The relationship between mycotoxin synthesis and isolate morphology in fungal endophytes of Lolium perenne


  • Sylvie Bony,

    1. Centre INRA de Clermont-Ferrand/Theix/Lyon, UMR 188, Toxicologie et Métabolisme Comparés des Xénobiotiques, Ecole Nationale Vétérinaire de Lyon, BP 83, 69280 Marcy-l’Etoile, France;
    Search for more papers by this author
  • Nathalie Pichon,

    1. Centre INRA de Clermont-Ferrand/Theix/Lyon, UMR Amélioration et Santé des Plantes, 234 Avenue du Brézet, 63039 Clermont-Ferrand cedex, France
    Search for more papers by this author
  • Catherine Ravel,

    1. Centre INRA de Clermont-Ferrand/Theix/Lyon, UMR Amélioration et Santé des Plantes, 234 Avenue du Brézet, 63039 Clermont-Ferrand cedex, France
    Search for more papers by this author
  • Andrée Durix,

    1. Centre INRA de Clermont-Ferrand/Theix/Lyon, UMR 188, Toxicologie et Métabolisme Comparés des Xénobiotiques, Ecole Nationale Vétérinaire de Lyon, BP 83, 69280 Marcy-l’Etoile, France;
    Search for more papers by this author
  • François Balfourier,

    1. Centre INRA de Clermont-Ferrand/Theix/Lyon, UMR Amélioration et Santé des Plantes, 234 Avenue du Brézet, 63039 Clermont-Ferrand cedex, France
    Search for more papers by this author
  • Jean-Jacques Guillaumin

    Corresponding author
    1. Centre INRA de Clermont-Ferrand/Theix/Lyon, UMR Amélioration et Santé des Plantes, 234 Avenue du Brézet, 63039 Clermont-Ferrand cedex, France
    Search for more papers by this author

Author for correspondence: Jean-Jacques Guillaumin Tel: +33 4 7362 4440 Fax: +33 4 7362 4459


  • • Variability in the fungal endophytes of 83 natural populations of Lolium perenne (perennial ryegrass) from Europe was assessed.
  • • One plant per population was used for endophyte isolation and mycotoxin analysis. Variability in three isozymes, colony morphology and growth rate on potato dextrose agar (PDA), and synthesis of ergovaline, lolitrem B and peramine was recorded.
  • • Three species were found among 94 strains isolated: Neotyphodium lolii, Neotyphodium sp. (LpTG-2) and Gliocladium-like. The most frequent species was N. lolii, which showed high variability. In 12 populations, a single plant harboured two different endophytes. One-third of the isolates of N. lolii did not produce ergovaline whereas a few isolates did not produce lolitrem B. Ergovaline and lolitrem B-deficient strains, but not the few peramine-deficient isolates, had characteristic morphologies on PDA. No isolate was deficient for both ergovaline and lolitrem B synthesis.
  • • Selection of ergovaline and lolitrem-deficient strains based only on the morphology of the isolates in culture may be possible.


The presence of fungal endophytes in the seeds of Lolium perenne L. was first observed by Guérin (1898). McLennan (1920) studied the distribution and development of the mycelium in the different organs of the grass. Sampson (1937) was the first to isolate an endophytic fungus from L. perenne and to grow it in pure culture.

In New Zealand, a damaging tremorgenic disease of sheep and cattle, ‘ryegrass staggers’, had long been reported and attributed to the consumption of perennial ryegrass by grazing animals (Cunningham & Hartley, 1959). Later on, ryegrass staggers was shown to be caused by alkaloids (Aasen et al., 1969) and the implication of the fungal endophyte in the toxicosis was eventually established by Fletcher & Harvey (1981).

Sampson (1933) observed the presence of two different endophytic fungi in Lolium perenne and Latch et al. (1984) isolated them. The first type was thin, poorly stainable and highly branched. It was generally sterile in culture, but the production of a few conidiophores resembling Gliocladium led to this species being called ‘Gliocladium-like’ (Philipson, 1989).

