Nat. Prod. Rep., 2002, 19(6), 742 - 760   DOI: 10.1039/b104971m	Review Article
Quinoline, quinazoline and acridone alkaloids

Joseph P. Michael
Molecular Sciences Institute, School of Chemistry, University of the Witwatersrand, Wits 2050, South Africa. E-mail: jmichael@aurum.chem.wits.ac.za

Covering: July 2000 to June 2001. Previous review: Nat. Prod. Rep., 2001, 18, 543.

This review covers the isolation, structure determination, synthesis and biological activity of quinoline, quinazoline and acridone alkaloids from plant, microbial and animal sources. The literature from July 2000 to June 2001 is reviewed, and 119 references are cited.

An impressively diverse range of novel quinoline alkaloids was reported during the period under review, and convincing spectroscopic evidence was presented for most of the new structures. These alkaloids and their sources are listed in Table 1, which also includes previously unreported sources of known alkaloids.^1–24

Species	Alkaloida	Ref.
Boronia pinnata	Dictamnine 77	1
	Evolitrine 78
	Folimine
	(–)-Pinolinoneb25
	Preskimmianine
Castanea mollissima	(–)-Chestnutamideb84	2
Dictamnus dasycarpus	Dasycarineb26	3
Galipea officinalis	N-Methyl-4-hydroxy-3-(2,3-epoxy-3-methylbutyl)quinolin-2-oneb27	4
Haplophyllum tuberculatum	(–)-Haplotubineb71	5
	rac-Haplotubinoneb33
Huperzia serrata	(+)-Huperzine Jb85	6
	(–)-Huperzine Kb86
	(+)-Huperzine Lb87
Lycopodium chinense	(–)-Senepodine Ab88	7
Oceanapia sp. (sponge)	Uranidine 125	8
	Uranidine derivativeb127
Orixa japonica	Eduline 2	9
	Isobalfourodine	10
	(–)-Isopteleflorineb31	11
	3-O-Methylorixineb29
	Orijanone (orixiarine) 30
Penicillium cf. simplicissimum	rel-(3R,4R)-(+)-3-Methoxy-4,5-dihydroxy-4-(4-methoxyphenyl)-3,4-dihydroquinolin-2-one 102	12
	rel-(3R,4R)-(–)-3-Methoxy-4-hydroxy-4-(4-methoxyphenyl)-3,4-dihydroquinolin-2-one 103
	(+)-Penigequinolone A 104
	(+)-Penigequinolone B 105
	(–)-Peniprequinoloneb106
Penicillium sp. 386	(–)-Penicillazineb107	13
Penicillium sp. EPF-6	(+)-Quinolactacin Ab108	14,15
	(–)-Quinolactacin Bb109
	(+)-Quinolactacin Cb110
Ruta chalepensis	Maculosidine	16
	4-Methoxy-1-methylquinolin-2-one
	2-[6-(3,4-Methylenedioxyphenyl)hexyl]-4-methoxyquinolineb4
	2-[6-(3,4-Methylenedioxyphenyl)hexyl]quinolin-4-oneb5
	Pteleine 79
Sarcomelicope megistophylla	Furomegistine Ib74	17
	Furomegistine IIb75
	(–)-Megistolactoneb8	18
	Megistonine Ib10	19
	Megistonine IIb11
	(+)-Sarcomejineb7	20
Stauranthus perforatus	Skimmianine 82	21
	Veprisine
Strychnos cathayensis (Loganiaceae)	4-Methoxyquinolin-2-one	22
Zanthoxylum davyi	Skimmianine	23
Zanthoxylum schinifolium	-Fagarine 80	24
	(+)-Platydesmine ent-58
	Robustine 81
aOnly new alkaloids and new records for a given species are listed in the Table. Structures of known alkaloids, if not specifically numbered, may be found in previous reviews in this series.  bNew alkaloids.

Funayama et al. have largely been responsible for recent studies on the alkaloidal constituents of Orixa japonica, an oriental plant which has medicinal applications. Their findings have been summarised in a short review describing the 24 known and new alkaloids (mainly hemiterpenoid quinoline derivatives) that they have isolated from this source, as well as some of their studies on the biosynthesis and biological activity of selected alkaloids.¹⁰ The review also recounts in some detail their investigations into the absolute stereochemistry of (–)-preorixine 1, an important biosynthetic intermediate for several of the more complex Orixa alkaloids.

The ability of Orixa japonica extracts to act as relaxants of rat jejunum smooth muscle has been traced to two well-known alkaloids, eduline 2 and japonine 3, the effects of which were comparable to that of the typical muscle relaxant papaverine.⁹ The identity of eduline, isolated for the first time from this plant source, was confirmed by synthesis from 6-methoxy-1-methylisatoic anhydride and acetophenone.

The novel alkaloids 4 and 5 isolated from root extracts of Saudi Arabian Ruta chalepensis are unique in possessing a 1,6-hexylene spacer between the ring moieties rather than the more common ethylene or butylene chains.¹⁶ Long-range relationships in the NMR spectra proved particularly useful for the unambiguous siting of substituents. The alkaloids were correlated chemically by the conversion of 5 into 4 upon treatment with iodomethane and sodium carbonate in acetone.

The bark of the New Caledonian tree Sarcomelicope megistophylla continues to yield an intriguing range of unprecedented quinoline alkaloids, all of which are conceivably formed from highly oxygenated acridone alkaloids such as melicopicine 6 by oxidative cleavage of ring A. The structural similarity is clearly apparent in the new alkaloid (+)-sarcomejine 7 (absolute configuration unknown).²⁰ An unusual long-range ¹H–¹⁵N heteronuclear shift correlation study at natural abundance showed a three-bond relationship between nitrogen and the side-chain methine proton, thereby excluding the alternative structure with the C-2 and C-3 substituents interchanged. Further degradation of the putative acridone ring A is apparent in the structure of the weakly cytotoxic alkaloid (–)-megistolactone, the absolute configuration of which could also not be determined.¹⁸ Structure 8 rather than the alternative structure 9 was assigned on the basis of a strong NOE interaction between the N-methyl group and the methine proton. The biosynthetic origins of megistonine I 10 and megistonine II 11, also unambiguously characterised by NMR spectroscopy, are less obviously ascribable to acridone oxidation.¹⁹ These two compounds possess a methoxy group at C-3, which is unusual in rutaceous quinoline alkaloids (cf. japonine 3), although less rare than the authors believe.

New approaches to the synthesis of toddaquinoline 12 by Harrowven et al.^25,26 have overcome the problems of regiochemical control apparent in their earlier synthesis, which involved intramolecular addition of an aryl radical to a pyridine²⁷ (cf. Ref. 28a). Whereas the azastilbene 13 (X = Br) yielded a mixture of the desired toddaquinoline methyl ether 14 (28%) and its regioisomer 15 (30%) on treatment with tributyltin hydride and AIBN in toluene at 80 °C, the use of a different radical initiator, sodium cobalt(I)salophen in THF at room temperature, yielded 14 (61%) and less than 5% of 15. Cobalt appears to play a dual role in this reaction, first initiating homolysis of the carbon–bromine bond, and then acting as a Lewis acid to enhance the electrophilicity at C-6 in the pyridine ring towards attack by the nucleophilic radical. By contrast, photochemical cyclisation of the azastilbene 16 (X = H, 1 1 mixture of geometrical isomers) was less selective, giving a mixture of 14 (20%) and 15 (54%), while cyclisation initiated by lithium–halogen exchange of 13 (X = Br) with n-butyllithium at –78 °C gave 14 (11%) and yet another regioisomer, the unstable 17 (32%). Ironically, it proved impossible to demethylate the methyl ether 14, necessitating a change of protecting group. The improved synthesis of toddaquinoline proceeded through the benzyl ether 18, prepared as shown in Scheme 1. Radical-mediated cyclisation with sodium cobalt(I) salophen gave toddaquinoline benzyl ether 19 in 61% yield, after which conventional hydrogenolytic debenzylation completed the synthesis of the target alkaloid 12.

	Scheme 1 Reagents and conditions: i, NaOBn, DMF, 65 °C; ii, n-BuLi, THF, –100 °C, 1 h, then DMF, –60 °C; iii, THF, 0 °C; iv, Na–Hg, Co(II)salophen, THF, rt, then add 18, THF, –78 °C to rt; v, H2(1 atm), 5% Pd–C, HOAc, rt.

