The chemistry of isoindole natural products

Introduction

Isoindole (2H-isoindole, 1), known since more than a century, consists of a fused benzopyrrole ring system and constitutes the regioisomer of the abundant 1H-indole heterocycle. The fully reduced member of the isoindole family is termed isoindoline (2,3-dihydro-1H-isoindole, 2). Formal oxidation to the 10π-system leads to isoindole (1), which is usually only stable when the labile ortho-quinoid structure is embedded in a π-system [1]. Incorporation of additional oxygen gives the isoindolinone (1,3-dihydro-2H-isoindole-1-one, 3) and phthalimide (1,3-dihydro-2H-isoindole-1,3-dione, 4) substitution pattern.

The isoindole structure has attracted scientists for decades and can be found in several natural and pharmaceutical compounds [2,3]. A number of structures were explored over the years and promising drug conjugates such as 5–11 could be developed (Figure 1).

Compared to the synthesis of indoles, where a number of named-reactions have been reported, only conventional methods are used for the related isoindole motif. For the construction of this rare skeleton inter- and intramolecular Diels–Alder reactions are one of the most powerful methods. This was also exemplified by medicinal chemists from AstraZeneca in the manufacturing route to an mGluR2 positive allosteric modulator [4]. In 2012, a programmable enantioselective one-pot synthesis of isoindolines was reported by Waldmann [5]. Several other strategies were pursued for the synthesis of isoindoline-type structures and the synthesis, chemical and spectroscopic properties of this substance class were reviewed elsewhere [6-8].

In the late 1950s, thalidomide (8), a phthalimide-based drug used by pregnant women against morning sickness, became the most infamous drug in history and caused thousands of fatal casualties as well as numerous severe birth defects. Modification of the phthalimide core led to the approval of lenalidomide (9) [9] in 2004 and pomalidomide (10) in 2013 by the Food and Drug Administration (FDA) as drugs against multiple myeloma. The phosphodiesterase 4 (PDE4) inhibitor apremilast (11), which lacks the glutarimide is currently in phase III clinical trials.

The first naturally occurring isoindole, 6-methoxy-2,5-dimethyl-2H-isoindole-4,7-dione (18), was isolated from the sponge Reniera sp. in 1982 [10]. The postulated structure was elucidated through extensive NMR studies and unambiguously confirmed by a four-step synthesis. In 1991, a more concise and elegant route to this antimicrobial metabolite was established by Schubert-Zsilavecz [11]. The reaction was initiated by heating paraformaldehyde (13) and sarcosine (14) in the presence of benzoquinone 16. This transformation proceeds via a 1,3-dipolar cycloaddition between the in situ formed azomethinylide 15 and the benzoquinone 16 to directly give 17 (Scheme 1). Spontaneous oxidation of the so-obtained cyclization adduct generates isoindole 18.

Isoindoles have also found application as dyes. Pigment yellow 139 (12), which is sold by BASF as Paliotol^® Yellow K 1841 belongs to the class of highly resistant and effective 1,3-disubstituted isoindoline dyes. Recently, the use of isoindoles as red to near-infrared fluorophores was reported [12]. Another interesting isoindole-based dye, 25, arises from the condensation of primary amines with o-diacylbenzene 19 (Scheme 2) [13]. After initial formation of 20, isomerization to 21 and 22 can occur through a sequential dehydration–hydration process. Dimerization of 21 and 22 generates 23, the substrate for a formal retro-Aldol reaction. Loss of formaldehyde gives 24, which is spontaneously oxidized to the intensive blue-violet pigment 25. This reaction sequence is characterized by its high sensitivity and has found application as a marker in analytical chemistry (e.g. staining of primary amines) [14,15].

The following review section highlights synthetic and biochemical aspects of selected isoindole-type natural products. The presented results are not meant to be exhaustive but should give a better understanding about the structurally and biologically most attracting structures. We intend to inspire today´s chemists and are convinced that these natural products have a huge potential for the development of new chemistry.

Review

Indolocarbazoles. Staurosporine (26) was the first member of the indolocarbazole alkaloid family to be discovered by Ōmura from Streptomyces staurosporeus at the Kitasato Institute in 1977 [16]. Over the past 35 years, more than 60 natural indolocarbazole compounds have been isolated from several bacteria and marine invertebrates either as their glycosides (26–29) or aglycones (30–33) [17]. Based on the number of glycosidic bonds linking the carbohydrate moiety to the isoindole framework, the latter can be divided into two subclasses. The first comprises two linkages, exemplified by (+)-staurosporine (26) and (+)-K252a (27), whereas the second class contains only one glycosidic bond as found in rebeccamycin (28) and holyrine A (29). Selected members of the indolocarbazole alkaloid family are depicted in Figure 2.

The strong antimicrobial activity against fungi and yeast of 26 was simultaneously reported with its isolation [16]. Moreover, with IC₅₀ values in the low nanomolar range (1–20 nM), staurosporine (26) is one of the most potent protein kinase inhibitors to date [18]. It took nine years to identify the crucial protein kinase inhibitory effects of 26 [18]. Solving the crystal structure of the catalytic subunit Cα of cAMP-dependent protein kinase bound to 26 gave pivotal insights in the binding mode [19]. Staurosporine (26) targets the active site of the adenosine-binding pocket of most protein kinases. This is achieved mimicking several aspects of the adenosine moiety, by induced-fit structural changes and the conformational flexibility of the enzyme residues (Figure 3). Together with several related analogues, 26 displays significant cytotoxic and antiproliferative effects [19].

Due to the key role of protein kinases in cellular signaling and their association with cancer, a tremendous effort in the development of selective protein kinase inhibitors was undertaken. This resulted in the discovery of the anticancer agent imatinib (Gleevec, 34) by rational drug design (Figure 4). Midostaurin (35), a semisynthetic derivative of staurosporine (26), is currently in clinical trials for the treatment of acute myeloid leukemia [20].

