Asymmetric Entry into 10b-aza-Analogues of Amaryllidaceae Alkaloids Reveals a Pronounced Electronic Effect on Antiviral Activity.

Development of a chiral pool-based synthesis of 10b-aza-analogues of biologically active Amaryllidaceae alkaloids is described, involving a concise reductive amination and condensation sequence, leading to ring-B/C-modified, fully functionalized ring-C derivatives. Differentiated anticancer and antiviral activities of these analogues are presented. Despite complete conformational and functional group overlap, the 10b-aza-analogues have diminished anticancer activity and no antiviral activity. These unprecedented electronic effects suggest a possible role for π-type secondary orbital interactions with the biological target.

Introduction

Amaryllidaceae alkaloids have attracted the attention of medicinal[1−4] and synthetic organic chemists[5−8] for many decades in view of the diverse array of biological activities demonstrated. Examples include (Figure A) galanthamine (1) used clinically in the treatment of Alzheimer’s disease,[9] the anticancer agent pancratistatin (2),[10] and the selective anticancer/antiviral agents trans-dihydronarciclasine (3)[11,12] and trans-dihydrolycoricidine (4).[13] The latter two alkaloids have recently been reported to exhibit potent activity against Zika virus,[12] herpes simplex virus type 1 (HSV-1), and varicella zoster virus.[13]

Figure 1

(A) Structure of representative Amaryllidaceae constituents galanthamine (1), pancratistatin (2), trans-dihydronarciclasine (3), and trans-dihydrolycoricidine (4) and (B) general quinazolidinone core structure (5), the novel hybrid heterocycle 6 and proposed retrosynthesis from an anthranilamide 7, and pentose l-arabinose (8).

Despite the common structural motifs apparent in compounds 1–4, these biological activities manifest through diverse mechanisms. Galanthamine is a potent inhibitor of acetylcholinesterase,[14] whereas pancratistatin selectively targets cancer cell mitochondria.[15] The antiviral profile is unusual with potent activity being reported against both DNA and RNA viruses,[11−13] and accordingly several antiviral targets have been postulated.[16,17] Significant efforts have been made to elucidate the anticancer pharmacophore of pancratistatin. This work has revealed that ring-A substituents and the stereochemistry of ring-C substituents are crucial for activity,[18−22] leading to the identification of analogues that are more potent than the natural products.[23,24] The ring-A 7-aza-analogue proved to be devoid of anticancer activity,[25] while the 10-aza-analogue was potent,[26] demonstrating the importance of the evaluation of aza analogues in revealing the subtle electronic requirements of the pharmacophore. The contribution of structural modifications on the antiviral activity is also critical[13] but much less established.[11−13] Only a single report on the structure–activity relationship of ring-B modified lycoranes (dengue virus) has appeared.[27]

The potent, selective antiviral activity of a series of quinazolinone derivatives (Figure , 5) was recently reported against HSV-1.[28,30] These compounds are accessible via a one-pot multicomponent coupling reaction of an amine, an aldehyde, and an anthranilamide derivative. High selectivity was observed for anti-HSV-1 activity in neuronal host cells for both compound 4 and the quinazolinones.[13,26] Given the synthetic challenges associated with the synthesis of fully functionalized alkaloids such as 3 and 4,[5−8,12,29] we have now explored a new hybrid quinazolinone–lycorane heterocyclic core shown as structure 6 (Figure ). We envisioned that compounds such as 6 could be elaborated in an asymmetric fashion from an anthranilamide such as 7 and a chiral-pool pentose (such as 8) in a few steps. In this communication, we report the success of this route and rapid synthesis of the compound of structural type 6, including 21 which is the 10b-aza-analogue of the potent antiviral agent trans-dihydrolycoricidine (4), from l-arabinose in six steps (carried out in four chemical procedures) with a 23% overall yield.

Results and Discussion

Two initial attempts to fuse pentoses such as d-ribose and l-arabinose onto anthranilamide 7b through N-alkylation of the protected 5-iodo analogues failed due to chemoselectivity incompatibilities, and we eventually settled on the process outlined in Schemes –3. The C5 primary alcohol of l-arabinose, possessing the desired stereochemistry, was selectively protected as the TBS–ether 9 (Scheme ), which, due to its hydrophilicity, was immediately engaged in the reductive amination with anthranilamide 7b in the same reaction vessel affording intermediate 10. After a short solvent partition sequence, the crude triol 10 was acylated to give 11, which was isolated and purified over silica. The overall three-stage telescoped reaction sequence was highly efficient giving access to 11 with 64% overall yield (based on l-arabinose) over three steps, requiring only a single silica-gel chromatographic purification.

