Keywords

2.1 Introduction

Plant natural products are frequently used as chemotaxonomic markers which are indicative of select members of a family, genus, and/or species. However, the erratic occurrence of some natural products raises questions about their biosynthetic origin and significance as chemotaxonomic markers. Four mechanisms could explain the sporadic occurrence of natural products in unrelated taxa [1]: First, plant taxa may have lost their ability to produce the natural product; second, the biosynthetic pathways of a natural product may have originated multiple times over evolutionary history; third, the genes responsible for the biosynthesis of a natural product may have been horizontally transferred between unrelated taxa [2]; or fourth, the natural product may be produced by a microbe associated with several, unrelated plant species.

Fungal endophytes are microbes that live within plants for part of their life cycle without causing any apparent disease symptoms [3]. Endophytes differ in their host range, tissues colonized, in planta biodiversity, in planta colonization, mode of transmission, and fitness benefit conferred to the host [4]. These criteria have been used to separate endophytes into different classes. For a more detailed review of the proposed classes of endophytes and their diagnostic criteria see Rodriguez et al. [4].

Recent research has shown that fungal endophytes associated with plants are a rich source of natural products [4, 5]. The objective of this review is to highlight natural products found in plants that are reported to be derived from fungal endophytes and, when appropriate, briefly comment on the plant-endophyte interaction. We will first summarize current knowledge on alkaloids synthesized by Clavicipitaceae endophytes (Sect. 2.2), then on diverse other secondary metabolites (Sect. 2.3). Specifically, we will highlight the role that fungal endophytes play in the synthesis of the indolizidine alkaloid swainsonine and the interaction between host and endophyte (Sect. 2.4). The review will not highlight plant mycorrhizal relationships.

2.2 Alkaloids Derived from Clavicipitaceous Endophytes

The most studied fungal endophyte system is the symbiotic association of the fungal endophytes in the genera Epichloë and Neotyphodium (Clavicipitaceae) with many cool season grasses (Poaceae). Neotyphodium species are asexual and grow within the intercellular spaces of their grass hosts, while Epichloë species represent the sexual states of several Neotyphodium species. Neotyphodium species are strictly vertically transmitted while Epichloë species are horizontally and vertically transmitted. Epichloë species are distinguished from Neotyphodium species because they are capable of exiting their plant hosts via the formation of sexual reproductive stroma on plant inflorescences.

Epichloë and Neotyphodium species may produce four classes of bioactive metabolites in their symbiotic associations with plants: ergot alkaloids , indole diterpenes, loline alkaloids, and peramine (Fig. 2.1). These four classes of alkaloids are derived from amino acid precursors, and the pathways are independent of one another. No individual fungal endophyte has been reported to produce representatives of all four classes; most produce metabolites belonging to one to three of the chemical classes [5].

Fig. 2.1
figure 1

Chemical structures of representative alkaloids produced by Clavicipitaceae endophytes associated with the grasses and morning glory families

Full size image

The diverse suite of metabolites in the ergot alkaloid family can be grouped as clavines, simple amides of lysergic acid, or ergopeptines with the classification based on the structural complexity and position in the pathway [6, 7]. Like the ergot alkaloids , the indole-diterpenes represent a suite of products including the terpendoles, lolitrems, and janthitrems derived from oxidation and prenylation of a shared biosynthetic precursor, terpendole I [8, 9]. The lolines represent a family of aminopyrrolizidine alkaloids, derived from homoserine and proline. Lastly, peramine represents a single alkaloid rather than a family of alkaloids and is derived from a dipeptide possibly made up of arginine and a precursor to proline. Epichloë and Neotyphodium endophytes of grasses and the alkaloids associated with this interaction affect herbivores and significantly impact ecological communities. For a more detailed commentary on each of these alkaloids and their biology and the effects of this symbiotic interaction, readers are referred to Schardl et al. [5, 10].