Several short syntheses of simple 2-substituted quinolines merit brief mention. A new route to 2-alkylquinolines, including the natural product 2-propylquinoline 20, by addition of Grignard reagents to N-[(isobutoxycarbonyl)oxy]quinolinium chloride 21 holds promise for preparing other 2-substituted quinoline alkaloids.²⁹ Reduction of 1-aryl-3-(2-nitroaryl)prop-2-en-1-ones 22 with low-valent titanium, prepared from samarium powder and titanium tetrachloride in THF, also provided an efficient route to 2-arylquinolines, amongst them the alkaloid 23.³⁰ Treatment of polymer-bound flavylium salts 24 with aqueous ammonia represents a versatile new route for the synthesis of analogues of the 2-phenylquinolin-4-one alkaloids.³¹ Finally, the alkylation of 2-phenylquinolin-4-ones with haloalkanes in the presence of sodium hydride and THF yielded a range of 4-alkoxy-2-phenylquinoline alkaloid analogues, some of which showed potent antiplatelet activity.³²

The 3,4-trans-diol structure of (–)-pinolinone 25, extracted from roots of the Australian shrub Boronia pinnata,¹ is unique amongst rutaceous quinoline alkaloids. This novel alkaloid is effectively a 3-prenylquinolin-2-one that has been oxidised at C-3 and reduced at C-4. The location of substituents was fixed by HMBC correlations and NOE studies, and in particular the spatial relationship of the hydroxy groups was deduced from an NOE interaction between H-4 and 3-OH. The absolute configuration was not established.

A more typical 3-prenylquinolin-2-one, dasycarine 26, was isolated from the roots of Chinese Dictamnus dasycarpus.³ Once again, long-range NMR spectroscopic methods were used to establish the location of the substituents.

Naturally occurring 3-prenylquinolin-2-ones are commonly modified by oxidation of the prenyl side chain. A case in point is the epoxyprenylquinolin-2-one 27 (incorrectly named by the authors as an epoxyisobutyl-2-quinolone), a novel metabolite of the trunk bark (angostura) of Venezuelan Galipea officinalis, which is well known for its medicinal properties.⁴ Several alkaloids isolated from leaves and stems of Orixa japonica and thoroughly characterised by spectroscopic techniques also show side-chain oxidation.¹¹ They include the known alkaloid N-demethyllunidonine 28, the novel compound (+)-3-O-methylorixine 29 and the apparently new compound orijanone 30. The latter alkaloid had, in fact, been reported in 1998 as a metabolite of Skimmia laureola³³ (cf. Ref. 28b) and assigned the name orixiarine, which should thus take precedence.³⁴ The tricyclic derivative (–)-isopteleflorine 31 (absolute configuration unknown) is, however, a new natural product from O. japonica, although it had previously been prepared by hydrolysis of another O. japonica alkaloid, 3-O-acetylisopteleflorine 32.³⁵

The growing family of 3-monoterpenoid quinoline alkaloids has been augmented by a most unusual member, haplotubinone.⁵ When a comprehensive suite of NMR spectroscopic experiments on this optically inactive metabolite from Saudi Arabian Haplophyllum tuberculatum failed to yield an unambiguous structure, single crystal X-ray analysis was used to establish the complete structure and relative stereochemistry shown in 33. Noteworthy features in the solid-state structure included hydrogen-bonding between the hydroxy group and the epoxide oxygen, the dimeric association of pairs of enantiomers by hydrogen bonds between the lactam N–H and C-2 carbonyl groups, and the folding back of the side chain to place the (Z)-methyl group of the prenyl unit over the benzene ring. The latter feature probably accounts for the unusual upfield shift ( 0.85) for this methyl group in the ¹H NMR spectrum.

A formal [3 + 3] cycloaddition between 4-hydroxy-1-methylquinolin-2-one 34 and ,-unsaturated iminium salts 35 in the sense shown in Scheme 2 (inset) underlies a conceptually simple route to ring C-substituted pyrano[3,2-c]quinolines 36 by McLaughlin and Hsung.³⁶ For example, the iminium species formed in situ from the geraniol-derived aldehyde 37, piperidine and acetic anhydride, reacted with 34 in toluene at 85 °C to give (±)-huajiaosimuline 38 directly in 79% yield. The lengthier synthesis of (±)-simulenoline 39 from aldehyde 40 proceeded through the adduct 41, which was then deprotected and oxidised to give the tricyclic product 42. Wadsworth–Emmons homologation with diethyl 2-oxopropylphosphonate and subsequent addition of methyllithium completed the synthesis of 39. Finally, prenal 43 was the precursor in a one-pot synthesis of N-methylflindersine 44 (63% yield). In this case, further transformation of the product 44 by dihydroxylation under appropriate conditions yielded the racemic cis- and trans-diols 45 and 46, respectively. Although neither of these is a natural product, their spectroscopic data provided useful analogies for confirming the tentatively assigned trans-configuration of a related natural product, zanthodioline 47.

	Scheme 2 Reagents and conditions: i, piperidine, toluene, 0 °C, 5 min, then Ac2O, 85 °C, 1 h, then 34, 85 °C, 48 h; ii, Cp2ZrCl2, AlMe3, n-BuLi, (CH2O)n; iii, Dess–Martin periodinane, CH2Cl2, rt; iv, HF–pyridine, THF, rt; v, (EtO)2POCH2COMe, NaH, THF, rt; vi, MeLi, THF–Et2O (1 1), –78 °C to rt; vii, magnesium monoperoxyphthalate, PriOH–H2O (2 1), rt; viii, OsO4(cat.), K3Fe(CN)3, K2CO3, ButOH–H2O (1 1), rt.

The significant contribution by Barr and Boyd on the enantioselective synthesis of various tricyclic hemiterpenoid alkaloids and the proof of their absolute configurations, communicated in 1994³⁷ (cf. Ref. 28c), has been published with full experimental details and several extensions.³⁸ The previous work described the conversion of the pyrano[2,3-b]quinoline 48, prepared in two steps from atanine 49, via bromohydrin ester 50 into several alkaloids, including (+)-(3S)-geibalansine 51, (–)-(3S)-ribalinine 52, (–)-(2S)-edulinine 53, (–)-(2R)-araliopsine 54, as well as the alkaloid analogue (+)-(3S)--ribalinine 55. The results permitted the correction of incorrect absolute stereochemistries recorded in the literature for edulinine and araliopsine. The new extensions include conversion of atanine into chiral diols 56 and ent-56 with AD-mix- and AD-mix-, respectively, and proof of absolute configuration by analysis of Mosher esters. Further transformation of 56 and its enantiomer into the bromoacetates 57 and ent-57 with acetoxyisobutyryl bromide prefaced transformation into (–)-(2R)-platydesmine 58 and its (+)-(2S) enantiomer, respectively, upon treatment with base. The N-methylplatydesminium methiodides were readily prepared from the free bases by treatment with iodomethane in ethanol. The (+)-(2S) salt 59 proved to be identical to a natural sample isolated from Skimmia japonica, a result that conflicted with the literature assignment of the (2R)-configuration to (+)-platydesmine methiodide. To clinch matters, single-crystal X-ray crystallography of (+)-platydesminium perchlorate unequivocally supported the (2S) absolute configuration. Thus the (2R) absolute configurations previously assigned to both (+)-platydesmine and its (+)-methosalt are incorrect, and the structure of the former alkaloid is ent-58. Other transformations of relevance included mild alkaline hydrolysis of (+)-platydesminium methiodide 59 to complete an alternative synthesis of (–)-(2S)-edulinine 53; and treatment of N-methylatanine 60 with AD-mix- to give the same (–)-alkaloid.

A short synthesis of racemic araliopsine (±)-54 by Parsons and co-workers exploited the addition of 4-hydroxy-1-methylquinolin-2-one 34 (see Scheme 2) to 2-methylbut-3-en-2-ol upon sonication in acetic acid at 60 °C in the presence of manganese(III) acetate.^39,40 The target alkaloid was obtained directly in 40% yield. This radical-mediated process appears to be general for the addition of 34 to alkyl-substituted alkenes, and several analogues of general structure 61 could be prepared in yields of 31–65%. With styrenes, however, approximately equal amounts of the angularly fused dihydrofuro[3,2-c]quinoline products 61 (R¹ = Ph) and the linearly fused isomers 62 were produced. The reaction of 34 with alkynes (phenylacetylene, 2-methylbut-1-en-3-yne) also yielded mixtures of isomeric tricyclic products, as did the reaction of quinoline-2,4-diol with alkenes.

