Reviewing Colchicaceae alkaloids - perspectives of evolution on medicinal chemistry.

The subject of chemosystematics has provided insight to both botanical classification and drug development. However, degrees of subjectivity in botanical classifications and limited understanding of the evolution of chemical characters and their biosynthetic pathways has often hampered such studies. In this review an approach of taking phylogenetic classification into account in evaluating colchicine and related phenethylisoquinoline alkaloids from the family Colchicaceae will be applied. Following on the trends of utilizing evolutionary reasoning in inferring mechanisms in eg. drug resistance in cancer and infections, this will exemplify how thinking about evolution can influence selection of plant material in drug lead discovery, and how knowledge about phylogenetic relationships may be used to evaluate predicted biosynthetic pathways.

INTRODUCTION

Using chemical information to define groups of plants is probably as old as botanical classification itself, a practice often referred to as chemosystematics or chemotaxonomy. In the modern sense it was largely introduced during the 1960s and 1970s, for example with the works of Bate-Smith [1,2], Swain [3] and Harborne [4], and by the founding of the journal now known as Biochemical Systematics and Ecology in 1973. In 1980 Dahlgren published the first systematic classification for flowering plants emphasizing chemical characters in defining relationships between families and higher taxa [5], visualizing the distribution of characters in the earlier introduced form of dahlgrenograms [6]. With the advent of molecular systematics based on comparison of nucleotide sequences, several of the groups which were previously to some extent defined by chemical characters, received even further support: the close relationships of the families containing betalain pigments instead of anthocyanins in Caryophyllales [7]; the occurrence of benzylisoquinoline alkaloids in the families of Dahlgren’s Magnoliflorae [8]; the “mustard oil bomb” of the order Brassicales [9]; and the iridoids in the asterids [10]. The collective literature has thus confirmed the statement of Helen Abbott in 1887 that “the evolution of chemical constituents follows parallel lines with the evolutionary course of plant forms, the one being intimately connected with the other…” [11].

Natural products have been hitherto a very successful source of new drugs, with slightly more than a third of “small compounds” launched as “new chemical entities” for the past 30 years belonging to this group [12]. At the same time it has been shown that natural products occupy a different and larger chemical space than synthetic drugs [13, 14, 15], and that they have a higher probability to pass through the pharmaceutical industry drug developmental pipeline [16]. This is today usually discussed within the concept of natural products being prevalidated for activity [14, 15, 17, 18, 19, 20, 21]. As these compounds per definition are derived within a biological setting they are today presumed to have some sort of function or raison d’être, and even if this function is unknown, unstudied or only based on available bioassays, it is clear that this inherent property is influenced by evolution [22, 23]. Following on the trends of utilizing evolutionary reasoning in inferring mechanisms in eg. drug resistance in cancer and infections, this contribution will exemplify how thinking about evolution can influence selection of plant material for drug lead discovery, and how knowledge about phylogenetic relationships may be used to evaluate predicted biosynthetic pathways. As a case in point, an approach of taking phylogenetic classification into account in evaluating the medicinally important and historically fascinating alkaloid colchicine and related phenethylisoquinoline alkaloids from the family Colchicaceae is applied.

DISCOVERY AND MEDICINAL USE OF COLCHICINE

Colchicine, present in and named after the autumn crocus (Colchicum autumnale L., Colchicaceae), is today a well-known compound recognized for its antimitotic activity and as a treatment for acute gout and familial Mediterranean fever, a hereditary inflammatory disorder [24, 25, 26]. But the history of colchicine is full of misunderstandings and ill-advised inferences. This drug has truly ancient origins, being present in the writings of several classical Greek physicians from Nicander of Colophon, over Dioscorides, Pliny the Elder and Galen, to Alexander of Tralles [27]. Its use as a medication was however heavily discouraged by its toxicity, and often it was described rather as a poison than a remedy. The true identification was also heavily debated and views on whether terms such as colchicon, ephemeron, hermodactylum and surugen were equivalent differed between scholars at various times in history. Autumn crocus corm was present in the London Pharmacopoeia between 1618 and 1639, after which it was not mentioned until the 1788 edition [28]. The modern introduction of colchicine-based medicines for gout is usually attributed to von Störck in 1763 [29], but see Hartung [27] for counter-arguments. Colchicine-based medicines have been continuously employed as a remedy for this condition since the very early 19th century. The use of colchicine has been based on clinical experience, and it was only in 2009 the U. S. Food and Drug Administration approved it based on clinical trials defining dosage and efficacy [30].

