Action of Molybdate Anion on d-Glucosone: Catalytic Conversion to Aldonates Involving C1–C2 Transposition

Abstract

Treatment of the osone (aldos-2-ulose), d-[1-¹³C]glucosone (1¹), with sodium molybdate at 90° for 30 h gave a mixture of d-[2-¹³C]gluconate (2²) and d-[2-¹³C]mannonate (3²) in an 85:15 ratio, indicating that the reaction proceeds with C1–C2 transposition similar to that observed previously with aldoses. Reactions with several singly and doubly ¹³C-labeled isotopomers of 1 confirmed this transposition. In contrast to the aldose reaction, the reaction with 1 is irreversible, presumably due to electrostatic repulsion between the negatively charged catalytically active bimolybdate species and the negatively charged aldonate product. The production of 2 and 3 is mediated by the formation of two structurally distinct bimolybdate complexes, one producing the gluco isomer and the other producing the manno isomer. Reaction byproducts were also observed, specifically d-[1-¹³C]arabinose (4¹) and [¹³C]formate (5), when 1¹ was used as the reactant. These byproducts appear to form from the breakdown of bimolybdate complexes via alternative pathways that compete with those responsible for aldonate formation.

1. Introduction

A remarkable carbon skeletal rearrangement in carbohydrate chemistry occurs during the C2-epimerization of aldoses catalyzed by molybdate anion involving C1–C2 transposition (Scheme 1).¹ The Bilik reaction²⁻⁷ has been proposed to proceed via initial complexation of the acyclic form of an aldose with bimolybdate in which O1, O2, O3, and O4 of the aldose are bound in the complex (Scheme 2). Subsequent reversible exchange of the oxidation states of C1 and C2 then occurs, promoted through reversible covalent bonding between C1 and either C2 or C3. The putative transition state (Scheme 2, structure B) shows two C–O and two C–C bonds either breaking or forming during the reaction. Importantly, this proposed mechanism explains the high stereospecificity of the reaction. In the interconversion of d-glucose and d-mannose, only the re and si faces of the gluco and manno C1 carbonyl groups, respectively, are available for bonding to C3 (Scheme 2). Consequently, the reaction interconverts an aldose with its C2-epimer with few byproducts under mild reaction conditions.¹⁻⁷ Aldose molybdate-catalyzed epimerization (MCE) plays a prominent role in the synthesis of stable isotopically labeled saccharides because of its ability to transfer isotopes initially introduced at C1 of an aldose to the C2 position of its C2-epimer in high yield in a single step.⁸⁻¹⁰

Scheme 1. Aldose Molybdate-Catalyzed C2 Epimerization (MCE) Reaction Showing the Interconversion of d-Glucose and d-Mannose Accompanied by C1–C2 Transposition.

The blue C and red H atoms are transferred intact. The green O atom can exchange with solvent water.

Scheme 2. Interconversion of Putative Bimolybdate Complexes of the Acyclic Aldehyde Forms of d-Glucose (A) and d-Mannose (C) Via Transition-State Structure (B).

Atom numberings in (A) and (B) derive from the gluco starting aldose and that in (C) corresponds to the manno product. R = –CHOH–CH₂OH.

The originally proposed structures¹ of the bimolybdate species shown in Scheme 2 have been challenged by subsequent crystallographic evidence showing that acyclic 1,1-gemdiol (hydrate) forms of aldoses and acyclic alditols complex with bimolybdate.¹¹ Alternative structures have been suggested in which the hydrate forms of aldoses are bound by bimolybdate (Scheme 3).

1

2

Scheme 3. Interconversion of Alternate Bimolybdate Complexes¹¹ in Aldose MCE Involving the Acyclic Hydrate (gem-diol) Forms of d-Glucose (A) and d-Mannose (B).

Arrows in Mo-complex A show S_N2-like insertion and expulsion of OH^– during its conversion to Mo-complex B.

The formation of bimolybdate (Mo₂O₇H₂) from molybdic acid (MoO₃) in aqueous solution and its subsequent complexation with an aldohexose acyclic hydrate to give a dianionic bimolybdate–sugar complex are described by eqs 1 and 2. The interconversion shown in Scheme 3 is initiated by S_N2-attack of OH^– at C2 of Mo-complex A, followed by cleavage of the C2–C3 bond and formation of the C1–C3 bond with expulsion of OH^– from C1 to give Mo-complex B. This process could occur in a concerted or stepwise manner.