The second, more frequent, type consisted of twisted, poorly branched hyphae. It was shown to belong to clavicipitaceae and described as a new species, Acremonium loliae Latch, Christensen and Samuels (later changed to Acremonium lolii). Within the genus Acremonium, this species was classified in a section Albo-lanosa created by Morgan-Jones & Gams (1982) to include the anamorphs of the clavicipitaceae. However, the fungi belonging to this section were reclassified by Glenn et al. (1996) into the new genus Neotyphodium, the clavicipitaceous endophytes of L. perenne being named N. lolii (Latch, Christensen & Samuels) Glenn, Bacon & Hanlin.

Christensen et al. (1991, 1993) introduced an additional distinction among the clavicipitaceous endophytes of perennial ryegrass on the basis of growth rate, isolation delay, sporulation at 23°C and isozyme pattern. The majority of isolates could be accommodated as Neotyphodium lolii. However, a few isolates showing higher growth were just named by the abbreviation LpTG-2.

The clavicipitaceous endophytes of perennial ryegrass are known to synthesize several mycotoxins, among which three are particularly important: lolitrem B, a tremorgenic molecule responsible for ‘ryegrass staggers’; ergovaline, a compound belonging to the ergopeptine group, which has vasoconstrictive effects and causes various diseases on grazing mammals (‘fescue-foot’ and ‘fescue toxicosis’); and peramine, a tripeptide which is repellent and toxic for insects but not for mammals. The strains of LpTG-2 which have been studied synthesized ergovaline and peramine, but not lolitrem B, while variation was observed among strains of N. lolii for the synthesis of lolitrem B, ergovaline and peramine (Christensen et al., 1993). Schardl et al. (1994) showed that LpTG-2 is a heteroploid species originating from interspecific hybridization between Neotyphodium lolii and the parasitic species Epichloë typhina.

The dual action of clavicipitaceous endophytes (beneficial effects on the host vs risk of toxicity for grazing animals) sets forage grass breeders special problems. Sometimes, these have been solved by removing the endophyte from the cultivars selected for pasture and by maintaining or introducing it into varieties selected for turf. Another strategy, developed mainly in New Zealand (Latch, 1989), proposed to select harmless clavicipitaceous endophytes (producing no or little ergovaline or lolitrem) and to inoculate these strains to commercial cultivars. This strategy involves an extensive study of the variability of the endophytes of Lolium perenne. Such research was conducted on a large scale in New Zealand (Latch, 1994; Fletcher & Easton, 1997).

Very few studies have been conducted on this subject in Europe, despite the fact that Lolium perenne probably originated from the Near East and has diversified in Europe (Balfourier et al., 2000). Maximum genetic variability would therefore be expected in Europe.

In France, 547 natural populations of Lolium perenne were collected from 1983 to 84 (Charmet et al., 1990, 1993; Balfourier & Charmet, 1991). About half of these populations were sampled at random and checked for the presence of endophytes (Lewis et al., 1997; Ravel, 1997). One, or several, endophytes were found in 188 populations out of 262 (72%). The present study was conducted on part of this material. The objectives were to identify the fungal species involved and to describe intraspecific variability, particularly as concerns the production of mycotoxins.

Materials and Methods


Seventy-three populations with endophytes were selected from the populations studied by Ravel (1997). They were conserved at INRA Clermont-Ferrand as seeds and/or living plants (Table 1a). The material was selected on geographical and ecological bases to achieve homogeneous cover of the country, and to include a wide range of ecological situations.

