Recent advances in the synthesis of azaphenalene alkaloids. First enantioselective approaches

The azaphenalene alkaloids are biosynthesised and segregated by diverse insects of the Coccinellidae family (ladybirds) and are believed to play an important role in the defensive mechanism against their natural predators. The particular unique framework of these alkaloids, along with their potential in the field of biological pest control, has led to several research groups to develop synthetic sequences to prepare these compounds. The main purpose of the present review is providing an update of the more recent synthetic progress towards these alkaloids, including the pioneer enantioselective approaches to the chiral congeners.


INTRODUCTION
The Coccinellidae family is comprised of more than 5000 species worldwide of small coleopteran insects, commonly named ladybugs, ladybirds or lady beetles. 1 Due to their voracious appetite for aphids and other harvest-damaging insects, some coccinellids have played an important role in biological pest control. 2 The main defence mechanism of coccinellids against their natural predators (such as birds and ants) is reflex bleeding, consisting in the secretion of an orange fluid (haemolymph) when they feel attacked. 3 The strong smell and bitter taste of this fluid is associated with the presence of specific alkaloids. 1,4 Among them, the most characteristic are the azaphenalenes, which confer the bitter taste to these insects. 5 Up to now, nine different monomeric alkaloids with perhydro-9b-azaphenalene skeleton have been isolated (Figure 1), which differ in their relative stereochemistry, the oxidation state of the nitrogen atom and the presence and location of a carbon-carbon double bond. In addition, several "dimeric" alkaloids containing at least one perhydro-9b-azaphenalene unit have also been discovered. 6 Precoccinelline (and its N-oxide, coccinelline) and myrrhine are meso compounds. In the rest of this family members the reflection symmetry is broken either by the presence of an endocyclic double bond (hippocasine, its N-oxide, propyleine and isopropyleine) or because of the trans position of protons H-3a and H-9a (hippodamine and convergine). Propyleine and isopropyleine are in equilibrium and have not been independently isolated. Besides their application for mimicking the effect of ladybirds in crops, it has been also described that some azaphenalene alkaloids are non-competitive antagonists for nicotinic acetylcholine receptor 7 and some derivatives are potent receptor antagonists of the neurotransmitter serotonin. 8 All the azaphenalene alkaloids were isolated and characterized during the 1970's. The corresponding studies, along with the early syntheses, were comprehensibly reviewed by King and Meinwald in 1996 in an article dealing with the "defensive chemistry of coccinellids". 4 The main purpose of the present review is providing an update of the more recent synthetic progress towards these alkaloids, which includes the pioneer enantioselective approaches to the chiral congeners.

Biogenetic studies
On the basis of some experiments performed with 14 C-labelled sodium acetate, it was initially proposed that coccinelline was biogenetically derived from a -polyketoacid produced by linear combination of seven acetate units, 9 but subsequent studies from the same laboratory with 3 H 3 -labelled stearic acid led to the formulation of the more plausible hypothesis depicted in Scheme 1. 10 Scheme 1. Biogenetic hypothesis for coccinelline. 10

Chronology
Probably attracted by the peculiar structure and apparent exclusiveness of their origin, along with their potential application in pest control, immediately after their isolation from various ladybird species, several groups undertook synthetic studies, which culminated in the total synthesis of all the known monomeric azaphenalene alkaloids, allowing to confirm their structure and relative stereochemical assignment. The plot displayed in Figure 2 shows the chronological evolution of the synthetic publications from 1976 to 2018. Analysis of this plot reveals that, after the early syntheses, there was essentially a lack of new contributions, while, in the last years, the interest for these compounds has resurfaced. All the alkaloids presenting a free amine functionality have been efficiently converted into the corresponding N-oxides and, hence, a unique synthetic sequence allows the preparation of each couple precoccinelline/coccinelline, hippocasine/hippocasine N-oxide and hippodamine/convergine. On the other hand, the selective preparation of propyleine or isopropyleine is not possible since, as already mentioned, these two enamines are in equilibrium. Some of the works included in the plot describe an improvement or an alternative preparation of an intermediate previously transformed into a targeted alkaloid. Up to 2010, all the reported syntheses ended up with achiral or racemic products and, only more recently, some enantioselective approaches have been described. Moreover, the syntheses of three "dimeric" azaphenalene alkaloids have been also disclosed.