Not only are ergot alkaloids found in cool season grasses, but they are also present in select taxa of the Convolvulaceae, the morning glory family [11]. The three major types of ergot alkaloids have been detected in the Convolvulaceae [11], as well as the lolines in one taxon of the Convolvulaceae, Argyreia mollis [12]. In regard to the other bioactive metabolites found in grasses, there are no published reports of any Convolvulaceae taxa containing the indole diterpenes or peramine. However, there are a number of reports of livestock having a tremorgenic syndrome caused by feeding on select Ipomoea species including Ipomoea muelleri and Ipomoea asarifolia [13, 14]. These cases suggest that Ipomoea species may contain the indole diterpenes, as they are known to be tremoregenic, but this remains to be verified. Recent research has demonstrated the presence of clavicipitaceous epibiotic fungal symbionts described as Periglandula species in Ipomoea taxa [1, 15–17]. These fungi are vertically transmitted and appear to have a narrow host range, as the described species to date occur on different plant hosts. Periglandula species are the only clavicipitaceous fungi known to be associated with a dicot host [15]. The ecological effects of this interaction have yet to be studied; however, effects are likely to be substantial due to the bioactivities of the endophyte-derived alkaloids.

2.3 Other Natural Products

Numerous other natural products in plants are reported to be produced by fungal endophytes associated with the host. Some examples include paclitaxel (also known as taxol) , podophyllotoxin, deoxypodophyllotoxin, camptothecin and structural analogues, hypericin and emodin, and azadirachtin (Fig. 2.2). In each of these cases, the compounds are thought to be produced by the plant as well as an endophyte that has been isolated from the plant host. Furthermore, in the cases of paclitaxel and camptothecin, the biosynthetic pathways in planta have been described, and several enzymes in the pathways have been characterized.

Fig. 2.2
figure 2

Chemical structures of other highlighted natural products reported to be produced by endophytes

Full size image

2.3.1 Taxol

Taxol, a taxane-type diterpene, is a tetracyclic lactam found in the genus Taxus [18]. All Taxus species are reported to contain taxanes. Taxol is an anticancer drug that acts as a microtubule inhibitor. Taxol was first reported to be produced by Taxomyces andreanae, an endophytic fungus associated with Taxus brevifolia [19]. Since then, more than 20 genera that produce taxol have been isolated, representing several species of unrelated endophytic fungi [20]. In several cases, more than one endophytic isolate from the same host has been reported to produce taxol. In one study of 109 endophytic isolates cultured from Taxus chinensis, 28 isolates were reported to produce taxol [21]. Taxol-producing endophytes have been isolated from various hosts including Taxus species, cypress [22], Wollemia pine [23], and a variety of other non-Taxus species [24, 25]. These hosts include some that are reported to contain taxol and some that are not. None of the taxol-producing endophytes has been reported to contain any intermediates that are detected in planta for the taxol biosynthetic pathway. Furthermore, there are no reports of plant taxol biosynthetic enzymes found in fungi. In the cases where the endophytic isolates were isolated from Taxus species, the authors propose that horizontal gene transfer was responsible for the biosynthetic machinery of the plant being transferred to the fungus, although no data are presented in support of this hypothesis. In cases where taxol-producing endophytic isolates were isolated from hosts that are reported not to produce taxol , the authors suggest that the endophytic isolate may have been associated with a Taxus species, but through time came to be associated with a different host [23].

2.3.2 Podophyllotoxin

Podophyllotoxin , podophyllotoxin glycoside, deoxypodophyllotoxin, and related analogues are present in four plant families: the Berberidaceae, Cupressaceae, Polygalaceae, and Linaceae [26]. These compounds are aryl tetralin lignans, and have various bioactivities including but not limited to antiviral [27], anticancer [28], antiproliferative [29], and insecticidal [30] properties. In fact, podophyllotoxin is in great demand as a precursor in the synthesis of topoisomerase inhibitors that are anticancer drugs. Recently, endophytes have been isolated from both Podophyllum and Juniperus species that are reported to produce these structurally related compounds [31–33]. In two Podophyllum species, two unrelated endophytes were identified: in Podophyllum hexandrum the endophyte Trametes hirsuta was identified that produces podophyllotoxin, podophyllotoxin glycoside, and deoxypodophyllotoxin [32], whereas in Podophyllum peltatum the endophyte Phialocephala fortinii was identified that produces podophyllotoxin [31]. In Juniperus communis, the endophyte Aspergillus fumigatus was identified and reported to produce deoxypodophyllotoxin [33]. Interestingly, in the latter two examples the authors do not mention if the endophyte produces any of the other structurally related analogues. In all three cases, the authors propose that horizontal gene transfer was responsible for the biosynthetic machinery from the plant being transferred to the fungus, although no data are presented in support of this hypothesis.