An electrophysiological study of the ability of various linearly and angularly fused dihydrofuroquinolines and dihydropyranoquinolines to block the voltage-gated potassium channel Kv1.3 in mouse fibroblasts necessitated the preparation of a range of relevant alkaloids and alkaloid analogues from the 3-prenylquinolin-2-ones 63 and 64 by reported transformations.⁴¹ The compounds in the study included, amongst others, the alkaloids dihydroflindersine 65, N-methylhaplamine 66, khaplofoline 67, isoplatydesmine 68 and O-methylribaline 69. The known based-induced skeletal rearrangements of 68 and 69 to give derivatives of araliopsine, ribalinine and -ribalinine (vide supra) were confirmed, and AM1 calculations on the heats of formation of the various isomers supported the experimental observation that angularly fused dihydropyranoquinolinones (e.g., -ribalinine) are the most stable isomers and linear dihydrofuroquinolinones (e.g., isoplatydesmine itself) the least. In the pharmacological studies themselves, angularly fused isomers were in general more potent channel blockers than their linear analogues, and furoquinolines more potent than dihydrofuroquinolines. The most active compounds, N-methylhaplamine 66 and 70, might thus function as templates for the development of novel immunosuppressants.

(–)-Haplotubine 71, a new furoquinoline alkaloid isolated from the aerial parts of Haplophyllum tuberculatum, contains a 6,7-dihydroxygeranyloxy substituent at C-7.⁵ This unusual substituent has only been encountered once before amongst the rutaceous quinoline alkaloids, namely, in the compound bucharaine 72. The connectivity in the side chain was established from HMBC spectra.

The only secofuroquinoline alkaloids to have been reported in the past were the rhoifolinic esters 73. Bark extracts of Sarcomelicope megistophylla have now yielded another two alkaloids of this class, furomegistine I 74 and furomegistine II 75, both of which are conceivably derived by oxidative cleavage of a more conventional furoquinoline alkaloid such as acronycidine 76.¹⁷ The propensity of S. megistophylla to produce oxidatively modified alkaloids has already been commented upon (vide supra). The structures of the furomegistines were deduced on the basis of their spectroscopic characteristics, long-range NMR spectroscopic experiments and NOE effects once again playing a pivotal role in the structural elucidation. Furomegistine II was found to be racemic, possibly indicating its formation from furomegistine I by an unselective intramolecular conjugate addition. Both alkaloids showed moderate cytotoxicity towards human lung carcinoma A549 and human colon adenocarcinoma HT29 cells (IC₅₀ 90 and 100 µM, respectively).

That many furoquinoline alkaloids are biologically active is well known. Dictamnine 77 and evolitrine 78 , amongst other compounds, have recently been shown to inhibit activation of Epstein–Barr virus early antigen (EBV–EA) in Raji cells.¹ Pteleine 79 showed moderate antimicrobial activity against Mycobacterium smegmatis, Bacillus subtilis, Staphylococcus aureus and Candida albicans (minimum inhibitory concentration 5–100 µg cm^–3),¹⁶ while antiplatelet aggregation activity has been demonstrated for (+)-platydesmine ent-58, dictamnine, -fagarine 80, robustine 81 and skimmianine 82.²⁴ Dictamnine, skimmianine and especially kokusaginine 83 were able to block potassium channel Kv1.3 currents in mouse fibroblasts.⁴¹ Dictamnine and evolitrine showed antifeedant activity against fourth instar larvae of the tobacco caterpillar, Spodoptera litura.⁴²

The flowers of Castanea mollissima (Fagaceae), an economically important variety of chestnut tree that is widely distributed in China, are the source of an unusual new alkaloid, chestnutamide 84.² Convincing spectroscopic evidence was presented for the unprecedented pyrrolizino[3,2-b]quinoline skeleton in this compound. In particular, comprehensive HMBC and NOESY correlations established the full connectivity in the assigned structure. The natural product was very slightly laevorotatory ([]_D –0.016, c 0.25, CHCl₃), but its absolute configuration was not determined.

The club mosses of the Lycopodiaceae have yielded several novel decahydroquinoline alkaloids in recent years. The latest additions to this growing group of compounds are huperzines J 85, K 86 and L 87, which were isolated from Huperzia serrata, a Chinese herbal plant used in the treatment of contusions, strains and swellings.⁶ The relative configurations in 85 and 87 were deduced from NOESY correlations, and that in 86, although not specifically elucidated, is expected to be the same. However, the CD spectra of J and L showed negative Cotton effects, while that in K was positive. These three alkaloids, the first examples of naturally occurring N-oxides in the genus Huperzia, belong to the well-known phlegmarane (C₁₆N₂) group of Lycopodium alkaloids.

The first member of a new class of C₂₂N₂Lycopodium alkaloids, (–)-senepodine A 88, has recently been isolated from Japanese specimens of L. chinense.⁷ The structure of this unique octahydroquinoline–quinolizidine dimer was amply supported by one- and two-dimensional NMR spectroscopy. Analysis of proton–proton coupling constants and NOESY data revealed the complex relative stereochemistry shown in 88; moreover, the latter also indicated that the quinolizidine system had a cis-fused ring junction between two piperidine chair conformations. The authors suggest that senepodine A is biogenetically derived from two identical precursors such as 89, themselves conceivably originating in pelletierine 90 (Scheme 3). The new alkaloid was cytotoxic towards murine lymphoma L1210 cells (IC₅₀ 0.1 mg cm^–3), but not towards human epidermoid carcinoma KB cells.

	Scheme 3

The revised structure of antidesmone 91,⁴³ a metabolite of Antidesma membranaceum (Euphorbiaceae) and probably identical to hyeronine A from Hyeronima oblonga, was described in last year's review in this series (cf. Ref. 28d). The revision resulted in part from biosynthetic feeding experiments in cell cultures of A. membranaceum with ¹³C- and ¹⁵N-labelled precursors (sodium acetate, glucose, ammonium nitrate, glycine, alanine, aspartic acid), which Bringmann and co-workers have now published in full.⁴⁴ The results showed that antidesmone is built up in a biosynthetically novel manner from a linear C₁₆ polyketide and a C₂ unit derived directly from glycine. Acetate and, unexpectedly, alanine and glycine were efficient sources of the polyketide chain (92, above the dashed line), which indicated conversion of both amino acids into acetyl-CoA. The transformation of glycine into acetyl-CoA apparently represents a new biosynthetic pathway. Glycine also served as the precursor of the methoxy group. Very significantly, [U-¹³C₂,¹⁵N]glycine appeared as an intact C₂N fragment in the pyridone ring (92, below the dashed line), its CO₂H group obviously undergoing an unusual change in oxidation state to form the 2-CH₃ substituent. Alanine, on the face of it a more plausible source of the C₂N unit, was not incorporated at these positions at all. A final noteworthy feature of the feeding experiments was the specific incorporation of [U-¹³C₄]aspartic acid into the 2-CH₃ substituent and, less prominently, into the acetate-derived positions, which suggests novel conversion of this amino acid into both acetyl-CoA and glycine.

The cinchona alkaloids have normally been surveyed with indole alkaloids and related tryptophan-derived metabolites in this Journal⁴⁵ rather than with the quinoline alkaloids. However, an exception must be made this year for a publication of uncommon significance: the first stereoselective total synthesis of (–)-quinine 93, by Stork and co-workers.⁴⁶ Previous attempts to synthesise this historically and medicinally important alkaloid, succinctly summarised in the prolegomenon to the article, go back almost 150 years, but up to now there has never been a satisfactory solution to the problem of controlling the relative stereochemistry at C-8 and C-9. Stork's approach (Scheme 4) commenced with the (S)-4-vinylbutyrolactone 94 and introduced a second stereogenic centre to give trans-disubstituted 95 (>20 1) by a circuitous but efficient route (58% overall yield). After one-carbon homologation and functional group interconversions, the key azido aldehyde 96 was treated with the lithium salt of 6-methoxy-4-methylquinoline 97. The diastereomeric mixture of alcohols thus produced was oxidised to the azido ketone 98, Staudinger reaction of which yielded tetrahydropiperidine 99. A crucial reduction of this compound with sodium borohydride proceeded with axial delivery of hydride to give the trisubstituted piperidine 100 as a single diastereomer in 91% yield, thereby ensuring the correct absolute stereochemistry at the future C-8 position. The formation of the quinuclidine ring system from 100 was easily achieved by removing the silyl protecting group and mesylating the liberated primary alcohol. The ensuing cyclisation afforded deoxyquinine 101 in 65% yield based on 100. The final phase of the synthesis exploited a stereoselective benzylic hydroxylation first reported by Gutzwiller and Uskokovi in 1978.⁴⁷ This entailed treating deoxyquinine with sodium hydride and oxygen in dry DMSO to give (–)-quinine 93 in 78% yield. The C-9 epimeric compound, epiquinine, was only a minor component of the final product (ca. 14 1).