Early approaches to achiral and racemic alkaloids
As stated before, the early syntheses of azaphenalene alkaloids were covered in a previous review 4 and will not be discussed here in detail. However, in order to provide a quick overview of these initial approaches, a summarized compilation is shown in Table  1, where the key precursors, total number of steps, global yields and corresponding literature references are supplied. Most of these syntheses started from symmetric substrates (entries 1-5) and, in some cases, by splitting of the sequence, a common intermediate was derived to different alkaloids (entries 1 and 5). The total number of steps from commercially available materials was in the range 8 to 15 and the highest global yield among the early total syntheses was 21% and corresponds to the last reported within this group (entry 7). The formal synthesis of the couple precoccinelline/coccinelline (entry 4) is merely an improvement of one of the steps in a previous synthesis (entry 3). All the final products were either meso forms or racemates, even in the case where the starting material was homochiral (entry 7).

Recent approaches to achiral and racemic alkaloids
In 2002, Takahata and collaborators disclosed a new methodology to prepare ketone 9, 18 a key intermediate in the previous syntheses of precoccinelline of Ayer and Furuichi, 12 Stevens and Lee 13 and Mueller et al., 15a,d which preparation had been later on improved by Langlois et al. 14 and Royer et al. 17 The synthetic sequence (Scheme 2) was initiated by treatment of glutaraldehyde, 1, with B-allyldiisopinocamphenylborane, followed by oxidation with hydrogen peroxide in basic medium to furnish diol 2. Aminocyclization of the corresponding ditosylates with benzylamine gave a mixture of the transand cis-2,6-diallylpiperidines, 3, in 64% and 14% yield, respectively, for the three steps. An exchange of N-protecting groups from benzyl to carbamate, followed by iodocarbamation produced oxazinone 5, which was oxidized under Wacker conditions to ketone 6. A two-step treatment consisting of reduction of the oxazolidinone and then N-protection as the Boc derivative furnished the allylpiperidine 7 in 81% yield. Oxidative cleavage of 7 and subsequent Wittig olefination provided an -unsaturated aldehyde, which was reduced to the dicarbonyl compound 8 in 59% yield for the three steps. Ketoaldehyde 8 was treated with TFA in order to liberate the amine and submit it to an intramolecular Mannich-type cyclization leading to the tricyclic ketone 9, precursor of precoccinelline. Overall, this preparation of 9 required 14 steps from glutaraldehyde and the In 2004, Stockman and co-workers described a new synthesis of (±)-hippodamine, through a two-directional strategy starting from the known symmetric alcohol 10 (Scheme 3). 19 Mitsunobu substitution of the alcohol by phtalimide furnished diene 11 in 97% yield. Homologation of 11 to diester 12 was best accomplished by osmium tetroxide catalysed oxidative cleavage followed by a double Horner-Wadsworth-Emmons (HWE) olefination. After extensive experimentation, the conditions of the planned key step consisting of a tandem N-deprotection/intramolecular double conjugate addition were conveniently adjusted to obtain the bicyclic amine 13 in good yield. To be effective, the next Dieckmann condensation required azeotropic removal of EtOH. Subsequent deethoxycarbonylation yielded ketone 14, which was olefinated to 15. The hydrogenation of alkene 15 was attempted in the presence of different additives and it was found that mesitylenesulfonic acid gave the best stereoselectivity. The synthesis of (±)hippodamine was accomplished in 9% overall yield. On the other hand, stereoselective hydroboration of 15 with 9-BBN furnished alcohol 16, which was converted to the corresponding iodide. Without isolation, this last intermediate was reduced with zinc delivering (±)-2epihippodamine, in 16% overall yield.
