ABSTRACT
Symbiotic Epichloë species are fungal endophytes of cool-season grasses that can produce alkaloids with toxicity to vertebrates and/or invertebrates. Monitoring infections and presence of alkaloids in grasses infected with Epichloë species can provide an estimate of possible intoxication risks for livestock. We sampled 3,046 individuals of 13 different grass species in three regions on 150 study sites in Germany. We determined infection rates and used PCR to identify Epichloë species diversity based on the presence of different alkaloid biosynthesis genes, then confirmed the possible chemotypes with high-performance liquid chromatography (HPLC)/ultraperformance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS) and gas chromatography-mass spectrometry (GC-MS) measurements. Infections of Epichloë spp. were found in Festuca pratensis Huds. (81%), Festuca ovina L. aggregate (agg.) (73%), Lolium perenne L. (15%), Festuca rubra L. (15%) and Dactylis glomerata L. (8%). The other eight grass species did not appear to be infected. For the majority of Epichloë-infected L. perenne samples (98%), the alkaloids lolitrem B and peramine were present, but ergovaline was not detected, which was consistent with the genetic evaluation, as dmaW, the gene encoding the first step of the ergot alkaloid biosynthesis pathway, was absent. Epichloë uncinata in F. pratensis produced anti-insect loline compounds. The Epichloë spp. observed in the F. ovina agg. samples showed the greatest level of diversity, and different intermediates of the indole-diterpene pathway could be detected. Epichloë infection rates alone are insufficient to estimate intoxication risks for livestock, as other factors, like the ability of the endophyte to produce the alkaloids, also need to be assessed.
IMPORTANCE Severe problems of livestock intoxication from Epichloë-infected forage grasses have been reported from New Zealand, Australia, and the United States, but much less frequently from Europe, and particularly not from Germany. Nevertheless, it is important to monitor infection rates and alkaloids of grasses with Epichloë fungi to estimate possible intoxication risks. Most studies focus on agricultural grass species like Lolium perenne and Festuca arundinacea, but other cool-season grass species can also be infected. We show that in Germany, infection rates and alkaloids differ between grass species and that some of the alkaloids can be toxic to livestock. Changes in grassland management due to changing climate, especially with a shift toward grasslands dominated with Epichloë-infected species such as Lolium perenne, may result in greater numbers of intoxicated livestock in the near future. We therefore suggest regular monitoring of grass species for infections and alkaloids and call for maintaining heterogenous grasslands for livestock.
INTRODUCTION
Grasslands are some of the most species-rich ecosystems in central Europe (1). In these grasslands, more than 100 grass species are naturally infected with fungal endophytes from the genus Epichloë, and the estimated number of possibly infected grass species is approximately 900 (2, 3). Endophytic Epichloë species live as systemic, asymptomatic symbionts inside the plant (4). The endophyte-plant interaction can range from a mutualistic to an antagonistic symbiosis (5, 6). In the often-mutualistic symbiosis, the grass provides the endophyte with nutrients, shelter, and dispersal, while the fungus can provide drought and herbivore resistance for the grass (6). The herbivore resistance is associated with the production of bioactive alkaloids produced by the endophytic fungi, which can be toxic to vertebrates and invertebrates. Additionally, Epichloë spp. have been shown to also enhance host plant immune response, as infection with the fungus can change expression of up to one-third of host plant genes (7, 8).
Epichloë spp. can produce up to four different alkaloid classes (6, 9). The 1-aminopyrrolizidines (including lolines) and peramine are toxic or deterrent to insects, whereas the indole-diterpenes and ergot alkaloids are known to cause toxicity to vertebrates, in addition to possessing anti-insect properties (6, 9). The production of alkaloids depends on the symbiotic Epichloë species in the host grass. Several studies screening for Epichloë species in specific grass species showed that endophyte diversity, and thus also alkaloid diversity, can be high within a single grass species (10–13). The genes encoding each class of alkaloid have been identified, and for most genes, the relative step they encode within each biosynthetic pathway is known (9, 14, 15). Peramine synthesis is encoded by a single gene, perA (16), whereas the ergot alkaloids (EAS), indole-diterpenes (IDT), and lolines (LOL) each have multiple genes associated in gene clusters (9, 17). Much of the alkaloid diversity present in Epichloë is due to presence or absence of the alkaloid biosynthesis genes, which makes it possible to predict synthesized alkaloids based on detection of genes encoding key pathway steps (10, 12, 13, 18, 19).
Lolitrem B, an indole-diterpene, is produced by Epichloë festucae var. lolii in Lolium perenne (perennial ryegrass) plants, as well as by Epichloë sp. FaTG-2, present in Festuca arundinacea (tall fescue) (20, 21). Lolitrem B is known to cause ryegrass staggers, which is characterized by tremors in grazing livestock (22). Most notably, New Zealand and Australia have experienced outbreaks of ryegrass staggers, which are responsible for severe economic loss of livestock (2). These outbreaks typically occur in summer and autumn, due to endophytes producing alkaloids in association with the plant in response to high temperatures, and peak in late summer (23, 24). Other indole-diterpenes such as paxilline, a related indole-diterpene, can also have tremorgenic effects on livestock and should be considered in the evaluation of toxicity (25). The ergot alkaloid ergovaline can be produced by E. festucae var. lolii and by Epichloë coenophiala in Festuca arundinacea (26), where high ergovaline concentrations are responsible for fescue toxicosis, which results in poor livestock performance and is often seen in regions of the United States where tall fescue is dominant (27).