Table 1.  Taxonomical identification, morphology and growth rate of the isolates. (a) Populations of French origin. (b) Populations of nonFrench origin. (c) Reference isolates
n° popul.IsolatesFungal species1MG2Linear growth3
1011910119Neotyphodium loliiIIIC7
1015110151N. loliiIIIC6
1015610156N. loliiIIIC6
1017010170N. loliiIIIC6
1018010180N. loliiIIC7
1018410184N. loliiIIIC4
1020810208N. loliiIC9
1022410224N. loliiIC9
1025110251N. loliiIIC7
1025610256 AN. loliiIIIC6
 10256 BN. loliiIVC2
1026110261N. loliiIIIC5
1026610266N. loliiIIC4
1030410304N. loliiIIIC6
1031910319N. loliiIIIC6
1035910359N. loliiIIIC6
1036110361 AN. loliiVIC7
 10361 BN. loliiIIC8
1036410364N. loliiIIC7
1036510365N. loliiIIC6
1036710367N. loliiIIC8
1037010370 AN. loliiIIC8
 10370 BN. loliiIIIC6
1040510405N. loliiIIC7
1045210452N. loliiIIIC6
1045810458N. loliiIIIC5
1050510505N. loliiIIC6
1050910509 AN. loliiIIC6
 10509 BN. loliiVIC5
1051510515 AN. loliiVC2
 10515 BN. loliiIIC7
1052010520N. loliiIIC6
1055110551N. loliiIIIC6
1055710557N. loliiIIC7
1057310573N. loliiIIIC6
1060610606N. loliiIIC6
1061110611N. loliiIIIC6
1062110621N. loliiIIC6
1062610626N. loliiIIC7
1065610656N. loliiIIIC4
1066710667N. loliiIIC6
1067410674N. loliiIIIC7
1067510675N. loliiIIIC7
1070210702N. loliiIIIC7
1070710707N. loliiIIIC5
1071010710N. loliiIIIC6
1071610716N. loliiIIIC7
1075310753N. loliiIIIC6
1076310763N. loliiIC9
1076910769N. loliiIIIC6
1077010770N. loliiIIIC5
1087310873N. loliiIC9
1088010880N. loliiIIC7
1090510905N. loliiIC9
1090610906N. loliiIIC9
1091010910N. loliiIIIC6
1091710917N. loliiIIIC6
1095410954N. loliiIIC8
1096310963 AN. loliiIVC1
 10963 BN. loliiVC4
1097210972N. loliiIIIC5
1110411104N. loliiIIC7
1111511115N. loliiIIC8
1111811118 AN. loliiIIIC6
 11118 BN. loliiVIC7
1115111151N. loliiIIIC7
1115411154 AN. loliiIIIC5
 11154 BGliocladium-likeVIIIC4
11163failure of isolation   
1117811178N. loliiIIIC9
1120511205 AN. loliiIIIC6
 11205 BN. loliiVIC9
1121411214N. loliiIIIC5
1126411264N. loliiIIIC7
1127411274N. loliiIIC6
1128011280N. loliiIIIC4
1130211302N. loliiIIIC1
1131411314 ALpTG-2VIIC4
 11314 BN. loliiIIIC6
1144211442 ALpTG-2VIIC3
 11442 BN. loliiIVC5
n° popul.CountryIsolatesFungal species1MG2Linear growth3
40137Italy40137 ALpTG-2VIIC4
  40137 BN. loliiIVC2
70036Poland70036N. loliiIIC6
110005Yugoslavia110005N. loliiVC2
110009Yugoslavia110009N. loliiIIC6
130003Greece130003N. loliiIVC1
210027Bulgaria210027N. loliiVC3
IsolatesFungal species1MG2Linear growth3
  1. 1 Identification from patterns Malate Dehydrogenase (MDH), Phosphoglucomutase (PGM) and Phosphoglucose Isomerase (PGI). 2 Morphological Groups (Fig. 2). 3Nonoverlapping groups of growth. Groups are ordered from C1, the group characterized by the highest linear growth, to C9, the group characterized by the lowest linear growth. The average growth in mm d−1 is 0.353 for C1, 0.304 for C2, 0.269 for C3, 0.231 for C4, 0.193 for C5, 0.150 for C6, 0.113 for C7, 0.077 for C8 and 0.019 for C9.

R 3N. loliiVIC9
R 4N. loliiVC4
R 5N. loliiVC3
R 6N. loliiIC8

For comparison, 10 populations of European, nonFrench origin, were included in the study (Table 1b). In addition, six standard isolates (four of Neotyphodium lolii and two of species LpTG-2) were obtained from G. C. M. Latch (from AgResearch, New Zealand (Table 1c). These last isolates originated from natural populations in Spain and Southern France.


General procedure

For each population, one plant had been selected in the field, and the presence of an endophyte had been verified. The 83 populations were represented by seeds originating from only one plant per population.