Colchicine was identified as a “mitotic poison” in the early 1930s and its use in production of polyploid plants was described in 1937 [31]. It was used in the determination of the chromosome number for the human species [32], and has since the explanation of its tubulin-interaction been an important tool in biochemistry and development of cancer drugs [26, 33, 34]. Colchicine itself is too toxic to be used as a treatment for cancer, but other alkaloids and derivatives have been investigated. One of the first examples is the use of the alkaloid demecolcine, a deacetylated and methylated derivative against myeloid leukemia [35], and recently the allocolchinoid phosphate-derivative ZD6126 has been studied [34].

Just as the medicinal use has a complicated history, this is equally true for the alkaloid itself. The first colchicine isolation is usually considered to be by Pelletier and Caventou in 1820 [36]. However, they considered the isolated compound to be the alkaloid veratrine, a steroid-derived alkaloid from plants in the death camas family Melanthiaceae, and Geiger was the first to coin the name colchicine after crystallization of the pure compound in 1833 [37]. The structure of the alkaloid proved to be very elusive, and it was not elucidated until the x-ray crystallography investigation of King and co-workers in 1952 [38], confirming the supposition of Dewar that the compound contain two 7-membered rings of which one is a tropolone [39]. A review of synthetic approaches to colchicine was published in 2004 [40].

BIOLOGICAL ORIGIN

Colchicine is considered to have a restricted distribution within the plant kingdom [41, 42], and is isolated commercially from the related ornamental species autumn crocus (Colchicum autumnale L.) and flame lily (Gloriosa superba L.). The medicinal importance of colchicine prompted early screenings of the plant kingdom to find suitable sources for the compound. From these early reviews a much wider distribution for colchicine was inferred [43]. However, refined chemical detection methods have discarded many reports as false positives [41, 44]. The more restricted distribution originally proposed coincided largely with the, at that time, recently introduced lilioid subfamily Wurmbaeoideae [45], today more familiar as the family Colchicaceae [46].

With the advent of molecular systematics based on comparison of DNA sequences it was shown that a few genera (such as Disporum Salisb., Uvularia L., Kuntheria Conran & Clifford, Schelhammera R.Br., and Tripladenia D.Don) previously placed in other families should be included in an expanded Colchicaceae family and a new classification has been adopted, (see Table ) [47, 48]. Investigation of crude alkaloid extracts with mass spectrometry has also confirmed that colchicine is present in, and restricted to, the genera of Colchicaceae in this wider circumscription [49].

An Australian species described under the name Kreysigia multiflora Rchb. has been shown to contain colchicine and other alkaloids [50]. However the exact identity of this species is confounded by the application of this name to plants belonging to Schelhammera as well as Tripladenia.

Just as genera have been included in the family based on phylogenetic analysis of nucleotide data, analysis of such data has also resulted in a notable exclusion. The genus Disporum was considered to be present in both Asia and North America, but results from genetic studies have shown that the North American representatives should be treated as a distinct genus, Prosartes D.Don, and be moved to the family Liliaceae [51]. The family Colchicaceae as currently circumscribed thus constitutes 15 genera and about 280 species, of which more than half belong to Colchicum [48].