In 2008, a publication from the same laboratory described the introduction of some changes in the sequence to improve the synthesis (Scheme 4). 20 The first modification concerned the oxidative cleavage of 11. This reaction produced large amounts of toxic osmium waste and, besides, the instability of the subsequent dialdehyde forced its immediate elaboration through the HWE reaction. As an alternative to convert diene 11 to diester 12, a double cross-metathesis with ethyl acrylate using Hoveyda-Grubbs second generation catalyst (H-G II) was accomplished in 88% yield and, hence, it was more efficient than the previous transformation. The second modification intended avoiding the use of the phtalimide protecting group and was achieved by developing a new tandem reductive amination/double intramolecular Michael addition on the previously known ketone 20. With this variation the original 44% overall yield of the key bicyclic intermediate 13 from commercially available compounds was increased to 46%.

Scheme 3. Synthesis of (±)-hippodamine and (±)-2epi-hippodamine by Stockman et al. 19
Scheme 4. Improvements on the synthesis of (±)-hippodamine and (±)-2epi-hippodamine by Stockman et al. 20 In 2006, Hsung and Gerasyuto described a stereodivergent synthesis of precoccinelline and hippodamine (and their N-oxides) using an aza-[3+3] annulation strategy (Scheme 5). 21 The synthetic route started with the alkylation of alkyne 21 with 1,3-dibromopropane, followed by chemoselective hydrogenation that furnished alkene 22. Reductive alkylation of4-methyl glutarimide with the magnesium reagent derived from 22 provided lactam 23 as a cis isomer. This lactam was efficiently converted into the tioether 24 by treatment with Lawesson's reagent and consecutive alkylation with α-bromo methyl acetate. Eschenmoser sulphide contraction led to the vinylogous urethane 25, which was desilylated by treatment with TBAF. Oxidation of the allylic alcohol 26 furnished aldehyde 27 as an approximately 7:1 mixture of the cis and trans isomers. The crucial aza-[3+3] annulation was accomplished by treatment of 27 with piperidinium trifluoroacetate in 51% yield with anti relative configuration at the ring junction. Due to its low stability, the intermediate diene 28 was directly subjected to Pd(OH) 2 catalysed chemoselective hydrogenation in a one-pot protocol that delivered the -unsaturated ester 29 in 43% yield. A second hydrogenation using Adams catalyst afforded 30 and 31 in a 2:1 ratio, which correlate with precoccinelline and hippodamine, respectively. The two isomers were resolved by selective alkaline hydrolysis of the equatorial ester 30, since the axial isomer 31 was unreactive towards aqueous KOH. Decarboxylation of 32 using Barton's methodology 22 led to precoccinelline. The synthesis of hippodamine required an additional step to convert ester 31 into its equatorial epimer. The yield of the sequence from 21 to the common intermediate 29 was 13.4%. Unfortunately, the last steps did not work very efficiently, furnishing the target alkaloids coccinelline and hippodamine in 2.8% and 2.2%, respectively , namely 5% overall.
Scheme 5. Synthesis of precoccinelline and (±)-hippodamine by Hsung and Gerasyuto. 21 One year later, the same research group published an article describing the synthesis of myrrhine by alternative elaboration of the intermediate 28 (Scheme 6). 23 This endeavour required the epimerisation of C-6a, which was achieved by aromatization of the dihydropyridine ring through oxidation with dichlorodicyanoquinone (DDQ), followed by stereoselective reduction of the pyridinium salt 34. The catalytic hydrogenation, best performed with Adams catalyst in AcOH, was concomitant with epimerisation of the -ester stereogenic centre. This epimerisation was not synthetically inconsequent because the successive decarboxylation of the equatorial carboxylic acid did not work well.