Isolations were carried out from the 83 seed lots. All the isolates were grown on PDA (Potato Dextrose Agar, Sigma Chemical, 39 g l−1) to observe their macroscopical morphology and measure their linear growth. Mycelium grown in liquid culture was checked for the isoforms of three enzymes, to allow the taxonomical identification of the isolates.

Straws from the 83 plants from which the seeds had been taken had been harvested in early August, frozen and lyophilized, and the analysis of ergovaline, lolitrem B and peramine was carried out on this material.


Seeds were sterilized for 4 h in commercial bleach (12°Cl.), then rinsed with sterile water and sterilized again for 3 min in Bayrochlor® (from Bayrochlor Company). They were rinsed three times with sterile water and the glumellas were removed. The seeds were air-dried and then sown in Petri dishes containing PDA plus penicillin and streptomycin (100 µg g−1 each). Approximately 10 seeds were placed in each dish. The Petri dishes were sealed with adhesive tape, placed on their side and incubated in the dark at 23°C, the seeds, slightly sunk in agar, were aligned horizontally, with their coleoptiles upwards (Fig. 1).

Figure 1.

Isolation of endophytes from germinating seeds of Lolium perenne. The fungus can be subcultured from the seed itself (Is), the coleoptile (Ic), the seminal roots (Ir).

Every 2 d the seeds were observed for germination and the appearance of mycelium on the medium. Under these conditions, the fungus was observed to grow from the cotyledon, the coleoptile and even the seminal roots. In the majority of cases, the coleoptile was cut into pieces with a microscalpel and the pieces subcultured on a new PDA medium. The fungi obtained were subcultured, paying particular attention to the appearance of any morphologically different colonies from the same seed lot: colonies showing a clearly different aspect or growth rate were subcultured separately.

Species identification by isozyme analysis

An unambiguous identification of the isolates was obtained by analysis of three isozymes: Malate Dehydrogenase (MDH), Phosphoglucose Isomerase (PGI) and Phosphoglucomutase (PGM). The isolates were grown in V8 liquid medium containing for 1 l: V8 juice (a commercial product constituted of extracts of 8 different vegetables): 100 ml; D-glucose: 10 g; L-asparagine: 2 g; KH2PO4: 1 g; MgSO4, 7 H2O: 0.5 g; KCl: 0.25 g; FeCl3: 0.01 g; CaCO3: 2 g; chloramphenicol: 0.05 g. The pH was adjusted to 6.0. The medium was distributed into Erlenmeyer flasks (100 ml in each) and sterilized in an autoclave (110°C for 30 min). Each isolate was represented by two replicates. The flasks were subjected to gentle shaking (70 rpm) in the dark. The temperature was 18°C for those isolates which, from their morphology, appeared to be Gliocladium-like, and 23°C for the other isolates.

The mycelium was removed after 3 wk and lyophilized. Extraction of the enzymes, gel preparation, electrophoresis and visualization were carried out according to the procedures described by Christensen et al. (1993), and Naffaa et al. (1998). The fungal species were identified from the description of electromorphs by the same authors.

Morphological description on PDA

Among the different morphological types which had been a priori subcultured from the isolations, only those that seemed stable in the course of the subsequent cultures were retained. The morphology of many isolates stabilized at the second subculture even when variation was present in the first subculture.

These isolates were grown, observed and described in standard conditions: the culture medium was 20 ml of PDA distributed in Petri dishes 9 cm in diameter. Incubation was conducted in the dark, at 23 ± 0.5°C. The inoculum consisted of a mycelium disk cut with a 4-mm borer from the margin of a 1-month-old colony also grown on PDA. The Petri dishes were sealed with adhesive tape. The linear growth of the isolates was measured on the same Petri dishes. The diameter of each colony was measured after 45 d from the average of two perpendicular diameters. The growth rate was evaluated in mm d−1. There were at least five replicates for each isolate.