Scheme 1

l-Arabinose (8) Selectively Protected at C5 Alcohol as TBS-Ether Was Coupled to Anthranilamide (7b) via a Reductive Amination: (a) TBS-Cl, Pyridine; (b) NaCNBH3, MeOH; and (c) Ac2O, DIPEA, DCM

Scheme 3

Synthesis of 21 and 22 from l-Arabinose (8) and 3,4-Methylenedioxyanthranilamide (7a): (a) TBS-Cl, Pyridine; (b) NaCNBH3, MeOH; (c) Ac2O, DIPEA, DCM; (d) HF·Py, THF; (e) IBX, DMSO, 60 °C; and (f) K2CO3, MeOH

Removal of the TBS-protecting group in 11 employing tetrabutylammonium fluoride in THF gave the primary alcohol 12 contaminated with products of acetyl migration, attributed to these basic conditions. Conversely, the use of HF·pyridine complex in THF resulted in clean deprotection with no acyl transfer, and the desired primary alcohol 12 was isolated in high yield (Scheme ). The selective oxidation of the primary alcohol to the aldehyde proved more challenging than expected and a number of different conditions were investigated. An attempted Swern oxidation unexpectedly resulted in dehydration of the amide in 12 leading to the nitrile derivative, while an attempted oxidation under Parikh–Doering conditions failed completely, and no conversion of the starting material was observed. Reaction of 12 using Dess–Martin periodinane resulted in rapid decomposition of the starting material. Finally, oxidation with IBX in dimethyl sulfoxide (DMSO) at 60 °C was found to proceed selectively, the intermediate aldehyde immediately condensing as expected,[28] to give the desired fused-ring products 13 and 14 as a 1:2 mixture of separable diastereomers. Saponification of each of the separated diastereomers with K2CO3 in MeOH afforded the desired individual triols 15 and 16 completing the synthesis of these new hybrid model compounds. We also determined that, separately, 13 and 14 do not interconvert under the cyclization conditions, indicating them to be the products of a kinetically controlled intramolecular condensation.

Scheme 2

Conversion of 11 via Deprotection, Oxidation, and Ring Closure to the Quinazolinone–Lycorane Hybrids15 and 16: (d) HF·Py, THF; (e) IBX, DMSO, 60 °C; and (f) K2CO3, MeOH; Bottom: ORTEP Diagram of 14

A single crystal X-ray structural analysis of the major diastereomer 14 was conducted, confirming the overall course of the chemistry as anticipated and confirming the stereochemistry of the major diastereomer as shown in Scheme . Interestingly, the 10b-aza heterocycle 14 was seen to adopt the alternative chair conformer as normally observed in the carbocyclic natural series. The C2- and C3-acetoxy and 4a-amino substituents of 14 adopt equatorial positions while the C4-acetoxy group projects axially. Diastereomer 14 crystallized as the thermodynamically more stable one of its chair conformers. A crystal structure of diastereomer 13 or deacetylated product 15 could not be obtained; however, the 1H NMR spectra of 15 is consistent (H4a, δ 4.59, d, J = 9.0 Hz) with the adoption of the natural phenanthridone-type chair conformer, having equatorial C4-hydroxyl and 4a-amino and diaxial C2, C3 alcohols. Thus, the diastereomeric pairs 13/14 or 15/16 adopt radically different three-dimensional projections of the crucial ring-C substituents with diastereomers 13 and 15 (and 19 and 21 vide infra) matching those of biologically active compounds such as 4.

The synthetic route was now repeated with the 3,4-methylenedioxy-substituted anthranilamide 7a (Scheme ), essentially as before. The three-step telescoped intermediate 17 was isolated in a highly efficient 68% yield. Desilylation of 17 and IBX-mediated oxidation of 18 gave two ring-fused diastereomers 19 and 20 in a ratio of 3:4 (by mass). These diastereomers also proved readily separable over the silica gel, and each compound was independently saponified yielding the methylenedioxy-substituted quinazolidinone–lycorane hybrids21 and 22. The chiral pool strategy thus allowed rapid, convergent fusion of l-arabinose onto the substituted anthranilamides giving access to 21, the direct 10b-aza-analogue (Figure , 6) of trans-dihydrolycoricidine, and analogues 13, 14, 15, 16, 19, 20, 21, and 22 (see Scheme ).