Although rebeccamycin (28) and its congeners only differ in the lack of the second glycosidic linkage, a completely different mode of action is effective. By stabilizing the topoisomerase I DNA-cleavable complex they prevent the replication of DNA and act as efficient antitumor compounds [21].

The extraordinary bioactivity of indolocarbazoles has drawn a lot of attention from chemists and biologists. The biosynthesis of staurosporine (26) and rebeccamycin (28), both being the most prominent representatives of this distinct class of natural products, has been studied extensively and reviewed by Knölker and Ōmura [17,22]. The biogenesis of staurosporine (26) is outlined in Scheme 3. Biosynthetically, the indolocarbazole scaffold is derived from two fused tryptophan molecules, whereas the sugar moiety originates from glucose and methionine [23]. In 1996, Steglich isolated lycogalic acid A (38), also known as chromopyrolic acid, along with staurosporinone (30) and traces of arcyriaflavin A (31) from the extracts of Lycogala epidendrum [24]. This finding and results from Hoshino, which demonstrated that 38 is derived from L-tryptophan (36) [25], indicate a close biosynthetic relationship between these alkaloids. After additional experiments, the authors concluded that lycogalic acid A (38) is the biosynthetic key intermediate for the biogenesis of indolocarbazoles. This hypothesis was verified through gene disruption studies and the isolation of putative intermediates [26-28]. The formation of the isoindole framework is initiated by the oxidation of L-tryptophan (36) to imino indolepyruvic acid 37, a process, which is catalyzed through the enzyme StaO. The subsequent condensation of two molecules of imine 37 is catalyzed by the oxidase StaD and gives lycogalic acid A (38) [26-28]. StaP and StaC convert 38 into staurosporinone (30) via an intramolecular aryl–aryl coupling and an oxidative decarboxylation [29,30]. Formation of the N-glycosidic bonds is carried out by StaG/StaN and after methylation, 26 is obtained [27].

As already mentioned before, indolocarbazoles are characterized by their excellent biological activities and unusual architecture and hence drawing an enormous interest from the synthetic organic chemistry community. This led to several syntheses of the natural occurring aglycone staurosporinone (30) [31-35] and culminated in the first total synthesis of staurosporine (26) by Danishefsky in 1995 [36]. The elegant syntheses of a series of closely related indolocarbazoles, staurosporine (26), K252a (27), RK-286c (49), MLR-52 (20) and staurosporinone (30), which were elaborated by Wood in 1996 and reviewed by Knölker in more detail [37], will be discussed below. Based on their previous results [38], the common intermediate 48 was recognized as an ideal branching point. The synthetic route to this component is characterized by the beautiful application of Rh(II)-catalyzed C–H and OH insertion reactions (Scheme 4). The preparation of both enantiomeric furanose building blocks commenced with the Rh₂(OAc)₄-catalyzed OH insertion of 39, respectively 40 into the α-diazo-β-ketoester 40. A tandem [3,3]/[1,2]-rearrangement cascade, followed by reductive ozonolysis and acid-promoted cyclization afforded (+)-41 and (−)-41 in an overall yield of 60% and 46%, respectively. For the synthesis of staurosporinone (30) and its 3,4-dimethoxybenzyl (DMB)-protected derivative 45, a ruthenium-catalyzed C–H insertion/electrocyclization cascade using 2,2’-bisindole 44 and diazolactams 43a/b was developed. The necessary cycloglycosidative coupling of 45 and (+)-41 was promoted by camphorsulfonic acid to give the DMB-protected K252a [(−)-46] as a 2:1 mixture, slightly favoring the desired diastereomer. Cleavage of the N-protecting group gave (−)-K252a [(−)-27] as its unnatural enantiomer. With eleven synthetic operations and a longest linear sequence of only seven steps, this route is highly convergent and employs a novel Rh-carbenoid-mediated formation of 30 as its key step. The synthesis of the natural enantiomer of (+)-K252a [(+)-27] was carried out in an analogous fashion using (R)-(−)-1-nonen-3-ol (42) instead of (S)-but-3-en-2-ol (39) to give the enantiomeric coupling partner of 45. Although no yields for the following coupling were reported, multigram quantities of optically pure (+)-46 could be obtained [37]. Further reduction and Moffat oxidation gave 47, which was believed to be the direct precursor for the common intermediate 48. Lewis acid catalyzed Tiffenau–Demjanov-type ring expansion of 47 gave 48 as a single product.

With α-hydroxy ketone 48 in hand, the stage was set for further modifications to access 26, 27, 49 and 50 (Scheme 5). During their attempts to methylate 48, an unprecedented benzilic acid-type rearrangement (ring contraction) was discovered. Exposure of 48 to copper(I) chloride in methanol furnished (+)-46 in 95% yield, which after removal of the protecting group gave (+)-K252a (27). Staurosporine (26) could be synthesized in five steps from 48 and in a longest linear sequence of 19 steps. For the preparation of (+)-RK-286c (49) and (+)-MLR-52 (50), ketone 48 was reduced and regioselectively methylated at the C3’ position. Cleavage of the protecting group gave 49, whereas 50 was obtained after dehydration using Martin’s sulfurane, dihydroxylation and a final deprotection step.

Macrocyclic polyketides: The first cytochalasan alkaloids, cytochalasin A (51) and B (52), originally named phomins [39], were isolated in 1966 [39-42]. Since then, the number of natural products belonging to this family has increased to over 80. Some representative members are depicted in Figure 5 [43]. Cyctochalasan alkaloids display a wide range of biological properties, such as cytotoxic, antimicrobial, antiviral and phytotoxic activities and were reviewed in detail by Hertweck in 2010 [44]. Structurally, they consist of one amino acid and a highly substituted hydroisoindolone moiety fused to a macrocyclic ring (ring size 9–14). In certain cases, the ring system can be oxidized to an unusual cyclic carbonate as observed for cytochalasin E (53). Depending on the amino acid incorporated in the isoindolone moiety, the cytochalasans can be further subdivided into cytochalasins [41] (phenylalanine), chaetoglobosins [45,46] (tryptophan), aspochalasins [47] (leucine), pyrichalasins [48] (tyrosine) or alachalasins [49,50] (alanine).

