Combined Enzyme- and Transition Metal-Catalyzed Strategy for the Enantioselective Syntheses of Nitrogen Heterocycles: (-)-Coniine, DAB-1, and Nectrisine.

The enantioselective syntheses of (-)-coniine, DAB-1, and nectrisine have been developed, utilizing a complementary strategy of enzyme- and transition metal-catalyzed reactions. The initial stereocenter was set with >99% enantioselectivity via an enzyme-catalyzed hydrocyanation reaction. Substrate incompatibilities with the natural enzyme were overcome by tactical utilization of ruthenium-catalyzed olefin metathesis to functionalize an enzyme-derived (R)-allylic fragment. The piperidine and pyrrolidine alkaloid natural products were obtained by a route that leveraged regio- and stereoselective palladium-catalyzed 1,3-substitutive reactions.

Introduction

Functionalized piperidine and pyrrolidine alkaloids are a broad and diverse class of molecules that typically exhibit high levels of biological activity. Nitrogen heterocycles are ubiquitous among natural products and are found in countless pharmaceutical compounds.[1−4] Importantly, the biological activity of these molecules is often tied to a specific enantiomer. Therefore, deeper understanding of the importance of piperidine and pyrrolidine alkaloids is reliant on the development of new strategies for their enantioselective synthesis. Enzymatic reactions are a prime source of enantioenriched material, and their use has been critical to the development of enantioselective reactions.[5,6] One significant drawback of enzymes, however, is their often-limited substrate compatibility toward small organic molecules. Our group has focused on augmenting the high enantioselectivities of enzyme-catalyzed reactions with transition metal-catalyzed processes that are able to stereospecifically alter the position and substitution of enzymatically derived stereochemistry.[7−9]

We targeted (−)-coniine (1), DAB-1 (2), and nectrisine (3) as platforms to demonstrate the general applicability of our synthetic strategy toward the larger classes of saturated nitrogen heterocycles (Figure ). Coniine has a storied past in toxicology circles as the active poison of spotted hemlock and is known most notably for its role in the death of Socrates.[10] Furthermore, the de novo chemical synthesis of coniine remains a touchstone for the synthesis of substituted piperidines.[11−17] DAB-1 (2) and nectrisine (3) are polyhydroxylated pyrrolidine iminosugars isolated from terrestrial sources.[18,19] DAB-1 is a potent inhibitor (IC50 = 9.3 μM) against a specific glycosidase, rabbit muscle glycogen phosphorylase, while being mostly ineffective toward glycogen synthesis enzymes.[20] Structurally related nectrisine (3) is also a strong inhibitor of α-glucosidases.[21] Selective inhibitors of this type are implicated in the treatment of diabetes as well as other sugar-related diseases. As a result, DAB-1 and nectrisine have been the targets of many synthetic efforts and remain useful targets to demonstrate stereoselective reaction methods in complex settings. A majority of these approaches utilized the chiral pool to establish many of the stereocenters;[22−31] however, a small subset of more recent syntheses has used other enantioselective methods to install the requisite stereochemistry.[32−35] As for our own approach, we envisioned the rapid, highly stereocontrolled syntheses of these molecules that would diverge from a single enantiopure precursor. More precisely, we would blend the utility of a highly enantioselective enzymatic reaction with the specificity of modern transition metal-catalyzed reactions to achieve this goal.

Figure 1

Target alkaloids: (−)-coniine (1), DAB-1 (2), and nectrisine (3).

Our unified synthetic plan for (−)-coniine and DAB-1 incorporates three catalytic methodologies for the construction of the heterocyclic rings and stereocenters (Scheme ). The six-membered ring of 1 would be synthesized by ruthenium-catalyzed ring-closing metathesis of diene precursor 4, while five-membered rings would be accessed by cyclization of a γ-amino ester such as 5 after appropriate functionalization of the alkene. We were encouraged by the validation of some of these late-stage approaches in synthetic efforts toward related nitrogen heterocycles.[36−39] The allylic amine stereocenters of 4 and 5 could be constructed by a palladium-catalyzed nucleophilic substitution reaction with concomitant 1,3-transposition from carbonate esters 6. In this reaction, a palladium π-allyl complex generated from 6 would be trapped by a nitrogen nucleophile at the distal position, providing the conjugated enoate 4 or 5, depending on the choice of nucleophile. This reaction would serve as a point of strategic divergence that would allow synthesis of piperidines and pyrrolidines from similar carbonate precursors. The stereocenter of 6 could then be established by an enzyme-controlled hydrocyanation of a requisite α,β-unsaturated aldehyde (7). The oxynitrilase enzyme is known to accomplish this transformation in good yield and with largely excellent enantioselectivities.[8] Additionally, we felt that any enzyme–substrate incompatibilities encountered during the syntheses could be surmounted through olefin cross-metathesis between 7 and the alkene displaying the conflicting group. Although there have been previous reports of allylic transposition to convert alcohol to amine stereocenters in the syntheses of nitrogen heterocycles, we believed our method would serve as a useful contribution because it is does not rely on stoichiometric or hazardous reagents, and the enantiopurity is installed and managed catalytically.[40−43]

Scheme 1

Retrosynthetic Analysis of (−)-Coniine and DAB-1

Results and Discussion

The syntheses began with the installation of the first chiral center by an enzyme-catalyzed hydrocyanation reaction (Scheme ).[44−48] The enzyme oxynitrilase [EC 4.1.2.10],[49] found in raw bitter almonds, has been shown to provide cyanohydrins of α,β-unsaturated aldehydes in good yields with remarkable enantiopurity.[50]trans-2-Hexenal (8a) and crotonaldehyde (8b) are both excellent substrates for this process. Exposure of these aldehydes to oxynitrilase in the presence of excess HCN furnished cyanohydrins 7 in good yields and >99% enantiomeric excess.[8] This reaction is particularly serviceable because of the low cost and ready availability of the enzyme from store-bought raw almonds, which, after grinding and removal of fats by extraction, are ready to use in the natural enzymatic matrix. The conversion of cyanohydrins 7 into their analogous α-hydroxyesters 9 was conducted under anhydrous Pinner conditions using TMSCl in ethanol. It is important to stress that the anhydrous Pinner conditions are compulsory if the erosion of enantiomeric purity that occurs under aqueous conditions is to be avoided. This functional group interconversion was necessitated by our earlier studies that found α,β-unsaturated cyanohydrin carbonates were poor substrates for stereo-controlled, palladium-catalyzed allylic substitution reactions. The corresponding ethyl ester carbonates (6), however, demonstrated excellent stereocontrol.[8,51] The more sterically demanding ethyl ester was better suited to attenuate π–σ–π isomerization of the π-allyl intermediate that led to competitive formation of the (S)-cis product than the more compact nitrile. Accordingly, α-hydroxyester 9a was converted to carbonate 6a with ethyl chloroformate without event.[8,9]