Preliminary anticancer investigations have been carried out on compound 4 and select 10b-aza-analogues. As the ring-A methylenedioxy substituent and free ring-C hydroxyl substituents are the known required features for anticancer activity,[11] the antiproliferative activity of compound 4 and the two hydroxyl derivatives 21 and 22 was investigated in four hematopoietic cancer cell lines. Compound 21 retained selective anticancer activity toward Karpas422 (9.72 μM), Toledo (18.9 μM), and K562 (13.23 μM) but was inactive toward MV4-11 cell lines. As expected, the synthetic compound 4 demonstrated potent broad-spectrum activity to Karpas422 (0.076 μM), Toledo (0.061 μM), and K562 (0.076 μM) and to the MV4-11 (0.073 μM) cell lines. The diastereomer 22 proved to be devoid of anticancer activity in all four cell lines investigated.

Preliminary anti-HSV-1 activity of select compounds has also been investigated with sharply defined results. As shown in Figure , neurons infected with HSV-1 were cultured at 2 h post infection and were treated with test compounds. While the synthetic alkaloid 4 proved to be a potent inhibitor of HSV-1 replication as expected,[13] the hybrid 10b-aza derivatives 19, 20, 21, and 22 demonstrated no inhibition of HSV-1.[28] The introduction of the 10b-aza nitrogen in 21 was not expected to change the ring-C conformation of the molecule significantly in comparison to 4; however, the introduced dipole in 21 (vinylogous amide) was expected to affect the electronic properties. Given the sharp differences observed in the anti-HSV-1 activity of the compounds 4 and 21, a detailed conformational analysis was conducted in an attempt to separate the electronic effects from the conformational/functional group attributes.

Figure 2

Anti-HSV-1 activity of synthetic compounds 4 and 19–22 in neurons infected with HSV-1 expressing EGFP linked to the ICP90 promoter at MOI = 0.3; Y-axis depicts GFP+ cell percentage. Vehicle = uninfected and viral-infected untreated cells and ACV = acyclovir-positive control.

In addition to the X-ray analysis of diastereomer 14 (see 14, Scheme ), further evidence in support of the conformations was obtained through investigation of density functional theory (DFT) calculations on compounds 4, 21, and 22, the calculated minima of which are depicted in Figure (see Supporting Information). Compounds 4 and 21 adopt the natural phenanthridone-type chair conformation in ring-C, whereas the aza-diastereomer 22 alone adopts the alternative conformer, identical to that observed in the X-ray structure of compound 14. The antiviral and anticancer results thus correlate with the adoption of the biased phenanthridone-type chair conformation as a critical conformational requirement. The differences in anti-HSV-1 activity seen between compounds 4 (cyclohexane) and 21 (vinylogous amide) are remarkable, given the structural overlap seen. This work demonstrates that substitution of the 10b carbon with nitrogen is not tolerated for antiviral activity and that the natural phenanthridone-type ring-C conformation is required for potent anticancer activity and is crucial for antiviral activity. The structure–activity dichotomy between structures 4 and 21 is quite startling. Molecular electrostatic potential heat maps (see graphical abstract) show that 4 has an electropositive hole over ring-C, and aza-analogue 21 is less electropositive over this surface, highlighting a possible role of π-type secondary orbital interactions with the antiviral target.

Figure 3

DFT-optimized molecular geometry (B3PW91/6-311+G(2d,p)) of compounds 4 (top left) and 21 (top right) illustrating identical ring-C phenanthridone-type conformations in contrast with diastereomer 22 (bottom).

Position 10b in these derivatives is the only position that may potentially affect the electrostatic potential of rings A, B, and C. The data show little effect on rings A and B, but a significant reduction in electron density in ring-C. The data draw attention to the different electronic distributions within the highest occupied molecular orbitals and lowest unoccupied molecular orbitals (see Supporting Information Figures S1–S3) of 4 and 21 as the only differences, suggesting increasing electron density toward position 10b as a method by which selective anticancer agents can be developed. The critical role of electronic properties (not conformational) of ring-C in analogue 4 for potent antiviral activity is unprecedented.

Conclusions

In conclusion, we report a rapid six-step (four processes) convergent route to quinazolinone–Amaryllidaceae alkaloid hybrid molecules. The use of readily available chiral pool pentoses allows absolute control of three stereogenic centers and minimizes the number of stereoselective reactions required to produce the functionally dense core. Although there are several reports in the literature describing the antiviral and anticancer pharmacophore of Amaryllidaceae analogues, there are few reports of B/C-ring fusion-modified derivatives. The selective anticancer and antiviral activities reported here highlight the requirement of a correct ring-C (natural phenanthridone-type) chair conformation and reveal the unprecedented subtle role of electrostatic interactions required for antiviral activity and potent anticancer activity. Further studies exploring electronic modifications of fully functionalized analogues and their selective broad-spectrum antiviral and anticancer activities are ongoing.

