoside, these energies are almost equal, and for both mannosides, a strong preference for the pyranoside form is observed (as expected).

Figure 3

Spatial orientation of vicinally located O-sulfate groups in per-O-sulfated methyl β-d-gluco-pyranoside and -furanoside: repulsions in 2,3- and 3,4-pairs of equatorial O-sulfates in 4C1 conformation (A), near to transorientation of 2,3- and 3,4-pairs of O-sulfates in 0S2 conformation (B) and C2-exo furanoside conformation (C).

Table 2

Calculated Differences between Total Energies and Natural Coulomb Electrostatic (NCE) Energies (kcal/mol) of the Furanoside and Pyranoside Forms of Different Monosaccharides

Conclusions

The described results demonstrate that for per-O-sulfated monosaccharides with galacto-, gluco-, fuco-, arabino-, and xylo-configurations, the furanoside forms are more stable than the corresponding pyranoside. The opposite situation occurs for unsulfated derivatives, which is consistent with the common knowledge that they predominantly exist as pyranosides in solution.[1,2] We suggest that this inversion of the energetic parameters serves as the driving force for the PIF rearrangement accompanied by the unexpected ring contraction process. For both α- and β-mannosides, the pyranoside forms are more stable even when all ring oxygen atoms are sulfated. The latter result suggests that mannosides do not undergo PIF rearrangement not only because of kinetics but also due to thermodynamic reasons.

Experimental Section

General Methods

1H and 13C NMR spectra were recorded on Bruker AV-600, Bruker AV-400, or Bruker Fourier 300HD spectrometers equipped with 5 mm pulsed-field gradient probes at 298–303 K. The resonance assignments in the 1H and 13C NMR spectra were made using COSY and HSQC 2D experiments. Chemical shifts are reported in ppm and are referenced to the DMF residual peaks in D2O (δ 3.01 for 1H and δ 37.54 for 13C).

Geometry optimization was performed using the ORCA 2.9.1 program.[19,20] RHF approximation with a 6-311++G** basis set was employed.[21] Sulfates in the studied structures were treated as anions. When the COSMO[22] model was applied, the built-in parameters for DMF were used. Geometry optimizations were performed until the RMS gradient reached a value less than 10–4. The resulting energies were taken after applying outlying charge correction. For the NCE calculations, ORCA 4.0 was used, and NBO 6.0 was called directly from single-point calculations of the previously optimized structures.

PIF Rearrangement (Typical Procedure)[7]

To a stirred solution of the monosaccharide substrate (0.05 mmol) in DMF (1 mL) was added the Py·SO3 complex (5 equiv per OH). The reaction mixture was kept for 10 min at 20 °C, and then, HSO3Cl (2 equiv per OH) was added dropwise under an inert atmosphere. The reaction mixture was stirred for the desired period of time, neutralized with aqueous NH4HCO3, concentrated in vacuo, and coevaporated with H2O and then with D2O. The residue was dissolved in D2O and analyzed by NMR spectroscopy.