Scheme 2

Synthesis of Building Blocks Utilizing the Oxynitrilase Enzyme

Our next challenge was to develop optimal conditions for maximum stereocontrol in the crucial palladium-catalyzed 1,3-chirality transfer step. Although many methods exist for the 1,3-chirality transfer of allylic alcohol derivatives,[40−43] we believed that the palladium-catalyzed method provides the greatest flexibility in terms of scope of nucleophile choice and general mildness of the reaction conditions.[52] Of special concern to us were the competing inner- and outer-sphere nucleophilic addition processes to the cationic palladium π-allyl intermediate generated by exposure of carbonate 6a to palladium(0) catalysts (Figure ). Specifically, amine nucleophiles that react directly with the chiral π-allyl complex 10a generate C–N bond formation products ((R)-11) with a net retention of stereochemistry (pathway A). On the other hand, Lewis basic amines that are inclined to coordinate to the metal (e.g., 10b) undergo an inner-sphere reductive elimination to give (S)-11 (pathway B). Products of pathway B would be delivered with a net inversion of configuration. We understood that this stereospecificity could be governed by hard–soft acid base theory, with softer nucleophiles reacting more preferentially with the π-allyl ligand directly (pathway A) and harder nucleophiles operating through an inner-sphere process (pathway B).[53] It was not clear to us at the outset whether amines would have the requisite hardness or softness to be selective for one pathway over the other, and it was thought that judicious choice of nitrogen-protecting groups could be used to modulate the stereospecificity.[54]

Figure 2

Competing π-allyl addition pathways.

Our initial attempts at palladium-catalyzed C–N bond formation began with the reaction between 3-butenylamine and allylic carbonate 6a (Table , entry 1). Although the yield of the resulting allylic amine (12a) was acceptable, all efforts to improve the stereospecificity of the reaction were unsuccessful. We, therefore, turned our attention to more highly resonance-stabilized versions of the 3-butenylamine moiety in order to soften the nucleophile and promote preferential outer-sphere substitution.[53] Although carbamates are significantly more acidified relative to primary amines, the reduced nucleophilicity of the bulky tert-butyl carbamate-protected 3-butenylamine provided only trace amounts of C–N bond formation (entry 2). We then further increased the acidity of the butenyl nitrogen, and therefore its base-mediated nucleophilicity, by converting it to the corresponding 2,4-dinitrobenzenesulfonamide (DNs) using 2,4-dinitrobenzenesulfonyl chloride. Upon exposure to the palladium catalyst in dichloromethane at room temperature, allylic carbonate 6a and the DNs-derived nucleophile quickly reacted to form 12c in 80% yield and 80% ee (entry 3). Although these numbers reflect similar results to those obtained under analogous conditions in the 3-butenylamine experiment, we found that the rate of reaction with the DNs-sulfonamide to be superior. We also looked into alternative ligands for the palladium catalyst, but the results were disappointing and detrimental to the yields (entries 4 and 5). Ultimately, we found that reducing the reaction temperature to −10 °C and employing the reactive sulfonamide nucleophile enabled us to obtain excellent enantiospecificity of 96% with only a modest reduction in overall yield (entry 6). Any further decrease in temperature resulted in poor yields (entry 7). We anticipate that secondary sulfonamides of this type will have broad utility in palladium-catalyzed allylic substitution reactions where high levels of stereospecificity are desired. The appeal of this reaction is further enhanced by the ease of both the synthesis and removal of the DNs group under mildly nucleophilic conditions.[55]

Table 1

Palladium-Catalyzed Amination of Carbonate 6a

entrya	R	ligand	t (°C)	12	yield (%)	ee (%)b
1	H	PPh3	25	12a	75c	74d
2	Boc	PPh3	25	12b	<5	nd
3	DNs	PPh3	25	12c	80	80
4	DNs	DPPE	25	12c	<5	nd
5	DNs	(R,R)-Trost ligand	25	12c	26	90
6	DNs	PPh3	–10	12c	67	96
7	DNs	PPh3	–20	12c	20	95

1.5 equiv amine, 0.25 M CH2Cl2.

Determined by enantioselective HPLC.

Determined after conversion to the t-butyl carbamate 12b (R = Boc).

Determined after conversion to the benzoyl amide (R = PhCO).

With optimized conditions for the palladium-catalyzed 1,3-chirality transfer in hand, we moved forward with the completion of the synthesis (Scheme ). Ring-closing metathesis with the second-generation Hoveyda–Grubbs catalyst excised the α,β-unsaturated ester as expected to provide DNs-protected unsaturated piperidine 13. In a single pot, 13 was rapidly deprotected with excess allyl amine followed by reprotection with Boc2O to afford the corresponding tert-butyl carbamate (14) in 75% yield. This functional group interconversion was initiated to avoid difficulties inherent with the purification and handling of volatile-free amines. Palladium-catalyzed hydrogenation of the remaining alkene readily yielded Boc-protected coniine 15. Because this intermediate has itself been transformed into (−)-coniine (1) via acid-mediated removal of the tert-butyl carbamate,[13] this inclusive synthetic effort, leading to (−)-15, represents a formal total synthesis of (−)-coniine (1). The measured optical rotation of 15 ([α]D20 −32.1 (c 0.895, CHCl3)); lit. [α]D23 −31.6 (c 0.86, CHCl3) is in convincing agreement with the reported value, suggesting a high level of enantiopurity initially derived from the enzyme-catalyzed hydrocyanation reaction.[11]