Most studies worldwide focus on the agriculturally important grass species L. perenne and F. arundinacea as model systems, because most reports of intoxication are in livestock grazing these two species (28). Endophyte infection rates for L. perenne appear low in Germany (8 to 28%) (23, 29, 30), in contrast to that in New Zealand and Australia (70%) (31). For German grasslands, only infection rates of L. perenne (30) and concentrations of alkaloids have been documented (23, 29, 32).
The aim of this study is to provide an overview on infection rates of Epichloë spp. in different regionally common grass species in German grasslands. We collected 13 grass species in three regions in Germany along a land use intensity gradient. Our focus was on the agronomically important grasses L. perenne, Festuca pratensis, F. arundinacea and Dactylis glomerata, which have been widely reported to be infected with Epichloë (33). We also sampled less-examined wild species that are reported to possibly be infected with Epichloë, for example, Festuca ovina aggregate (agg.) and Holcus lanatus (34). We determined infection rates and linked the endophyte genetic diversity based on presence or absence of alkaloid pathway genes to the actual alkaloid chemotype. Finally, based on our observations, we speculate on the risks for grazing livestock on the studied grasslands.
RESULTS
Endophyte infection rates.In total, 1,147 F. pratensis, 1,109 L. perenne, 164 F. ovina agg., and 133 D. glomerata plants were sampled and tested from all three regions. Festuca arundinacea (176 plants) was not found in the Schwäbische Alb (ALB) location and was only sampled from Hainich National Park (HAI) and Schorfheide-Chorin (SCH) regions. For the other plant species, a subset of the sampled plants from each region was tested (Table 1).
Epichloë infection rates, determined by PCR, of plants collected across all study sitesd
Using PCR, we determined the infection rates of F. pratensis, L. perenne, F. ovina agg., D. glomerata, and Festuca rubra as 81%, 15%, 73%, 8%, and 15%, respectively. It should be noted that the amplification of the L. perenne and F. ovina agg. samples resulted in faint bands, and the percentage of infected samples may be higher than we have reported. The eight other grass species did not show Epichloë infections. In total, of the populations tested, 92.3% of F. pratensis, 42.4% of L. perenne, 100% of F. ovina agg., 26.3% of D. glomerata, and 75% of F. rubra samples were infected with Epichloë (Table 1). Infection rates differed slightly between regions.
Endophyte genotypic diversity.All 926 infected F. pratensis samples were consistent with respect to the markers, which were expected for infection with Epichloë uncinata. Both mating types were present (mtAC and mtBA), and all PER markers (perA 5, perA-T2, and perA-ΔR) and LOL markers (lolC, lolA, lolO, and lolP) were present (Table 2). EAS and IDT genes were not detected.
Genotypes of different Epichloë spp. in the host grass species
Two genetic profiles were observed for Epichloë in L. perenne plants (Table 2). Both genetic profiles were identical with respect to the presence of mating type B (mtBA), the perA gene for peramine biosynthesis, and idtG, ltmQ, and ltmE of the indole-diterpene pathway (IDT), but differed with respect to the ergot alkaloid locus (EAS). In 98% of the samples (161), the dmaW marker, used to detect the gene encoding the first step in the EAS pathway, was not present, but the other EAS genes were detected. Only three samples (2%) contained dmaW, and these likely contained a functional EAS pathway, as the other EAS genes were present (Table 2). To verify that the lack of dmaW was not due to mismatched primers, another primer (dmaW10; Table 3) for this gene was tested, but also did not result in amplification of a PCR product.
Primers used in PCR to genotype endophytesa
F. ovina agg. samples showed high variability with the alkaloid gene profiles of the endophyte (Tables 2 and 4). Both mating types were present within locations, but more mating type B samples were observed. The majority of samples contained all of the perA markers, but for some samples (23%), the region encoding the perA reductase domain was not detected. The EAS genes were present in 88% of samples. IDT genes were present in 27% of samples, but not all of these contained idtG, which encodes the first step of the IDT pathway (Tables 2 and 4). In total, nine different combinations of alkaloid biosynthesis genes could be observed in the endophyte-infected F. ovina agg. samples (Table 4). The most common samples (58%) contained all of the perA and EAS markers and were predicted to produce peramine and ergot alkaloids. We predicted that 22% of the samples would be able to produce early pathway indole-diterpenes, from paspaline through to terpendole C, but due to the low template concentration, we could not confirm the presence of additional IDT genes.
Alkaloid biosynthesis gene diversity, predicted chemical profile, and detected IDTs of F. ovina agg. samples
We found two different genotypes in the six F. rubra samples. Both genotypes were mating type A, lacked the region encoding the perA reductase domain, and contained all of the EAS markers, but they differed with respect to the IDT markers. Five of the samples lacked IDT genes, and one sample contained idtG and ltmQ (Table 2).
Three different genetic profiles were present in the 15 D. glomerata samples (Table 2). Both mating types were present; 10 were mating type A and five were mating type B. Variation was associated with the presence (seven samples) or absence (three samples) of the region encoding the perA reductase domain.