ALKALOID DIVERSITY

A number of reviews on phenethylisoquinoline alkaloids can be found in the literature [eg. 52]. However, they usually only treat a particular group of a specific structural type and will be exemplified under the respective alkaloid type below. The present contribution provides a review of Colchicaceae alkaloids of all types as well as perspectives of their origin, evolution and hypothetical biosynthetic pathways.

In the Combined Chemical Dictionary [53] there are about 150 entries with structurally elucidated alkaloids reported from Colchicaceae genera. They are biosynthetically derived from condensation of phenylalanine and tyrosine to a phenethylisoquinoline precursor, which has been shown by several feeding experiments in elucidation of colchicine biosynthesis [54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65].

Based on proposed biosynthesis, feeding experiments and analogies of pathways elucidated for benzylisoquinoline alkaloids, as presented in Fig. (), we discern eight distinct structural types of alkaloids in the family Colchicaceae: phenethylisoquinolines, homoproaporphines, homoaporphines, androcymbines (homomorphinans), colchicines, allocolchicines, lumicolchicines and homoerythrinans. Phenethylisoquinolines and homoerythrinans have also been isolated from Dysoxylum in the mahogany family Meliaceae, which also contain a dibenzo[d,f]azecine alkaloid [53, 66, 67, 68, 69]. The homoerythrinan alkaloids also have additional disjunct occurrences in the monogeneric New Caledonian family Phellinaceae [53, 70] and the gymnosperm families Taxodiaceae, Podocarpaceae and Cephalotaxaceae [53, 71, 72]. Another group of alkaloids derived from phenethylisoquinoline precursors are the cephalotaxines unique for the family Cephalotaxaceae, which has recently been reviewed [73]. Thus, in total there are ten structural types of alkaloids with a possible common precursor. The hypothetical network of their interrelations as presented in Fig. (), builds on possible transformations known to occur in biochemical reactions and transforming one structural type into another. The included diaryl-3-azahexanoid precursors could constitute alternative biosynthetic pathways to homoerythrinan alkaloids explaining their disjunct distribution.

Phenethylisoquinolines (Type A)

Six phenethylisoquinoline alkaloids have been described from Colchicaceae, see Fig. (), of which autumnaline is proposed as a key intermediate. There are only few reports of isolation of these alkaloids [74, 75, 76, 77, 78], perhaps due to their proposed functions as precursors and thus efficient transformation into other structural types. They do not seem to have been tested for biological activity, but phenethylisoquinolines from Dysoxylum have shown effects on isolated rat cardiac muscle [79].

Homoproaporphines (Type B)

These minor alkaloids, see Fig. (), have been isolated primarily from the Eurasian species of Colchicum [80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96,] but are also present in African [97] and Australian taxa [98]. Several homoproaporphines have been investigated for cholinergic activity [99, 100], showing moderate anticholinesterase activities.

Homoaporphines (Type C)

These alkaloids were last reviewed in 1989, see Fig. (), [101], and very few have been investigated for biological activity [100].

Androcymbines (Type D)

This group, see Fig. (), is sometimes called homomorphinandienones. The isolation and structure elucidation of androcymbine and O-methylandrocymbine proved to be essential for understanding the biosynthetic origin of colchicine, finally linking this “alkaloid sui generis” to the phenethylisoquinoline precursors [74, 102].

Colchicines (Type E)