Scheme 3

Piperidine Synthesis by Ring-Closing Metathesis

Our approach to polyhydroxylated iminosugar DAB-1 (2) required additional considerations to account for the introduction of the stereodefined dihydroxyl groups and exocyclic hydroxymethylene functionality. The first challenge could be satisfied with stereoselective dihydroxylation, but the second one required more thought. It was decided that a protected version of the hydroxymethylene group should be incorporated directly into the unsaturated aldehyde starting material with oxygen substitution in the 4-position. Unfortunately, early experiments revealed that oxynitrilase was not able to hydrocyanate aldehydes with tolerable enantioselectivities when adorned with the requisite oxygen substitution.[50] In order to overcome this complication, we envisioned that a ruthenium-catalyzed cross-metathesis between a protected allyl alcohol and an enzyme-derived (R)-(β,γ)-unsaturated-(α)-hydroxyester should furnish our ideal starting material (Scheme ). Consequently, we were delighted to find that the cross-metathesis reaction amongst hydroxyester 9b and allyl benzyl ether supplied the needed alkene 9c in good yield and pleasing trans/cis selectivity. Other partners, such as allyl acetate, allyl alcohol, or cis-2-butene-1,4-diol derivatives, were either less efficient or provided unsatisfactory trans/cis ratios. Although the optimized metathesis reaction leading to (R)-9c generated a trans/cis ratio of 95:5, this inseparable material could not be carried forward to the palladium transposition step. Any cis isomer present would undergo π–σ–π equilibration of the palladium π-allyl intermediate, ultimately eroding enantiopurity with the formation of the undesired enantiomer. Fortunately, the undesired cis isomer was separable by careful chromatographic purification after conversion to the carbonate (6b). Palladium-catalyzed allylic azidation of carbonate 6b was achieved following our previously reported conditions.[8,51] Stereospecificity in this allylic transposition was high, providing azide 16 in good yield and enantiopurity. This sequence provided the key amine stereocenter in short order from the enzymatically derived chiral fragment.

Scheme 4

Catalytic Reactions to Generate the Key Nitrogen Stereocenter

The next stage of the synthesis focused on the introduction of the two remaining alcohol stereocenters (Scheme ). Conversion of azide 16 to carbamate 17 prior to dihydroxylation was necessary to exert a greater influence on the stereoselectivity of the crucial dihydroxylation step. Under the standard Upjohn conditions, however, the diastereoselectivity ratio was a disappointing 1.3:1, marginally favoring the unwanted diol 19. Fortunately, dihydroxylation with AD-mix β delivered the desired diastereomer (18) in good yield and excellent diastereoselectivity (>20:1 dr).

Scheme 5

Synthesis of Polyhydroxylated Carbamate 18

With all heteroatoms and stereocenters correctly positioned, completion of the DAB-1 synthesis rested on three remaining steps: lactam formation, amide reduction, and benzyl deprotection (Scheme ). One-pot cyclization of aminoester 18 to lactam diol 20 was efficaciously accomplished in 87% yield via facile removal of the Boc group followed by a cyanide-catalyzed lactamization. Reduction of the resultant lactam (20) was achieved under classic borane-mediated conditions. Finally, deprotection of the benzyl-protecting group was performed under a hydrogen atmosphere with a 10% loading of palladium on carbon. We were gratified to discover that this two-step reaction sequence produced our target molecule, DAB-1 (2), in a 63% yield, isolated as the hydrochloride salt. The observed optical rotation of synthetic DAB-1 was in good agreement with literature-reported values.[31]

Scheme 6

Syntheses of DAB-1 and Nectrisine

Nectrisine, a related iminosugar with important biological activity, was also accessed from lactam 20 by a modified procedure. Hydrogenolysis of the benzyl ether (20) was accomplished to provide lactam triol 21. This triol is a known intermediate in the synthesis of nectrisine.[34,56] This formal synthesis underscores the general utility of our route to substituted pyrrolidine alkaloids.

In conclusion, we have developed a novel combined enzyme- and metal-catalyzed strategy for the enantioselective syntheses of three notable alkaloid natural products. The inexpensive and readily available oxynitrilase enzyme introduced the first stereocenter in each instance with remarkable enantioselectivity. Alkene cross-metathesis proved a resourceful method to overcome the substrate limitations of the enzyme. Palladium-catalyzed 1,3-allylic substitutive transpositions translated this stereochemistry into the key amine stereocenters using both azide and novel DNs-protected nitrogen nucleophiles. This combined strategy provided functionally dense and stereochemically rich molecular building blocks that were readily elaborated into (−)-coniine, DAB-1, and nectrisine. We anticipate that this strategy could be employed as a template for the synthesis of other piperidine and pyrrolidine alkaloids by taking advantage of the broad scope of the catalytic reactions described as well as a source of functionally dense synthetic intermediates.

Experimental Section

General Remarks

All reactions were carried out in a dry glassware under an argon atmosphere using standard Schlenk techniques unless otherwise specified. Ethanol was purified by distillation from magnesium turnings and iodine. All other solvents were purified by the passage through solvent purification columns. CDCl3, CD3OD, and D2O were used as received. Compounds 7a, 7b, 9a, 9b, and 6a were prepared according to previously reported literature procedures.[8] The Hoveyda–Grubbs second-generation complex was generously provided by Materia, Inc. All other commercial reagents were used as received.

1H and 13C spectra were recorded on a 300 or 400 MHz spectrometer. Residual CHCl3 solvent peaks were referenced to 7.27 and 77.23 ppm for 1H and 13C NMR, respectively. Residual CH3OH solvent peaks were referenced to 3.31 and 49.00 ppm for 1H and 13C NMR, respectively. Infrared spectra were recorded on an FT-IR spectrometer as a thin film on KBr plates. High-resolution mass spectra (HRMS) were provided by the University of California, Irvine Mass Spectrometry Facility. All HRMS were by positive-ion CI or ESI.

N-(But-3-en-1-yl)-2,4-dinitrobenzenesulfonamide

To a stirred suspension of 3-buten-1-amine hydrochloride (0.500 g, 4.65 mmol) in CH2Cl2 (23 mL, 0.2 M) under nitrogen was added distilled triethylamine (1.62 mL, 11.6 mmol) in drop-wise fashion. When dissolution was complete, the solution was cooled in an ice/water bath, and 2,4-dinitrobenzenesulfonyl chloride (1.49 g, 5.60 mmol) was added portion-wise. The ice bath was then removed, and the reaction was permitted to warm to room temperature for 1 h. The mixture was then diluted with EtOAc (50 mL) and washed with saturated aqueous NaHCO3 (3 × 10 mL) and brine (10 mL). The organic layer was dried over MgSO4, rinsed through a glass-fritted plug of silica gel and MgSO4, and concentrated. The residue was triturated in hot diethyl ether, then back-triturated with an equal volume of hexane, and filtered on a fine glass frit to afford 767 mg of N-(but-3-en-1-yl)-2,4-dinitrobenzenesulfonamide as slightly off-white crystals (55%): mp 91.8–92.4 °C; TLC R = 0.36 (hexanes/EtOAc, 3:1); IR (KBr) 3358, 3330, 3112, 3101, 1608, 1555, 1530, 1362, 1349, 1341, 1173, 748 cm–1; 1H NMR (400 MHz, CDCl3): δ 8.66 (d, J = 2.0 Hz, 1H), 8.56 (dd, J = 8.6, 2.2 Hz, 1H), 8.35 (d, J = 8.6 Hz, 1H), 5.64 (ddt, J = 17.0, 10.3, 6.8 Hz, 1H), 5.38 (br s, 1H), 5.08 (br d, J = 10.2 Hz, 1H), 5.06 (dd, J = 17.0, 1.0 Hz, 1H), 3.23 (quasi br q, 2H), 2.29 (br q, 2H); 13C NMR (75 MHz, CDCl3): δ 149.9, 148.3, 139.3, 133.7, 132.7, 127.4, 120.9, 118.8, 43.1, 33.9; HRMS (ESI): calcd for C10H15N4O6S ([M + NH4]+), 319.0707; found, 319.0711.