Experimental Section

All reagents were obtained from MilliporeSigma and used as received unless otherwise specified. THF was distilled over sodium metal with a benzophenone indicator. Dichloromethane, methanol, and diisopropylethylamine were distilled over calcium hydride for use in reactions performed under anhydrous conditions. Reagent grade methanol, diethyl ether, dichloromethane, ethyl acetate, and hexane were used without further purification and for chromatography. Thin layer chromatography (TLC) was performed using aluminum sheets precoated with silica gel 60F254 (MACHEREY-NAGEL) and visualized using 254 nm UV light. 1H and 13C NMR spectra were recorded on a Bruker AV 600 spectrometer using CDCl3 or d4-MeOH as solvents. Chemical shifts (δ) are reported in ppm and coupling constants (J) are expressed in hertz.

(2S,3R,4S)-1-((tert-Butyldimethylsilyl)oxy)-5-((2-carbamoylphenyl)amino)pentane-2,3,4-triyl Triacetate (11)

A round-bottom flask was charged with l-arabinose (150 mg, 1.00 mmol, 1 equiv) and pyridine (1.20 mL, 15.0 mmol, 15 equiv). To this suspension, t-butyldimethylsilylchloride (163 mg, 1.05 mmol, 1.05 equiv) was added in portions. This reaction mixture was stirred at room temperature for 2 h and then concentrated to a total volume of 0.5 mL to afford protected sugar 9. A solution of anthranilamide (130 mg, 0.95 mmol, 0.95 equiv) in dry methanol (1.0 mL) was added to the flask. After 10 min, NaCNBH3 (120 mg, 1.91 mmol, 1.9 equiv) was added. Upon completion of the reaction as determined by TLC (typically 5–10 min), water was added to quench the reaction mixture. This aqueous solution was extracted with dichloromethane (3 × 15 mL). The combined organic phase was dried over sodium sulphate and concentrated to dryness under reduced pressure to afford the crude product 10, which contained an inseparable sugar impurity. In a round-bottom flask, crude 10 was dissolved in dry dichloromethane (3.3 mL) to which were added diisopropylethylamine (1.7 mL, 9.9 mmol, 10 equiv), acetic anhydride (566 μL, 6.0 mmol, 6 equiv), and N,N-dimethylaminopyridine (6.0 mg, 0.05 mmol, 0.05 equiv). This reaction mixture was stirred for 3 h at room temperature, and then quenched by the addition of water. This aqueous solution was extracted with dichloromethane (3 × 15 mL). The combined organic phase was dried over sodium sulphate and concentrated to dryness under reduced pressure. This material was then purified by gradient elution silica-gel chromatography (80:20 hexane/EtOAc → 30:70 hexane/EtOAc) to afford 11 as a yellow oil in a 64% isolated yield over the three-step sequence. 1H NMR (600 MHz, CDCl3): δ 7.44 (dd, J = 7.9, 1.5 Hz, 1H), 7.36 (ddd, J = 8.6, 7.2, 1.6 Hz, 1H), 6.93 (d, J = 8.4 Hz, 1H), 6.70 (t, J = 7.5 Hz, 1H), 5.43 (dd, J = 8.4, 2.4 Hz, 1H), 5.38 (ddd, J = 8.1, 5.8, 2.4 Hz, 1H), 5.07 (ddd, J = 8.3, 5.0, 3.2 Hz, 1H), 3.73 (dd, J = 11.5, 3.2 Hz, 1H), 3.66 (dd, J = 11.5, 5.0 Hz, 1H), 3.39–3.31 (m, 2H), 2.13 (s, 3H), 2.09 (s, 3H), 2.01 (s, 3H), 0.85 (s, 9H), 0.01 (s, 3H), −0.00 (s, 3H); 13C NMR (151 MHz, CDCl3): δ 171.54, 170.75, 170.16, 170.02, 134.70, 133.81, 128.56, 116.96, 113.51, 108.00, 71.01, 69.66, 68.76, 61.82, 60.55, 44.35, 25.85, 21.10, 20.96, 20.89, 18.33, −5.42. ESI HRMS: calcd for C24H38N2O8Si [M + H]+, 511.2470; found, 511.2470; [α]D20 +14.4° (CHCl3, c 0.016).