Since H1 and H2 are sterically unhindered in bimolybdate complexes, they can be replaced by R-groups. Molybdic acid catalyzes the interconversion of branched-chain aldoses with 2-ketoses,¹² such as 2-C-(methyl)-d-erythrose with 1-deoxy-d-xylulose (Scheme 4A).¹³ In a more structurally challenging case, β-d-galactopyranosyl-(1→2C)-2-C-(hydroxymethyl)-d-erythrose has been shown to interconvert with β-d-galactopyranosyl-(1′→1)-d-threo-2-pentulose (Scheme 4B) using a Mo resin catalyst.¹⁴

Scheme 4. Two Examples of MCE Reactions Showing the Interconversion of 2-C-Substituted d-Erythroses and 2-Ketoses.

The known substrate requirements of aldose MCE led us to consider the following question: can aldos-2-uloses (osones) like d-arabino-hexos-2-ulose (1) (d-glucosone) serve as substrates in the MCE reaction (Scheme 5)? The dihydrate form of 1 contains the required hydroxyl groups for complexation with bimolybdate, and internal reciprocal redox changes at C1 and C2 should give an aldonate product. Since dihydrate 1 is prochiral at C2, two C2-epimeric aldonates would be produced. This report describes the behavior of 1 in the MCE reaction determined from NMR studies of selectively ¹³C-labeled isotopomers and presents structural models that explain the reaction stereochemistry.

Scheme 5. Interconversion of the Dicarbonyl and Dihydrate Forms of d-Glucosone (1) in Aqueous Solution and MCE Reaction of the Latter to Potentially Give C2-Epimeric Aldonates.

2. Experimental Section

2.1. Preparation of ¹³C-Labeled d-Glucosones (1)

d-[1-¹³C]Glucosone (1¹) (the superscript on a compound number denotes the carbon(s) labeled with ¹³C), d-[2-¹³C]glucosone (1²), d-[4-¹³C]glucosone (1⁴), d-[1,2-¹³C₂]glucosone (1^1,2), d-[1,3-¹³C₂]glucosone (1^1,3), and d-[2,3-¹³C₂]glucosone (1^2,3) were prepared from d-[1-¹³C]glucose, d-[2-¹³C]glucose, d-[4-¹³C]glucose, d-[1,2-¹³C₂]glucose, d-[1,3-¹³C₂]glucose, and d-[2,3-¹³C₂]glucose, respectively. The ¹³C-labeled d-glucoses were obtained from Omicron Biochemicals, Inc., South Bend, Indiana. ¹³C-Labeled d-glucoses were converted to ¹³C-labeled d-glucosones using pyranose 2-oxidase (glucose 2-oxidase, EC 1.1.3.10; Sigma), and osones were purified as described previously.¹⁵ Enzyme reactions were typically conducted with ∼200 mg of ¹³C-labeled d-glucose (1.1 mmol) and typically gave ∼170 mg (0.94 mmol, 85% yield) of ¹³C-labeled 1 in >95% purity based on ¹³C{¹H} NMR and HPLC assays after chromatography.

2.2. MCE Reaction Conditions

A reaction mixture containing ¹³C-labeled d-glucosone (1) (∼50 mM) and reagent-grade sodium molybdate (5 mM) in distilled and deionized H₂O at pH 4.5 was stirred in a sealed glass vial at 90 °C for 30 h, after which time the pH of the reaction mixture had dropped to ∼2.2. A Na-phosphate solution at pH 7.4 (0.50 M) was added to adjust the final concentration of Na-phosphate to 100 mM, and the reaction mixture was incubated at 37 °C for 2 d to promote lactone hydrolysis.

2.3. NMR Spectroscopy

High-resolution 1D ¹³C{¹H} NMR spectra of reaction mixtures were obtained on 90% H₂O/10% ²H₂O (v/v) solutions at 22 °C. Samples were analyzed in 5 mm NMR tubes on a Varian DirectDrive 600 MHz FT-NMR spectrometer equipped with a 5 mm ¹H–¹⁹F/¹⁵N–³¹P AutoX dual broadband probe. ¹³C{¹H} NMR spectra (150 MHz) were collected with ∼36,000 Hz spectral windows and ∼3.0 s recycle times. FIDs were processed to optimize spectral S/N, and final spectra had digital resolutions of ∼0.14 Hz/pt. Chemical shifts were referenced externally to sodium 4,4-dimethyl-4-silapentane-1-sulfonate (DSS).