Mycotoxin analysis

Lolitrem B was analysed by an adaptation of the HPLC method of Gallagher et al. (1985). Ground freeze-dried straw was mixed with chloroform-methanol (2 : 1) for 1 h on a horizontal shaker. After centrifugation, the supernatant was filtered and evaporated under a nitrogen stream. The dry residue dissolved in dichloromethane was cleaned on a Si Bond Elut column (Varian) conditioned with dichloromethane. Lolitrem B was then eluted from the column with dichloromethane-acetonitrile (80 : 20). The elution solution was evaporated under a nitrogen stream and the residue was dissolved in the HPLC mobile phase. The HPLC analysis was performed using a Thermoquest liquid chromatograph system equipped with a silica column (250 mm × 4.6 mm, 20 µl injection) and a fluorescence detector (excitation 268 nm, emission 440 nm). The mobile phase was dichloromethane-acetonitrile (85 : 15) with a flow rate of 1 ml min−1. Calculations were performed with an external lolitrem B standard solution (supplied by R. G. Gallagher, AgResearch, New Zealand). The determined limit of quantification was 0.18 µg g−1 of dry plant material.

Ergovaline was analysed according to an adaptation of the method of Hill et al. (1993). Samples were extracted by alkaline chloroform with an internal standard (ergotamine 500 ng ml−1 in chloroform) and shaken for 2 h at room temperature. Extracts were filtered and purified on a silica column (Ergosil Analtech Newark, DE, USA) prewashed with chloroform. Pigments were removed by washing with chloroform-acetone (1 : 3). Ergot alkaloids were eluted with methanol and the eluent was evaporated under a nitrogen stream at 30°C. The dry residue was dissolved in methanol. The HPLC analysis was performed on a Thermoquest liquid chromatograph system equipped with a Si C-18 (Zorbax XDB) column (150 mm × 4.6 mm, 20 µl injection) and a fluorescence detector (excitation 250 nm, emission 420 nm). The mobile phase was acetonitrile (36, 5%) in aqueous ammonium carbonate (200 mg l−1) run at 1.1 ml min−1.

A standard was prepared by adding ergovaline (from F. Smith, Department of Pharmacal Sciences, Auburn University, AL, USA) and ergotamine (Sigma Chemical) to 500 mg of a noninfected ryegrass sample up to 500 ng g−1 and treated as described above. The limit of quantification of the method was 0.15 µg g−1 of dry plant material.

Peramine was analysed by a method supplied by B.A. Tapper (AgResearch, New Zealand). Dried ground straw was mixed with 30% propan-2-ol for 30 min at 90°C. After centrifugation, the crude extract was cleaned-up on a Bond Elut CBA column (Varian) conditioned with 80% aqueous methanol containing 2% ammonium hydroxide and then with pure methanol. Peramine was eluted from the column with a 5% formic acid −80% methanol solution. The eluent was dried under a nitrogen stream and the residue resuspended in methanol. The HPLC analysis was performed using a Thermoquest liquid chromatograph system equipped with a silica column (250 mm × 4.6 mm, 20 µl injection) and an UV detector (280 nm). The mobile phase was a buffer consisting of 50 mM ammonium acetate, 5 mM guanidinium carbonate and 0, 2% acetic acid in 18% aqueous methanol at a flow rate of 1 ml min−1. An endophyte-free ryegrass sample supplemented with peramine and extracted as described above was used as external standard for calculation. The limit of quantification of this method was 2 µg g−1 of dried plant material.

Statistical procedures

Linear growth data were subjected to variance analysis using the General Linear Model (GLM) procedure of SAS. In a first analysis, the main effect was the endophyte strains and the means were grouped by the cluster method developed by Scott & Knott (1974). The advantage of this method is to partitionate the mean values into nonoverlapping groups. In a second variance analysis, the morphological groups were designated as the main effect and the means were separated by the test of Bonferroni, which is adapted to unequal cell-size.

The production of lolitrem B for two distinct morphological groups of endophytes was compared by a two-sample t-test (procedure TTEST in SAS). The distribution obtained for the peramine was not Gaussian, leading to compare the medians by a nonparametrical test (proc. NPAR1WAY in SAS).


Success of isolation

Isolates were obtained in all cases except one. The delay in the isolate emergence was extremely variable, from 7 to 33 d (median value, 14 d) but generally consistent within seed lots.