Alkaloid gene profiles can be also used for identification of Epichloë spp. and their host grass species. In total, 15 of 1,109 samples (1.4%) of L. perenne were likely misidentified, as the endophyte gene profiles were consistent with the F. pratensis samples. Other possible misidentified samples were seven F. pratensis samples (0.6%), five F. ovina agg. (3.0%), and one D. glomerata sample (0.7%). Misidentifications of samples are possible, as a large proportion of the grass samples were collected vegetatively, occasionally after a mowing event or on strongly grazed study sites. We used the complete sampled plant material for the analyses of genotypes and chemotypes of the Epichloë fungi. Therefore, the identification of the plant species with plant marker genes is not possible. For the calculation of infection rates, possible misidentified samples were excluded.
Determining phylogenetic placement by evaluating mating type genes.Mating type genes (mtBA and mtAC) were chosen for phylogenetic analyses, as the template quantity for each sample was limited and these fragments amplified well.
The F. pratensis endophytes had both mating type genes, mtAC and mtBA, and the sequence grouped with the expected E. uncinata reference sequences in either the Epichloë typhina subsp. poae clade (mtAC) or the Epichloë bromicola clade (mtBA). Hence, the F. pratensis endophytes were identified as Epichloë uncinata (Fig. 1). Interestingly, all E. uncinata mtAC genes, including the reference sequence for this species, are considered pseudogenes, as a sequence transition results in a premature stop codon.
Phylogenetic analyses of mating type genes of Epichloë isolates from selected plant samples (bold). (a) Phylogenetic tree derived from maximum likelihood analysis of mtAC introns, including Epichloë isolates from Festuca pratensis (FP A17 1, FP A17 2, FP H17 1, FP H17 2, and FP S27 1, FP S27 2). (b) Phylogenetic tree derived from maximum likelihood analysis of mtBA introns, including Epichloë isolates from Dactylis glomerata (DG A15 1a, DG A15 1b, DG A15 2, and DG A15 3), Festuca pratensis (FP A17 1, FP A17 2, FP H17 1, FP H17 2, FP S27 1, and FP S27 2), Festuca ovina agg. (FO H17 1, FO H17 2, and FO H17 3) and Lolium perenne (LP H17 1, LP H17 2, LP A17 1, LP S27 1, and LP S27 2). Trees are rooted at the left edges according to other published housekeeping gene trees (80). The first two letters for each isolate designation signify the grass species, as follows: DG, Dactylis glomerata; FP, Festuca pratensis; FO, Festuca ovina agg.; LP, Lolium perenne. H, Hainich National Park; A, Swabian Alb; S, Schorfheide-Chorin; 17, sampled in 2017; 15, sampled in 2015; 1, 2, or 3, sample number. The samples were clustered with reference gene sequences from NCBI. Sequences which are from hybrids are marked with a single-letter abbreviation required for the allele, as follows: a, Epichloë amarillans; o, Epichloë bromicola; p, Epichloë typhina subsp. poae; f, Epichloë festucae; b, Epichloë baconii and Lolium-associated endophyte (LAE) clade.
All sequenced L. perenne samples grouped in the Epichloë festucae clade (mtBA) along with F. ovina agg. (Fig. 1), which is consistent with the expected E. festucae var. lolii for L. perenne, although it should be noted that there is no current way of distinguishing E. festucae from E. festucae var. lolii based on sequence data.
The D. glomerata samples grouped in the E. typhina clade (mtBA), but the samples segregated into two branches. The D. glomerata samples segregated based on the presence or absence of the region encoding the perA reductase domain (Fig. 1).
Analysis of simple sequence repeats of E. uncinata.Since all of the E. uncinata samples from F. pratensis had the same genetic profile based on presence or absence of alkaloid and mating type genes, further genetic diversity was investigated with simple sequence repeat (SSR) primers. We selected and analyzed 90 F. pratensis samples (7.85% of the infected samples) representing all regions of the study. B10 was the most informative marker and resulted in three different patterns associated with previously documented variation (Table 5 and Table S2) (35). In total, 30 samples were classified as ecotype 1, 56 as ecotype 3, and four as ecotype 4. For the analyzed subset of samples, ecotype 4 was only identified in HAI, whereas ecotype 3 occurred in all three regions, and ecotype 1 occurred in HAI and SCH. Ecotypes 3 and 4 have been previously reported for samples found in Germany (Table 5) (35).
Classification of Epichloë uncinata from Festuca pratensis into ecotypesa
Chemical diversity.Based on the genetic profile of E. uncinata in F. pratensis, we predicted that the E. uncinata-infected plants would produce the lolines N-formylloline (NFL) and N-acetylloline (NAL). Gas chromatography-mass spectrometry (GC-MS) analyses confirmed that all 24 tested samples contained NFL and all but one contained NAL (Table S3). Other metabolites in the loline pathway were also detected, such as N-acetylnorloline (NANL), loline, and N-methylloline (NML), but 1-acetamidopyrrolizidine (AcAP) was not detected. The total concentrations of lolines varied from 18 to 3,515 μg/g. Significant differences were not found between ecotypes 1 and 3 for total concentrations of lolines (Mann-Whitney-Wilcoxon test: P = 0.92, W = 54) or for the presence or absence of different lolines (tested individually, generalized mixed effect model: NANL, P = 0.54, Z = 0.61, df = 1,22; loline, P = 0.75, Z = 0.32, df = 1,22; NAL, P = 0.10, Z = 0.003, df = 1,22; NML, P = 0.41, Z = −0.83, df = 1,22; NFL, P = 1, Z = 0, df = 1,22; ACAP, P = 1, Z = 0, df = 1,22; combination tested, permutational multivariate analysis of variance [PERMANOVA]: P = 0.66, df = 1,22, F = 0.42). There was a significant positive correlation of the presence of different loline combinations and the total concentration of lolines (PERMANOVA: P = 0.01 [***]; df = 1,22; F = 15.16).