	Scheme 4 Reagents and conditions: i, Et2NH, Me3Al; ii, TBDMS–Cl, imidazole, DMF; iii, LDA, –78 °C, then ICH2CH2OTBDPS; iv, PPTS (0.3 equiv.), EtOH; v, xylenes, reflux; vi, DIBAL-H, –78 °C; vii, Ph3PCHOMe; viii, Ph3P, DEAD, (PhO)2P(O)N3; ix, HCl (5 M), THF–CH2Cl2, rt; x, 97+ LDA, THF, –78 °C, then add 96, THF, –78 °C; xi, Swern oxidation; xii, Ph3P, THF, reflux; xiii, NaBH4, MeOH–THF (1 1); xiv, HF, MeCN; xv, MeSO2Cl, pyridine, CH2Cl2; xvi, MeCN, reflux; xvii, NaH, DMSO, then O2.

The nematicidal activity of the culture metabolites of Penicillium cf. simplicissimum was traced by bioassay to several structurally related compounds, among them the known natural products 102 and 103, penigequinolones A and B, 104 and 105, and the novel compound (–)-peniprequinolone 106.¹² Comprehensive ¹H and ¹³C NMR spectra were recorded for all these products, and similar Cotton effects in their CD spectra indicated that they all possess the same relative stereochemistry at C-3 and C-4. All but 103 showed nematicidal activity (at concentrations of greater than 100 mg dm^–3) towards Pratylenchus penetrans, a parasitic nematode that causes root lesions in various economically important crops. The penigequinolones in particular may prove to be useful for controlling parasitic nematodes, since they were found to be non-toxic to a free-living nematode, and to lettuce and rice seedlings; in fact, both peniprequinolone and the penigequinolones actually accelerated root growth in rice seedlings.

A strain of the fungus Penicillium sp. (strain 386) collected from sand in a marine habitat in the South China Sea has yielded (–)-penicillazine 107, a novel quinolin-2-one that incorporates an unprecedented 5,6-dihydro-4H-1,2-oxazine substituent.¹³ Various spectroscopic techniques were used to establish the gross structure of this unique natural product, and the stereochemical relationships of the hydrogen atoms in ring D were deduced from analysis of coupling constants and ROESY correlations. Unusual changes in chemical shifts revealed by variable temperature ¹H NMR spectroscopy were ascribed to intermolecular hydrogen bonding effects. However, it required single crystal X-ray diffraction on penicillazine monohydrate to reveal the complete structure shown in 107, with cis-fusion of the dihydrooxazine and cyclohexene rings. The absolute stereochemistry was not determined.

Cultured broth of the entomopathogenic fungus Penicillium sp. EPF-6, isolated from larvae of the mulberry pyralid moth, yielded three new quinolone antibiotics, (+)-quinolactacin A 108, (–)-quinolactacin B 109 and (+)-quinolactacin C 110.^14,15 The novel 2,3-dihydro-1H-pyrrolo[3,4-b]quinoline-1,9(4H)-dione structures were established uneventfully with the aid of spectroscopic techniques. Disappointingly, the quinolactacins were inactive towards a wide range of bacteria, fungi and yeasts; very weak activity was apparent only against Aspergillus niger. However, quinolactacin A inhibited the production of tumour necrosis factor (TNF) induced by lipopolysaccharide in murine peritoneal macrophages (IC₅₀ 12.2 µg cm^–3) and in macrophage-like J774.1 cells. This interesting biological effect prompted Tatsuta and co-workers to devise a short biomimetic approach to the synthesis of members of this group of compounds (Scheme 5).⁴⁸ Anthranilic acid 111 was converted in five steps into the -keto thiolester 112, reaction of which with L-valine methyl ester yielded the -keto amide 113. Hydrogenolysis of the benzyloxycarbonyl protecting group was followed by Dieckmann-like condensation to give the intermediate tetramic acid 114, which on exposure to silica gel afforded racemic quinolactacin B, (±)-109. Analogues 115–117 were similarly prepared by replacing valine with suitable derivatives of proline, glutamic acid and arginine, respectively. Quinolactacin A could presumably be made in the same way with isoleucine as precursor.

	Scheme 5 Reagents and conditions: i, ClCO2Bn, Na2CO3, THF–H2O, rt; ii, MeI, NaH, DMF, rt; iii, KOH, MeOH–H2O, 65 °C; iv, 2,2-dipyridyl disulfide, Ph3P, THF, rt; v, MeCOSBut, LiHMDS, THF, –78 °C to rt; vi, Et3N, CuI, THF, rt; vii, H2 Pd–C, EtOH, rt; viii, NaOMe, MeOH, reflux; ix, SiO2.

Boger and Ichikawa communicated syntheses of the antitumour antibiotic thiocoraline 118 and the related octadepsipeptide BE-22179 119 in 2000⁴⁹ (cf. Ref. 28h). Full details of the syntheses have since appeared in a publication that also included an evaluation of their ability to bind to duplex DNA by bisintercalation, and their exceptionally potent cytoxicity towards the L1210 cell line at subnanomolar concentrations.⁵⁰ Syntheses of two larger cyclic decadepsipeptides, luzopeptin C 120⁵¹ and luzopeptin E2 121⁵² by Ciufolini and coworkers involved macrocyclodimerisation of the key pentapeptides 122 and 123, respectively. Deprotection of the N-Boc group and acylation with 3-hydroxy-6-methoxyquinaldic acid 124 was a late and comparatively trivial step in the synthesis of the target antibiotics.

When the Australian sponge Oceanapia sp. was screened for activity against mycothiol S-conjugate amidase (MCA), a recently discovered mycobacterial enzyme, bioactivity-guided fractionation led to the isolation of the known sponge metabolite uranidine 125, several bromotyrosine alkaloids, e.g. (–)-pseudoceratine 126, and the interesting uranidine–bromotyrosine hybrid (–)-127.⁸ The elucidation of the structure was facilitated by obvious spectroscopic similarities between the three alkaloids, as well as various long-range NMR spectroscopic correlations. The (1S,6R) absolute stereochemistry of (–)-127 was ascertained by comparing its CD spectra with those of (1S,6R)-(–)-126 and its enantiomer.

Recent refutations of structure 128 for the sponge metabolite haliclorensin have cast a cloud over the proposed structure 129 for the more complex alkaloid halitulin, which was found in the same organism (Haliclona tulearensis).⁵³ Heinrich and Steglich synthesised both enantiomers of the putative alkaloid 128, but found that the optical rotation and ¹³C chemical shifts differed considerably from those reported for the natural product.⁵⁴ Similarly, a synthesis of (±)-128 by Banwell and co-workers showed that its properties were different from those of the natural product.⁵⁵ The revised structure 130 for haliclorensin has been proposed very recently and confirmed by synthesis, and indications from chiroptical properties are that the natural product is a 3 1 mixture of the (S)- and (R)-enantiomers.⁵⁶ This unusual diazacyclotetradecane system can obviously not be a component of the halitulin structure, and structure 129 for halitulin may well still turn out to be correct.

Kibayashi and co-workers have synthesised (–)-lepadins A, B and C, 131–133, three complex decahydroquinoline alkaloids isolated from the tunicate (sea squirt) Clavelina lepadiformis and a predatory flatworm that feeds on it, by a route that includes their hallmark reaction, a stereocontrolled intramolecular acylnitroso Diels–Alder reaction.^57,58 Several key steps in the lengthy synthesis are shown in Scheme 6. Enantiocontrol originated from benzylidene-protected (2S)-2,4-dihydroxybutanal 134, a readily prepared derivative of (S)-malic acid. The key cycloaddition was initiated by oxidation of the hydroxamic acid 135 with tetrapropylammonium periodate in water–DMF (50 1), which gave the two oxazino lactams 136 and 137 (90%) in a ratio of 1 6.6. The aqueous conditions proved crucial for maximising the formation of the desired diastereomer. After hydrogenation of 137, the next important step was the stereoselective -hydroxylation of the lactam, which was optimally achieved by treating the lithium enolate with (+)-[(8,8-dichlorocamphoryl)sulfonyl]oxaziridine. After silylation, compound 138 was obtained in a ratio of 17 1 with its -silyloxy epimer. Treatment of lactam 138 with methylmagnesium bromide followed by sodium cyanoborohydride afforded the methylated product 139 as the sole diastereomer because of the approach of hydride from the less hindered face of the iminium ion intermediate. Construction of the quinoline framework was most satisfactorily achieved by intramolecular aldol reaction of 140 with catalytic amounts of piperidine and acetic acid to give 141 (87%). However, dehydration of the 3-hydroxycarbonyl product was accomplished only after the aldehyde had been converted into the corresponding methyl ester. The product 142 yielded a 1 2 mixture of the labile alcohol 143 and the C-5 epimer 144 in 87% yield upon desilylation with tetrabutylammonium fluoride in THF over 5 days. The yield of the desired 5-isomer 144 could be increased to 75% by equilibrating 143 under the same conditions. The final stereogenic centre was introduced by catalytic hydrogenation of the amino alcohol 145 over palladium on carbon, which gave exclusively the cis-fused decahydroquinoline 146 (85%) after N-Boc protection. Oxidation to the aldehyde and Takai methylenation with iodoform and chromium(II) chloride yielded the (E)-vinyl iodide 147, which proved to be the pivotal intermediate in the synthesis of all three lepadins. Completion of the synthesis of (–)-lepadins A 131 and B 132, for example, entailed Suzuki coupling of 147 with (E)-hex-1-enylboronic acid followed by appropriate manipulations at the alcohol and amine groups. For (–)-lepadin C 133, the Suzuki cross-coupling was performed with (E)-5-hydroxyhex-1-enylboronic acid, and oxidation of the side chain preceded the functional group manipulations in the bicyclic nucleus. While (–)-lepadin B has previously been synthesised, this is the first reported synthesis of (–)-lepadins A and C. The spectra of the trifluoroacetate salts of the synthetic compounds proved identical to those obtained on the natural products, but the optical rotations were larger.