Ethyl (R,E)-4-(But-3-en-1-yl(tert-butoxycarbonyl)amino)hept-2-enoate (12b)

To a solution of carbonate 6a (100 mg, 1 equiv) in tetrahydrofuran (1.6 mL, 0.25 M) under a nitrogen atmosphere was added 3-buten-1-amine (56 μL, 1.5 equiv) followed by Pd(PPh3)4 (24 mg, 5 mol %). The solution was maintained at ambient temperature until complete consumption of 6a was determined by TLC (∼45 min). The solution was then filtered through a small pad of silica and concentrated. The crude oil was then redissolved in 5:1 THF/H2O (4 mL) and sodium bicarbonate (172 mg, 5 equiv) was added followed by di-tert-butyldicarbonate (268 mg, 3 equiv). The resulting mixture was stirred at ambient temperature for 24 h. The reaction mixture was then diluted with dichloromethane (5 mL) and extracted with ammonium chloride (5 mL). The aqueous phase was extracted with dichloromethane (2 × 5 mL), and the phase was then dried over Na2SO4, filtered, concentrated, and purified by automated silica gel chromatography to provide 100 mg (75%) of 12b as colorless oil. IR (KBr): 2975, 2875, 1723, 1697, 1642 cm–1; 1H NMR (300 MHz, 298K, CDCl3): δ 6.89 (dd, J = 15.9, 5.6 Hz, 1H), 5.83 (dd, J = 15.9, 1.7 Hz, 1H), 5.72 (ddt, J = 15.1, 9.0, 5.1 Hz, 1H), 5.04 (d, J = 15.1, 1H), 4.99 (d, J = 9.0 Hz, 1H), 4.2–4.6 (br s, 1H), 4.19 (q, J = 7.1 Hz, 2H), 3.03 (br s, 2H), 2.11–2.39 (m, 2H), 1.61 (q, J = 7.2 Hz, 2H), 1.45 (s, 9H), 1.41–1.13 (m, 5H), 0.93 (t, J = 7.3 Hz, 3H); 13C NMR (75 MHz, CDCl3): δ 166.5, 155.5, 147.8, 135.5, 121.8, 116.6, 80.1, 60.6, 56.2, 44.3, 36.8, 34.1, 28.6, 19.6, 14.4, 14.0; HRMS (ESI): calcd for C18H31NO4H ([M + H]+), 326.2326; found, 326.2326.

Ethyl (R,E)-4-(N-(But-3-en-1-yl)benzamido)hept-2-enoate (SI-1)

Amine 12a (16 mg, 0.0711 mmol, 1 equiv), taken unpurified from the first part of the procedure above, was dissolved in pyridine (0.07 mL, 1M) and cooled to 0 °C. Benzoyl chloride (21 μL, 0.178 mmol, 2.5 equiv) was added, and the solution was maintained at 0 °C for 2 h. The mixture was then dissolved in ether (5 mL) and extracted with saturated aqueous NH4Cl (5 mL). The organic phase was dried over MgSO4, filtered, and concentrated to provide clear oil that was of sufficient purity for HPLC analysis. 1H NMR (300 MHz, 298K, CDCl3): δ 7.42–7.35 (m, 5H), 6.91 (br s, 1H), 5.86 (br s, 2H), 5.06 (br s, 2H), 4.35 (br s, 1H), 4.23 (q, J = 6.2 Hz, 2H), 3.34 (br s, 2H), 2.64–2.25 (bm, 1H), 1.76–1.42 (bm, 4H), 1.32 (t, J = 6.2 Hz, 3H), 0.86 (br s, 3H); 74% ee (AD Chiralcel HPLC column, 5% IPA in hexane, R = 18.9 min, 20.4 min).

Ethyl (R,E)-4-((N-(But-3-en-1-yl)-2,4-dinitrophenyl)sulfonamido)hept-2-enoate (12c)

A −10 °C solution of dba3Pd2·CHCl3 (3.2 mg; 1.5 mol %) and triphenylphosphine (3.2 mg; 6 mol %) in THF (0.20 mL, 1 M) was added dropwise to a −10 °C solution of allylic carbonate 6a (50 mg; 0.205 mmol), triethylamine (0.082 mL, 0.615 mmol, 3 equiv), and N-(but-3-en-1-yl)-2,4-dinitrobenzenesulfonamide (92 mg; 0.307 mmol, 1.5 equiv) in THF (approx. 0.6 mL, ∼0.3 M). The flask was protected from light with an aluminum foil wrap and held at −10 °C for the duration. The progress of the reaction was monitored by TLC using UV visualization. After 14 h, the reaction was diluted with ethyl acetate (2.5 mL) and rinsed through a short plug of silica gel (60–200 mesh) and MgSO4 with additional ethyl acetate. The solvent was removed under reduced pressure, and the crude product was chromatographed on silica gel (hexanes/EtOAc, 6:1). Concentration in vacuo afforded 60 mg (67%) of adduct 4 as pale yellow oil: TLC R = 0.40 (hexanes/EtOAc, 3:1); [α]D25 +37.6 (c 1.73, CHCl3) ee = 96% (AD-H Chiralcel HPLC column, 7.5% IPA in hexane, R = 8.3 min, 9.4 min); IR (film): 3101, 2962, 2874, 1718, 1555, 1539, 1367, 1352, 1171, 746 cm–1; 1H NMR (400 MHz, CDCl3): δ 8.49 (dd, J = 2.2, 8.6 Hz, 1H), 8.46 (d, J = 2.1 Hz, 1H), 8.24 (d, J = 8.6 Hz, 1H), 6.76 (dd, J = 6.3, 15.8 Hz, 1H), 5.85 (dd, J = 1.4, 15.8 Hz, 1H), 5.64–5.76 (m, 1H), 5.03–5.12 (m, 2H), 4.53 (br q, 1H), 4.16 (q, J = 7.1 Hz, 2H), 3.20–3.40 (m, 2H), 2.39–2.50 (m, 1H), 2.26–2.38 (m, 1H), 1.64–1.76 (m, 2H), 1.25–1.45 (m, 2H), 1.26 (t, J = 7.1 Hz, 3H), 0.93 (t, J = 7.3 Hz, 3H); 13C NMR (100 MHz, CDCl3): δ 165.6, 149.8, 148.1, 144.5, 138.9, 133.9, 132.6, 126.3, 124.0, 119.9, 117.8, 60.9, 59.0, 44.9, 35.6, 34.4, 19.4, 14.2, 13.7; HRMS (ESI): calcd for C19H25N3O8NaS ([M + Na]+), 478.1255; found, 478.1250.