The genetic profile of E. festucae var. lolii in L. perenne indicated that peramine and indole-diterpenes could be produced. The samples (2%) that contained dmaW were predicted to produce ergovaline, which was confirmed by ultraperformance liquid chromatography-mass spectrometry (UPLC-MS) analyses. Of the 10 samples tested for ergovaline, the two samples with dmaW produced ergovaline and peramine, and one also produced lolitrem B, whereas the eight samples without dmaW only produced peramine and lolitrem B. Ergot alkaloids were not produced in the majority of the samples, as they lacked a functional copy of the dmaW gene, which encodes the first step of the ergot alkaloid biosynthesis pathway. These data provided evidence that the PCR screen is effective for identifying endophyte-infected samples that differ with respect to potential alkaloid production.
We examined 33 F. ovina agg. plants from all three regions, of which we could detect at least one IDT gene, either idtG or idtP, in 17 samples. Different indole-diterpene patterns could be detected in samples that contained IDT genes, but this was not always consistent with what we were expecting for the sample and may be the result of a small amount of sample. For the samples that produced IDTs, we detected emindole SB and paspaline, both early pathway alkaloids, but also a paxilline isomer, terpendole C, terpendole E, a terpendole I isomer, and a 13-desoxypaxilline isomer (Table 4). We detected different patterns of alkaloid production, some of which, but not all, could be explained by the presence of IDT genes.
DISCUSSION
In this study, we determined the level of endophyte infection of 13 grass species across a land use intensity gradient. By utilizing PCR with Epichloë-specific primers, we were able to determine the infection rates of each host species and obtain information on the alkaloid potential for each endophyte-infected plant. We were able to confirm the Epichloë species found in four host species and linked genotypic diversity of alkaloid pathway genes present in the endophyte to alkaloid production. L. perenne, F. arundinacea, and F. pratensis are agriculturally important forage grasses known to harbor Epichloë species (2). However, endophyte-infected L. perenne and F. arundinacea have also caused significant livestock production losses due to the production of the alkaloids lolitrem B and ergovaline, which are toxic to vertebrates (2). Cases of livestock toxicity have been more limited to Australia, New Zealand, and the United States, where these grasses are more heavily cultivated and dominated by a single species (36). Livestock intoxication associated with European pastures has been rarely reported, likely due to greater plant diversity and lower infection rates compared to those on other continents. Reports of ryegrass staggers from the United Kingdom, France, Germany, and the Netherlands date back to 1990 (2, 37–39), with only one report from France in 2004 (38, 40). No other livestock intoxications have been reported in the international scientific literature, although informal reports about animals showing “strange behavior” exist (38).
Infection rates varied across host species.We showed that five of 13 grass species, L. perenne, F. pratensis, F. ovina agg., F. rubra, and D. glomerata, were infected with Epichloë. These grass species are all used as forage or lawn grasses. No endophyte infection was found in F. arundinacea, Agrostis stolonifera, Alopecurus pratensis, or Pheleum pratense nor in the wild grasses Cynosurus cristatus, Holcus lanatus, Bromus erectus, and Bromus hordeaceus. In our study, the average infection rate for L. perenne (15%) corresponded with those reported by other studies in Germany of between 8% and 28% (23, 29, 30), whereas infection rates in New Zealand and Australia for L. perenne are higher (70%) (31). We found higher endophyte infection rates of F. pratensis (81%) and F. ovina agg. (73%) than those reported in other European countries, with that of F. pratensis ranging between 42% and 74% and that of F. ovina agg. between 24% and 29%. Interestingly, our F. arundinacea samples were not infected, whereas in other European countries its infection rate has varied between 32% and 98%. The infection rate of D. glomerata (8%) in our study was consistent with those reported in other European countries (0% to 17%) (41, 42). Typically, Epichloë-infected D. glomerata will produce stromata, fruiting bodies of sexually transmitted Epichloë fungi, that are observed during flowering (4, 34). However, as we often sampled prior to flowering, we did not observe stromata on any D. glomerata samples in the field. The endophyte infection rate for F. rubra was 15% in German grasslands, which is lower than reported infection rates in Finland (32%) (41) or Poland (41%) (42). Infection rates can vary in different regions or countries due to differences in environmental conditions. Abiotic conditions like temperature, precipitation, and geology affect infection frequencies with Epichloë species (30, 42). Grasses infected with Epichloë are also better adapted under drought (43, 44).