2.3.3 Hypericin

Hypericin is a naphtodianthrone derivative of significant medicinal value isolated from St. John’s Wort, Hypericum perforatum [34]. Hypericin is reported to have a number of bioactivities including but not limited to being an antidepressant [35], anti-inflammatory, and antiviral [36]. Hypericin is a photodynamic compound activated by light, and animals consuming St. John’s Wort can suffer from primary photosensitization due to hypericin. Emodin is thought to be a biosynthetic precursor of hypericin [36, 37]. Recently, an endophyte isolated from the inner stem tissues of Hypericum perforatum and found to produce both emodin and hypericin was identified as Thielavia subthermophila [37].

2.3.4 Camptothecin

Camptothecin , a pentacyclic quinoline alkaloid, has been reported from several genera in unrelated angiosperm families [38]. For example, camptothecin has been reported in Camptotheca spp. (Nyssaceae) [39], Ophiorrhiza spp. (Rubiaceae) [40], Nothapodytes foetida, and Apodytes dimidiate (Icacinaceae) [41], and two genera of the Apocynaceae and Gelsemiaceae [42, 43]. Camptothecin and its structural analogues 9-methoxycamptothecin and 10-hydroxycamptothecin are potent neoplastic compounds that are important because they inhibit the enzyme topoisomerase I which is required for DNA replication and transcription [38]. In fact, two semisynthetic drugs targeting different cancers are derived from camptothecin [38]. Endophytes isolated from Nothapodytes foetida, Apodytes dimidiate, and Camptotheca acuminata have been reported to produce camptothecin [44–47]. Two unrelated endophytes have been isolated from Nothapodytes foetida that produce camptothecin [44, 45]. Interestingly, separate strains of Fusarium solani that produce camptothecin were isolated from two unrelated plants (Apodytes dimidiate and Camptotheca acuminata) [46, 47]. In two of the above examples, the isolated endophytes produce camptothecin and its structural analogues [46, 47] while in the other two examples camptothecin is produced, but the authors do not mention if the endophyte produces other structurally related compounds [44, 45].

The endophyte isolated from Camptotheca acuminata was particularly intriguing because it lost its ability to produce camptothecin at the third subculture and continued as such to the seventh generation [46]. To further investigate the loss of camptothecin synthesis in the endophyte isolated from Camptotheca acuminata, the authors took advantage of the fact that the biosynthetic pathway of camptothecin is known and several enzymes in the pathway have been described [48]. The authors identified geraniol 10-hydroxylase, secologanin synthase, and tryptophan decarboxylase in the Fusarium solani strain isolated from Camptotheca acuminata [48], which were 100 % identical to the host enzymes. The authors found no evidence of the strictosidine synthase in the endophyte, the enzyme in the biosynthethic pathway of camptothecin that forms strictosidine from tryptamine and secologanin. The authors proposed a cross-species biosynthetic pathway where the endophyte uses the host strictosidine synthase. Furthermore, the authors showed that the genes involved in camptothecin biosynthesis in the endophyte had undergone random nonsynonymous mutations from generation 1 to generation 7 to further explain the inability of the endophyte to produce camptothecin and its precursors [48].

2.3.5 Azadirachtin

Azadirachtin A and its structural analogues are highly oxygenated tetranortriterpenoids isolated from the Indian neem or Indian lilac, Azadirachta indica [49]. These compounds are reported to have various bioactivities, including as insecticides that act as antifeedants and have growth-regulating properties [50]. They are found in all parts of the plant, with the highest concentrations in the seeds [49]. Recently, an endophytic fungus identified as Eupenicillium parvum was isolated from the plant host, Azadirachta indica , that was capable of producing azadirachtin A and B but not the other structural analogues [51].