(2S,3R,4S)-1-((2-Carbamoylphenyl)amino)-5-hydroxypentane-2,3,4-triyl Triacetate (12)

Compound 11 (107 mg, 0.20 mmol, 1.0 equiv) was dissolved in dry THF (3.2 mL) to which were added HF–pyridine (274 μL). This reaction mixture was stirred for 1 h at room temperature and then quenched by the addition of solid NaHCO3. After 15 min, water and dichloromethane were added to form an aqueous–organic partition. The aqueous was extracted exhaustively with dichloromethane and then the combined organic phases were dried with sodium sulphate and concentrated to dryness. This afforded 12 as a pale yellow oil in 92% yield, which was used immediately in the next transformation.

(2S,3R,4S,4aR)-6-Oxo-2,3,4,4a,5,6-hexahydro-1H-pyrido[1,2-a]quinazoline-2,3,4-triyl Triacetate (13)

2-Iodoxybenzoic acid (12 mg, 0.043 mmol, 1.2 equiv) was dissolved in reagent grade DMSO (100 μL) and heated to 60 °C for 30 min. A solution of compound 12 (15 mg, 0.036 mmol, 1.0 equiv) in DMSO (100 μL) was then added. This reaction mixture was stirred for 15 min at 60 °C and then purified directly by silica gel chromatography (100:0 Et2O/EtOAc → 90:10 Et2O/EtOAc) to afford 13 and 14 as a 1:2 mixture of diastereomers in 36% yield. 13 was obtained as a clear, colorless oil; 1H NMR (600 MHz, CDCl3): δ 7.95 (dd, J = 7.7, 1.7 Hz, 1H), 7.40 (ddd, J = 8.7, 7.3, 1.7 Hz, 1H), 6.92 (td, J = 7.5, 0.9 Hz, 1H), 6.83 (s, 1H), 6.75 (d, J = 8.4 Hz, 1H), 5.49–5.45 (m, 1H), 5.29 (dd, J = 9.5, 3.0 Hz, 1H), 5.03 (dt, J = 4.1, 2.1 Hz, 1H), 4.95 (dd, J = 9.6, 1.9 Hz, 1H), 4.06 (dd, J = 14.5, 1.9 Hz, 1H), 3.26 (dd, J = 14.5, 2.1 Hz, 1H), 2.99 (s, 1H), 2.14 (s, 3H), 2.10 (s, 3H), 2.06 (s, 3H); 13C NMR (151 MHz, CDCl3): δ 170.10, 169.92, 169.21, 163.68, 147.93, 134.33, 129.18, 119.78, 116.41, 112.17, 69.67, 68.01, 67.09, 65.93, 44.03, 21.19, 20.94, 20.88. ESI HRMS: calcd for C18H20N2O7 [M + H]+, 377.1343; found, 377.1342; [α]D20 +69.9° (CHCl3, c 0.0029).

(2S,3R,4S,4aS)-6-Oxo-2,3,4,4a,5,6-hexahydro-1H-pyrido[1,2-a]quinazoline-2,3,4-triyl Triacetate (14)

Colorless solid. 1H NMR (600 MHz, CDCl3): δ 7.93 (dd, J = 7.8, 1.7 Hz, 1H), 7.45 (ddd, J = 8.3, 7.4, 1.7 Hz, 1H), 6.95–6.87 (m, 2H), 6.31 (s, 1H), 5.46 (dd, J = 3.2, 1.7 Hz, 1H), 5.22 (td, J = 10.3, 4.9 Hz, 1H), 5.11 (dd, J = 10.2, 3.2 Hz, 1H), 4.92 (dd, J = 2.5, 1.7 Hz, 1H), 4.26 (dd, J = 13.4, 5.0 Hz, 1H), 2.76 (dd, J = 13.4, 10.5 Hz, 1H), 2.13 (s, 3H), 2.02 (s, 3H), 1.89 (s, 3H); 13C NMR (151 MHz, CDCl3): δ 170.47, 170.45, 170.10, 163.48, 147.10, 134.82, 128.70, 119.85, 116.42, 111.79, 70.90, 68.79, 65.64, 53.57, 46.01, 21.05, 20.79, 20.55. ESI HRMS: calcd for C18H20N2O7 [M + H]+, 377.1343; found, 377.1344; [α]D20 +77.2° (CHCl3, c 0.0067). A small sample suitable for X-ray diffraction was crystallized from hexanes/EtOAc (50:50) by slow evaporation in the refrigerator.