3. Results and Discussion

3.1. Reaction of Sodium Molybdate with d-[¹³C]Glucosones–Formation of ¹³C-Labeled C2-Epimeric Aldonates

A reaction mixture containing d-[1-¹³C]glucosone (1¹) (∼50 mM) in 5 mM sodium molybdate in ²H₂O was incubated at 90 °C for 30 h, after which time the reaction mixture was made 100 mM in Na-phosphate at pH 7.4 and incubated at 37 °C for 2 d. The latter incubation converted 1,4- and 1,5-aldonolactones that formed during the molybdate reaction to acyclic aldonates to simplify NMR analyses of the products. The ¹³C{¹H} NMR spectrum of the resulting reaction mixture contained signals arising from five ¹³C-labeled products: d-[2-¹³C]gluconate (2²) (76.65 ppm), d-[2-¹³C]mannonate (3²) (76.42 ppm), d-[1-¹³C]arabinose (4¹) (α-pyranose, 99.34 ppm; β-pyranose, 95.14 ppm), and [¹³C]formate (5) (173.61 ppm) (Figure 1; Scheme 6). Weak signals were observed at 73.54 and 63.79 ppm but not identified. These results show that the MCE reaction converts 1 to two C2-epimeric aldonates 2 and 3 with accompanying C1–C2 transposition and that the aldonate products are produced in different amounts, with gluco2 more abundant (∼85%) than manno3 (∼15%). As a control, when the same reaction with 1¹ was conducted in the absence of sodium molybdate, aldonate formation was not detected (Figure S2, Supporting Information).

Figure 1.

(A) ¹³C{¹H} NMR spectrum (150 MHz) of the ¹³C-labeled products generated from the reaction of d-[1-¹³C]glucosone (1¹) with sodium molybdate: (a) α-d-[1-¹³C]arabinopyranose (4¹α) (99.3 ppm); (b) β-d-[1-¹³C]arabinopyranose (4¹β) (95.1 ppm); (c) d-[2-¹³C]gluconate (2²) (76.6 ppm); (d) d-[1-¹³C]mannonate (3²) (76.4 ppm); (e) [¹³C]formate (5) (173.6 ppm). (B) Expanded upfield region of (A) showing the closely spaced C2 signals of the [2-¹³C]aldonates. In (A) and (B), only signals arising from ¹³C-labeled carbons are shown.

Scheme 6. Products Generated from the MCE Reaction Using d-[1-¹³C]Glucosone (1¹) Observed by ¹³C{¹H} NMR.

To determine whether the reaction was reversible, authentic d-[1-¹³C]gluconate (2¹) and d-[1-¹³C]mannonate (3¹) were treated separately with molybdate under the same reaction conditions, but neither the expected product d-[2-¹³C]glucosone (1²) nor the C2-epimeric d-[1-¹³C]aldonate were detected. Failure of the reverse reaction is probably due to the inability of the negatively charged bimolybdate species to complex with negatively charged aldonates (pK_a ∼ 3–4). Conversely, in the forward direction, a significant driving force for the breakdown of the aldonate–bimolybdate complex is likely to be electrostatic repulsion. The MCE reaction with 1 differs from that with aldose reactants in that the former involves irreversible (unidirectional) reactions (see more discussion below), while the latter involves equilibration between the starting aldose–bimolybdate complex and that of its C2-epimer (Scheme 2).

Confirmation of C1–C2 transposition accompanying the conversion of 1 to 2 and 3 was obtained by conducting reactions with d-[2-¹³C]glucosone (1²), d-[4-¹³C]glucosone (1⁴), d-[1,3-¹³C₂]glucosone (1^1,3), and d-[2,3-¹³C₂]glucosone (1^2,3). ¹³C{1H} NMR analyses of the product mixtures (Figures 2 and S1, Supporting Information) showed the presence of C2-epimeric d-[1-¹³C]aldonates, d-[4-¹³C]aldonates, d-[2,3-¹³C₂]aldonates, and d-[1,3-¹³C₂]aldonates, respectively, as expected if C1–C2 transposition occurred (Figures S3–S8, Supporting Information). Reactions using different ¹³C-isotopomers of 1 were also used to confirm the 85% gluco/15% manno product ratio of the reaction (Table S1, Supporting Information).