The biosynthesis of cytochalasans was established on the basis of various isotope labeling experiments using cytochalasin B (52) as a model system [51-55]. It was hypothesized that the carbon backbone, which is connected to an amino acid, most likely originates from a polyketide synthase (PKS)/nonribosomal peptide synthetase (NRPS) hybrid machinery [56]. The discovery of a gene locus for a PKS-NRPS synthetase of the chaetoglobosin producing fungus Penicillium expansum led to a more sophisticated biosynthetic insight of chaetoglobosin A (57) (Scheme 6). The stepwise assembly of the nonaketide 61 is realized from activated tryptophan (36), one acetyl-CoA (59) starter and eight malonyl-CoA extender (60) units. The remaining methyl groups are believed to be installed by S-adenosyl methionine (SAM). An intramolecular Aldol condensation generates the pyrrolinone 62, which reacts via an [4 + 2]-cycloaddition to prochaetoglobosin I (63) [56]. Recently, gene disruption studies of Chaetomium globosum revealed the enzymes involved in the final oxidative transformations leading to chaetoglobosin A (57) [57].

The diverse biological activities and unique molecular architecture of the cytochalasan alkaloids have made them a famous synthetic target. Total syntheses of several members of the cytochalasans were reported, which were reviewed by Hertweck and Bräse [44,58]. In line with the focus on the construction of the isoindole component, one can identify the Diels–Alder reaction as the most popular strategy. A linear, biomimetic synthesis using a late stage intramolecular Diels–Alder reaction was implemented in the Stork synthesis of cytochalasin B (52) [59] and the Thomas synthesis of cytochalasin H, D, G and O [60-63]. An intermolecular reaction was reported for the synthesis of aspochalasin B (58) by Trost and Vedejs [64-67]. The synthesis of cytochalasin D (70) by Thomas is outlined in Scheme 7. The intermediate 67, which contains all of the carbon atoms of the natural product, was synthesized in eight steps starting from the advanced building blocks 64, 65 and 66. Selective deprotonation and trapping the so-formed enolate with phenylselenyl chloride gave an intermediate selenide. Oxidation and elimination of the corresponding selenoxide at elevated temperature generated the necessary dienophile, which directly underwent an intramolecular endo-Diels–Alder reaction to give the isoindolinone 68. For the synthesis of 69 another ten steps were necessary to discriminate between the three remaining double bonds, functionalize the full carbon skeleton and set the remaining stereocenters. Five steps, including the exchange of protecting groups and one oxidation, completed the synthesis of cytochalasin D (70).

In comparison to the syntheses mentioned before (Stork, Thomas, Trost, Vedejs), the convergent approach reported by Myers allows a modular entry to diverse members of the cytochalasin alkaloid family [68]. As proof of concept, the macrolactone cytochalasin B (52) and the carbocyclic cytochalasin L-696,474 (78), a potent HIV-1 protease inhibitor [69-72], were synthesized (Scheme 8). Strategic bond disconnections revealed the common isoindolinone precursor 73. The synthesis of the latter commenced from N,N-dibenzylphenylalanine (71) to afford the Diels–Alder substrate 72 in four steps. The envisioned intramolecular Diels–Alder cyclization of the silyl enol ether 72 provided the depicted endo-diastereomer in good yield. Exchange of the N-benzyl for a Boc-protecting group and cleavage of the silyl enol ether gave the corresponding ketone, which was first converted to an enol-triflate and then to the tricyclic alkene 73. At this stage, the syntheses of cytochalasin B (52) and cytochalasin L-696,474 (78) diverged. For 78, another five steps were necessary to form the isoindolinone 74. After oxidation of the primary alcohol with Dess–Martin periodinane, the obtained aldehyde was coupled with the readily available N-phenyltetrazole sulfone 75 via a Julia–Kocienski olefination. Installation of the phosphonate and desilylation gave 76, which, after oxidation, reacted in the presence of sodium 2,2,2-trifluoroethanol (NaOTFE) in 2,2,2-trifluoroethanol (TFE) via an intramolecular Horner–Wadsworth–Emmons reaction to 77. Cytochalasin L-696,474 (78) was obtained from 77 via reduction, acetylation and a magnesium sulfate induced rearrangement of the epoxide to the allylic alcohol. The synthesis of cytochalasin B 52 (15 steps from 73) proceeded in a similar fashion and was highlighted in detail by Hertweck [44]. The authors state that this approach is highly diversifiable and allows for a rapid access to a variety of members of the cytochalasin alkaloid family.

The power of the intramolecular Diels–Alder strategy for the synthesis of the isoindolinone moiety was also recognized during the synthesis of aspergillin PZ (91) [73]. This secondary metabolite has a remarkable architecture and was first isolated from the fungus Aspergillus awamori in 2002 [74]. The highly functionalized pentacyclic skeleton comprises a unique 12-oxatricyclo[6.3.1.0^2,7]dodecane ring system and ten stereogenic centers. Its carbon framework is closely related to other members of the aspochalasin/cytochalasin alkaloids such as cytochalasin D (70), which lacks the unusual oxatricyclo ring system. The novel structure and cytotoxic activities of natural analogues of aspergillin PZ (91) against the HL-60 cancer cell line (IC₅₀ = 50–80 nM) initiated Tanis’ synthetic program to further evaluate the therapeutic potential of 91 and derivatives thereof (Scheme 9) [73]. The synthesis could be accomplished in 26 steps and includes two key steps: the 8-oxabicyclo[3.2.1]octane unit was synthesized via an unprecedented 2-oxonia-[3,3]-sigmatropic aldol reaction cascade and the isoindolinone moiety was assembled using an intramolecular Diels–Alder reaction. The synthetic endeavor began with the conversion of dihydropyran 79 to the acetal 80 (Scheme 9). Activation of the anomeric position with tin(IV) chloride presumably led to the formation of the oxonium ion 81. This intermediate was envisioned to undergo a Prins–Pinacol rearrangement via intermediate 84 to give the cis-aldehyde 85. However, only the diastereomeric trans-aldehyde 83 was isolated in moderate yields from the reaction mixture. The authors concluded that a competing 2-oxonia-[3,3]-sigmatropic aldol pathway via 82a and 82b might be operative for this system. The stereochemical outcome could be attributed to the sterically less demanding transition state 82b. The epimerization of 83 to 85 proceeded via an intermediate lactone and extended the route by seven steps.