This is the largest group of alkaloids within Colchicaceae as of today, see Fig. (). A statement that is true even if, as done here, the allo- and lumicolchicine derivatives are excluded. They commonly interact with tubulin and are considered to be a main toxic principle of Colchicaceae plants. They have also been evaluated for anticholinesterase activity [103]. This is the first isolated group of alkaloids containing a tropolone ring in their structure, a fact that has been used to distinguish them from other alkaloids in simple TLC screenings due to a positive Oberlin-Zeisel reaction when sprayed with FeCl3 after treatment with HCl vapor [41], and are thus sometimes known as tropolonic alkaloids. This name would however also include the unrelated tropoloisoquinoline alkaloids described from Abuta in the moonseed family Menispermaceae [104, 105, 106]. Colchicines were reviewed by Boyé and Brossi in 1992 [107], and have not been updated since. A further ten colchicine-type alkaloids have been described as minor components from Colchicum and Gloriosa species since that review [78, 108, 109, 110, 111]. Even though colchicine itself is deemed to be too toxic as an anti-cancer drug, investigation of compounds binding to the same site and interacting with tubulin is an active field of research [112]. There is also a continued interest in modification of the colchicine-type core to modify toxicity and activity of these compounds [113, 114, 115, 116].

Allocolchicines (Type F)

These alkaloids are also called dibenzocycloheptylamines, see Fig. (), and have been suggested to be catabolic metabolites of colchicines [117, 118]. They usually retain the colchicine-type alkaloids ability to bind to tubulin but are less toxic, which has yielded some interest as potential antitumor drugs [34, 119].

Lumicolchicines (Type G)

There is a debate whether these alkaloids, see Fig. (), are produced in the plant or simply artifacts from the isolation since they can also be formed as photoisomers from colchicine-type alkaloids under influence by light. In contrast to allocolchicines the lumicolchicine-derivatives lack specific binding to tubulin.

Homoerythrinans (Type I)

This group of alkaloids has a very narrow distribution within Colchicaceae and has only been found in the Australian genera Schelhammera and Kuntheria [120, 121, 122, 123, 124], of which the latter is a relatively recent taxonomic split from the former [125]. However, among the phenethylisoquinoline alkaloids this is a very widespread group present in Dysoxylum of Meliaceae [66], Phelline of Phellinaceae [70], Athrotaxis of Taxodiaceae [71], Manoao of Podocarpaceae [72], and Cephalotaxus of Cephalotaxaceae [73]. Some homoerythrinan alkaloids can even be found in several of these families, eg. taxodine and 3-epischelhammericine. The ten homoerythrinan-type alkaloids found in Colchicaceae, see Fig. (), have not been studied for biological activity in any great detail. The alkaloid 3-epischelhammericine as well as other alkaloids from Dysoxylum have shown molluscicidal activity and cardiac effects [67, 79], and dyshomoerythrine from Manoao is reported to have insecticidal activity [126].

Miscellaneous Alkaloids

Five alkaloids, see Fig. (), reported from Colchicaceae have structures not completely consistent with any of the previous groups. They are all isolated from different species of Colchicum [110, 127, 128, 129].

Phylogeny of Colchicaceae and Alkaloid Distribution

A phylogenetic hypothesis of plant relationships is dependent on sampling size, which is a function of the number of investigated plant species and the number of characters in the matrix. In molecular systematic studies based on nuclear sequences the characters will equal the number of nucleotides in the sequence used for analysis. For Colchicaceae a family wide investigation based on chloroplast gene sequences, representing fourteen out of fifteen genera (the single species of Kuntheria was not sampled) has been published [47]. A genus-level tree based on those results, with Kuntheria placed together with Schelhammera and Tripladenia, is shown in Fig. ().

Chemosystematic prediction is likewise dependent on sampling size, here as a function of the number of investigated plant species and the limits of detection for the compounds of interest using a specific method. As the Colchicaceae alkaloids span from alkaline compounds, over aprotic or quaternary nitrogen compounds insensitive to pH-changes, to phenolic alkaloids even within the different structural types the isolation method may bias the detection of compounds [130]. Another bias in reviewing Colchicaceae alkaloids may be the strong focus on detection of colchicine itself in the literature. This alkaloid is by far the most commonly reported compound, and it has been shown to be present in all investigated genera of the family by nanospray-MS (again with the single species of Kuntheria unsampled) [49].