Genotypes and chemical diversity of different grass endophytes.Many studies have used endophyte infection but only focused on pathway end products such as production of ergovaline, lolitrem B, peramine, and N-formylloline (20, 23, 33). Yet in undisturbed native environments, more alkaloid diversity can be observed among and within populations of endophyte-infected grasses (10, 12, 13, 45), even for F. arundinacea and L. perenne (20, 21, 46). In some grass-endophyte associations, alkaloids can be induced, for example, by grazing animals as a defense reaction (47–49). In addition, environmental conditions and plant genotype can influence production of alkaloids (50). Recently, more emphasis has been placed on evaluating the genetic capability of Epichloë species to produce alkaloids by determining which alkaloid biosynthesis genes are present within the endophyte (9, 15, 51, 52). Understanding which alkaloid biosynthesis genes are present can affirm the alkaloid potential of any given isolate, as much of the alkaloid diversity observed in Epichloë species is due to gene presence/absence polymorphisms that can be easily observed by PCR, providing an effective tool to rapidly evaluate samples from large population sizes (10, 12, 13, 19). Alkaloid chemotypes that do not match the predictions based on PCR have been shown to have altered gene expression (little or no expression) that silences the pathway, or a gene can lack functionality due to the presence of nonsense mutations that would result in a different pathway end product (10, 15, 19, 52, 53).
F. pratensis is an agriculturally important grass species that was widely distributed across the collection sites. All infected F. pratensis samples, apart from 0.6% of samples that may have been misidentified at sampling, were infected with E. uncinata, and the genetic profile of mating type and alkaloid biosynthesis genes was consistent with that described for occurrence in meadow fescue (46, 51, 53, 54). Epichloë uncinata is known to produce NFL (55), which was consistent with our samples. Total loline concentrations of up to 5,500 μg/g (dry weight) are reported in literature (56), and the toxicity threshold of lolines for invertebrates ranges from 50 μg/g (dry weight) (57) up to 100 μg/g (dry weight) (58). Most samples showed concentrations above 100 μg/g (dry weight), with a maximum of 3,515 μg/g (dry weight) (Table S3). Although E. uncinata appears to have the perA gene, it has been previously described as nonfunctional due to independent frameshift mutations in both alleles (53). Despite the fact that we did not find any differences in the alkaloid gene profiles of E. uncinata-infected F. pratensis samples, we were able to distinguish additional diversity using the previously published SSR marker B10, which is found in different ecotypes from Europe (35, 59). Regional differences were observed between the three sampling regions. The occurrence of different ecotype distributions might be explained by unknown environmental factors.
The endophytes found in L. perenne are known to have different capabilities to produce the alkaloids peramine, lolitrem B, and ergovaline (60). Perhaps the most documented strains are those that have caused livestock toxicity through the production of lolitrem B, a neurotoxin that causes ryegrass staggers (61). However, there are also commercial cultivars released with safe endophytes, such as AR1, which do not produce vertebrate toxic alkaloids (36, 62, 63). Lolium perenne was well represented in our collections, and the majority of the endophyte-infected samples were capable of producing lolitrem B and peramine. Interestingly, these samples also contained many of the genes required for ergot alkaloid biosynthesis but lacked dmaW, which encodes 4-γ,γ-dimethylallyltryptophan synthase, the determinant step in ergot alkaloid biosynthesis (64). As expected, we were unable to detect ergovaline in these samples; therefore, there should be no risk of ergovaline intoxication for grazing animals on these study sites. However, since lolitrem B was detected in the L. perenne samples, intoxication of animals grazing on those pastures may be possible.
Festuca ovina agg. is a plant with high genetic variation and large gene flows (65). The species is variable and contains many subspecies, which are hard to distinguish (66). We detected high genotypic diversity of E. festucae in F. ovina agg. We also detected both mating types within the populations, and we conclude that these may represent sexually active populations. Populations that are recombining could also result in greater alkaloid gene diversity. The greatest diversity within the endophyte-infected F. ovina agg. samples was associated with variation of IDT genes. Epichloë festucae samples are known to differ in the presence of IDT alkaloid pathway genes (9, 52), and we detected different pathway intermediates of the IDT alkaloid pathway in our samples. In addition, we only detected IDTs in half of the IDT gene-positive samples. It is possible that the gene-positive samples in which IDTs were not detected do contain IDTs but at concentrations that were below the limits of detection. As alkaloid production is known to be induced by herbivory (47, 48), we assume that the genes were not upregulated or possibly could be nonfunctional. Naturally occurring pathway variation can result from a combination of altered gene expression and/or sequence variations, including sequence polymorphisms within a gene, resulting in nonsynonymous changes or the complete absence of the gene (52). Terpendole C, as well as paxilline, is known to be tremorgenic in mice (67). We detected a paxilline isomer; however, it is unknown if this isomer is tremorgenic, as relatively small changes in the structure of the molecule can change its toxicity (68). The grass F. ovina agg. often occurs on nutrient-poor grassland, which can be found in the regions of HAI and ALB, where most of the samples were collected. F. ovina agg. shows good drought resistance, but due to its poor growth it is only used as lawn grass and not as forage for livestock (69). The intoxication risk for grazing animals by infected F. ovina agg. is presumably low, but further research is needed.
Variation was also present with respect to the peramine biosynthesis gene in the F. ovina agg., F. rubra, and D. glomerata endophyte-infected samples. The region encoding the peramine reductase domain has been previously shown to be required for peramine biosynthesis (53). Interestingly, Epichloë typhina in D. glomerata is not known to produce alkaloids (56), which was confirmed recently with D. glomerata infected with Epichloë typhina from Oregon, USA, where the region encoding the perA reductase domain was not detected by PCR (70). We observed variation with the perA markers in our endophyte-infected D. glomerata samples, but since choke in D. glomerata in the USA has only been reported since 1996 (71), genetic variation of E. uncinata in D. glomerata in European grasslands may be greater.