2.4 Swainsonine

Several species in the legume (Fabaceae) genera Astragalus, Oxytropis, and Swainsona are toxic to grazing livestock in the Americas, Asia, and Australia [52–55]. Locoism, a disease induced by prolonged grazing of some Astragalus and Oxytropis species generally called locoweeds, was first noted by the Spanish conquistadors, and again during the settlement of western North America by pioneers [52, 53, 56]. Clinical signs and pathology of locoism, a neurologic disease, are similar in animals intoxicated by locoweed species and Swainsona species [57, 58]. Swainsonine (Fig. 2.3), a trihydroxyindolizidine alkaloid, was first identified as the bioactive principle of the neurologic disease in Swainsona canescens, a legume native to Australia [59], and subsequently identified as the active principle in locoweeds [60]. Swainsonine inhibits the enzymes α-mannosidase and mannosidase II, resulting in lysosomal storage disease and altered glycoprotein synthesis [61, 62].

Fig. 2.3
figure 3

Chemical structure of swainsonine

Full size image

A fungal endophyte, Undifilum oxytropis, previously described as an Embellesia species [63], is reported to produce swainsonine in locoweeds [64, 65]. The Undifilum genus (Pleosporaceae) is related to the genera Alternaria, Embellesia, and Ulocladium [65]. In general, Undifilum species are associated with swainsonine-containing Astragalus and Oxytropis species in North America and China [65–67]. Other species of Undifilum have been found and characterized in the swainsonine-containing plant species Astragalus lentiginosus and Astragalus mollissimus [67]. Undifilum bornmuelleri, a pathogen of the legume Securigera varia [65], is reported to not produce swainsonine. Undifilum species associated with locoweeds are vertically transmitted and have no apparent sexual stage [68]. Undifilum species appear to have a narrow host range, as different plant species are associated with unique Undifilum species [65, 67].

Swainsonine concentrations vary greatly among plant species, varieties and populations, often ranging from not being detected to greater than 0.2 %. For example, Astragalus locoweed species generally have greater swainsonine concentrations than Oxytropis locoweed species in North America, although this depends on the specific taxa being compared [69]. Swainsonine concentrations vary greatly among different varieties of some Astragalus and Oxytropis species. For example, some Oxytropis lambertii and Astragalus mollissimus varieties contain swainsonine, while others contain very little or no swainsonine [69, 70]. Additionally, swainsonine concentrations vary greatly among populations, for example some populations of O. lambertii var. bigelovii contain swainsonine while others contain very little or no swainsonine. Furthermore, within toxic populations of locoweeds, swainsonine concentrations vary greatly among individual plants.

Initial reports demonstrated that the endophyte usually could be cultured only from locoweed samples, with swainsonine concentrations greater than 0.01 %, but could be detected by polymerase chain reaction (PCR) in all samples regardless of swainsonine concentration [69]. To further describe the plant–endophyte–swainsonine relationship and to investigate the differences in swainsonine concentrations among individual plants in locoweed populations, a method was developed to quantitate the endophyte in plant samples. Using quantitative polymerase chain reaction (qPCR), the amount of endophyte was measured in the Astragalus and Oxytropis locoweeds with different swainsonine concentrations. Two chemotypes of plants were identified, namely chemotype 1 plants, which contain swainsonine concentrations greater than 0.01 %, and chemotype 2 plants, which have concentrations below 0.01 % (generally near 0.001 % or not detected) (Fig. 2.4a) [71, 72]. These two chemotypes were determined to differ significantly in the amount of endophyte they contain, which may help to explain the difference in swainsonine concentrations of the respective host plants (Fig. 2.4b) [71, 72].