After having secured the cis-epimer 85, the diene moiety was introduced in four steps to give aldehyde 86, which was reacted with the lithium enolate of lactam 87 to furnish the aldol adduct 88 (Scheme 10). Oxidation of the secondary alcohol to the corresponding ketone and selective selenation with phenylselenyl chloride gave 89. Following the reported sequence for the synthesis of cytochalasin D (70), namely oxidation of the selenide to the selenoxide, elimination and intramolecular Diels–Alder reaction, isoindolinone 90 could be obtained in moderate yield. Removal of the protecting groups then yielded aspergillin PZ (91).

Though having established a rather long and low yielding synthesis, ample quantities could be obtained for a preliminary evaluation of the biological activity of 79. In a first screen against A2058 melanoma and DU145 prostate cancer cell lines, 79 showed to be inactive [73].

Isoindolinones derived from isoquinoline alkaloids: The aporhoeadane alkaloids, a term designated by Shamma [75], are abundantly found in South American members of the botanical family of Berberis (Berberidaceae). Along with the ever present isoquinoline berberine (92), the isoindole-containing benzazepines chilenine (93) [76], lennoxamine (94) [77] and chilenamine (95) [78], the isoquinoline nuevamine (96) [77] and the benzazocine magallanesine (97) [79] were isolated from different Berberidaceae species (Figure 6).

Chilenine (93) was the first isoindolobenzazepine alkaloid isolated from Berberis empetrifolia in 1982. The biogenesis proceeds most likely via a Pictet–Spengler reaction of dopamine (99) with 4-hydroxyphenylacetaldehyde (100), both derived from L-tyrosine (98) (Scheme 11). After oxidation and O-methylation, which is carried out by S-adenosylmethionine (SAM), (S)-reticuline (101) is obtained. Oxidation of the N-methyl group to the iminium ion and cyclization gives the tetracyclic carbon skeleton, which upon further methylation and oxidation furnishes berberine (92) [80]. Prechilenine (102), itself isolated as its O-methyl ether is formed via oxidation [78]. A base-catalyzed semipinacol-type rearrangement yields 93 [76]. Chilenine (93) can be further transformed to the [4.4.0]-tricycle lennoxamine (94) and the [5.3.0]-ring system chilenamine (95) [77-79]. Other members such as nuevamine (96) and magallanesine (97) contain the corresponding [4.3.0] and [6.3.0]-scaffold.

Since their first isolation in the 80’s of the last century, a myriad of syntheses of these small, sometimes highly functionalized and mainly biological inactive alkaloids have been described in the literature. Most of this work is summarized in an excellent review by Leonard [81]. We confine this section to the syntheses of magallanesine (97) by Danishefsky and Kurihara, which were also mentioned by Evans and Bentley [82,83].

Magallanesine (97), which was isolated in 1985 from Berberis darwinii, a plant native to southern Chile and Argentina, is the first known isoindolobenzazocine alkaloid. The seminal total synthesis of 97 was achieved in the group of Danishefsky (Scheme 12) using a dimethylformamide acetal-mediated cyclodehydration of 105. The isoindole unit of 105 was prepared from the condensation of 103 and 104. Oxygen–sulfur exchange provided the thiophthalimide 105, which could be converted to magallanesine (97) via an intramolecular aldol condensation in one step [84].

In 1996, Kurihara (Scheme 13) reported a [1,2]-Meisenheimer rearrangement followed by an intramolecular Heck cyclization to elaborate the isoindolobenzazocine moiety [85]. The synthesis commenced with the preparation of ester 106 from piperonal according to a known procedure [86]. A four-step sequence yielded azetidine 107. After oxidation with hydrogen peroxide, the resulting N-oxide cleanly underwent a [1,2]-Meisenheimer rearrangement upon heating in tetrahydrofuran. The so-formed azocine 108 was converted to amine 109 by hydrogenolysis of the N–O bond. Amide formation with acid chloride 110 gave 111, which upon sequential oxidation yielded enaminone 112. Finally, the isoindolinone moiety was generated via an intramolecular Heck reaction using tetrakis(triphenylphosphine)palladium(0) and thallium acetate, to give magallanesine (97) in excellent yield [85].

Aporphine alkaloids: The aristolactam alkaloids are classified as members of the aporphine alkaloid family and contain a characteristic phenanthrene lactam core. Aristolactams (113–116) along with aristolochic acids (117,118) and 4,5-dioxoaporphines (119,120) were mainly isolated from Aristolochiaceae but can also be found in several other plant species (Table 1) [87-92]. Several comprehensive overviews of various aristolactams were published by Shamma, Parmar and Bentley [87,92-95].

The co-occurrence of aristolactams, aristolochic acids and 4,5-dioxoaporphines led to the assumption of a close biogenetic relationship between these three classes [87,96]. The biosynthesis of aristolochic acid I (117) was elucidated via labeling experiments and is depicted in Scheme 14 [97-100]. First, two molecules of the amino acid L-tyrosine (98) are transformed to (R)-orientaline (121) in a similar fashion as described for the biogenesis of (S)-reticuline (101) (compare Scheme 11). A de-aromatizing spirocyclization of 121 leads to (R)-orientalinone (122), which, after reduction of the ketone to the secondary alcohol 123, undergoes a [1,2]-alkyl migration with concomitant loss of water. Re-aromatization and dioxolane formation gives stephanine (124). It is hypothesized that direct oxidation of 124 forms 4,5-dioxoaporphine 125, which upon benzilic acid rearrangement, N-methyl cleavage and decarboxylation provides aristolactam I (113) [96]. Additional oxidation of the amide function furnishes aristolochic acid I (117).