Conclusion.We have evaluated grasses represented in three diverse locations in Germany. Our study has looked beyond endophyte infection rates and used a combination of genetic and chemotypic evaluation to determine endophyte diversity. A significant finding of the study was that although the endophyte infection rates for some host species were higher than that in other regions in Europe, not all of these samples could be considered toxic to vertebrates. The host with the highest endophyte infection rates was F. pratensis, but it has no known livestock toxicity, although insect-toxic properties can be an advantage for farmers. L. perenne samples were able to produce the vertebrate toxin lolitrem B; however, ergovaline was not of concern because the gene encoding the determinant step for ergot alkaloid biosynthesis was absent in most samples. Interestingly, none of the F. arundinacea samples were infected, whereas infected F. ovina agg. samples contained various alkaloid biosynthesis precursors for which the toxicity potential is not clear. Further studies are necessary to estimate the influence of abiotic and biotic effects on infections, alkaloid gene expression, and alkaloid production on managed grasslands. Previous studies showed no influence of land use intensity on infection rates or alkaloid production in Epichloë-infected L. perenne plants (23). Other studies showed that abiotic factors like soil type or soil moisture can alter infection rates of grasses infected with Epichloë species (30, 48). We conclude that some grasses in Germany are infected with endophytes that could be toxic to livestock, and these should be monitored to make sure they will not become dominant grasses in the pastures. We showed that PCR screens provide a robust way of testing the endophyte status of pasture grasses and could be a helpful tool for future monitoring. It is necessary to regularly record Epichloë infection rates, as well as alkaloid profiles and concentrations, for a large number of cool-season grass species in natural and agricultural grasslands to predict the risks and chances of grass endophytes in a changing climate, and with new breeds of Epichloë-grass varieties by international companies.
MATERIALS AND METHODS
Plant material and sampling.In 2015 and 2017, tillers of 13 grass species were collected in three geographically separated regions in Germany. In 2017, Lolium perenne L., Festuca pratensis Huds. [synonym Schedonorus pratensis (Huds.) P. Beauv.], Festuca ovina L. aggregate (agg.), Festuca arundinacea Schreb. (synonym Schedonorus arundinaceus Roem. & Schult.), Festuca rubra L., Phleum pratense L., Bromus erectus Huds., Bromus hordeaceus L., and Agrostis stolonifera L. were collected. We collected the grass species in the northern region, Schorfheide-Chorin (SCH) in June 2017; in the middle region, Hainich National Park (HAI) in June and July 2017; and in the southern region, Schwäbische Alb (ALB) in July 2017. Additionally, samples from a collection in 2015 from Dactylis glomerata L., Alopecurus pratensis L., Holcus lanatus L., and Cynosurus cristatus L. were used. These samples were collected in Hainich National Park (HAI) and Schwäbische Alb (ALB) in June 2015 and in Schorfheide-Chorin in July 2015. The grass species were selected due to their abundance in the sampling regions, and these species were previously reported as Epichloë infected (34).
We sampled 150 different study sites (Fig. 2), 50 per region, that are part of the German Biodiversity Exploratories project (www.biodiversity-exploratories.de). The three regions span a latitude of 800 km from north to south Germany and cover different landscape types (72). The UNESCO Biosphere region Schorfheide-Chorin (SCH) is located in the state of Brandenburg, in northeast Germany, and is defined by glacially formed landscapes (72). Hainich National Park (HAI), located in central Germany (Thüringen), and the UNESCO Biosphere region Schwäbische Alb (ALB) located in southwest Germany (Baden-Württemberg), are both characterized by calcareous bedrock (72) (Fig. 2). Study sites were located inside and in the surrounding areas of the protected zones.
Study site locations. (a) Topographic map of Germany with the three study regions, Schorfheide-Chorin (SCH) in the north, Hainich National Park (HAI) in central Germany, and Schwäbische Alb (ALB) in the south. Each region is depicted, enlarged, with corresponding land cover. Black dots indicate the 50 grassland study sites per region. Map created in ArcGIS with land cover data based on corine (https://land.copernicus.eu/pan-european/corine-land-cover). (b) Pictures of one intensively and one extensively managed grassland in SCH taken in June 2017. Intensively managed grasslands have 3 to 6 cuts or more per year and are heavily fertilized (up to 400 kg N/ha), and extensively managed grasslands are 1- to 3‐cut meadows or pastures with low stocking densities and low fertilization (1).
All study sites are part of real-life grasslands that are conventionally managed by farmers. The study sites can be divided into meadows, pastures, and mown pastures, with both fertilized and unfertilized study sites (72). The 50 grassland study sites in each region have been selected along a land use intensity gradient, but land use intensity was not used as an explanatory variable in this study (72). The sites with the highest intensity had four or five cuts per year and high stocking rates (70 livestock units/ha; cattle, horse, sheep, or goat), and a large amount of fertilizer was applied (up to 400 kg/ha/year) (1). The lowest-intensity sites were species-rich seminatural calcareous grasslands and wetlands. For each specific study site, a 50-m by 50-m plot within the grasslands was selected (72).