(2S,3R,4S,4aR)-2,3,4-Trihydroxy-1,2,3,4,4a,5-hexahydro-6H-pyrido[1,2-a]quinazolin-6-one (15)

Compound 13 (3.5 mg, 0.01 mmol, 1.0 equiv) and K2CO3 (0.6 mg, 0.005 mmol, 0.5 equiv) were dissolved in a 3:1 mixture of methanol/water (100 μL). After 30 min, this mixture was purified directly by silica gel chromatography (100:0 DCM/MeOH → 80:20 DCM/MeOH) to afford 15 as a white, amorphous solid in 92% yield. 1H NMR (600 MHz, methanol-d4): δ 7.81 (dd, J = 7.7, 1.7 Hz, 1H), 7.43 (ddd, J = 8.8, 7.2, 1.7 Hz, 1H), 6.98 (d, J = 8.4 Hz, 1H), 6.89–6.85 (m, 1H), 4.59 (d, J = 9.0 Hz, 1H), 4.02–3.98 (m, 2H), 3.95 (t, J = 3.5 Hz, 1H), 3.70 (dd, J = 13.0, 2.4 Hz, 1H), 3.13 (dd, J = 13.0, 2.3 Hz, 1H); 13C NMR (151 MHz, methanol-d4): δ 166.49, 151.39, 135.48, 129.12, 119.89, 117.76, 113.87, 71.61, 70.29, 69.77, 68.93, 46.82; [α]D20 +116.9° (MeOH, c 0.00092).

(2S,3R,4S,4aS)-2,3,4-Trihydroxy-1,2,3,4,4a,5-hexahydro-6H-pyrido[1,2-a]quinazolin-6-one (16)

Compound 14 (8.0 mg, 0.02 mmol, 1.0 equiv) and K2CO3 (1.5 mg, 0.011 mmol, 0.5 equiv) were dissolved in a 3:1 mixture of methanol/water (200 μL). After 30 min, this mixture was purified directly by silica gel chromatography (100:0 DCM/MeOH → 80:20 DCM/MeOH) to afford 16 as a white, amorphous solid in 95% yield. 1H NMR (600 MHz, methanol-d4): δ 7.76 (dd, J = 7.7, 1.7 Hz, 1H), 7.39 (ddd, J = 8.4, 7.3, 1.7 Hz, 1H), 6.90 (d, J = 8.3 Hz, 1H), 6.80 (td, J = 7.5, 0.9 Hz, 1H), 4.65 (d, J = 1.7 Hz, 1H), 3.97 (dd, J = 12.7, 5.2 Hz, 1H), 3.93–3.88 (m, 2H), 3.42 (dd, J = 9.4, 3.1 Hz, 1H), 2.53 (dd, J = 12.7, 10.6 Hz, 1H); 13C NMR (151 MHz, methanol-d4): δ 166.57, 150.30, 135.46, 128.99, 119.25, 117.31, 112.64, 75.47, 73.27, 71.57, 66.80, 49.85; [α]D20 −129.0° (MeOH, c 0.0038).

(2S,3R,4S)-1-((tert-Butyldimethylsilyl)oxy)-5-((6-carbamoylbenzo[d][1,3]dioxol-5-yl)amino)pentane-2,3,4-triyl Triacetate (17)