Fig. 2.4
figure 4

Comparison of swainsonine and endophyte concentrations among plant parts and between two chemotypes of Oxytropis sericea. Mean swainsonine (%) and endophyte (pg/ng total DNA) concentrations ( ± SE) a from different plant parts (root, crown, scape, leaf, and flower/pod) and b from two chemotypes of O. sericea plants [71]

Full size image

Swainsonine and endophyte concentrations have been shown to be influenced by the plant part [71, 72] and phenological stage of the plant [73, 74]. Swainsonine was found in all plant parts, with greater concentrations in aboveground parts than in below-ground parts (Fig. 2.4a); the endophyte Undifilum was also found in all plant parts, again with greater concentrations in aboveground parts than in the root (Fig. 2.4b) [71, 72]. The root crown has endophyte amounts equivalent to aboveground parts, which we believe may serve as a reservoir for the endophyte for the following year’s growth, as many locoweeds are perennial plants [71, 72]. Swainsonine concentrations are greatest in floral parts and seeds [73]. This is consistent with the optimal defense theory of protecting reproductively valuable plant and endophyte tissues. Swainsonine and endophyte amounts in aboveground parts increase throughout the growing season until the plant reaches maturity [73, 74]. As plants senesce, swainsonine concentrations decrease significantly [73]. Swainsonine and endophyte amounts are highly correlated over the growing season, suggesting the endophyte amount plays a critical role in determining swainsonine concentration in the plant [73].

Three mechanisms were proposed that may explain the origin of chemotype 2 plants in Astragalus and Oxytropis locoweeds [72]. First, different endophyte genotypes distributed across host plants may account for differences in swainsonine accumulation. However, recent work does not support this suggestion, as we have reported that the internal transcribed spacer (ITS) of the rDNA region of Undifilum spp. from three locoweed species are identical between chemotype 1 and 2 plants [72].

Second, plants with no detectable or very low swainsonine concentrations may have arisen, due to imperfect transmission of some critical amount of the endophyte to the seed or seedling [72, 75] that is required for new plants to be colonized and produce concentrations of swainsonine characteristic of chemotype 1 plants. To test this, the transmission of the endophyte in these two chemotypes was investigated. Seeds derived from chemotype 1 and chemotype 2 plants were germinated and the chemical phenotype of the resulting plants was evaluated. Progeny derived from the seeds of chemotype 1 plants had swainsonine concentrations consistent with both chemotype 1 and chemotype 2 plants, demonstrating imperfect transmission. All progeny derived from seeds of a chemotype 2 plant had swainsonine concentrations consistent with chemotype 2 plants [76], suggesting that these seeds may lack some critical amount of endophyte.

Third, low swainsonine chemotype 2 plants may suppress endophyte growth and thus alkaloid production due to a plant genotype by endophyte interaction similar to that observed in grasses [72, 77, 78]. To investigate this possibility further, swainsonine and endophyte amounts were investigated in seeds derived from chemotype 1 and chemotype 2 plants. Seeds derived from a chemotype 1 plant had greater swainsonine and endophyte amounts than seeds derived from a chemotype 2 plant. We hypothesized that seed endophyte amounts were a critical factor in determining the swainsonine concentration in the resulting plant. To address this hypothesis, we investigated if plants derived from chemotype 1 seeds could have swainsonine concentrations like chemotype 2 plants by reducing the amount of Undifilum in chemotype 1 seeds [79]. Furthermore, we investigated if plants derived from chemotype 2 seeds could have swainsonine concentrations like chemotype 1 plants by inoculating chemotype 2 embryos with Undifilum [79]. Plants derived from chemotype 1 seeds, where the amount of endophyte was reduced by seed-coat removal and fungicide treatment, had swainsonine and endophyte amounts consistent with chemotype 2 plants (Fig. 2.5). On the other hand, plants derived from Undifilum-inoculated chemotype 1 and chemotype 2 embryos had swainsonine and endophyte amounts consistent with chemotype 1 plants (Fig. 2.5). Similar results were seen for A. lentiginosus and A. mollissimus. This suggests that endophyte amount at seed germination is a determining factor if a plant exhibits a swainsonine chemotype 1 or 2. If the plant endophyte by genotype interaction is a key determinant, one may expect that plants derived from inoculated chemotype 2 embryo would have swainsonine and endophyte amounts similar to a chemotype 2 plant.