In 70 plants, the colonies growing from different seeds of the same lot were homogenous in their morphology and growth rate and were considered to belong to one isolate. However, for 12 plants, two different morphologies could be distinguished among the colonies from the same seed lot. These differences in morphology remained stable when the colonies were subcultured. The two variants were considered as different isolates, distinguished by the letters A and B in Table 1. Thus, the total number of isolates studied was 70 + (12 × 2) = 94.

Species identification

Species identification was carried out by the isozyme patterns of MDH, PGI and PGM. All the isolates were identified by this method: 83 were found to belong to Neotyphodium lolii, 7 to LpTG-2 and 4 to Gliocladium-like (Table 1). Seeds from 12 plants harboured two different isolates. In one case, an isolate of Gliocladium-like was obtained as well as with an isolate of N. lolii. In three cases, isolates of LpTG-2 and N. lolii were obtained. In eight cases, two different isolates of N. lolii were found in the same seed lot.

Morphology of cultures on PDA

The isolates were assigned to eight stable and well-characterized Morphological Groups (MG) on PDA (Fig. 2).

Figure 2.

Macromorphology of the eight Morphological Groups (MG) on PDA: a–f: Neotyphodium lolii, g: LpTG-2, h: Gliocladium-like.

Morphological Group I (MG I, Fig. 2a) consisted of slow-growing isolates which develop vertically more than horizontally. The mycelium is strongly aggregated. Vertical development is accompanied by the formation of irregular crests and convolutions. This morphology is often described as ‘brain-like’. The margin of the colony is vertical. The colony is waxy, without aerial mycelium.

In MG II, the centre of the colony resembles MG I, but the margin is made of horizontal, smooth mycelium (Fig. 2b). Aerial mycelium is absent or scarce.

MG III is the most heterogeneous; it includes two subtypes which both are originally flat and smooth. In subtype A, domes or warts made of aerial, white mycelium erupt at the surface after 10–15 d of growth (Fig. 2c), they can later fill the centre of the colony or constitute sectors. In subtype B, the centre of the colony is white and slightly raised; after 15 d, it appears scored by furrows which are both radial and concentrical. These two morphologies have been included in one group because some isolates alternately showed one or the other morphology according to the subcultures.

MG IV is characterized by a high radial growth rate. The colony is flat and smooth, with very few or no aerial mycelium (some tufts in the central part). The margin, made of intramatricial mycelium, appears diffuse (Fig. 2d).

In MG V, as in the previous type, the colony is flat, with a diffuse margin. After 15 d, it is covered in its major part with aerial, powdery or granular mycelium (Fig. 2e). This group resembles group VIII (Gliocladium-like), however, the lower side is tinted with light grey, not with olive green.

The colony characterizing MG VI is flat, smooth, waxy, almost mucose, without aerial mycelium or white warts. The centre of the colony is typically grooved by curved, radial furrows (Fig. 2f).

The colony characterizing MG VII has an irregular outline. Its surface is cottony, and entirely covered with aerial mycelium. The margin is diffuse, consisting of superficial hyphae (Fig. 2g).

In the last group, MG VIII, the colony is circular, white, flat, and covered with powdery or granular aerial mycelium (Fig. 2h). The agar on the under side is tinted with olive green (permitting a clear distinction from group V).

The allocation of the 94 isolates obtained to Morphological Groups is shown in Table 1.

The seven isolates belonging to species LpTG-2 all belonged to MG VII and were the only isolates in this group. The four isolates identified as Gliocladium-like were the only ones in MG. The isolates identified as Neotyphodium lolii were allocated between MG I to VI, with the following distribution: MG I: five isolates; MG II: 26; MG III: 39; MG IV: five; MG V: four; MG VI: four. The six reference isolates obtained from AgResearch Grassland (R1–R6) had the following distribution: R1 and R2 (LpTG-2): MG VII; R3, R4, R5 and R6 (N. lolii): MG VI, V, V and I, respectively.

Growth rate

Analysis of the linear growth rate of the isolates led to the definition of nine nonoverlapping groups as shown in the last column of Table 1.