Scheme 6. Synthesis of myrrhine by Hsung and Gerasyuto. 23 In 2010, Spring and co-workers published a new synthesis of myrrhine (Scheme 6). 24 Their starting material and strategy were identical to those in the previous work of the Stockmann group, but they developed a methodological alternative, which allowed to construct the tricyclic framework from an acyclic precursor in a single operation. The appropriate substrate 36, analogous to 12, was prepared from alcohol 10 in a similar manner. Their main challenge was to find a Lewis acid able to deprotect the nitrogen atom and promote the double conjugate addition and Dieckman cyclisation cascade. This objective was accomplished by the use of 0.5 equivalents of tin(II) triflate, delivering ketoester 37 in 72% yield, with the relative configuration of myrrhine at the ring junctions, in contrast with the relative stereochemistry of intermediate diester 13 of the Stockmann sequence. According to the authors explanation, based on some experimental results, when using Sn(OTf) 2 in catalytic amounts the enolate formed after the first conjugate addition undergoes a Dieckmann condensation before the second conjugate addition. The synthesis of myrrhine from ketoester 37 was completed through three additional steps, including ester hydrolysis/decarboxylation, Wittig olefination and catalytic hydrogenation under acidic conditions to direct the facial selectivity. The synthesis of myrrhine was thus completed in 7 steps and 6.5% total yield.
In 2014, Kuznetsov and collaborators described the synthesis of (±)-hippocasine, second after the early approach of Mueller, 15d starting from 4-picoline, 39, (Scheme 8). 25 Their strategy encompasses a one-pot nucleophilic alkylation/allyl boration, an intramolecular allylic amination and a ring-closing metathesis (RCM) reaction. The organolithium nucleophile 40 was generated by reaction of t BuLi with the iodide precursor, prepared in two steps and 63% yield from commercially available materials. Sequential incorporation of 4-picoline and triallylborane to an ethereal solution of 40 resulted in the formation of the 2,6-trans tetrahydropyridine 42 in 74% yield. Protection of the nitrogen, followed by acetal hydrolysis and then addition of vinylmagnesium bromide, delivered the allylic alcohol 43 as a unique isomer, provided that the reaction was run at -90ºC. Acylation of alcohol 43 and removal of the Boc group led to the free amine 44, which by palladium promoted cyclisation was converted into the azabicyclic isomers 45 and 46 with low diastereoselectivity. The relative configuration of hippocasine correlates with that of the minor isomer 45, but complementary assays with other palladium or iridium catalysts did not improved these results. After separation, diene 45 was treated with Grubbs second generation catalyst (G II) to furnish the tricyclic amine 47. Partial hydrogenation of 47 afforded (±)-hippocasine. The overall yield was 7.4%. Longer reaction times produced exhaustive hydrogenation, affording (±)-2epi-hippodamine, although the yield of this transformation was not indicated.

Enantioselective approaches
The first enantioselective synthesis of an azaphenalene alkaloid was developed by Katsumura and co-workers in 2013 (Scheme 9). 26 They devised a target-oriented strategy addressed to hippodamine which relied on an asymmetric azaelectrocyclisation reaction. This methodology furnished oxazolidine 52 from amino-alcohol 48, 27c vinylstannane 49 and (Z)-2-iodo-4oxobuenoate, 50, in a one-pot operation, as a single stereoisomer, in 81% yield. The starting materials 49 and 50 (readily available from commercial precursors) 27a,b provided the appropriate carbon framework, while the amino-alcohol 48 provided the nitrogen atom and acted as sacrificial chiral auxiliary. Chemo-and stereoselective hydrogenation of the -unsaturated ester 52, followed by a three-step conversion of the ester into a methyl group, delivered compound 54 in a quite efficient fashion. The subsequent nucleophilic vinylation led to a 1:7 mixture of diastereoisomers 55, the relative configuration of the major one correlating with hippodamine. The observed diastereoselectivity was attributed to the directing effect of the alkoxide coordinated to the Grignard reagent in the intermediate iminium ion, whose indanol moiety is orientated to the opposite face of -alkyl chain. Oxidative removal of the indanol moiety and then protection of the amine with Cbz allowed the separation of the minor isomer, furnishing pure 56 in good yield. A sequence of four conventional reactions converted the vinyl substituent into a methyl ketone (57), which was next involved in the intramolecular Mannich reaction leading to the tricycle 59. Finally, deoxygenation of the acetal was accomplished via dithioacetalisation-desulphuration, as in Ayer's synthesis of racemic hippodamine. 11 The synthetic hippodamine was levorotatory with a measured [α] D value of -1.2 (c 1.1, CHCl 3 ) and had the same absolute configuration as the alkaloid extracted from the insects. The optical activity of natural hippodamine had not been described before, but its absolute configuration was established through X-ray analysis of convergine hydrochloride, which showed optical activity values very small and strongly solvent dependent, even with changes of the sign. 28 The overall yield of (-)-hippodamine from 48, 49 and 50 was 10.3%.