However, the similarities between the systematic and chemical sampling on genus level makes Colchicaceae a good case study in evolution of chemical characters. The genera Uvularia and Disporum have been treated as part of the now defunct family Uvulariaceae, which by some authors also have included Schelhammera (including the later segregated Kuntheria) and Tripladenia, though these latter genera have also been placed in the former Convallariaceae [46, 131]. Burchardia has been reported to be devoid of tropolonic alkaloids [132], and its relationship to the rest of Colchicaceae have been questioned and sometimes it is instead treated in a monogeneric family Burchardiaceae [133]. The alkaloid chemistry of the early diverging lineages, Burchardieae and Uvularieae, has not been investigated in any detail according to standard literature searches, but the three genera are confirmed to contain colchicine [49].

The chemistry of the tribe Tripladenieae, which is restricted to eastern Australia extending to New Guinea, have in comparison been fairly extensively investigated. The genera Schelhammera and Kuntheria are unique in Colchicaceae by producing homoerythrinan instead of colchicine-type alkaloids as major components [120, 121, 122, 123, 124]. Further investigations are needed to see if this is the case for Tripladenia as well. However, as stated above a plant under the name of Kreysigia multiflora has been shown to contain appreciable amount of homoaporphine, homoproaporphine and colchicine-type alkaloids [50, 134, 135], and this plant is either Tripladenia or a species of Schelhammera. The collection data for the material used by Badger and Bradbury [134] indicate that their original isolation is done on Tripladenia cunninghamii. Our nanospray mass spectrometry screening has confirmed the presence of colchicine in both Schelhammera and Tripladenia [49].

The three remaining tribes constitute the classical subfamily Wurmbaeoideae [45] or Colchicaceae in its strict sense [46] and several reviews of their alkaloid contents have been published [41, 44, 52, 107, 136]. Besides colchicine- and lumicolchicine-type alkaloids Iphigenieae also contain homoaporphine and homomorphine alkaloids [101, 137, 138, 139], while only colchicine- and lumicolchicine-type alkaloids have been reported from the few species investigated from the tribe Anguillarieae [44, 132]. The most species-rich tribe is Colchiceae, which also contains Colchicum and Gloriosa the two most studied genera, and it is reported to contain all structural types present in Colchicaceae except for homoerythrinan alkaloids [97, 101, 108, 109, 111, 117, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156].

Presence of Pathways

Combining this distribution data with the hypothetical biosynthetic network in Fig. () indicates that the pathway to colchicine-type alkaloid is present in all genera of Colchicaceae. Since this pathway contains the androcymbine-type alkaloids as an intermediate, these are also expected to have a family-wide distribution. The homoaporphine and homoproaporphine alkaloids are both present in the tribe Colchiceae, while only homoaporphines are present in the tribe Iphigenieae. This could indicate that the two types are biosynthesized by independent pathways, and that homoproaporphines are not precursors for homoaporphines as would be suggested by analogy with benzylisoquinoline alkaloids. There is some experimental evidence that this is the case [57].

The homoerythrinan alkaloids are often discussed together with their erythrinan counterparts [157, 158, 159], and their biosynthesis is considered to be analogous to these. The two possible pathways to homoerythrinan alkaloids in Fig. () could explain the disjunct distributions of these alkaloids without co-occurrence of androcymbine- or colchicine-type alkaloids outside Colchicaceae. A pathway analogous to that proposed for erythrinan alkaloids [159, 160] would include androcymbine-type intermediates, while a pathway from diaryl-3-azahexenoid-type precursors would by-pass these and directly produce dibenzo[d,f]azecines.

CONCLUSION AND FURTHER STUDIES

The raison d’être for alkaloids is still something of an enigma, and hypotheses have included views such as metabolic waste or end products, as a source of stored nitrogen, and as defense compounds or ecological adaptors [161, 162]. Regardless of the validity of any of these hypotheses, the large structural variation and the high proportion and wide range of biological activities associated with alkaloids make them a highly interesting compound-type for medicinal chemists. However, the research focus is often limited to a very small number of compounds with extraordinary activity, such as colchicine in the present example. This might be due to presence of low concentrations of minor components, bias in selection of screening assays for biological activity, or skewed sampling of plant material. We argue that by accepting the notion that natural products are prevalidated for biological activity and taking evolutionary aspects into account, these potential problems can be attenuated and open new avenues of interesting research.