Fig. 2.5
figure 5

Swainsonine concentrations in Oxytropis sericea across different seed treatments. Mean swainsonine concentrations (%, ± SE) from plants grown after each seed treatment: C1 CTL (plants derived from control chemotype 1 seeds), C2 CTL (plants derived from control chemotype 2 seeds), C1 RSC (plants derived from chemotype 1 seeds where the seed coat was removed), C1 FUNG (plants derived from chemotype 1 fungicide-treated seeds), C1 HEAT (plants derived from chemotype 1 heat-treated seeds), C1 RSC INOC (plants derived from endophyte-inoculated chemotype 1 embryos), and C2 RSC INOC (plants derived from endophyte-inoculated chemotype 2 embryos). Y axis shows a break from 0.005 % swainsonine to 0.1 %. Different letters above each bar represent significant differences at P < 0.05 [77]

Full size image

Concentrations of swainsonine are reported to differ among populations of Oxytropis sericea. For example, mean concentrations were approximately three times greater in plants from Cuchara, CO (0.40 %) than plants from Park Valley, UT (0.15 %) [80]. Differences in concentrations among populations may be due to the environment, genotype of the plant, and/or endophyte, or some combination of these factors. To identify the contributing factor(s) responsible for the observed swainsonine differences, plants were grown in a common garden. Swainsonine concentrations were different between the two populations in plants grown in a common garden [80]. The endophyte cultured from each population produced different amounts of swainsonine in vitro when grown under common culture conditions [80]. Furthermore, a cross-inoculation experiment was performed [80], wherein the endophyte from one population was inoculated to the embryo from the other location, and vice versa. Results from this experiment and others clearly demonstrated that the endophyte amount was responsible for differences in swainsonine concentration between the two populations, not plant genotype or differences in endophyte amount.

Limited studies exist regarding the ecological benefit or role of swainsonine and the influence of environmental conditions on swainsonine production. Swainsonine does not deter grazing; in fact, animals take 2–3 weeks to show clinical signs and continue grazing locoweeds after becoming intoxicated [81]. Furthermore, swainsonine concentrations are reported not to change in response to the clipping used to simulate herbivory [82]. Swainsonine has not been reported to be bioactive against insects, fungi, or bacteria; one preliminary study suggested that it has no effect on a native weevil herbivore of locoweeds [83]. Legumes are known for forming symbioses with nitrogen-fixing bacteria, and investigators found that swainsonine concentrations were greater in plants inoculated with one strain of Rhizobium but not others, which suggested an interaction between the two classes of symbionts [84]. An alternative interpretation is that the host may have increased substrate availability for swainsonine production due to the improved nitrogen status. However, another report showed no consistent differences in swainsonine concentrations across locoweed taxa with deficient or adequate nitrogen, when nitrogen was supplied through fertilizer [85]. Lastly, water stress was shown to result in increased swainsonine concentrations in some locoweed species, but not others [86]. Thus, the ecological role of swainsonine for the endophyte and/or plant remains unclear, and swainsonine production does not appear to respond in a consistent manner to variation in environmental conditions experienced by the plant.

The following observations support that swainsonine is a fungal-endophyte-derived secondary metabolite in locoweeds, and is not produced by the host plant: (1) Astragalus and Oxytropis species that contain swainsonine are infected with Undifilum species; (2) plants derived from Astragalus and Oxytropis embryos where the seed coat was removed have no detectable swainsonine, or have concentrations less than 0.001 % [68, 79]; (3) plants derived from fungicide-treated Astragalus and Oxytropis seeds have no detectable swainsonine or have concentrations less than 0.001 % [79]; (4) Undifilum species isolated from locoweeds produce swainsonine in vitro [64]; (5) plants derived from Undifilum-inoculated embryos have swainsonine concentrations greater than 0.01 % [79]; and (6) rats fed U. oxytropis developed lesions and clinical signs similar to those fed swainsonine-containing O. lambertii [87].