(R)-1-((2,4-Nitrophenyl)sulfonyl)-6-propyl-1,2,3,6-tetrahydropyridine (13)

A solution of Hoveyda–Grubbs second-generation catalyst (42.5 mg; 5 mol %) in CH2Cl2 (5 mL) was added dropwise to an ice-cooled, nitrogen-purged solution of 12c (447 mg, 0.98 mmol) in CH2Cl2 (20 mL) to give a final diene concentration of 0.04 M. The ice bath was removed, and the flask was allowed to warm to ambient temperature for 3 h. The reaction was diluted with CH3OH (1 mL) and EtOAc (15 mL) and then rinsed through a glass-fritted plug of silica gel (60–200 mesh) and MgSO4 with additional EtOAc. The solution was concentrated, and the resulting brown oil was purified by silica gel chromatography (hexanes/EtOAc, 6:1) to afford 297 mg (84%) of 13 as pale yellow oil: TLC R = 0.38 (hexanes/EtOAc, 3:1); [α]D25 −195.1 (c 1.445, CHCl3) ee = 95% (AD-H Chiralcel HPLC column, 7.5% IPA in hexane, R = 7.7 min, 10.4 min); IR (film): 3103, 2960, 2935, 1605, 1553, 1537, 1364, 1352, 1167, 1111, 748 cm–1; 1H NMR (400 MHz, CDCl3): δ 8.46 (dd, J = 2.2, 8.6 Hz, 1H), 8.40 (d, J = 2.2 Hz, 1H), 8.19 (d, J = 8.6 Hz, 1H), 5.65–5.8 (m, 2H), 4.31 (app br s, 1H), 3.80–3.95 (m, 1H), 3.24–3.34 (m, 1H), 1.88–2.08 (m, 2H), 1.52–1.67 (m, 2H), 1.26–1.47 (m, 2H), 0.90 (t, J = 7.3 Hz, 3H); 13C NMR (100 MHz, CDCl3): δ 149.5, 148.0, 139.9, 132.0, 127.8, 126.3, 124.8, 119.7, 54.8, 39.0, 37.2, 23.9, 19.4, 13.9; HRMS (ESI): calcd for C14H18N3O6S ([M + H]+), 356.0911; found, 356.0918.

tert-Butyl (R)-6-Propyl-3,6-dihydropyridine-1(2H)-carboxylate (14)

Allyl amine (0.796 mL; 0.605 g, 10.6 mmol) was added dropwise to an ice-cooled, stirred solution of the DN-protected piperidine 13 (188 mg, 0.531 mmol) in anhydrous CH2Cl2 (5.3 mL, 0.1 M). After 18 min, the initially red solution had turned orange and was carefully concentrated in vacuo (approx. 400 Torr) at room temperature to furnish the volatile dehydroconiine intermediate. To assist in the removal of the excess allyl amine, the crude product was twice dissolved in CH2Cl2 (10 mL) and sequentially evaporated as before. The liquid residue was dissolved in p-dioxane (4.3 mL) and H2O (5.3 mL) and then treated with NaHCO3 (178 mg, 2.12 mmol) and a solution of Boc2O (347 mg, 1.59 mmol) in p-dioxane (1 mL). After 18.5 h of stirring at room temperature, the mixture was diluted with EtOAc (approx. 10 mL) and then quenched by the dropwise addition of H2SO4 (1 M) until the solution reached a pH of 1. The aqueous layer was separated and extracted three times with EtOAc. The combined organic extractions were washed with saturated aqueous NaHCO3 (1 × 10 mL), 0.1 M aqueous NaOH (3 × 10 mL), dried over MgSO4, and rinsed through a short plug of silica gel (60–200 mesh) and MgSO4. The solution was concentrated, and the resulting oil was purified on silica gel using radial chromatography (hexanes/EtOAc, 25:1) to afford 90 mg (75%) of (R)-14 as colorless oil: TLC R = 0.29 (hexanes/EtOAc, 10:1); [α]D20 −209.1 (c 1.865, CHCl3); [α]D20 lit. ref. (S)-14 +138.8 (c 1.73, CHCl3); IR (film): 3032, 2961, 2930, 2872, 1691, 1651, 1418, 1391, 1363, 1250, 1172, 766, 704 cm–1; 1H NMR (400 MHz, CDCl3): δ 5.73–5.85 (m, 1H), 5.60–5.73 (m, 1H), 4.35 (br s, 1H), 4.12 (br s, 1H), 2.72–2.98 (br m, 1H), 2.10–2.28 (br m, 1H), 1.92 (br dt, 1H), 1.30–1.57 (m, 4H), 1.47 (s, 9H), 0.94 (t, J = 7.2 Hz, 3H); 13C NMR (100 MHz, CDCl3): δ 154.8, 128.9, 125.0, 79.2, 51.8, 37.5, 36.4, 28.5, 25.1, 19.4, 14.2. The 1H NMR spectrum is an exact match with that previously published in the literature.[41]

tert-Butyl (R)-2-Propylpiperidine-1-carboxylate (15)