A round-bottom flask was charged with l-arabinose (50 mg, 0.35 mmol, 1 equiv) and pyridine (400 μL, 15 mmol, 15 equiv). To this suspension, t-butyldimethylsilylchloride (54 mg, 0.36 mmol, 1.05 equiv) was added in portions. This reaction mixture was stirred at room temperature for 2 h, and then concentrated to a total volume of 0.5 mL. A solution of 3,4-methylenedioxyanthranilamide (60 mg, 0.33 mmol, 0.95 equiv) in dry methanol (320 μL) was added to the flask. After 10 min, NaCNBH3 (42 mg, 0.66 mmol, 1.9 equiv) was added. Upon completion of the reaction as determined by TLC (typically 2–5 min), water was added to quench the reaction mixture. This aqueous solution was extracted with dichloromethane (3 × 5 mL). The combined organic phase was dried over sodium sulphate and concentrated to dryness under reduced pressure. This mixture was dissolved in dry dichloromethane (1.0 mL) to which were added diisopropylethylamine (575 μL, 3.3 mmol, 10 equiv), acetic anhydride (187 μL, 1.98 mmol, 6 equiv), and N,N-dimethylaminopyridine (2.0 mg, 0.016 mmol, 0.05 equiv). This reaction mixture was stirred for 3 h at room temperature and then quenched by the addition of water. This aqueous solution was extracted with dichloromethane (3 × 15 mL). The combined organic phase was dried over sodium sulphate and concentrated to dryness under reduced pressure. This material was then purified by chromatography (80:20 hexane/EtOAc → 30:70 hexane/EtOAc) to afford 17 as a yellow oil in a 68% isolated yield over three steps. 1H NMR (600 MHz, CDCl3): δ 6.85 (s, 1H), 6.42 (s, 1H), 5.90 (s, 2H), 5.58 (s, 2H), 5.42 (dd, J = 8.4, 2.4 Hz, 1H), 5.33 (ddd, J = 7.7, 5.5, 2.4 Hz, 1H), 5.07 (ddd, J = 8.4, 5.0, 3.3 Hz, 1H), 3.73 (dd, J = 11.5, 3.2 Hz, 1H), 3.66 (dd, J = 11.4, 5.0 Hz, 1H), 3.33 (dd, J = 15.4, 5.9 Hz, 1H), 3.25 (dd, J = 14.5, 7.2 Hz, 1H), 2.13 (s, 3H), 2.08 (s, 3H), 2.02 (s, 3H), 0.86 (s, 9H), 0.02 (s, 3H), 0.01 (s, 3H); 13C NMR (151 MHz, CDCl3): δ 171.44, 170.59, 170.15, 169.99, 152.60, 147.91, 138.51, 107.12, 105.24, 101.43, 93.95, 71.08, 69.70, 68.89, 61.80, 44.26, 25.86, 21.13, 20.92, 20.89, 18.33, −5.41. ESI HRMS: calcd for C25H38N2O10Si [M + H]+, 555.2368; found, 555.2366; [α]D20 −8.7° (CHCl3, c 0.013).

(2S,3R,4S)-1-((6-Carbamoylbenzo[d][1,3]dioxol-5-yl)amino)-5-hydroxypentane-2,3,4-triyl Triacetate (18)

Compound 17 (95 mg, 0.17 mmol, 1.0 equiv) was dissolved in dry THF (2.7 mL) to which were added HF–pyridine (233 μL). This reaction mixture was stirred for 20–30 min at room temperature and closely monitored by TLC to avoid the rearrangement of acetate-protecting groups. When the reaction was complete (typically 15–25 min), it was quenched by the addition of solid NaHCO3. After 15 min, water and dichloromethane were added to form an aqueous–organic partition. The aqueous was extracted exhaustively with dichloromethane, and then the combined organic phases were dried with sodium sulphate and concentrated to dryness. This afforded 18 as a pale yellow oil in 95% yield, which was used immediately in the next transformation.

(2S,3R,4S,4aR)-6-Oxo-2,3,4,4a,5,6-hexahydro-1H-[1,3]dioxolo[4,5-g]pyrido[1,2-a]quinazoline-2,3,4-triyl Triacetate (19)

2-Iodoxybenzoic acid (54 mg, 0.20 mmol, 1.2 equiv) was dissolved in reagent grade DMSO (500 μL) and heated to 60 °C for 30 min. A solution of compound 18 (72 mg, 0.16 mmol, 1.0 equiv) in DMSO (200 μL) was then added. This reaction mixture was stirred for 15 min at 60 °C and then purified directly by silica gel chromatography (100:0 Et2O/EtOAc → 90:10 Et2O/EtOAc) to afford 19 and 20 as a 3:4 mixture of diastereomers in 45% yield. 19 was obtained as a clear, colorless oil; 1H NMR (600 MHz, CDCl3): δ 7.39 (s, 1H), 6.36 (s, 1H), 6.34 (s, 1H), 5.96 (d, J = 1.2 Hz, 1H), 5.45 (t, J = 3.6 Hz, 1H), 5.29 (dd, J = 9.2, 3.1 Hz, 1H), 5.06–5.04 (m, 1H), 4.84 (dd, J = 9.2, 1.5 Hz, 1H), 3.84 (dd, J = 14.2, 2.5 Hz, 1H), 3.23 (dd, J = 14.2, 2.4 Hz, 1H), 2.13 (s, 3H), 2.09 (s, 6H); 13C NMR (151 MHz, CDCl3): δ 170.07, 169.24, 163.56, 153.24, 145.57, 141.70, 131.05, 107.88, 101.87, 94.62, 69.38, 67.77, 66.94, 66.27, 44.97, 21.14, 20.92, 20.86. ESI HRMS: calcd for C19H20N2O9 [M + H]+, 420.1169; found, 420.1238; [α]D20 −13.4° (CHCl3, c 0.0031).