	Scheme 6 Reagents and conditions: i, Pr4NIO4, H2O–DMF (50 1), 0 °C; ii, H2, Pd–C, THF; iii, LiHMDS, THF, –78 °C, then (+)-[(8,8-dichlorocamphoryl)sulfonyl]oxaziridine; iv, TBDPSCl, imidazole, DMF, rt; v, MeMgBr, THF, 0 °C, then NaBH3CN, AcOH, THF, 0 °C; vi, Zn, 90% AcOH, 60 °C; vii, PhCOCl, then 5% aq. KOH; viii, CS2, NaH, imidazole, then MeI, THF; ix, Bu3SnH, AIBN, C6H6, reflux; x, PPTS, ButOH, reflux; xi, H2, Pd(OH)2, MeOH; xii, (COCl)2, DMSO, Et3N, –78 °C to 0 °C; xiii, piperidine (0.2 equiv.), AcOH (0.2 equiv.), C6H6, reflux; xiv, PDC, DMF, then CH2N2, Et2O; xv, SOCl2, Et3N; xvi, Bu4NF, THF, rt, 5 d; xvii, TBDMSCl, imidazole, DMF; xviii, LiAlH4, THF, reflux; xix, H2(5 atm), 10% Pd–C, THF, then (Boc)2O, CH2Cl2, 0 °C to rt; xx, CHI3, CrCl2, THF, rt; xxi, (E)-C4H9CHCHB(OH)2, Pd(Ph3P)4(5 mol%), aq. KOH (2 M), THF, 50 °C; xxii, Bu4NF, THF; xxiii, TFA, CH2Cl2, then K2CO3.

Oppolzer and Flaskamp reported a short synthesis of the frog skin alkaloid (–)-cis-decahydroquinoline 195A 148, commonly referred to as pumiliotoxin C, in 1977.⁵⁹ Methodological refinements⁶⁰ (Scheme 7) have now resulted in an improved overall yield of 148 to 25% – approximately 20 times that of the original procedure – based on the chiral amino alcohol 149. The improvements entailed ring opening of the aziridine 150 with propargylmagnesium bromide followed by a one-pot conversion of product 151 into the (E)-amine 152, and an optimised imine formation and N-acylation to give dienamide 153. The subsequent intramolecular Diels–Alder cycloaddition afforded the octahydroquinoline 154 as a mixture of diastereomers. After catalytic hydrogenation and removal of the N-acyl group, the decahydroquinoline 148, diastereomer 155 and an unidentified isomer were obtained in a ratio of 62 37 1. The target alkaloid 148 could be crystallised from the mixture as the hydrochloride salt (57%). Chromatography of the mother liquors yielded a further quantity of 148 (3.5%) and the more polar diastereomer 155, also isolated as the hydrochloride salt (14%).

	Scheme 7 Reagents and conditions: i, NaH, p-TsCl, THF, 0 °C to rt; ii, HCCCH2MgBr, Et2O, 0 °C to rt; iii, BuLi, NH3, –78 °C to –30 °C, then MeI, –30 °C, then Na; iv, MeCHCHCHO, 4 molecular sieves, Et2O, 0 °C to rt; v, Me2CHCOCl, Et3N, CH2Cl2, –78 °C to 5 °C; vi, toluene, sealed tube, 230–240 °C; vii, H2, 10% Pd–C, MeOH, rt; viii, BuLi, THF, –78 °C to rt, then recrystallisation as HCl salt.

The tricyclic frog skin alkaloid gephyrotoxin 156 and its perhydro analogue 157 have been popular synthetic targets for over 20 years. Three formal syntheses of these compounds have appeared recently. The approach to 157 by Mehta and Reddy,⁶¹ employing ideas they had developed in a prior route to pumiliotoxin C, involved the conversion of the protected cyclopentadienone dimer 158 via the cis-hydrindanone 159 into the decahydroquinolinone 160, which had previously featured in a synthesis of perhydrogephyrotoxin 157 by Ibuka and Chu.⁶² Pearson's continuing explorations of the intramolecular Schmidt reaction of azides with carbocations has recently been extended to include the preparation of an impressive range of benzo-fused 1-azabicycloalkanes.⁶³ Extensive model studies aimed at the synthesis of gephyrotoxin itself culminated in the synthesis of 161 (45%) and the regioisomer 162 (10%) by treatment of azido alkene 163 with trifluoromethanesulfonic acid followed by reduction of iminium ion intermediates with L-Selectride and replacement of bromide by hydroxy. Compound 161 had previously featured as an intermediate in a prior route to gephyrotoxin 156 by Ito et al.,⁶⁴ which itself converged with a pioneering synthesis of the racemic alkaloid by the Kishi group.⁶⁵ The route to (+)-156 by Hsung and co-workers⁶⁶ exploited a formal [3 + 3] cycloaddition similar to that already depicted in Scheme 2 (Section 1.3, vide supra). In this case, intramolecular cycloaddition resulting from treatment of precursor 164 with piperidinium acetate yielded a mixture of isomers of the tricyclic vinylogous amide 165. If the alcohol was not protected, only the unwanted -H isomer was obtained. Removal of the silyl protecting group from 165 and chromatographic separation afforded the -H tricyclic compound 166, a central intermediate in the 1981 synthesis of (+)-156 by Fujimoto and Kishi.⁶⁷

A list of new quinazoline alkaloids, and known quinazolines isolated from new sources, is presented in Table 2.^68–76 The simple alkaloid samoquasine A 167, isolated from seeds of the custard apple Annona squamosa (Annonaceae), is especially interesting in being the first naturally occurring compound to possess the benzo[h]quinazoline ring system.⁶⁸ The structure was established by spectroscopic methods and corroborated by preparation of the O-methyl derivative by treatment with trimethylsilyldiazomethane. The authors were apparently unaware that this compound was first prepared almost 40 years ago by reaction between 1-naphthylamine and ethoxymethyleneurethane, although no spectroscopic details are given in this earlier work.^77,78 Samoquasine A showed significant cytotoxicity against murine lymphoma L1210 cells (IC₅₀ 0.38 µg cm^–3).

A resurgence of interest in the potent antimalarial alkaloids (+)-isofebrifugine 168 and (+)-febrifugine 169 (see Section 2.2) has resulted in reinvestigations of their main plant sources, the genera Dichroa and Hydrangea (Saxifragaceae). Chinese workers re-isolated these two alkaloids (to which they appended the old-fashioned names -dichroine and -dichroine, respectively) as principal components from extracts of the leaves of Dichroa febrifuga, and in addition obtained much smaller amounts of quinazolin-4-one 170 (3H tautomer shown), 2-(4-hydroxybutyl)quinazolin-4-one 171 and the interesting new quinazolinone–quinolizidine dimer (+)-neodichroine 172, which was isolated as a crystalline solid.⁷¹ While the authors made no specific comments about compound 171, it also appears to be a new natural product, although it has previously been synthesised.⁷⁹ Evidence for the structure of neodichroine 172 came from ¹H and ¹³C NMR spectra, recorded in deuterated pyridine, together with COSY and NOE data. The trans-diaxial disposition of 9-H and 9a-H in the quinolizidine ring was apparent from the large coupling constant (J 10.1 Hz), and another large coupling constant for 3-H (J 11.3 Hz) indicated that the quinazolinyl substituent was equatorial. Neodichroine also formed an acetate that gave a well-resolved ¹H spectrum. However, definitive evidence for the structure came from a semi-synthesis by Mannich reaction between febrifugine and formaldehyde at pH 4. Although this direct correlation with putative (2S,3R)-(+)-febrifugine led the authors to propose the (9R,9aS) absolute configuration for 172, it seems that they were unaware that the absolute configuration of (+)-febrifugine was recently revised to (2R,3S) as a result of Kobayashi's unambiguous total synthesis⁸⁰ (cf. Ref. 28f,g). (+)-Neodichroine is thus more likely to be the (3R,9S,9aR) enantiomer, as shown in 172.