A nitrogen-purged solution of alkene 14 (51 mg; 0.23 mmol) in EtOH (1.5 mL, 0.15 M) was exposed to the hydrogenation catalyst 10% Pd/C (2.5 mg; 5% w/w). A hydrogen-filled balloon was attached to the flask via a rubber septum/needle arrangement, and the mixture was allowed to stir at room temperature for 14 h. The reaction was diluted with ether (approx. 10 mL) and filtered through a short plug of silica gel (60–200 mesh) and MgSO4. The solution was concentrated, and the resulting oil was purified by column chromatography (hexanes/EtOAc, 25:1) to afford 42 mg (82%) of 15 as pale yellow oil: TLC R = 0.38 (hexanes/EtOAc, 10:1); [α]D20 −32.1 (c 0.895, CHCl3); [α]D23 lit. −31.6 (c 0.86, CHCl3); IR (film): 2932, 2866, 1691, 1416, 1366, 1173, 1148, 926, 878, 768 cm–1; 1H NMR (400 MHz, CDCl3): δ 4.18 (br s, 1H), 3.94 (br d, 1H), 2.72 (br t, 1H), 1.16–1.70 (m, 10H), 1.42 (s, 9H), 0.89 (t, J = 7.2 Hz, 3H); 13C NMR (100 MHz, CDCl3): δ 155.3, 79.0, 50.3, 38.3, 32.1, 28.6, 25.9, 19.6, 19.2, 14.2; the 1H and 13C NMR spectra are an exact match with that previously published in the literature.[11]

Ethyl (R,E)-5-(Benzyloxy)-2-hydroxypent-3-enoate (9c)

To a flame-dried round-bottomed flask hydroxyester 9b (1.59 g, 11.03 mmol, 1.0 equiv) was dissolved in dichloromethane (36.8 mL, 0.3 M) at 0 °C. The Hoveyda–Grubbs second-generation catalyst (173 mg, 0.276 mmol, 2.5 mol %) was added, and the solution was degassed by sparging argon for 10 min. Allyl benzyl ether (6.82 mL, 44.10 mmol, 4.0 equiv) was added dropwise, and sparging was continued for an additional hour. The reaction was then maintained at 0 °C for 4 h and, upon completion, quenched with the addition of ethyl vinyl ether (0.1 mL). The crude mixture was diluted with diethyl ether (20 mL), filtered through silica gel, and concentrated under vacuum. Purification by SiO2 flash chromatography (16% EtOAc/hexanes) afforded 1.9 g (69% yield) of 9c as pale brown oil. 1H NMR (300 MHz, 313 K, CDCl3): δ 7.36–7.28 (m, 5H), 6.06 (dtd, J = 15.5, 5.3, 1.5 Hz, 1H), 5.90–5.82 (m, 1H), 4.69 (t, J = 5.4 Hz, 1H), 4.54 (s, 2H), 4.35–4.19 (m, 2H), 4.08 (d, J = 5.3 Hz, 2H), 3.00 (d, J = 5.9 Hz, 1H), 1.32 (t, J = 7.1 Hz, 3H); 13C NMR (75 MHz; CDCl3): δ 173.4, 138.3, 129.6, 128.68, 128.58, 127.91, 127.85, 72.4, 70.9, 69.8, 62.4, 14.3; FTIR: 3421, 2981, 2916, 2848, 1732 cm–1; [α]D25 −24.3 (c 1.02, CHCl3); HRMS: calcd for C14H18O4Na ([M + Na]+), 273.1103; found, 273.1095.

Ethyl (R,E)-5-(Benzyloxy)-2-((ethoxycarbonyl)oxy)pent-3-enoate (6b)

To a flame-dried 100-mL round-bottomed flask was added hydroxy ester 9c (850 mg, 3.40 mmol, 1.0 equiv) and diluted with pyridine (2.2 mL, 1.5 M). The reaction mixture was cooled to 0 °C and maintained for 10 min before slow drop-wise addition of ethyl chloroformate (0.65 mL, 6.79 mmol, 2.0 equiv). The reaction stirred for 1 h before quenching by pouring into a separatory funnel with saturated aqueous NH4Cl (20 mL). The mixture was diluted with EtOAc (10 mL), the two layers were separated, and the organic layer was washed with saturated NH4Cl (3 × 20 mL) and then saturated NaCl (1 × 15 mL). The organic layer was then dried with Na2SO4, filtered, and concentrated under vacuum. The crude residue was purified using SiO2 flash chromatography (16% EtOAc/hexanes). This material was further purified by radial silica gel chromatography (gradient; 100% hexanes to 5% EtOAc to 10% EtOAc in hexanes) to remove the cis isomer and afford 950 mg (87% yield) of 6b as colorless oil: 93% ee (OD-H Chiralcel HPLC column, 5% IPA in hexane, R = 6.8 min, 8.8 min), 1H NMR (300 MHz, 313 K, CDCl3): δ 7.36–7.30 (m, 5H), 6.11 (dtd, J = 15.6, 5.0, 1.1 Hz, 1H), 5.93–5.86 (ddt, J = 15.6, 6.3, 1.2 Hz, 1H), 5.41 (dd, J = 6.3, 0.9 Hz, 1H), 4.53 (s, 2H), 4.25 (q, J = 7.1 Hz, 4H), 4.08 (d, J = 5.1 Hz, 2H), 1.31–1.27 (m, 6H); 13C NMR (101 MHz; CDCl3): δ 168.4, 154.4, 138.1, 132.9, 128.6, 127.9, 123.5, 75.3, 72.6, 69.4, 64.8, 62.0, 14.3, 14.2; FTIR: 2981, 2854, 1747 cm–1; [α]D25 −31.6 (c 1.17, CHCl3); HRMS: calcd for C17H22O6Na ([M + Na]+), 345.1314; found, 345.1320.

Ethyl (S,E)-4-Azido-5-(Benzyloxy)pent-2-enoate (16)

In a flame-dried 20 mL scintillation vial, carbonate 6b (900 mg, 2.79 mmol, 1.0 equiv) was dissolved in anhydrous dichloromethane (9.3 mL, 0.3 M). The resulting solution was then cooled to 0 °C in an ice/water bath, and Pd(PPh3)4 (161 mg, 0.140 mmol, 5 mol %) was added to the vial. Azidotrimethylsilane (0.741 mL, 5.58 mmol, 2.0 equiv) was then added dropwise, and the solution was allowed to warm to ambient temperature for 4 h. The reaction mixture was quenched by dilution with diethyl ether (10 mL), filtered through silica gel, and concentrated under vacuum. The crude residue was purified by SiO2 flash chromatography (9% EtOAc/hexanes to 25% EtOAc/hexanes) to afford 600 mg (78%) of 16 as colorless oil: 93% ee (OD-H Chiralcel HPLC column, 5% IPA in hexane, R = 13.1 min, 15.2 min); 1H NMR (300 MHz, 313 K, CDCl3): δ 7.38–7.31 (m, 5H), 6.80 (dd, J = 15.6, 5.9 Hz, 1H), 6.12 (dd, J = 15.6, 1.6 Hz, 1H), 4.60 (d, J = 1.3 Hz, 2H), 4.30 (dddd, J = 7.4, 5.9, 4.3, 1.6 Hz, 1H), 4.22 (q, J = 7.1 Hz, 2H), 3.65 (dd, J = 9.9, 4.3 Hz, 1H), 3.54 (dd, J = 9.9, 7.4 Hz, 1H), 1.31 (t, J = 7.1 Hz, 3H); 13C NMR (75 MHz; CDCl3): δ 165.7, 141.3, 137.5, 128.7, 128.1, 127.9, 124.5, 73.7, 71.8, 61.8, 60.9, 14.3; FTIR: 2253, 2104, 1717 cm–1; [α]D25 −11.8 (c 1.01, CHCl3); HRMS: calcd for C14H17N3O3Na ([M + Na]+), 298.1168; found, 298.1165.