The aristolactams and aristolochic acids have a broad spectrum of biologically interesting properties. Although both can be isolated from the same plants, their biological activity is highly variable. Herbal drugs containing aristolochic acids display cytotoxic activities and inhibitory effects on platelet aggregation and were traditionally used in folk medicine [101]. However, since the early 1980’s aristolochic acids are associated with the development of several cancer types and nephropathy (Chinese herbs nephropathy). The mechanism for the formation of covalent DNA adducts is shown in Scheme 15 [102,103]. The DNA damage is initiated by metabolic reduction and cyclization of 117 to N-hydroxyaristolactam 126. After loss of water, the aristolactam nitrenium ion 127, which is believed to be the main carcinogen, reacts with DNA bases such as adenosine, to give the corresponding 7-(deoxyadenosin-N⁶-yl)aristolactam I (dA–AAI, 128).

Interestingly, aristolactams are not mutagenic themselves but show immunosuppressant [104], antiplatelet [105], antimycobacterial [106], neuroprotective [107] activities and are excellent inhibitors of cyclin-dependent kinases (CDKs) with IC₅₀ values in the nanomolar range [108,109]. CDKs are regulatory proteins, which play an important role in the cell cycle and are considered to be a potential target for anticancer treatment. This makes the aristolactams promising lead components for further drug discovery [108]. The similarity of the aromatic moiety of aristolactams and staurosporines, which are also known to be potent kinase inhibitors could explain the observed inhibition of cyclin-dependent kinases [109]. Over the past years, considerable synthetic efforts have been dedicated to the aristolactams and the development of non-natural analogues.

Pioneering work was carried out in the groups of Castedo [110] and Couture [111,112]. A beautiful application of a directed ortho-metalation (DoM) strategy was reported in the synthesis of piperolactam C (131) by Snieckus (Scheme 16) [113]. Treatment of 129 with base afforded aminophenanthrene 130 via a remote lateral metalation–cyclization sequence. Metalation of 130 with excess n-butyllithium followed by carbonation then yielded piperolactam C (131).

In 2008, Heo and coworkers reported the synthesis of several aristolactams employing a one-pot cross-coupling/aldol condensation cascade reaction (Scheme 17) [91]. The synthesis commenced with the preparation of the isoindolinone building block 134. Friedel–Crafts acetylation of 132, followed by a Lieben haloform degradation gave the corresponding acid, which after methylation yielded 133. Bromination and lactam formation afforded isoindolinone 134. The following one-pot Suzuki–Miyaura/aldol condensation of 134 with boronic acid 135 was carried out at 150 °C in a microwave reactor and gave aristolactam BII (115) in good yield. By variation of the coupling partners, Heo was able to efficiently synthesize a library of natural and non-natural analogues.

Cularine alkaloids: Another prominent class of isoquinoline alkaloids, typically found in the botanical family of Fumariaceae and in particular in Sarcocapnos enneaphylla, are the cularines (136–142) (Figure 7) [114-116]. In comparison to the related aporphine alkaloids, the cularine skeleton contains a benzoxepine ring system. The parent alkaloid (+)-cularine (138) was first isolated by Manske in 1938 [117], and its molecular skeleton was assigned on the basis of NMR and X-ray crystallographic studies. The structure elucidation revealed a twist-boat conformation of the dihydrooxepine ring [118-120]. Within this family, aristoyagonine (136), which was first isolated by Castedo in 1984, is the only known naturally occurring member of the cularine alkaloids bearing an isoindolinone moiety [121]. Aristocularine (137), a synthetic derivative of aristoyagonine (136), is known to display cytotoxic activities in the micromolar range [122].

Biosynthetically, the cularines can be traced back to L-tyrosine (98) (Scheme 18) [123,124]. Incorporation of radioactively labeled L-tyrosine into crassifoline (144), which is an established cularine precursor, most likely occurs via alkaloid 143. Hydroxylation and sequential methylation of 143 leads to 144. Oxidative phenol coupling then forms the cularine alkaloid skeleton. The generated cularine enneaphylline (145) and the regioisomeric iso-cularine sarcocapnidine (141) could serve as branching points for the synthesis of further cularine derivatives. Sequential oxidations of 141, similar to the biosynthesis of the aristolactams (Scheme 14), yields yagonine (146), which could be converted to aristoyagonine (136) via a benzilic acid-type rearrangement [121].

The first syntheses of aristoyagonine (136) were reported by Castedo and Suau (Scheme 19) [121,122,125-127]. The approach carried out by Castedo relied on 3-hydroxysarcocapnine (147), which could either be isolated from natural sources or prepared via an Ullmann coupling of a 1-(2-bromobenzyl)-8-hydroxyisoquinoline derivative [121,125]. Oxidation of 147 with 2,3-dichloro-5,6-dicyano-1,4-benzoquinone (DDQ) gave the dioxocularine yagonine (146). The bioinspired, base-mediated transformation of 146 to 136 obviously proceeds via a benzilic acid rearrangement–oxidation sequence.

In 1996, the group of Suau accessed aristocularine 137 along with 149 in one step from acetamide 148. In the presence of oxalyl chloride and the Lewis acid tin(IV) chloride, a tandem cyclization (Bischler–Napieralski/Friedel–Crafts acylation reaction) was triggered to directly give 137 and 149 [122,127].