Figure 3 shows the distribution of linear growth for the eight Morphological Groups (MG). The main effect of MG on growth rate was highly significant (P = 0.001). According to the comparison of means, the different MG are classified in four groups of growth. MG IV and MG V belong to the group showing the significantly highest growth rate. LpTG-2 (MG VII) and Gliocladium-like (MG VIII) belong to a second group which overlaps the set including MG IV and MG V. MG II and MG III, which contain the majority of isolates, are classified in a third, distinct group of intermediate growth. MG I and MG VI constitute a fourth group characterized by significantly lower growth rates.

Figure 3.

xRelationship between increase in colony diameter (mm d−1) and morphological groups. The horizontal, white line in each box is the median of the data; the height of the box is equal to the interquartile distance (IQD). The verticle line corresponds to a distance 1.5 × IQD from the median. Data points falling outside this interval are indicated by horizontal, black lines.


The analyses were carried out for a total of 79 plants representing 79 populations. For lolitrem B, the results of the 12 plants from which two different fungal isolates had been isolated were considered separately. Among the 67 plants which harboured only one isolate 11 did not contain lolitrem B and 56 contained lolitrem B at concentrations higher than the quantification threshold (0.18 µg g−1). Of the 11 plants without lolitrem B, three plants harboured a Gliocladium-like fungus; that is no plant harbouring only a Gliocladium-like isolate synthetized lolitrem B. A further four plants harboured an isolate of LpTG-2. That is, no plant harbouring only an isolate of LpTG-2 synthetized lolitrem B. Another four plants harboured an isolate of Neotyphodium lolii belonging to MG I. That is, no plant with an isolate in MG I synthetized lolitrem B.

The 56 plants containing lolitrem B did so at concentrations ranging from 0.8 to 5.75 µg g−1 (Fig. 4a) with an average value 2.90 µg g−1 All these isolates belonged to N. lolii and were distributed between MG II (22 isolates), MG III (31 isolates) and, marginally, MG IV (1) and MG V (2). MG IV and V were very small and the mean values for lolitrem B concentrations were compared only between MGs II and III, with MG II isolates producing significantly more lolitrem B (P = 0.0063).

Figure 4.

Frequency distribution of mycotoxin concentrations. (a) lolitrem B; (b) ergovaline; (c) peramine.

In nine of the 12 plants from which two different isolates had been obtained, both isolates belonged to N. lolii. In these plants, lolitrem B was present at concentrations ranging from 0.80 to 4.30 µg g−1. In three plants where there was one isolate of N. lolii and one isolate of LpTG-2 no lolitrem was detected in two cases, and one plant contained 2.1 µg g−1 lolitrem.

Among the 67 plants that harboured only one isolate, 41 contained ergovaline, at concentrations higher than the quantification threshold of 0.15 µg g−1 The plants without ergovaline included three plants harbouring Gliocladium-like isolates and 23 plants harbouring N. lolii. Among these 23 isolates, 22 belonged to MG II and only one to MG IV. Correspondingly, ergovaline was never detected in plants in which only one isolate of MG II had been found.

The 41 plants containing ergovaline, did so at concentrations ranging from 0.24 to 3.46 µg g−1 with an average value of 1.14 µg g−1 Among these, four belonged to LpTG-2 and 37 to N. lolii, and were distributed between MG I (four isolates), MG III (31 isolates) and MG V (two isolates). Interestingly, the highest concentrations were observed in two plants harbouring a MG I isolate.

For the plants containing two different isolates, in three cases a LpTG-2 isolate was associated with an isolate of N. lolii of MG III or MG IV, and these plants contained ergovaline. Four associations involved two isolates of N. lolii belonging to MG groups other than MG II and ergovaline was also present. In four situations in which MG II was involved in association with isolates of other groups, ergovaline was absent in three cases and present in one case. The last association was between Gliocladium-like and an isolate of N. lolii MG III and the plant contained ergovaline.