In addition to the Astragalus and Oxytropis species, swainsonine has been reported in the Fabaceae genus Swainsona and two other plant families, the Convolvulaceae and the Malvaceae. As mentioned previously, swainsonine was first isolated from S. canescens, and some other Swainsona species including Swainsona galegifolia and Swainsona greyana contain swainsonine [88]. In the Convolvulaceae, some Ipomoea and Turbina species, including Ipomoea carnea, Ipomoea riedelli, Ipomoea sericophylla, and Turbina cordata have been reported to contain swainsonine [89–91]. Only a single species of Malvaceae, Sida carpinifolia, has been reported to contain swainsonine [92]. Like the legumes, swainsonine was identified in the plant species associated with these families due to livestock poisoning and subsequent economic impact. The presence of swainsonine in these species may be a case of convergent evolution or horizontal gene transfer; however, due to the sporadic occurrence of swainsonine in these genera, it seems likely that a swainsonine-producing fungal endosymbiont is associated with these taxa that contain swainsonine.

Swainsonine is reported to be produced by Undifilum species (Pleosporales) associated with locoweeds, and additionally by two phylogenetically disjunct fungi not associated with plants as endophytes. Rhizoctonia leguminicola (Cantharellales) is a fungal pathogen of red clover (Trifolium pratense) that causes black patch disease in the plant, and Metarhizium anisopliae is an entomopathogen that attaches to the outside of an insect, grows internally, and causes death (Hypocreales) [93, 94]. As a result, we hypothesized that a fungal endosymbiont may be associated with swainsonine-containing Swainsona and Ipomoea species. Furthermore, we speculated that it may be vertically transmitted and thus associated with seeds. To investigate this possibility, seeds from S. canescens and I. carnea were surface sterilized and placed on media to culture any seed-associated endophytes. Both plant species yielded a single endosymbiont that produced swainsonine in vitro. Each endosymbiont was detected by PCR and recovered by culturing from swainsonine-containing plant material from each respective species. Lastly, preliminary data suggest that these endophytes are vertically transmitted and have a narrow host range. The endosymbiont from S. canescens was characterized by morphologic and phylogenetic methods and represents a novel species of Undifilum . The endosymbiont from I. carnea belongs to the Ascomycete order Chaetothyriales and is not phylogenetically similar to any described fungi [95]. Furthermore, I. carnea plants derived from fungicide-treated seeds lack swainsonine, further demonstrating that the endosymbiont, but not the host plant, produces swainsonine [95]. These studies [95] provide more evidence that the presence of swainsonine in plants is associated with the presence of a vertically transmitted endosymbiont.

2.5 Summary

Recent research has increasingly shown that fungal endophytes associated with plants are a rich source of diverse natural products found in plants. The alkaloids, derived from the Clavicipitaceae endophyte association with grasses, and the morning glories represent one of the more well-studied examples of natural products derived from endophytes. The indolizidine alkaloid swainsonine and the association between swainsonine-containing plant hosts with the associated endophytic fungi represents a parallel example where only the endophyte, not the host plant, produces the alkaloid. One of the most striking differences between the two systems is that all vertically transmitted endophytic fungi that produce the ergot alkaloids, indole diterpenes, lolines, and peramine are derived from Clavicipitaceae , regardless of the plant family with which they are associated. In contrast, vertically transmitted endophytes that produce swainsonine are derived from different fungal orders that form relationships with specific plant families. In the light of this observation, the evolutionary history of the swainsonine biosynthetic pathway in these diverse fungi is particularly intriguing.

In regard to the other natural products discussed here, future work is needed to further describe the interaction between the plant hosts and fungal endophytes . In most of these examples, an endophyte was simply isolated from the plant host that produces the natural product of interest. Additionally, in most of these examples there was evidence that the plant host produced the natural product as well, so horizontal transfer of the biosynthetic pathway to the fungal endophyte was proposed. If these cases are verified as reported, this represents a tremendous example of horizontal gene transfer to various unrelated fungi in which all the biosynthetic genes are functionally transferred. These examples are thus fundamentally different from the Clavicipitaceae endophyte association and the Undifilum endophyte association wherein the endophytes found in related plant taxa are also related.