Of the phenethylisoquinoline alkaloids of Colchicaceae only colchicine-type alkaloids have been, and is still, extensively investigated and the activity under scrutiny has almost exclusively been that of cytotoxicity. This leaves seven structural groups of alkaloids within the family for which we today have very little information on possible biological activities and roles. Especially interesting from this perspective are the homoerythrinan alkaloids, seemingly absent in most of the family and replacing colchicine-type alkaloids in Tripladenieae. With the presence of these alkaloids in several distantly related plant groups it is imperative to predict that they have biological activities of potential for human use, and the reported molluscicidal and insecticidal activities can be considered as a starting point and need further investigations.

From an evolutionary standpoint it is also obvious that the sampling of Colchicaceae has been very skewed. Almost 70% of investigated species are from the genus Colchicum, and several genera lack in depth studies of their alkaloid content. This is notable for the early lineages as well as the second most species rich genus Wurmbea. Compared to the structural diversity of known benzylisoquinoline and Amaryllidaceae alkaloid types, which have at least analogous biosynthesis patterns, the phenethylisoquinoline alkaloids of Colchicaceae is a small group. This could be the result of undersampling of different phylogenetic lineages within the family.

It has been argued that we only take advantage of a very small fraction of the plant biodiversity around us. As an example De Luca and co-workers propose in depth collaborative investigations of plant metabolomes, using new technologies to elucidate and modify biosynthetic pathways to create larger yields or new chemically diverse biological active compounds [163]. Another option for finding new compounds or activities, as argued here, is taking advantage of evolutionary evidence in selection of the plants or compounds to be studied. The phylogenetic distribution of specific compounds predicts at least one of either presence of biosynthetic pathways or advantageous biological activity, thus identifying less known or unstudied chemistry. Based on the known phylogenetic hypothesis in Fig. () a number of hypotheses for further study can be identified, such as: the need to re-investigate Burchardia, including identification of the reported non-tropolonic alkaloid; perform studies on the presence or absence of alkaloids in the tribe Uvularieae, which at present is practically uninvestigated; an in depth study of the homoerythrinan alkaloids of the tribe Tripladenieae and their biological activities; and instigate molecular and biochemical comparison between species from Tripladenieae and Colchicaceae sensu stricto to investigate biosynthetic pathways.

In conclusion we have shown how thinking about evolution can influence selection of plant material in drug lead discovery, and how knowledge about phylogenetic relationships may be used to evaluate predicted biosynthetic pathways and identify priority species for further study.

Table 1.

Classification of the Family Colchicaceae [47], Including Selected Generic Synonyms Present in the Chemical Literature.

Tribe	Genus	Synonyms
Burchardieae	Burchardia R.Br.
Uvularieae	Disporum Salisb.
	Uvularia L.
Tripladenieae	Kuntheria Conran & Clifford
	Schelhammera R.Br.
	Tripladenia D.Don
Iphigenieae	Camptorrhiza E.Phillips	Iphigeniopsis Buxb.
	Iphigenia Kunth
Anguillarieae	Baeometra Salisb. ex Endl.
	Wurmbea Thunb.	Anguillaria R.Br. Dipidax Lawson ex Salisb. Neodregea C.H.Wright Onixotis Raf.
Colchiceae	Colchicum L.	Androcymbium Willd. Bulbocodium L. Merendera Ramond
	Gloriosa L.	Littonia Hook.
	Hexacyrtis Dinter
	Ornithoglossum Salisb.
	Sandersonia Hook.