Figure 2.

(A) ¹³C{¹H} NMR spectrum (150 MHz) of the ¹³C-labeled products generated from the reaction of 1² with sodium molybdate: (a) d-[1-¹³C]mannonate (3¹) (181.74 ppm); (b) d-[1-¹³C]gluconate (2¹) (181.22 ppm). The weak signal at 162.89 ppm may arise from the presence of H¹³CO₃^–. (B) ¹³C{¹H} NMR spectrum of the ¹³C-labeled products generated from the reaction of 1^1,3 with sodium molybdate: (a) [¹³C]formate (5) (173.6 ppm); (b,g) α-d-[1,2-¹³C₂]arabinopyranose (4^1,2α) (¹J_C1,C2, 45.5 Hz); (d,j) β-d-[2,3-¹³C₂]arabinopyranose (4^1,2β) (¹J_C1,C2, 45.7 Hz); (e,h) d-[2,3-¹³C₂]gluconate (2^2,3) (¹J_C2,C3, 40.4 Hz); (f,i) d-[2,3-¹³C₂]mannonate (3^2,3) (¹J_C2,C3, 41.0 Hz); (c) β-d-[1,2-¹³C₂]ribopyranose (¹J_C1,C2, 47.0 Hz). In (A) and (B), only signals arising from ¹³C-labeled carbons are shown.

3.2. Mechanistic Implications of C1–C2 Transposition during MCE of 1

Acyclic aldoses form two structurally distinct complexes with bimolybdate (Schemes 2 and 3). These complexes were modeled from X-ray structures of aldoses and alditols complexed with bimolybdate (Scheme S1, Supporting Information).^11,16 By analogy, putative complexes involving the acyclic monohydrate form (at C2) of 1 are shown in Schemes 7 and 8, but the acyclic dihydrate form of 1 could also be involved. Complex 1′ leads to the formation of 2 (Scheme 7), and Complex 2′ leads to 3 (Scheme 8). In both complexes, OH^– initiates the reaction to promote C2–C3 bond cleavage and the formation of a new C1–C3 bond. Upon release of the aldonate, OH^– is regenerated. Structural constraints in both complexes allow only one face of the C1 carbonyl carbon to be attacked during C1–C3 bond formation, leading to the observed product stereochemistry. Present data are insufficient to establish a quantitative kinetic model of the reaction, but the formation of Complexes 1 and 2, and their interconversion with Complexes 1′ and 2′, respectively, are likely to be reversible, whereas the release of the aldonates from Complexes 1′ and 2′ is expected to be irreversible.

Scheme 7. Putative Bimolybdate Complexes in the Conversion of 1² to 2¹ with Concomittant C1–C2 Transposition.

The initially formed Complex 1 has O1 of 1² bound to both molybdenum atoms (bridging oxygen). Complex 1 leads to product 2¹ via Complex 1′.

Scheme 8. Putative Bimolybdate Complexes in the Conversion of 1² to 3¹ with Concomittant C1–C2 Transposition.

The initially formed Complex 2 has the pro-R O2 of 1² bound to both molybdenum atoms (bridging oxygen). Complex 2 leads to product 3¹ via Complex 2′.