In 2006, another approach to aristoyagonine (136) was reported by the group of Couture (Scheme 20) [128]. For the preparation of 136, their previously developed procedures for the syntheses of aristolactams were adopted [111,112]. The synthesis commenced with the generation of the amide 152 from the coupling of 150 with 151. Exposure of 152 to potassium bis(trimethylsilyl)amide gave the isoindolinone 153 via an intramolecular addition–elimination reaction. In the presence of the aldehyde 154, 153 underwent a Horner–Wittig reaction to generate 155 as a 1:2.3 mixture of E- and Z-isomers. Treatment of 155 with boron trichloride not only liberated the phenol but also isomerized the E-double bond to the Z-isomer. Finally, a copper(II) triflate-catalyzed Ullmann coupling furnished 136.

Meroterpenoids: The term meroterpenoids describes a family of natural products with a mixed biosynthetic origin, partially derived from terpenoids and polyketides [129]. Several members of this class containing an isoindolinone motif, for instance compounds 156, 157 and 159–161, can be extracted from Stachybotrys fungi (Figure 8) [130-135]. Memnobotrin A (158) is found in the closely related Memnoniella echinata fungus [136], and hericenone B (162) and erinacerin A (163) were isolated from the mushroom Hericium erinaceum [137,138]. Another member of this family, aspernidine A (164), was recently isolated form the fungus Aspergillus nidulans by Hertweck [139]. A comprehensive review about meroterpenoids produced by fungi was published by Simpson in 2009 [140]. The isolated molecules display a broad spectrum of interesting biological properties, such as high antiviral activity (156) [141], inhibition of the HIV-1 protease (157) [132], antibacterial and antifungal activities (159–160) [133] or neuritogenic properties (161) [134]. So far, only a few total syntheses were accomplished and the exact biosynthetic pathways are unknown.

The sesquiterpenoid stachyflin (156) was first isolated from the fungus Stachybotrys sp. RF-7260 by Shionogi & Co., Ltd., Japan, in 1997 [130] and shows outstanding antiviral activity against the influenza A subtype H1N1 (IC₅₀ = 3 nM), outperforming current drugs such as amantadine and zanamivir [141]. The putative biosynthesis of 156, which contains a pentacyclic ring system with a cis-fused decalin is depicted in Scheme 21. Farnesyl pyrophosphate (167), which originates from the terpenoid pathway through the condensation of dimethylallyl pyrophosphate (DMAPP, 165) with two units of isopentyl pyrophosphate (IPP, 166), and orsellinic acid (168), which is derived from a fungal iterative type I polyketide pathway [142], are connected to give 169. This substrate is already poised for a polyene cyclization cascade, which only has to be triggered via activation of the epoxide. After the decalin moiety has formed, two sigmatropic, stereospecific [1,2]-shifts of intermediate 170 provide the tertiary carbocation 171, which could be trapped by the phenolic alcohol to give stachyflin (156) [143].

In 2011, the first enantioselective total synthesis of (+)-stachyflin (156) via a Lewis acid induced domino epoxide-opening/rearrangement/cyclization cascade was accomplished by the group of Katoh (Scheme 22) [144]. The synthesis commenced with the conversion of dimethyl 2,6-dihydroxyterephthalate (172) to nitrile 173 within five consecutive steps. Hydrogenation and lactam formation of 173 gave isoindolinone 174, which was converted to bromide 175. Reductive alkylation of the protected Wieland–Miescher ketone 176 with bromide 175 using Birch conditions gave 177 as a single diastereoisomer. Epoxide 178 could be prepared in seven steps from 177, but was obtained as a mixture of inseparable diastereoisomers. The Lewis acid induced key step, a domino epoxide-opening/rearrangement/cyclization cascade, most likely proceeded in a stepwise manner via intermediate 179. Activation of the epoxide with boron trifluoride etherate induces the planned sigmatropic [1,2]-methyl and [1,2]-hydride shifts to generate the tertiary carbocation at the ring junction. This is then trapped by the phenol to give the pentacyclic compounds 180a and 180b. The former diastereoisomer could be converted to 180b via inversion of the secondary alcohol. Cleavage of the methyl ether and the 3,4-dimethoxybenzyl (^3,4DMB) group of 180b led to (+)-stachyflin (156). The synthesis proceeded in 25 steps with an overall yield of 2.6%.

Another novel class of isoindolinone-containing natural products, categorized as spirodihydrobenzofuranlactams or stachybotrylactams (157, 181, 182), was isolated from different Stachybotrys species by Jarvis and Roggo between 1995 and 1996 (Figure 9) [131,132]. These compounds are antagonists of endothelin and display strong immunosuppressant activities. The pseudosymmetric dimer stachybocin A (182), which was isolated by Ogawa in 1995 [145], constitutes the most potent representative [132]. Oxidation of the isoindolinone unit of 157 gives the relatively scarce phthalimide motif, which is present in stachybotrylactam V (181).

To date, only one enantioselective total synthesis of spirodihydrobenzofuranlactam I 157 has been published (Scheme 23) [146]. The synthesis carried out by the group of Guo commenced with trans-decalone 183, which was derived from the Wieland–Miescher ketone. For the conversion of 183 to the α,β-unsaturated aldehyde 184 six steps were necessary. Halogen–metal exchange of 185 with n-butyllithium followed by the addition of 184 gave the carbinol 186 in good yield. Removal of the benzyl groups, cleavage of the tert-butyl group with concomitant formation of the methoxy ester (COOt-Bu → COOMe), and global deprotection gave 187. Acid-catalyzed (Amberlyst 15) spiroannulation afforded a 1.7:1 mixture of the benzofuran and the benzopyran (90% overall conversion). For the installation of the isoindolinone, a seven-step sequence similar to the one described for stachyflin (156) was used (Scheme 23). The arene appendage was desymmetrized via monobromination and cross coupling with copper cyanide provided a nitrile. Hydrogenation and lactam formation under basic conditions completed the synthesis of stachybotrylactam I (157).