In 2014, Snyder and collaborators published an article describing the second enantioselective approach to azaphenalene alkaloids, making use of a chiral pool strategy. 29 They accomplished the total or formal syntheses of several monomeric alkaloids from the common intermediate 65, which contains the complete carbon skeleton of the targets (Scheme 10). The synthesis started with the commercially available amino-alcohol 60 encompassing a key stereogenic centre that was intended to induce diastereoselectivity in the sequential transformations. Although the authors developed two alternative routes to convert 60 into compound 65, only the most practical of them is displayed in Scheme 10. After protection of the alcohol in 60 as the tert-butyldimethylsilyl ether, a copper-mediated allylation provided alkene 61. Oxidative cleavage of the olefin in 61 led to ketone 62 that, without isolation, was submitted to a second nucleophilic allylation and then another oxidative cleavage to produce hydroxyketone 63 as a synthetically inconsequent 2:1 mixture of diasteroisomers in good total yield. This mixture was dehydrated to enone 64 by trifluoroacetate formation and DBU-promoted elimination. The preparation of 65 was concluded by stereoselective reduction of enone 64 with aluminium tris(2,6-diphenylphenoxide) (ATPH) in combination with L-selectride, followed by acid promoted desilylation, in 79% yield for the two steps and with a diastereoisomeric ratio of 8:1.  1, CHCl 3 ), allowed to establish the absolute configuration of the natural occurring material, which was reported to be levorotatory too. 30 Reduction of the propyleine/isopropyleine mixture with NaBH(OAc) 3 produced precoccinelline and hippodamine in a 3.7:1 ratio and a total yield close to 80% that was not precisely determined due to the difficulties associated with the isolation of these compounds, which are highly volatile. 1,20 Alternatively, α-carbonyl oxidation of 65 with BzONHMe·HCl (Scheme 12), followed by an analogous sequence of steps as above, by Snyder et al. 29 In 2015, Fustero and co-workers developed the first asymmetric synthesis of (+)-2epihippodamine and the third of (-)-hippodamine. 31 They employed a chiral N-sulfinyl amine both as a chiral inducer and nucleophilic nitrogen source (Scheme 13) and applied the strategy reported by Stockman for the synthesis of the same alkaloids in racemic form (Scheme 3). 19 The starting material was the readily available symmetric ketone 19 (see Scheme 4). Its condensation with (R)-N-tert-butanesulfinamide, followed by treatment with sodium borohydride, led to the reductive amination product 72 bearing the chiral auxiliary. A double cross-metathesis with ethyl acrylate furnished diester 73. This substrate was submitted to a desymmetrisation process by means of an intramolecular aza-Michael reaction that originated the first ring of the alkaloid. Although this transformation produced a mixture of three isomers, 95% of the product presented S configuration at C-2 and the cis/trans relative geometry C2/C-6 is synthetically inconsequent. Thus, after the second intramolecular aza-Michael reaction, due to its inherent C-2 symmetry, both isomers deliver the same bicyclic product 75. In our laboratory, we developed a new strategy for the stereoselective synthesis of azaphenalene alkaloids starting from the chiral material 76, readily available in either enantiomeric form through the palladium-catalysed asymmetric allylic alkylation of glutarimide in the presence of the DACH-naphtyl Trost ligand 32 (Scheme 14). Our strategy relied on an iterative methodology encompassing two nucleophilic allylations and two RCM processes. The starting acylaminal (1ʹS)-77, easily prepared from glutarimide in two steps and 90% yield, 32a was acetylated and, without isolation of the intermediate acetate, trimethyl(2-methylallyl)silane and a catalytic amount of TMSOTf were added, resulting in the formation of diene 78 