(2S,3R,4S,4aS)-6-Oxo-2,3,4,4a,5,6-hexahydro-1H-[1,3]dioxolo[4,5-g]pyrido[1,2-a]quinazoline-2,3,4-triyl Triacetate (20)

Clear, colorless oil. 1H NMR (600 MHz, CDCl3): δ 7.38 (s, 1H), 6.48 (s, 1H), 6.20 (s, 1H), 5.97 (d, J = 1.9 Hz, 2H), 5.46 (t, J = 2.7 Hz, 1H), 5.21 (td, J = 9.6, 4.7 Hz, 1H), 5.10 (dd, J = 9.6, 3.3 Hz, 1H), 4.83 (d, J = 2.2 Hz, 1H), 4.01 (dd, J = 13.0, 4.8 Hz, 1H), 2.74 (dd, J = 13.1, 9.6 Hz, 1H), 2.13 (s, 3H), 2.04 (s, 3H), 2.01 (s, 3H); 13C NMR (151 MHz, CDCl3): δ 170.48, 170.33, 170.05, 163.47, 153.25, 144.82, 141.96, 110.42, 107.65, 101.86, 95.15, 70.04, 69.74, 68.62, 66.06, 46.71, 21.04, 20.82, 20.74. ESI HRMS: calcd for C19H20N2O9 [M + H]+, 420.1169; found, 420.1247; [α]D20 +10.0° (CHCl3, c 0.094).

(2S,3R,4S,4aR)-2,3,4-Trihydroxy-1,2,3,4,4a,5-hexahydro-6H-[1,3]dioxolo[4,5-g]pyrido[1,2-a]quinazolin-6-one (21)

Compound 19 (4.7 mg, 0.011 mmol, 1.0 equiv) and K2CO3 (0.8 mg, 0.006 mmol, 0.5 equiv) were dissolved in 3:1 mixture of MeOH/water (130 μL). After 30 min, this mixture was purified directly by silica gel chromatography (100:0 DCM/MeOH → 80:20 DCM/MeOH) to afford 21 as a white, amorphous solid in 81% yield. 1H NMR (600 MHz, methanol-d4): δ 7.24 (s, 1H), 6.63 (s, 1H), 5.97 (s, 2H), 4.53 (d, J = 8.8 Hz, 1H), 4.03–3.99 (m, 2H), 3.94 (t, J = 3.6 Hz, 1H), 3.58 (dd, J = 12.9, 2.6 Hz, 1H), 3.13 (dd, J = 12.9, 2.3 Hz, 1H); 13C NMR (151 MHz, methanol-d4): δ 172.42, 161.63, 156.86, 141.98, 115.76, 110.87, 104.49, 79.50, 77.63, 77.21, 55.91, 49.53. ESI HRMS: calcd for C13H14N2O6 [M + H]+, 295.0925; found, 295.0917; [α]D20 +34.2° (MeOH, c 0.0023).

(2S,3R,4S,4aS)-2,3,4-Trihydroxy-1,2,3,4,4a,5-hexahydro-6H-[1,3]dioxolo[4,5-g]pyrido[1,2-a]quinazolin-6-one (22)

Compound 20 (12.0 mg, 0.029 mmol, 1.0 equiv) and K2CO3 (1.9 mg, 0.014 mmol, 0.5 equiv) were dissolved in a 3:1 mixture of methanol/water (250 μL). After 30 min, this mixture was purified directly by silica gel chromatography (100:0 DCM/MeOH → 80:20 DCM/MeOH) to afford 22 as a white, amorphous solid in 88% yield; 1H NMR (600 MHz, methanol-d4): δ 7.20 (s, 1H), 6.55 (s, 1H), 5.94 (d, J = 1.2 Hz, 2H), 4.55 (d, J = 1.9 Hz, 1H), 3.95–3.91 (m, 2H), 3.82 (dd, J = 12.4, 5.1 Hz, 1H), 3.43–3.41 (m, 1H), 2.49 (dd, J = 12.4, 10.4 Hz, 1H); 13C NMR (151 MHz, methanol-d4): δ 154.65, 148.18, 142.19, 129.60, 128.88, 107.69, 102.96, 95.32, 75.14, 72.39, 71.74, 67.08, 50.65. ESI HRMS: calcd for C13H14N2O6 [M + H]+, 295.0925; found, 295.0917; [α]D20 +60.2° (MeOH, c 0.0045).