In each study site, all selected grass species were collected when present. From each individual plant, three tillers of approximately 3-cm height were collected (23). We sampled the lowest part of the grass tiller between the ground and the first leaf because this region is where the highest concentrations of alkaloids and the Epichloë hyphae accumulate (23, 73). For each tiller, we only selected tissue that was still green. We sampled 20 different individuals, but on some sites, this was reduced to 10 due to small populations of these species. Individual plants growing at least 3 m apart were sampled to avoid representing the same individual twice. The samples were collected in 2-ml Eppendorf tubes, kept on dry ice in the field, and stored at −20°C before further processing.
Endophyte detection by multiplex PCR.A multiplex PCR method for all collected species was performed to detect the endophyte and determine the presence of alkaloid biosynthesis genes (10). With this method, the endophyte infection rates for all collected grass species were determined. An initial multiplex PCR (M1) with Epichloë-specific primers was performed to determine if the samples were infected (primers listed in Table 3). Further PCR analyses were only performed for infected samples. The numbers of plant samples examined per species were as follows: 1,109 L. perenne, 1,154 F. pratensis, 164 F. ovina agg., 176 F. arundinacea, 133 D. glomerata, 40 F. rubra, 24 A. pratensis, 23 C. cristatus, 45 P. pratense, 24 H. lanatus, 50 B. erectus, 50 B. hordeaceus, and 45 A. stolonifera.
For Epichloë-infected samples, five multiplex PCRs (M1 to M5) with different primer combinations (Table 3) were performed to predict the genotypic diversity of the different Epichloë species. All alkaloid biosynthesis primers were designed, using available information from Epichloë fungal genome and gene sequences, to bind to conserved gene regions (10, 21). We isolated total genomic DNA from the freeze-dried and ground grass tillers with the MagAttract 96 DNA plant core kit (Qiagen, Inc., Valencia, CA). PCRs were performed with a volume of 25 μl in a total volume containing 3 μl DNA (1 ng/μl), 5 μl 5× Green GoTaq reaction buffer containing 1.5 mM MgCl2, 0.5 μl of deoxynucleoside triphosphates (dNTPs, 10 mM; Promega Corp., Madison, WI), 0.2 μl 1.0-U GoTaq DNA polymerase (Promega Corp., Madison, WI), and 0.5 μl of each target-specific primer (10 μM) (Table 3). The PCR parameters were 2 min of initial denaturation at 94°C, 30 cycles of 15 s at 94°C, 30 s at 56°C, and 1 min at 72°C, followed by 10 min at 72°C. For samples with faint PCR products in the first multiplex PCR (M1) the number of cycles was increased to 40, which was necessary for all L. perenne samples and some F. ovina agg. samples. For multiplex 5 (M5) the annealing temperature step during the cycles was changed from 56°C to 60°C for 30 s, and only 30 cycles were applied because 40 cycles decreased the quality of the amplicons.
PCR products were visualized by gel electrophoresis with 1.5% agarose gel in 1× Tris-borate-EDTA (TBE) buffer following ethidium bromide staining and UV transillumination.
Phylogenetic analyses.Samples were selected for phylogenetic placement using the sequences of the mating type genes, as we were able to obtain better amplification for the polyploid species we evaluated. We selected five samples of L. perenne, six of F. pratensis, nine of F. ovina agg., and four of D. glomerata. The mating type genes mtBA and mtAC (primers in Table 3) were sequenced for these 24 representatives. The PCRs were performed in 50 μl containing 8 μl DNA (1 ng/μl), 10 μl 5× Green GoTaq reaction buffer, 0.4 μl 1.0-U GoTaq DNA polymerase, 1 μl of dNTPs (10 mM), and 1 μl of the target-specific primers (10 μM). The same PCR parameters as those for the multiplex PCR were used, but the extension time was increased to 2 min to account for the longer amplicon sizes. PCR amplicons were visualized as described previously. Only samples showing a single PCR product were selected for purification. PCR products were purified with the QIAquick PCR purification kit (250) (Qiagen, Venlo, the Netherlands) according to the manufacturer's instructions. BigDye Terminator chemistry (version 3.1, Applied Biosystems, Foster City, CA) was used for sequencing. Due to the low DNA concentration of the template, we could only generate sequence data for the mating type genes. Geneious version 11.1.5 (Biomatters Ltd., Auckland, New Zealand) was used for alignment and sequence editing. Forward and reverse reads were joined and trimmed, and the final consensus sequences were then aligned with reference sequences from NCBI (Table S1) using MUSCLE (version 3.7) and default settings. RaxML version 8.2.11 (74) was used to infer a phylogenetic tree with 1,000 fast bootstraps.
Analysis of simple sequence repeats in Epichloë uncinata.As the genetic profiles of the F. pratensis infected samples were all similar, we used SSR primer pairs B10 and B11 (59) on a subset of 96 samples to determine if endophyte diversity could be detected in these samples. Total genomic DNA (1 ng/μl) was diluted 10×, and 4 μl diluted DNA was used in the PCR. The total reaction volume was 10 μl containing 1 μl 10× PCR buffer (without MgCl2) (Invitrogen Life Technologies), 0.3 μl MgCl2 (50 mM), 0.2 μl of dNTPs (5 mM), 0.2 μl of each primer (10 μM), and 0.15 μl Platinum Taq (Invitrogen, 5 U/μl). PCR products from B10 and B11 were labeled with 2′-chloro-7′phenyl-1,4-dichloro-6-carboxy-fluorescein (VIC) and NED fluorescent dyes, respectively (46). The PCR parameters were 4 min at 94°C, 35 cycles of 30 s at 94°C, 30 s at 60°C, and 30 s at 72°C, followed by 7 min at 72°C. Samples were kept at 4°C overnight. PCR products were diluted 10-fold, and an aliquot of 1.5 μl PCR product, 9.9 μl formamide, and 0.1 μl 500 LIZ size standard (Applied Biosystems/Invitrogen Life Technologies) was added per reaction (46). PCR products were denatured at 96°C for 5 min and immediately placed on ice for 20 min. An Applied Biosystems 3730 DNA analyzer was used to separate the PCR fragments, and Peak Scanner 1.0 (Applied Biosystems) was used to visualize fragment sizes. Only the B10 data were informative for determining ecotypes based on published data (35).