Ethyl (S,E)-5-(Benzyloxy)-4-((tert-butoxycarbonyl)amino)pent-2-enoate (17)

To a flame-dried 20 mL scintillation vial was added allylic azide 16 (575 mg, 2.09 mmol, 1.0 equiv) and anhydrous EtOH (10 mL, 0.1 M). The solution was cooled to 0 °C, and SnCl2 (792 mg, 4.18 mmol, 2.0 equiv) was added quickly. The reaction mixture was warmed to ambient temperature for 3 h before the solvent was removed under reduced pressure. The residue was then dissolved in 1,4-dioxane (9 mL, 0.9 M) and H2O (1 mL, 0.1 M). NaHCO3 (702 mg, 8.35 mmol, 4 equiv) was then added followed by Boc2O (684 mg, 3.13 mmol, 1.5 equiv). The vial was maintained at room temperature for 16 h. The reaction was diluted with EtOAc, and the solution was transferred to a separatory funnel followed by the addition of 2 M KHSO4 until a pH of 1 was reached. Next, the aqueous phase was extracted with EtOAc (3 × 10 mL). The organic layers were combined and washed with saturated aqueous NaHCO3 (3 × 10 mL) and saturated aqueous NaCl (1 × 10 mL). The organic phase was dried with MgSO4, and the reaction mixture was washed through a plug of layered MgSO4 and SiO2. The solvent was removed under reduced pressure, and the residue was purified using SiO2 flash chromatography (50% EtOAc/hexanes) to afford 600 mg (82%) of 17 as colorless oil: 1H NMR (300 MHz, 313 K, CDCl3): δ 7.38–7.30 (m, 5H), 6.93 (dd, J = 15.7, 5.0 Hz, 1H), 5.99 (dd, J = 15.7, 1.8 Hz, 1H), 5.03 (br s, 1H), 4.52 (d, J = 1.4 Hz, 2H), 4.48 (br s, 1H), 4.20 (q, J = 7.1 Hz, 2H), 3.58 (t, J = 3.5 Hz, 2H), 1.45 (s, 9H), 1.29 (t, J = 7.1 Hz, 3H); 13C NMR (75 MHz; CDCl3): δ 166.3, 155.3, 146.2, 137.7, 128.6, 128.02, 127.83, 122.2, 80.0, 73.5, 71.4, 60.6, 51.5, 28.5, 14.4; FTIR: 3345, 2979, 2950, 1714 cm–1; [α]D25 −3.3 (c 1.02, CHCl3); HRMS: calcd for C19H28NO5 ([M + H]+), 350.1967; found, 350.1975.

Ethyl (2S,3R,4R)-5-(Benzyloxy)-4-((tert-Butoxycarbonyl)amino)-2,3-dihydroxypentanoate (18)

To a flame-dried 20 mL scintillation vial was added unsaturated carbamate 17 (575 mg, 1.65 mmol, 1.0 equiv), methane sulfonamide (157 mg, 1.65 mmol, 1.0 equiv), potassium carbonate (250 mg, 1.81 mmol, 1.1 equiv), (DHQD)2PHAL (51.3 mg, 0.066 mmol, 4 mol %), potassium osmate dihydrate (36.4 mg, 0.099 mmol, 6 mol %), and AD-mix β (2 g). Then, equal parts H2O and BuOH were added (3.3 mL, 0.5 M), and the biphasic reaction mixture was stirred vigorously for 16 h at ambient temperature. The solution was then diluted with EtOAc (5 mL) and quenched by the addition of sodium thiosulfate (2.47 g, 15.63 mmol, 9.5 equiv) followed by stirring for 1 h. The two layers were separated, and the aqueous layer was extracted with EtOAc (6 × 5 mL). The combined organic layers were then dried over Na2SO4, filtered through silica gel, and concentrated under vacuum to afford crude yellow oil. Purification by SiO2 flash chromatography (25% EtOAc/hexanes) afforded 500 mg (79%) of diol 18 as pale yellow oil: 1H NMR (300 MHz, 313 K, CDCl3): δ 7.38–7.31 (m, 5H), 5.35 (d, J = 8.4 Hz, 1H), 4.56 (q, J = 9.6 Hz, 2H), 4.32–4.27 (m, 2H), 4.25–4.23 (m, 1H), 4.06 (td, J = 9.4, 1.3 Hz, 1H), 3.95 (dd, J = 9.3, 2.6 Hz, 1H), 3.76 (tt, J = 8.6, 5.7 Hz, 1H), 3.62 (dd, J = 9.3, 3.8 Hz, 1H), 2.71 (d, J = 10.2 Hz, 1H), 1.45 (s, 9H), 1.33 (t, J = 7.1 Hz, 3H); 13C NMR (75 MHz, CDCl3): δ 172.4, 156.9, 137.8, 128.6, 127.96, 127.84, 80.6, 76.8, 73.6, 72.1, 71.0, 69.0, 61.8, 52.2, 28.4, 14.3; FTIR: 3391, 2978, 1736 cm–1; HRMS: calcd for C19H29NO7Na ([M + Na]+), 406.1842; found, 406.1844; [α]D25 −11.4 (c 1.01, CHCl3).