Another more recently reported quinazolinone–quinolizidine dimer, (+)-hydrachine A 173, was isolated as a semi-solid by Taiwanese workers from the roots of Hydrangea chinensis, where it occurred together with quinazolin-4(3H)-one 170.⁷²¹H and ¹³C NMR spectra, recorded in deuterated chloroform, were supplemented by COSY and HMBC correlations to establish the atomic connectivity. Central to the structural assignment were HMBC correlations between 4-H and both carbonyl groups, C-2 on the quinazolinone, and signals ascribed to C-3 (mistakenly listed as C-5), C-6 and C-9a; these were taken as evidence for siting the quinazolinyl substituent at C-4 on the quinolizidine ring. The stereochemistry was determined from NOESY correlations and analysis of coupling constants; in particular, 4-H, 9-H and 9a-H all showed large couplings consistent with axial orientations. Bohlmann bands in the IR spectrum indicated trans-fusion of the quinolizidine ring. While the evidence presented for the structure of 173 seems irreproachable, one cannot but be intrigued by the tantalising similarity between the structures proposed for neodichroine and hydrachine A and their relationship to febrifugine. The ¹H NMR spectroscopic data for the two compounds are not directly comparable, since they were recorded in different solvents, and other reported spectroscopic and physical properties (IR, UV, MS, optical rotation) show similarities and differences that may or may not be significant. However, the ¹³C NMR spectroscopic data are in remarkable agreement (±1.3 ppm in the quinolizidine ring) despite the difference in solvents; the only discrepancy in interpretation is that the authors interchange the assignments of the doublet signals for the bridgehead position C-9a and the carbon bearing the quinazoline substituent. Is it conceivable that the two alkaloids, both originating in genera known for their production of febrifugine, are in fact the same? If so, the conversion of febrifugine into neodichroine must be regarded as deciding the issue.

The identity of the new alkaloid peganol N-oxide 174, isolated from aerial parts of Nitraria komarovii, was established spectroscopically, and by reduction with zinc in hydrochloric acid to give a mixture of peganol 175 and deoxypeganine (deoxyvasicine) 176.⁷⁴ More interesting are the two dimeric peganine derivatives dipegine 177 and dipeginol 178, which were obtained from Peganum harmala.⁷⁶ Although dipegine was originally reported as long ago as 1974, comprehensive NMR data in the present publication permitted unambiguous location of the bridge between C-4 and C-9, rather than C-11 as had previously been proposed. However, the relative configurations at these stereogenic sites could not be ascertained directly. Molecular mechanics calculations on the most stable conformers of both possible diastereomers led the authors to predict average vicinal coupling constants between 4-H and 9-H. While the predicted ³J value of 2.7 Hz for the diastereomer shown in 177 was reasonably close to the experimentally determined coupling constant (J 2.1 Hz), this stereochemical assignment for dipegine should still be regarded as tentative. The gross structure of the novel alkaloid dipeginol 178 was also revealed by analysis of its NMR spectra and those of its mono-acetate derivative, but no attempts were made to assign the relative configuration. Since both dimeric alkaloids are high-melting solids, the structural ambiguities could probably be resolved by X-ray crystallography.

The only new quinazoline alkaloid reported from a fungal source during the review period was (–)-circumdatin G 179, extracted together with (–)-circumdatin F 180 from the culture broth of Aspergillus ochraceus.⁶⁹ Circumdatin F had previously been isolated in such small quantities that its optical rotation and absolute configuration could not be determined.⁸¹ The present work permitted the measurement of optical rotations for both 180 ([]_D –18.9, c 0.11, MeOH) and 179 ([]_D –21.7, c 0.19, MeOH), as well as the acquisition of comprehensive spectroscopic data. Their (S)-absolute configurations were assigned by analogy with (S)-(–)-circumdatin C 181, the stereochemistry of which had previously been determined by degradation to L-alanine.⁸²

A critical review of the ethnopharmacology and toxicology of the Indian medicinal plant Adhatoda vasica (Acanthaceae), the principal alkaloid of which is vasicine 182, has called into question several reports on potentially adverse effects both of the plant extract and of vasicine itself.⁸³ While traditional applications in the treatment of various disorders, especially ailments of the respiratory tract, appear to be well documented, claims of oxytocic and abortifacient effects and of acute and general toxicity seem to be based on inappropriate testing methods and unreliable data. In the meantime, medicinal uses of the vasicine alkaloids continue to be documented, and patent applications have been filed for the use of deoxypeganine 176 in the treatment of nicotine dependence,⁸⁴ drug dependence,⁸⁵ alcoholism⁸⁶ and Alzheimer's dementia,⁸⁷ and in the treatment of poisoning by an organophosphorus cholinesterase inhibitor.⁸⁸

Tryptanthrin 183 has been found to inhibit the production of both nitric oxide and prostaglandin E₂ in murine macrophage RAW 264.7 cells activated by interferon- and lipopolysaccharide.⁸⁹ In the former case, the mechanism appears to involve suppression of inducible NO synthase expression, and in the latter case the alkaloid inhibits cyclooxygenase activity. The results suggest that tryptanthrin may be a useful anti-inflammatory agent. The alkaloid has also shown good antibacterial activity against Helicobacter pylori both in vitro and during in vivo studies with Mongolian gerbils infected with the ulcer-causing pathogen.⁹⁰

Samarium(II) iodide in THF has been shown to mediate the reaction between N,N-diethyl-o-nitrobenzamide and various benzonitriles or phenylacetonitrile to yield 2-substituted quinazolin-4(3H)-ones in yields of 55–75%.⁹¹ Among the products formed was the alkaloid glycosminine (glycophymine) 184. The reaction failed with acetonitrile.

Kamal et al. found that the intramolecular aza-Wittig reaction of N-(2-azidobenzoyl)lactams 185 gave deoxyvasicinone 186 and analogues in quantitative yield within 10–15 minutes when the precursors were treated with chlorotrimethylsilane and sodium iodide in acetonitrile at room temperature.⁹² This mild transformation provides a useful alternative to the standard cyclisation conditions, which entail treatment of the azide precursors with triphenylphosphine or tributylphosphine in boiling toluene for several hours. Even more interestingly, the reductive cyclisation of 185 proceeded in yields of 25–50% when catalysed by bakers' yeast in a water–ethanol mixture at 37.5 °C. With convenient access to deoxyvasicinone, the authors embarked on a lipase-mediated resolution of vasicinone 187 itself. The 3-bromo derivative 188 was prepared by free radical bromination of 186 with NBS (57%), and this product was converted into (±)-acetylvasicinone 189 by treatment with potassium acetate and 18-crown-6 in acetonitrile (100%). Enzymatic hydrolysis of the ester with lipase PS Amano in a mixture of acetonitrile and phosphate buffer (pH 7) was highly enantioselective, yielding (R)-(+)-vasicinone and the (S)-(–)-acetate in 98% enantiomeric excess (ee). Alternatively, chemical hydrolysis of 189 yielded racemic vasicinone, which could then be resolved by enzymatic transesterification with vinyl acetate in THF in the presence of lipase PS in THF to yield (S)-(–)-vasicinone, as well as the (R)-(+)-acetate in better than 99% ee.

Deoxyvasicinone 186 was the common precursor in the short divergent syntheses of isaindigotone 190 and luotonin A 191 by Molina et al. (Scheme 8).⁹³ Straightforward condensation of 186 with 4-acetoxy-3,5-dimethoxybenzaldehyde in acetic anhydride followed by hydrolysis of the ester completed the first reported synthesis of 190 in 64% overall yield. The formal synthesis of 191 simply entailed the oxidation of 186 with selenium dioxide to give the dione 192 (42%). Alternatively, a two-step reaction sequence commencing with condensation of 186 with benzaldehyde followed by ozonolysis afforded 192 in similar yield (43%). The synthesis is formal because Kelly and co-workers recently converted 192 into luotonin A 191 by base-catalysed reaction with 2-aminobenzaldehyde⁹⁴ (cf. Ref. 28h).