Alkaloid analyses.Alkaloids of a representative subset of 10 L. perenne, 24 F. pratensis, and 35 F. ovina agg. samples were analyzed. Quantitation of ergovaline, peramine, and lolitrem B in frozen L. perenne samples (fresh plant weight) was performed by ultraperformance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS) using a published protocol (75). We prepared the samples as described in König et al. (23). Homoperamine and ergotamine were used as internal standards. For a subset of 35 freeze-dried F. ovina agg. samples, we analyzed indole-diterpene (IDT) alkaloids by high-performance liquid chromatography-mass spectrometry (HPLC-MS) using a published protocol (76). Endophyte-infected, freeze-dried F. pratensis samples were analyzed for loline alkaloids. For each F. pratensis sample, loline alkaloids from 75 mg lyophilized tissue were extracted and prepared for GC-MS analysis using a Varian CP-3800 gas chromatograph (Agilent Technologies, Santa Clara, CA, USA) coupled to a Varian Saturn 2200 mass spectrometer (Agilent Technologies) (77, 78). The standard calibration curve was performed using N-formylloline (NFL). The concentrations of other loline alkaloids were calculated based on the NFL standard curve, and thus concentrations are considered relative to NFL concentrations.
Due to insufficient sample amounts for F. pratensis and F. ovina agg., we pooled samples that originated from the same region and that appeared genetically similar based on the marker analyses. The pooling strategy resulted in analyses for F. pratensis of seven ecotype 1 samples, one ecotype 4 sample, and 16 ecotype 3 samples. Additionally, we compared differences in loline alkaloid production of ecotypes 1 and 3 with the Mann-Whitney-Wilcoxon test for total amount of lolines and with a generalized mixed model with binomial errors for presence or absence of the different loline alkaloids (tested individually). Influence of presence and absence of different loline alkaloid combinations on the ecotype and the amount of lolines was tested with a PERMANOVA with 999 runs and a Bray-Curtis similarity matrix (79). Data were analyzed with R version 3.5.2.
Data availability.The 24 partial mating type gene sequences have been deposited in GenBank under the accession numbers MK992983 to MK993006. The related raw data were deposited on the BExIS database of the Biodiversity Exploratories (www.bexis.uni-jena.de) with the following data set IDs: 24187, 24387, 24386, 24388, 24369, 24749, and 25406.
ACKNOWLEDGMENTS
We thank Julia König for collecting grass species in 2015. Furthermore, we thank the Noble Research Institute (Ardmore, OK, USA), especially Nikki Charlton and the rest of Carolyn A. Young’s mycology lab, for the opportunities to evaluate our collection and benefit from their knowledge. We also thank Christopher Schardl from the Department of Plant Pathology, University of Kentucky, for advice on loline alkaloids; Daniel Cook from the USDA (UT, USA) for initiating and supporting a cooperation with Stephen T. Lee and for advice on indole-diterpenes; and Alexander Keller from the University of Würzburg for support with the phylogeny tree; as well as Jie Zhang for her help with ArcGIS, Danush Taban for assistance with laboratory work; and Fabian Bötzl for statistical support. We thank the managers of the three Exploratories, Kirsten Reichel-Jung, Iris Steitz, and Sandra Weithmannand (Swabian Alb), Katrin Lorenzen and Juliane Vogt (Hainich), and Miriam Teuscher (Schorfheide), and all former managers for their work in maintaining the plot and project infrastructure; Christiane Fischer for giving support through the central office; Andreas Ostrowski for managing the central database; and Markus Fischer, Eduard Linsenmair, Dominik Hessenmöller, Daniel Prati, Ingo Schöning, François Buscot, Ernst-Detlef Schulze, Wolfgang W. Weisser, and the late Elisabeth Kalko for their role in setting up the Biodiversity Exploratories project.
The work was funded by the DFG Priority Program 1374 “Infrastructure-Biodiversity-Exploratories” (DFG-Refno. KR 3559/3-2). The Noble Research Institute, LLC, provided support for Carolyn A. Young.
Fieldwork permits were issued by the responsible state environmental offices of Baden-Württemberg, Thüringen, and Brandenburg.
FOOTNOTES
- Received 25 February 2019.
- Accepted 14 June 2019.
- Accepted manuscript posted online 21 June 2019.
Supplemental material for this article may be found at https://doi.org/10.1128/AEM.00465-19.
- Copyright © 2019 American Society for Microbiology.
REFERENCES
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.↵
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.↵
- 17.↵
- 18.↵
- 19.↵
- 20.↵
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.↵
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.↵
- 53.↵
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵
- 81.
- 82.
- 83.
- 84.