in 97% yield and 10:1 diastereoisomeric ratio. A first RCM reaction provided, after desilylation, alcohol 79 as a pure isomer in 86% yield for the two steps. The relative configuration of 79 was established by X-ray diffraction analysis. Oxidation of alcohol 79 by treatment with Dess−Mar n periodinane (DMP) furnished the corresponding aldehyde, which slowly decomposed to unidentified products, and, hence, was immediately submitted to the next alkenylation step. This transformation was better accomplished through a Wittig ethylenation that provided a synthetically inconsequent mixture of Z/E isomers. The next endeavour was the reduction of the lactam to hemiaminal, in order to generate an iminium ion suitable for the second nucleophilic allylation. After extensive experimentation, it was found that a one-pot protocol consisting of reduction with Red-Al, followed by addition of allylmagnesium bromide, delivered triene 81 in 83% yield. A second RCM reaction then furnished diene 82, which was amenable for partial or total hydrogenation. Although its full characterization data were not described, compound 82 had been previously prepared as a racemate and converted into the meso alkaloids precoccinelline and coccinelline. 12,15a Our synthetic efforts were therefore devoted to the partial hydrogenation that should provide a chiral alkaloid. This transformation was successfully accomplished by treatment of 82 with hydrogen (2 atm) in the presence of Pd/C and acetic acid in methanol as the solvent for 48 h. The first synthesis of (-)-9aepi-hippocasine, [α] −0.71 (c 0.27, CHCl 3 ), was completed in 11 steps from glutarimide and 26% total yield. The low specific rotation is in agreement with the values observed for related azaphenalene alkaloids. 28

"Dimeric" alkaloids
As anticipated above, several alkaloids isolated from the same natural sources present a more complicated structural framework, where at least one half is an azaphenalene unit ( Figure 3). In exochomine and chilocorines A and B, the azaphanalene moiety correlates with hippodamine, and the other half is a pyrroloquinolizinone attached through different linkages. Psylloborine A and isopsylloborine A present truly dimeric azaphenalene structures and are tautomeric enamines. Up to date, there are only two publications concerning synthetic efforts toward these "dimeric congeners" and both come from the group of Snyder. Initially, the synthesis of psylloborine A and isopsylloborine A was attempted by direct dimerization of two azaphenalene units derived from propyleine/isopropyleine (Scheme 15). 29 However, these efforts resulted in the formation of a non-natural regioisomer that was named as psylloborine B.
Scheme 15. Synthesis of psylloborine B by Snyder et al. 29 In view that the putative natural monomeric precursors afforded only a non-natural analogue, an "intramolecular dimerization" strategy was devised (Scheme 16). The starting material 60 was the same as for the monomeric alkaloids. The alcohol was protected and then allylated following the same methodology as above (Scheme 10). In the present sequence, the oxidative cleavage of 85 led to an aldehyde, which, without isolation, was treated with methyl (triphenylphosphoranylidene)acetate furnishing the unsaturated ester 86. Ester 86 was converted into the phosphonate 87 in a one-pot protocol, including catalytic hydrogenation, followed by condensation with dimethyl methylphosphonate. In a parallel sequence, aldehyde 88 was prepared by oxidation of the previously prepared key intermediate 65 (see Scheme 10). The reaction between phosphonate 87 and aldehyde 88 assembled the two halves of the target. The carbamate protecting groups in the resulting enone 89 were differentiated, taking advantage of neighbouring group participation, upon treatment with TFA at -78ºC, transforming one of them into a base-labile carbamate through cyclisation onto the enone and leaving intact the other one.