	Scheme 8 Reagents and conditions: i, 4-Acetoxy-3,5-dimethoxybenzaldehyde, Ac2O, reflux; ii, NaOH, EtOH, reflux; iii, SeO2, H2O, dioxane, reflux; iv, PhCHO, Ac2O, reflux; v, O3, CH2Cl2, 3 min, then Me2S.

The recent upsurge of activity in the synthesis of febrifugine 169 and related antimalarial compounds shows no signs of abating, as evidenced by a review in the Japanese literature by Takeuchi and Harayama.⁹⁵ Following hard on the heels of their stereoselective synthesis of racemic febrifugine⁹⁶ (cf. Ref. 28g), Takeuchi and co-workers have devised the enantioselective modification shown in Scheme 9.^97,98 The central feature was the use of bakers' yeast and sucrose in an ethanol–water solvent for the enantioselective reduction of the 2-allylpiperidin-3-one (±)-193 to give a separable mixture of unreduced (R)-(–)-193 (31%, 93% ee) and the (+)-alcohol 194 (41%, 97% ee). However, since racemisation of (–)-193 was readily accomplished with potassium carbonate, a reductive dynamic optical resolution could be achieved by adding this base to the fermentation mixture. In this way the yield of (+)-194 was boosted to 62% (97% ee), while (–)-193 was recovered with diminished optical activity (14% yield, 14% ee). Intramolecular bromoetherification of (+)-194 with NBS in acetonitrile afforded 195 as a 3 1 mixture of diastereomers. Elimination of hydrogen bromide from this product followed by methoxybromination with NBS in methanol yielded the ketal 196, this time as a 4 1 mixture of diastereomers. Hydrolysis of the ketal, reaction with the anion of quinazoline-4(3H)-one and removal of the benzyloxycarbonyl protecting group from nitrogen completed this green synthesis of (+)-isofebrifugine 168. Furthermore, since isomerisation of isofebrifugine to febrifugine has been well documented, the authors were able to form a 1 2 equilibrium mixture of 168 and (+)-febrifugine 169 simply by heating the former in water at 80 °C for 15 minutes. Isomerisation did not occur under acidic conditions. Oddly enough, when the deprotection of racemic Cbz-isofebrifugine 197 with aqueous hydrochloric acid was investigated, mixtures of the piperidine-cleavage products 198, 199 and 200 were obtained.⁹⁹ The reaction of 197 with boron trifluoride–diethyl ether yielded 198 as the sole product (74%). Further antimalarial tests on the products confirmed that (+)-febrifugine was almost 100 times as active towards Plasmodium falciparum as chloroquine, and also showed that it was twice as potent as its hydrochloride salt and about ten times as potent as (+)-isofebrifugine.⁹⁸ In an interesting corollary to this work, Takeuchi's team synthesised the racemic regioisomers 201 and 202 from the 4-allylpiperidin-3-one analogue of 193, and proved that they have negligible antimalarial activity.¹⁰⁰

	Scheme 9 Reagents and conditions: i, Bakers' yeast, sucrose, EtOH–H2O (1 10), K2CO3, 15 °C, 90 h; ii, NBS, MeCN, rt; iii, KOBut, THF, rt; iv, NBS, MeOH, rt; v, 10% aq. HCl, MeCN, rt; vi, quinazolin-4(3H)-one, K2CO3, DMF, rt; vii, H2, 20% Pd(OH)2–C, MeOH, rt; viii, H2O, 80 °C, then 10% aq. HCl; ix, BF3·Et2O, MeCN, reflux, 0.5 h; x, aq. HCl (5 M), MeCN, reflux, 0.5–2 h.

A stereocontrolled synthesis of (+)-febrifugine by Taniguchi and Ogasawara¹⁰¹ (Scheme 10) commenced with the chiral building block (–)-203, which was prepared from furfural by reported methods. A key step in this route was the ring-closing metathesis of 204 with Grubbs catalyst to give the dehydropiperidine 205 in 89% yield. The quinazoline substituent was introduced at a late stage by treating the epoxide 206 with the anion of quinazolin-4(3H)-one to give the alcohol 207 as a mixture of diastereomers. Oxidation with Dess–Martin periodinane and removal of the protecting groups under acidic conditions completed the synthesis of (+)-febrifugine 169 in 24 steps and 11% overall yield from (–)-203.

	Scheme 10 Reagents and conditions: i, Bu4NF, THF; ii, MeSO2Cl, Et3N; iii, LiI, THF; iv, Zn, EtOH–HOAc (10 1); v, NaBH4, EtOH; vi, (Cy3P)2Cl2RuCHPh (5 mol%); vii, H2, PtO2; viii, (PhS)2, Bu3P, pyridine; ix, BnBr, NaH; x, 30% H2O2; xi, CaCO3, Ph2O, reflux; xii, OsO4(cat.), NMO, aq. THF; xiii, p-TsCl, pyridine; xiv, K2CO3, MeOH; xv, quinazolin-4(3H)-one, KOH, MeOH; xvi, Dess–Martin periodinane; xvii, HCl (6 M), reflux.

In the synthesis of the febrifugine alkaloids by Hatakeyama and co-workers,¹⁰² the key sequence of steps involved ozonolysis of the chiral alkene 208 followed by condensation of the resulting aldehyde with hydroxylamine hydrochloride in allyl alcohol as solvent (Scheme 11). The unsaturated alcohol trapped the intermediate nitrone 209 to give the three cycloadducts 210–212 in a ratio of 64 10 26 and an overall combined yield of 74%. Although these adducts could be separated, it was more convenient to take the mixture through the subsequent hydrogenolysis, protection and epoxidation to give the epoxide diastereomers 213, treatment of which with the anion of quinazolin-4(3H)-one followed by Dess–Martin oxidation gave the cis-and trans-2,3-disubstituted piperidines 214 in a ratio of 22 78, respectively. Acid-induced deprotection with boiling 6 M hydrochloric acid yielded a mixture of (+)-isofebrifugine 168 (27%) and (+)-febrifugine 169 (58%), the 33 67 ratio of which indicated that partial epimerisation had taken place. In a result that casts light on this epimerisation process (presumably a reversible Michael reaction) and reinforces Takeuchi's observations⁹⁷ (vide supra), a purified sample of cis-214 afforded only (+)-isofebrifugine when heated with hydrochloric acid, whereas trans-214 yielded an 84 16 mixture of (+)-febrifugine and (+)-isofebrifugine. However, isofebrifugine could be isomerised to febrifugine under neutral conditions in boiling methanol.

	Scheme 11 Reagents and conditions: i, H2CCHOAc, Novozym 435, Pri2O; ii, H2, Lindlar catalyst, MeOH, then K2CO3; iii, TBDPSCl. imidazole, DMF; iv, Li naphthalenide, THF, –25 °C; v, MeSO2Cl, Et3N, CH2Cl2; vi, O3, then Me2S, NaHCO3, CH2Cl2. –78 °C; vii, HONH2·HCl, Et3N, H2CCHCH2OH; viii, H2, PdCl2, MeOH; ix, (Boc)2O, Et3N, CH2Cl2; x, N-Ts-imidazole, NaH, THF; xi, quinazolin-4(3H)-one, KH, DMF; xii, Dess–Martin periodinane, CH2Cl2; xiii, HCl (6 M), reflux; xiv, MeOH, reflux.

The important enantioselective synthesis of febrifugine and isofebrifugine by Kobayashi and co-workers⁸⁰ (cf. Ref. 28f,g) has already been mentioned. Patent applications based on the original route^80a have recently been filed.¹⁰³ However, these workers subsequently devised alternative syntheses of both alkaloids based on the Lewis acid-catalysed reaction of silyl enol ethers and related nucleophiles with acyliminium ions prepared in situ from N,O-acetals. After an extensive series of model studies with N-Cbz-protected 2-methoxypiperidines and 2-acyloxypiperidines, scandium(III) triflate was found to give the best yields of 2-acylmethylpiperidine products, and also resulted in high 2,3-trans/cis selectivity with 2,3-diacyloxypiperidine substrates.^104,105 The synthesis of febrifugine itself (Scheme 12) employed the (3S)-substrate 215, which was prepared via the Weinreb amide 216 either from L-ornithine or by a route involving an asymmetric aldol reaction.¹⁰⁵ In this case the tin(II) enolate of the quinazolinone-substituted ketone 217 was required in order to maximise the formation of the 2,3-trans-disubstituted product 218 (55%) in relation to its cis-isomer (14%). A disappointing two-step deprotection of 218 (25% yield) completed the synthesis of (+)-febrifugine 169. For isofebrifugine, reaction between the trimethylsilyl enol ether of 217 and the racemic semicyclic N,O-acetal 219 (a 74 26 mixture of isomers) was catalysed by trimethylsilyl triflate, and yielded the ring-opened product 220, almost exclusively as the syn diastereomer (93 7), in 51% yield.¹⁰⁶ The piperidine ring was formed by oxidation of the terminal alcohol to the aldehyde and reduction of the ensuing cyclic acyliminium ion, after which removal of the protecting groups completed the synthesis of racemic isofebrifugine (±)-168.