Chlorinated isoindolinone alkaloids: Pestalachloride A (193) was isolated from the plant endophytic fungus Pestalotiopsis adusta as a racemate and displays potent antifungal activities against Fusarium culmorum (IC₅₀ = 0.89 µM) [147]. Pestalone (192), a natural product and synthetic precursor of 193 shows strong antibiotic activity against methicillin-resistant Staphylococcus aureus (MRSA, MIC = 84 nM) and vancomycin-resistant Enterococcus (VRE, MIC = 178 nM) [148]. The structural similarity between pestalachloride A (193) and pestalone (192), which both contain the same prenylated polyketide core imply a direct biosynthetic relationship [147]. The reaction of 192 with an equivalent of ammonia gives a cyclic iminohemiaminal, which first tautomerizes to the hydroxy isoindole and then to the isoindolinone 193. This biosynthetic transformation was also used in the total synthesis of 193 by Schmalz (Scheme 24) [149]. Reaction of lithiated 188 (three steps from commercially available 5-methylresorcinol) with the aldehyde 189 (double bromination of 3,5-dimethoxybenzaldehyde) gave the racemic alcohol 190.

Oxidation to the benzophenone, installation of the prenyl side chain and introduction of the formyl group was accomplished within five steps to give 191a (Scheme 25). The envisioned deprotection of 191a using Lewis acidic conditions led to an unexpected and unprecedented metal-free carbonyl–olefin metathesis. Coordination of boron trifluoride etherate to 191a promotes the formation of the tertiary carbenium ion 194, which isomerizes via the oxetane intermediate 195 to the benzylic cation 196. Expulsion of acetone then yields indene 197 in excellent yield. Avoiding this side reaction could only be achieved by exchanging the methyl ether protecting group for a more labile methoxymethyl ether at the benzophenone stage. The following prenylation and formylation proceeded smoothly under the same conditions to give 191b (Scheme 24). Cleavage of the MOM ethers afforded pestalone (192), which could be converted into 193 in a single step by treatment of 192 with ammonia in aqueous ammonium chloride solution (pH 8).

Muironolide A (204) was isolated by Molinski from the marine sponge Phorbas sp. in 2009 [150]. The molecular framework of this unique natural product has a hexahydro-1H-isoindolinone-triketide, a trans-2-chlorocyclopropane and a trichlorocarbinol ester. A first biological evaluation showed that 204 displays antifungal activity. However, only 90 µg could be isolated and further biological screening was not possible [151]. Several sponge-derived macrolides have proven to be effective cytotoxic agents [152,153] and one could speculate that muironolide A (204) shares this potential. A first attempt to access 204 in the laboratory was reported by Molinski shortly after the isolation (Scheme 26) [151]. Coupling of the dienophile precursor 198 with sorbic acid (199) gave a tertiary amide, which was converted to 200 via a reduction–oxidation sequence. The asymmetric intramolecular Diels–Alder reaction was then catalyzed by Kristensens’ catalyst (201) to give 202. A base-promoted epimerization of the aldehyde and a Horner–Wadsworth–Emmons reaction furnished 203, the most advanced intermediate reported so far.

Just recently, the group of Zakarian reported another approach to the fully elaborated isoindolinone core of 204 [154].

Chlorizidine A (208) (Scheme 27), a cytotoxic metabolite from a marine Streptomyces species shows an unprecedented 5H-pyrrolo[2,1-a]isoindolinone ring system, which is connected to a dichlorinated pyrrolizine [155]. Chlorizidine A (208) and its semisynthetic derivatives display significant cytotoxic activities against various human cancer cell lines. The IC₅₀ value of 208 against the colon cancer cell line HCT-116 was determined to be in the micromolar range (3.2–4.9 μM) [155]. A first biosynthesis was postulated on the basis of the structural similarity between 208 and marinopyrrole A, another secondary metabolite derived from a marine Streptomyces species [156]. The common biosynthetic precursor 206 stems from a mixed nonribosomal peptide synthetase (NRPS)/polyketide synthase (PKS) pathway. The amino acid proline is first oxidized and chlorinated by an FADH₂-dependent halogenase to give 205, which after extension by the polyketide synthase gives the polyketides 206 and 207, respectively. The intermolecular condensation of these two components yields chlorizidine A (208) [155].

Anthraquinone-type alkaloids: Lactonamycin (215) and lactonamycin Z (217) were isolated from Streptomyces rishiriensis and Streptomyces anglieri in 1996 and 2003 [157,158]. Both compounds are potent antibacterial agents against Gram-positive bacteria, including methicillin-resistant Staphylococcus aureus (MRSA) and vancomycin-resistant Enterococcus (VRE) [159]. Beyond that, the modest cytotoxic activity of 215 and 217 against human cancer cell lines makes them interesting lead components for drug discovery. The IC₅₀ values of 215 for different leukemia cell lines range from 0.11–0.22 µM. For 217, an IC₅₀ value of 0.32 µM against gastric adenocarcinoma was observed [157,158]. The hexacyclic aglycone of 215 and 217, lactonamycinone (214), consists of a densely oxygenated fused hydrofuran–hydrofuranone and a naphta[e]isoindole ring system. The core of lactonamycin (215) is decorated with the 2-deoxysugar α-L-rhodinopyranose (216), whereas α-L-2,6-dideoxyribopyranose (218) is found in 217. The structure of the aglycone 214 is related to tetracenomycins, a family of tetracyclic aromatic polyketides, which are produced by several Streptomyces species [159]. Biosynthetic investigations revealed a striking similarity of the gene cluster responsible for the biogenesis of lactonamycinone (214) to those clusters found in tetracenomycin-producing bacteria. Cloning experiments and incorporation of various labeled precursors resulted in a first hypothesis for the biogenesis of 215 and 217 (Scheme 28) [159]. This led to the assumption that the aglycone 214 is assembled from nine acetate units and glycine, respectively N-methylglycine derivative, which contributes both carbon atoms and the nitrogen to the skeleton of 210 (position 12 and 12a). The latter are uncommon starter units in a type II polyketide assembly line [160]. Upon oxidative cleavage of the D-ring of the putative tetracenomycin-like intermediate 210, the aldehyde 211 is reduced to the corresponding alcohol 212. Acetalization and introduction of the 1,2-cis-diol moiety furnishes 213. Sequential formation of the lactone and the isoindolinone generates 214. A final glycosylation step leads to lactonamycin (215) and lactonamycin Z (217).