Scheme 16. Synthesis of psylloborine A and isopsylloborine A by Snyder et al. 29
The synthesis of the key intermediate 90 was then completed through oxidation of the alcohol to the corresponding aldehyde and subsequent one-carbon homologation by reaction with a sulphanyl phosphonate. Treatment of 90 with 1,1,3,3tetramethylguanidine (TMG) triggered a cascade sequence involving carbamate cleavage, enone regeneration, condensation, enamine equilibration to the exocyclic isomer and a terminating Michael addition. The resulting product 91 was obtained as a mixture of two diastereoisomers embodying the first azaphenalene moiety, which presented low stability and was carried forward immediately to a second cascade of events. To this end, 91 was treated with TFA to remove the remaining Boc group and then the free amine was heated to promote condensation, Michael closure and a terminating Mannich reaction. This second cascade sequence delivered the expected heptacyclic product 92 in 15% yield for the three steps (first cascade, Boc removal, second cascade). Finally, reaction of 92 with Na/Hg amalgam transformed it into psylloborine A in 46% yield. This alkaloid could be converted into isopsylloborine A by TFA treatment, but repeated attempts to purify the crude material were unsuccessful due to its instability, and had to be characterised as such. The total synthesis of exochomine was published in 2016. 33 Conversely to the later alkaloids, the two moieties of exochomine present different skeletal framework and hence were prepared through unrelated synthetic routes. The starting material for the azaphanalene moiety (Scheme 17) was the known N-proptected alcohol 93. 34 When alcohol 93 was submitted to Swern oxidation, followed by Julia-Kocienski olefination and then catalytic hydrogenation, the elongated ketal 95 was produced in 66% overall yield. A Cu(I)-promoted allylation of 95 furnished alkene 96, along with its cis diastereoisomer, which could not be separated. Oxidative cleavage of the double bond in 96 produced the corresponding aldehyde, concomitant with partial deprotection of the ketone. The crude product was then treated with TFA, effecting Boc deprotection and condensation of the free amine with the ketone (and/or remaining ketal) to form an enamine, which reacted with the aldehyde furnishing the expected tricyclic, unsaturated iminium ion, which was reduced in situ with Hantzsch ester to produce, after acid treatment, the salt 98. Addition of KCN to 98 delivered nitrile 99 as a mixture of two isomers. Despite the individual steps from 96 to 99 were monitored through 1 H-NMR analysis, the involved intermediates were not isolated and the sequence was carried forward with crude reaction products. Reduction of the mixture of nitriles 99 with LiAlH 4 and then treatment with concentrated H 2 SO 4 led to the isolation of aldehyde 100 as a single isomer in good yield.

Scheme 17.
Synthesis of the azaphenalene moiety of exochomine by Snyder et al. 29 The synthesis of the pyrroloquinolizinone moiety (Scheme 18) started from ketone 101, which was prepared from ethyl L-glutamate, according to a previous literature reference. 35 The carbonyl group in 101 was protected as the dithiolane and then ester 102 was homologated to 103 by reduction to the aldehyde followed by Wittig olefination. The carbon-carbon double bond was reduced by in situ formed copper hydride and the resulting ester 104 was coupled with the aldehyde 100 by means of an aldol reaction. Without isolation of the aldol product, an acid-promoted, intramolecular Friedel-Crafts reaction furnished ketone 105 with the "dimeric" skeleton of the targeted alkaloid. The overall transformation of 104 to 105 was accomplished in 32% yield. The remaining steps to exochomine included: regioselective reduction of the enone, nucleophilic methylation, dehydration and cleavage of the dithiolane to deprotect the masked carbonyl group.
Scheme 18. Preparation of the pyrroloquinolizinone moiety and completion of the synthesis of exochomine by Snyder et al. 33

CONCLUSIONS
The peculiar structure and putative application in pest control of azaphenalene alkaloids make these compounds attractive targets for synthetic chemists. The early syntheses were mainly motivated by the corroboration of their structural assignment and the investigation of the biogenetic pathways. Then, there was essentially a lack of new contributions on the field, but, in the last years, the interest in these alkaloids has resurfaced, giving rise to several new, imaginative synthetic strategies. Moreover, among the recent reports, the first enantioselective approaches to the chiral congeners have been published, as well as pioneer syntheses of some "dimeric" members of the family. However, there is still a lot of room for improvement, particularly in terms of efficiency and stereoselectivity, while the total synthesis of some "dimeric" azaphenalene alkaloids remains undisclosed.