3.3. Formation of Byproducts

Prior studies have shown that d-glucosone (1) degrades in aqueous phosphate buffer (pH 7.5) primarily through a C1–C2 cleavage mechanism to give formate (5) and d-ribulose (6) (Scheme 9).¹⁵ However, C1–C2 transposition is also observed during this degradation; when doubly ¹³C-labeled 1(1,3) was used as the reactant, ∼10% of the d-ribulose was labeled at both C1 and C2 and the remaining d-ribulose was labeled only at C2.¹⁵ As discussed above, the treatment of d-[1-¹³C]glucosone (1¹) with molybdate gives d-[2-¹³C]gluconate (2²) and d-[2-¹³C]mannonate (3²) via putative Complexes 1/1′ and 2/2′, respectively (Schemes 6–8). However, d-[1-¹³C]arabinose (4¹) and [¹³C]formate (5) were also observed as byproducts (Figure 1). When 1² was treated with molybdate, d-[1-¹³C]ribulose (6¹) was not observed, indicating that degradation of 1 does not occur to any appreciable extent as a competing process. These results suggest that the formation of 4 and 5 is associated with a Mo-catalyzed reaction. This prediction was tested by reacting 1^1,3 with molybdate, which produced 2^2,3, 3^2,3, d-[1,2-¹³C₂]arabinose (4^1,2), and [¹³C]formate. The fact that 1¹ gave 4¹ (and [¹³C]formate) and 1^1,3 gave 4^1,2 (and [¹³C]formate) (Table 1) indicates that arabinose is generated subsequent to C1–C2 transposition; C1–C2 bond cleavage of 1¹ prior to C1–C2 transposition would give unlabeled 4 while that of 1^1,3 would give 4². Additional determinations of ¹³C labeling patterns in 2–5 using six ¹³C isotopomers of 1 (Table 1) gave results consistent with the reaction pathways shown in Scheme 10. The intramolecular processes in Scheme 7 apparently compete with those in Scheme 10 where, for example, OH^– attack at C1 in Complex 1 leads to C1–C2 bond cleavage to give 4 and [¹³C]formate; that is, C1–C2 bond cleavage occurs prior to C1–C2 transposition. If, however, the C1–C2 transposition pathway out-competes the C1–C2 bond cleavage pathway to give Complex 1′, followed by OH^– attack at C2 of the skeleton-rearranged product and subsequent C1–C2 bond cleavage, then 4¹ and unlabeled formate would be produced. Related reactions are possible for Complex 2 (Scheme S2, Supporting Information), which compete with those in Scheme 8.

Scheme 9. Degradation Pathway of d-[2-¹³C]Glucosone (1²) Showing the Formation of Unlabeled Formate (5) and d-[1-¹³C]Ribulose (6¹) via 2,3-Enediol and 1,3-Dicarbonyl Intermediates¹⁵.

Table 1. ¹³C Isotopomeric Products Observed during MCE Using Different ¹³C Isotopomers of 1.

13C isotopomer of 1	13C isotopomeric products
	formate (5)	d-arabinose (4)	d-gluconate (2)	d-mannonate (3)
1-13C	13C	1-13C	2-13C	2-13C
2-13C	13C	unlabeled	1-13C	1-13C
4-13C	nda	3-13C	4-13C	4-13C
1,2-13C2	13C	1-13C	1,2-13C2	1,2-13C2
1,3-13C2	13C	1,2-13C2	2,3-13C2	2,3-13C2
2,3-13C2	13C	2-13C	1,3-13C2	1,3-13C2

^and = not detected.

Scheme 10. Reaction Pathways That Compete with Those Shown in Scheme 7 in Which d-Arabinose and Formate are Produced from 1¹.

³C labeling in either byproduct depends on whether C–C bond cleavage occurs in Complex 1 or Complex 1’.

Two additional competing processes are possible in Scheme 10 in which OH^– attacks C2 of Complex 1 or C1 of Complex 1′, producing d-arabinonate and [¹³C]formaldehyde, or [1-¹³C]arabinitol and unlabeled bicarbonate, respectively. Similar reactions apply to Complexes 2 and 2′ in Scheme S2 (Supporting Information). There is some experimental evidence indicating that these side reactions occur but at a much lower level than those that produce 4 and 5.

The bimolybdate complexes shown in Schemes 7, 8, and 10 assume that oxygen atoms attached to C1–C4 of 1 are involved in complexation. However, arguing by analogy to reactions of branched-chain aldoses with bimolybdate to give 2-ketoses (Scheme 4), osone 1 could be viewed as a ketosugar analogue in which oxygen atoms attached to C2–C5 instead of C1–C4 are involved in bimolybdate complexation. The primary product of this alternate complexation would be the branched-chain aldopentose 6, and the reaction of 1¹ would give 6². No evidence of the production of 6 was found. If the reaction in Scheme 11 does occur at a level too low to detect, the equilibrium is expected to favor the osone.¹²⁻¹⁴

Scheme 11. Alternative Reaction of 1¹ with Bimolybdate Gives the C2-Labeled Branched Chain Aldopentose 6².