	Scheme 12 Reagents and conditions: i, LiAlH4, Et2O, 0 °C; ii, Ac2O, Et3N, DMAP (cat.), rt; iii, Sn(OTf)2(2 equiv.), Pri2NEt (2 equiv.), CH2Cl2, 0 °C to reflux; iv, add 217(0.5 equiv.), Sc(OTf)3(0.1 equiv.), CH2Cl2, reflux; v, 25% HBr in HOAc, 0 °C, then piperidine; vi, NaOMe, MeOH, rt; vii, Me3SiOTf (2 equiv.), Pri2NEt (2 equiv.), CH2Cl2, 0 °C to rt; viii, Me3SiOTf (2.5 equiv.), MeCN, rt; ix, SO3·pyridine, DMSO, Et3N; x, Et3SiH, BF3·OEt2; xi, aq. HCl (6 M), reflux.

N-Sulfinylanthraniloyl chloride 221 was the preferred starting material for Witt and Bergman's assembly of the tripeptides 222 (R = H, OBn), key intermediates en route to the fungal metabolites (–)-circumdatin F 180 and (–)-circumdatin C 181 (Scheme 13).¹⁰⁷ Cyclisation of 222 with triphenylphosphine and iodine in the presence of Hunig's base gave the 4-imino-4H-3,1-benzoxazines 223 (R = H, OBn), aminolysis of which with piperidine produced the amidines 224. The target alkaloids 180 ([]_D –55, c 0.94, CHCl₃) and 181 ([]_D –91, c 0.17, MeOH) were obtained after deprotection of 224 with hydrobromic acid in acetic acid followed by treatment with a tertiary amine and silica gel. It should be noted that the optical rotation of synthetic (–)-circumdatin F is considerably larger than that recently recorded on a natural sample (cf. Section 2.1). A different synthesis of circumdatin F and the related alkaloid sclerotigenin 225 by Snider and Busuyek,¹⁰⁸ also shown in Scheme 13, entailed the selective acylation of the benzodiazepinediones 226, without the need for protecting groups, at the more acidic anilide position with 2-azidobenzoyl chloride, followed by aza-Wittig cyclisation of the resulting imides 227 with tributylphosphine. The optical rotation for circumdatin F, incorrectly reported in this publication, has subsequently been corrected ([]_D –52.9, c 0.5, CHCl₃).¹⁰⁹ It is also intriguing that the ¹H NMR spectrum of this alkaloid showed the presence of two conformers in the ratio 99 1, arising from flipping of the seven-membered ring.

	Scheme 13 Reagents and conditions: i, Methyl anthranilate (R = H) or methyl 5-benzyloxyanthranilate (R = OBn), toluene, rt; 48 h; ii, N-Cbz-L-Ala, DCC, CH2Cl2, 0 °C to rt; iii, Ph3P, I2, Pri2NEt, CH2Cl2, rt; iv, 20% piperidine in EtOAc, rt; v, 45% HBr in HOAc, 60 °C; vi, Et3N (for R = H) or Pri2NEt (for R = OH), EtOAc, rt; vii, Et3N, DMAP, THF, then 2-N3C6H4COCl, THF, 20 °C (for R= Me); viii, Et3N, DMAP, DMSO–CH2Cl2, then 2-N3C6H4COCl, CH2Cl2, 20 °C (for R= H); ix, Bu3P, C6H6, rt to 60 °C.

A synthesis of (+)-fumiquinazoline G reported some years ago by Snider and He¹¹⁰ (cf. Ref. 28i) suffered from poor yields during the removal of a 2,4-dimethoxybenzyl protecting group. The difficulty has now been overcome by using the photolabile 2-nitrobenzyl protecting group.¹⁰⁸ Thus the final steps, shown in Scheme 14, entailed acylation of the protected diketopiperazine 228 with 2-azidobenzoyl chloride and aza-Wittig cyclisation of the product to give 229. Photolytic removal of the 2-nitrobenzyl group from a dilute solution of 229 in methanol at 254 nm afforded an 87% yield of ent-fumiquinazoline G 230. It is of interest that Avendaño and co-workers recently experienced problems in the removal of a 2,4-dimethoxybenzyl protecting group at the end of a synthesis of the fumiquinazoline regioisomer 231 by a similar aza-Wittig protocol.¹¹¹

	Scheme 14 Reagents and conditions: i, NaH, THF, –5 °C, then 2-N3C6H4COCl, THF, rt; ii, Bu3P, C6H6, rt to 75 °C; iii, MeOH, Pyrex, h(254 nm).

The more complex (–)-fumiquinazolines A, B and I, 232–234, have also been synthesised by Snider's group by routes in which most of the effort was, understandably, devoted to constructing the 3-oxotetrahydro-1H-imidazo[1,2-a]indol-9-yl substituents.¹¹² Formation of the 2H-pyrazino[2,1-b]quinazoline-3,6(1H,4H)-dione moieties was left to the final stages of the synthesis, and involved methodology similar to that shown in Scheme 13 (cf. steps 222224181). In the case of fumiquinazoline A, for example, treatment of the precursor 235 with triphenylphosphine and bromine in the presence of triethylamine followed by aminolysis of the resulting 3,1-benzoxazine with piperidine and final cyclisation gave a mixture of the Cbz-protected product 236 and its C-4 epimer in overall yields of 49% and 14%, respectively. Removal of the Cbz protecting group from the former by hydrogenolysis over palladium completed the synthesis of (–)-fumiquinazoline A 232 in 90% yield. The overall yields for (–)-fumiquinazolines B and I from the appropriate precursors analogous to 235 were 42% and 52%, respectively.

The synthesis of ent-alantrypinone 237 communicated by Hart and Magomedov in 1999¹¹³ and summarised in last year's review in this series (cf. Ref. 28j) has been published as a full paper with experimental details and additional observations on aspects of the chemistry.¹¹⁴ A related article by these authors on the Morin rearrangement of sulfoxides such as 238 with trifluoroacetic acid described the isolation of bridged products such as 239 (21–35%) and 240 (22–25%), amongst others, as well as the putative mechanism of the process.¹¹⁵

The known alkaloid isogravacridonechlorine 241 has been isolated from root extracts of Ruta chalepensis.¹⁶

Oxidation of the potentially valuable antitumour alkaloid acronycine 242 and its nitro derivative 243 with meta-chloroperoxybenzoic acid in toluene has yielded the Baeyer–Villiger ring-expansion products 244 (20% and 30% yields respectively), as well as the hydroxylated compounds 245 (10%).¹¹⁶ The pyran ring remained unaffected. However, similar oxidation of the cis-diol 246 yielded only the triol 247 (29%), whereas oxidation of 246 with lead tetraacetate followed by treatment with sodium borohydride afforded the ring-D expanded hemiacetal 248 (30%), which could in turn be oxidised to the lactone 249 with pyridinium chlorochromate (25%). The products showed varying degrees of in vitro cytotoxicity towards L-1210 leukemia cells, with 248 and 249 being approximately as active as acronycine itself. In related studies, the acronycine-inspired benzo[b]xanthenone derivative 250 was found to be more active than the parent alkaloid in inhibiting the proliferation of the same cell line.¹¹⁷ Cognate work on benzo[b]-fused acridones and pyranoacridones, some of which has previously been discussed in this series of reviews, has since been patented.¹¹⁸

In efforts to improve the bioavailability of the potent antitumour agent glyfoline 251, several water-soluble derivatives 252–255 were prepared by derivatising the parent alkaloid.¹¹⁹In vitro tests proved that all the derivatives were less cytotoxic than glyfoline towards nasopharyngeal carcinoma (NPC) cells. However, only 253 (n = 3) showed significant in vivo activity in mice bearing NPC xenografts. Although the effective dose was about half that of glyfoline itself, this compound might still prove to be a useful prodrug in combination chemotherapies.

Footnotes

The IUPAC name for propargyl is prop-2-ynyl.

The IUPAC name for triflate is trifluoromethanesulfonate.