The remarkable structure and the promising biological activities have attracted many synthetic groups. Several total syntheses of 215 and 217 were reported in recent years [161] (and references therein). The group of Danishefsky was the first to report a diastereoselective synthesis of lactonamycinone (214) employing a Diels–Alder reaction [162,163]. In 2010, Tastuta completed the first total synthesis of lactonamycin (215) by using a sequential conjugate addition, a stereoselective glycosylation reaction and a Michael–Dieckmann-type cyclization [164]. The recently published total synthesis of lactonamycin (215) and lactonamycin Z (217) by Saikawa and Nakata is based on a late-stage glycosylation strategy (Scheme 29). This enables the specific variation of the sugar components and gives access to various lactonamycin derivatives [161].

Starting from alcohol 219, the diene precursor 220 could be prepared in seven steps. The following Diels–Alder reaction with chloroethynylquinone 221, which was synthesized according to a known procedure [165] proceeded in a highly regioselective manner to give antharaquinone 222. Introduction of the fused hydrofuran–hydrofuranone moiety was accomplished via deprotection, palladium-catalyzed cyclization–methoxycarbonylation [166] and an acid-catalyzed lactone formation to afford 223. For the generation of the isoindolinone via a Bischler–Napieralski reaction, a chloroacetyl (CA)-protected phenol was essential to avoid competing carbamate formation of the starting material. Upon exposure of 223 to phosphorus pentoxide, the desired isoindolinone 224 was formed in 75% yield. Deprotection led to lactonamycinone (214), which was the substrate for a ytterbium triflate-catalyzed glycosylation to give 215 and 217, respectively.

Miscellaneous: At first glance, the hetisine alkaloids do not show any structural similarities with the compound classes described before. However, though lacking an “intact” isoindole core, one can spot the corresponding perhydro motif. Since their first discovery more than 70 years ago, more than 100 different hetisine alkaloids have been isolated from different species of the plants Aconitum and Delphinium, and to a lesser extend from Rumex, Consolida and Spiraea [167,168]. These plants have been widely used in traditional herbal medicine and further pharmacological studies revealed that hetisine alkaloids are highly bioactive compounds [167,168]. The diverse spectrum includes vasodilatoring, anti-arrhythmic, immunomodulating and analgesic activities. Hetisines can be classified, according to Wang and Liang, as C₂₀-diterpenoid alkaloids, which belong to the family of atisane alkaloids [167]. The hetisines comprise a heptacyclic core and are structurally the most complex members derived from the atisine skeleton. The most prominent representatives, nominine (225) [169,170], kobusine (226) [171,172], pseudokobusine (227) [173,174] and hetisine (228) [174-176] are depicted in Figure 10.

The putative biosynthetic pathway for the formation of the hetisine alkaloids is derived from phytochemical data and basic biochemical transformations (Scheme 30) [167]. Cyclization of geranylgeranyl pyrophosphate (229), involving the ent-copalyl diphosphate synthase, could give ent-copalyl diphosphate (230). After loss of the pyrophosphate group and double cyclization, a 1,3-hydride shift and Wagner–Meerwein rearrangement leads to the naturally occurring ent-atisir-16-ene (231). Oxidative incorporation of nitrogen, which might be derived from β-aminoethanol, generates the atisine-type skeleton 232. Oxidative C–C-bond formation gives the hetidine core (233) and C–N-bond formation the hetisine skeleton (234).

The first total synthesis of a hetisine-type alkaloid was accomplished by Muratake and Natsume in 2004 [177]. In their seminal work, (±)-nominine (225) was synthesized within 40 steps and 0.15% overall yield. Two years later, a more concise and efficient access to (±)-nominine (225), featuring a oxidoisoquinolinium-1,3-dipolar cycloaddition and a dienamine-Diels–Alder reaction, was accomplished by Gin (Scheme 31) [178]. Coupling of 235 and 236, both synthesized within three steps from simple starting materials, via a Staudinger–aza-Wittig reaction gave amine 237 as a mixture of four diastereoisomers. Conversion into 4-oxido-isoquinolinium betaine 238 could be achieved by an acid-catalyzed methanol extrusion and isomerization. Betaine 238 served as the substrate for the following 1,3-aza-dipolar cycloaddition. Carrying out the reaction at 180 °C in tetrahydrofuran provided a separable mixture of pyrrolidine isomers 241 and 242 (1:3.6). The undesired cycloadduct 242 could be equilibrated to 241 due to the reversibility of the reaction. Conversion to the β,γ-cyclohexenone 243 was accomplished within 6 consecutive steps. Generation of the dienamine 244 with pyrrolidine in methanol at 60 °C triggered an intramolecular Diels–Alder reaction to provide the full carbon skeleton 245. The final transformations of the synthesis involved a Wittig olefination of the ketone and a diastereoselective allylic oxidation to provide (±)-nominine (225). In 2008, the same group reported the first total synthesis of (+)-nominine by introducing the desired stereochemical information to the nitrile 235 [179].

Conclusion

The outlined syntheses of natural products containing the uncommon isoindole skeleton are still far away from being ideal. Low yielding multistep strategies dominate for more complex molecules and often cannot provide the amounts necessary for further investigations. Several biologically active members are still precluded from extensive biological studies due to their low abundance in nature. This and the fact, that many compounds are underexplored offers plenty opportunities for the development of innovative chemical transformations. The biosynthesis of many isoindole natural products is still uncertain and has yet to be unraveled. The presented examples could serve as an inspiration for the development of novel synthetic methods, new biosynthetic insights and drug development. Considerably more discoveries remain to be uncovered through exciting projects.