4. Conclusions

MCE of aldoses²⁻⁷ catalyzes chemical exchange between an aldose and its C2 epimer. The reaction is accompanied by C1–C2 transposition in which, for example, d-[1-¹³C]glucose is in equilibrium with d-[2-¹³C]mannose not d-[1-¹³C]mannose.¹ Putative bimolybdate complexes with the starting aldose have been proposed as catalytically active species, although other types of complexes have been proposed (see discussion below). 2-C-Substituted aldoses also serve as substrates, with C1–C2 transposition giving 2-ketose products.¹²⁻¹⁴

The present study was undertaken to determine whether osones (aldos-2-uloses) such as d-glucosone (1) could serve as substrates in molybdate-catalyzed skeletal rearrangement. The previously proposed structural requirements for complexation of aldoses with bimolybdate¹ could be met if the osone bound bimolybdate in its hydrate forms. We found that 1 serves as a substrate, that the reaction proceeds with C1–C2 transposition, and that C2-epimeric aldonates are produced. However, unlike the aldose reaction, which is reversible, the reaction with osones appears to be unidirectional. The lack of reversibility is likely caused by electrostatic repulsion between the negatively charged aldonates and the negatively charged bimolybdate. To account for the formation of two C2-epimeric aldonates in an irreversible process, two different bimolybdate complexes form with osones, each giving a single aldonate product. The relative rates of formation of two initial complexes, and/or of the transposition reactions themselves within the complexes, will influence the final ratio of C2-epimeric aldonates.

Reaction byproducts are observed when MCE is applied to aldoses,¹ but they are relatively minor in abundance, do not appear to be generated by a molybdate-mediated process, and can be reduced by modifying the reaction conditions, notably through the use of a molybdate resin catalyst.¹⁷ With osones, byproducts are also observed but appear to form by alternate routes of breakdown of the bimolybdate complexes. When osone 1 is the reactant, arabinose and formate are produced in the reaction. The fact that d-ribulose is not observed indicates that osone degradation via conventional hydrolytic pathways in solution¹⁵ is minimal under the MCE reaction conditions.

Other types of catalysts have been shown to promote aldose C2-epimerization with concomitant C1–C2 transposition, including various Lewis acid metal(III) chlorides¹⁸ and molybdenum-based polyoxometalate (POM) Keggin (H₃PMo₁₂O₄₀) and Dawson (H₆P₂Mo₁₈O₆₂) clusters.^19,20 Recent detailed computational studies indicate that these catalysts bind O1 and O2 of aldohexoses to form catalytically active complexes; O3–O6 are not involved in binding.¹⁹ In the present work, the conversion of glucosone to aldonates occurs during the first phase of the reaction, where only osone and molybdate are present in aqueous solution. Under these conditions, Keggin- and Dawson-like clusters cannot form. While the latter catalysts might form during the subsequent lactone hydrolysis step,²¹ they play no role in the conversion of osone to aldonate. While the aldose C2-epimerization reaction mechanism proposed for the Keggin and Dawson clusters may apply to reactions involving molybdate and its oligomers, we chose to invoke bimolybdate complexation to model the conversion of osone to aldonates and to explain the formation of reaction byproducts, since it is uncertain at present whether aldoses and osones behave similarly with regard to their mode(s) of binding molybdate and molybdenum-based POM clusters.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.4c04139.

• Expansion of the ¹³C{¹H} NMR spectrum shown in Figure 2B; ¹³C{¹H} NMR spectrum of products generated from a reaction of 1¹ incubated in the absence of sodium molybdate; ¹³C{¹H} NMR spectrum of ¹³C-labeled products generated from the reaction of 1⁴ with sodium molybdate; ¹³C{¹H} NMR spectra of the ¹³C-labeled products generated from the reaction of 1^1,2 with sodium molybdate; ¹³C{¹H} NMR spectrum of the ¹³C-labeled products generated from the reaction of 1^2,3 with sodium molybdate; ¹³C{¹H} NMR spectra (150 MHz) of the ¹³C-labeled products generated from the reaction of 1¹ with sodium molybdate before adjusting the solution pH with phosphate buffer; percentages of 2 and 3 determined from signal integration of ¹³C{¹H} NMR spectra of molybdate reaction mixtures using different ¹³C-isotopomers of 1; crystal structure of d-mannitol complexed with bimolybdate; and reaction pathways that compete with those in Scheme 8 in which 4 and 5 are produced from 1¹ (PDF)

The authors declare no competing financial interest.