Ethyl (2R,3S,4R)-5-(Benzyloxy)-4-((tert-butoxycarbonyl)amino)-2,3-dihydroxypentanoate (19)

Potassium osmate dihydrate (11 mg, 0.030 mmol, 0.1 equiv) and N-methylmorpholine oxide (134 mg, 1.14 mmol, 4 equiv) were added to a 20 mL vial equipped with a stir bar. The solids were dissolved in water/tert-butanol (1:1, 3 mL, 0.1 M), and alkene 17 (100 mg, 0.29 mmol 1 equiv) was added. The reaction mixture was stirred at ambient temperature. At completion (TLC, 3 h), the mixture was cooled with an ice/water bath, diluted with ethyl acetate (10 mL), and sodium thiosulfate (430 mg, 2.70 mmol, 9.5 equiv) was added. The mixture was allowed to warm to ambient temperature with vigorous stirring over 1 h. The layers were separated, and the aqueous phase was extracted with ethyl acetate (6 × 5 mL). The combined organic phases were dried over MgSO4, filtered through a silica gel plug, and concentrated. The crude oil was purified by silica gel chromatography (3:1 hexanes/EtOAc) to provide 82 mg (80%) of a 1.3:1 mixture of diastereomers 19 and 18. See the Supporting Information for a 1H NMR spectrum of this mixture, which is too complex to tabulate; 13C NMR (major diast., 75 MHz, CDCl3): δ 173.0, 157.0, 137.7, 128.6, 128.1, 127.9, 80.2, 73.5, 73.2, 71.5, 70.0, 62.2, 52.5, 28.5, 14.3; HRMS: calcd for C19H29NO7Na ([M + Na]+), 406.1842; found, 406.1829.

(3S,4R,5R)-5-((Benzyloxy)methyl)-3,4-dihydroxypyrrolidin-2-one (20)

To a flame-dried 50 mL round-bottomed flask at 0 °C was added dihydroxylated carbamate 18 (650 mg, 1.7 mmol, 1.0 equiv) and a solution of hydrochloric acid in dioxane (4 M, 21.2 mL, 50 equiv), which was maintained for 1 h. Upon completion via TLC, the solvent was removed under reduced pressure. The residue was redissolved in EtOH (20 mL), and K2CO3 (351 mg, 2.54 mmol, 1.5 equiv) was added. The solution was stirred for 10 min at 0 °C before addition of potassium cyanide (11 mg, 0.170 mmol, 10 mol %). After stirring for an additional 10 min, the reaction mixture was warmed to ambient temperature, and stirring was continued for 24 h. The mixture was then diluted with methanol (15 mL), filtered through Celite, and then concentrated under vacuum. The crude residue was purified by SiO2 flash chromatography (10% MeOH/DCM) to afford 350 mg (87%) of benzylated lactam 20 as a colorless foam: 1H NMR (300 MHz, CD3OD): δ 7.34–7.25 (m, 5H), 4.53 (s, 2H), 4.06 (d, J = 7.6 Hz, 1H), 3.89 (t, J = 7.1 Hz, 1H), 3.67 (dd, J = 9.4, 2.5 Hz, 1H), 3.49–3.38 (m, J = 4.6 Hz, 2H); 13C NMR (75 MHz; MeOD): δ 176.5, 139.5, 129.6, 129.03, 128.90, 77.5, 77.1, 74.5, 71.2, 58.8; FTIR: 3266, 1707 cm–1; [α]D25 +9.5 (c 1.04, MeOH); HRMS: calcd for C12H15NO4Na ([M + Na]+), 260.0899; found, 260.0906.

1,4-Dideoxy-1,4-imino-d-arabinitol-HCl (DAB-1) (2)

Lactam 20 (20 mg, 0.084 mmol, 1.0 equiv) was dissolved in a solution of borane in THF (1M, 2.1 mL, 2.1 mmol, 25 equiv) under an atmosphere of argon. The vial was sealed and heated to 50 °C. After 48 h, the reaction mixture was quenched by careful addition of MeOH (5 mL) and then concentrated under reduced pressure. The resulting oil was dissolved in 5 mL of MeOH and concentrated three additional times to ensure complete removal of borane. The resulting mixture was then dissolved in MeOH (1 mL, 0.08 M), and 10% palladium on carbon (22 mg) was added, followed by a solution of HCl in dioxane (4 M, 5 μL, 0.25 equiv). The atmosphere of the reaction vessel was exchanged for argon and then for hydrogen by three vacuum-refill cycles. After 17 h, the atmosphere was again exchanged for argon before exposing the vessel to air. The solution was filtered through a pad of Celite with excess methanol and concentrated under reduced pressure. The resulting oil was dissolved in methanol (5 mL), and HCl in dioxane (4 M, 1 mL) was added, and the resulting solution was concentrated under reduced pressure to provide 9 mg (63%) of 2 as pale yellow oil. 1H NMR (300 MHz, D2O): δ 4.06 (dt, J = 5.5, 3.3 Hz, 1H), 3.79 (t, J = 4.1 Hz, 1H), 3.67 (dd, J = 11.8, 5.0 Hz, 1H), 3.56 (dd, J = 11.9, 7.3 Hz, 1H), 3.24–3.09 (m, 2H), 2.93 (dd, J = 12.3, 3.2 Hz, 1H); 13C NMR (75 MHz; D2O): δ 76.6, 75.2, 67.5, 59.9, 50.9; [α]D25 +31.1 (c 0.28, H2O) lit. +33.3 (c 0.18, H2O). Spectroscopic data were an exact match with that previously published in the literature.[31]

(3S,4R,5R)-3,4-Dihydroxy-5-(hydroxymethyl)pyrrolidin-2-one (21)

Palladium (10%) on carbon (70 mg) and HCl in dioxane (4 M, 5 μL, 0.25 equiv) added to a solution of diol 20 (50 mg, 0.21 mmol, 1 equiv) in methanol (0.5 mL, 0.4 M) in a round bottom flask equipped with a stir bar. The atmosphere of the flask was exchanged successively for argon and then for hydrogen by three vacuum-refill cycles. The reaction mixture was stirred for 17 h before the atmosphere was exchanged again to argon before the flask was opened to air. The mixture was filtered through Celite with excess methanol, and the eluent was concentrated under reduced pressure to provide 30 mg (97%) of triol 21 as colorless oil. 1H NMR (300 MHz, D2O): δ 4.33 (dd, J = 8.0, 1.2 Hz, 1H), 4.03 (t, J = 7.6 Hz, 1H), 3.81 (dd, J = 12.2, 3.1 Hz, 1H), 3.64 (ddd, J = 12.2, 4.9, 1.2 Hz, 1H), 3.53–3.43 (m, 1H); 13C NMR (75 MHz; D2O): δ 178.2, 78.2, 77.2, 62.7, 60.6; FTIR: 3426, 2927, 1740, 1701 cm–1; HRMS: calcd for C5H9NO4Na ([M + Na]+), 170.0429; found, 170.0430; [α]D25 +10.6 (c 0.63, H2O) lit. +15.4 (c 0.12). Spectroscopic data were an exact match with that previously published in the literature.[34]