The quantification threshold was higher for peramine than for the other two mycotoxins (c. 2 µg g−1). Thus, in some cases peramine was detected, but could not be quantified. Among the 66 plants which harboured only one isolate, seven had no detectable peramine, six showed only traces of the mycotoxin (concentrations < 2 µg g−1) and 53 contained higher concentrations. The absence of peramine was not linked to absence or presence of ergovaline or lolitreme and could not be related to a special morphology of the isolate in culture. 53 plants contained peramine at concentrations ranging from 2.0 to 52.8 µg g−1, according to an asymmetrical distribution, with a median value 6.30 µg g−1. (Fig. 4c). Two plants showed a particularly high concentration of peramine (52.8 and 41.3 µg g−1). Peramine production was significantly higher for the isolates of MG III than for those of MG II.

For those plants from which two different isolates had been obtained, the peramine concentrations showed no consistent pattern.


Isolation of endophytes from germinating seeds was very successful. The 94 isolates obtained belonged to the three endophyte species which had already been described in Lolium perenne: Neotyphodium lolii, LpTG-2 and the ‘Gliocladium-like’ fungus. N. lolii was the most common, representing 90% of the isolates. No other species were detected. The specific identification of these three species was based on their isozyme patterns for the two enzymes MDH and/or PGM, as described by Christensen et al. (1993) and Naffaa et al. (1998). The presence of conidia in culture for LpTG-2 at 20–24°C (N. lolii is always sterile at this temperature), is another criterion advocated to distinguish between these two clavicipitaceous species (Christensen et al., 1993). The nine isolates of LpTG-2 isolated in our study were checked for the presence of conidia on PDA at 23°C. Conidia were observed for four isolates only. On the other hand, they were not observed on the cultures of five isolates of N. lolii which had been drawn by lot. The presence of conidia in cultures of LpTG-2 at room temperature seems to be a sufficient, but not necessary, condition for identification.

Pure culture macromorphology may be useful for species identification, since LpTG-2 and Gliocladium-like show a specific and very homogenous morphology (MG VII and MG VIII, respectively). Neotyphodium lolii varies greatly in morphology and in growth rate. Linear growth was preferred to biomass production as being potentially more discriminant. The strong correlation observed between macromorphology and linear growth is not surprising: the very slow apical growth of the mycelial hyphae of some strains is concomitant with a high rate of branching, the result being a three-dimensional growth of the colony, which is stroma-like, being as high as wide. This morphology, which has sometimes been named ‘brain-like’, characterizes MGs I and II; in MG II, the central prosenchyma is surrounded by a band of flat mycelium while in MG I, it falls abruptly into the culture medium. These two groups are clearly distinct from each other and from all the other morphological types which can be described within N. lolii. The morphologies I and II appear stable with subculturing and are retained after a storage at low temperatures (−80°C). The flat aspect of the young colony characterized all the other morphological groups described; the MGs IV, V and VI show typical morphological aspects, and contain only a few isolates. MG III is less homogeneous; in this group, the centre of the colony can undergo different development after 20 d of growth: for example appearance of cottony warts or of radial and concentric furrows. This type is the one that contained the highest number of isolates. The typical appearance of the colonies of LpTG-2 on PDA has already been described by Christensen et al. (1991). These authors had also reported the high variability of the isolates of N. lolii and published photos resembling our groups II and III.

The variability within N. lolii was also considerable with respect to the synthesis of the three main mycotoxins: lolitrem B, ergovaline and peramine. Four isolates, all of French origin (6% of the isolates) did not synthetize lolitrem B. 23 isolates (37%) did not synthetize ergovaline. Of these 23, 20 originated from France, one from Poland, one from Yugoslavia and one from Greece. Six isolates (10%) (excluding those with traces of peramine lower than the quantification threshold) did not synthetize peramine. Among these six, all of French origin, one was also lolitrem B-deficient, two were ergovaline-deficient, and three synthetized both ergovaline and lolitrem B.

The most striking observation was the close link between the deficiency for ergovaline or lolitrem B in N. lolii and the morphology of the isolates: all four isolates which did not produce lolitrem B were from MG I and no isolate of MG I produced lolitrem B. Among the 23 isolates that did not produce ergovaline, 22 belonged to MG II. In MG II, no isolate produced ergovaline. By contrast, the peramine-deficient strains distributed among several morphological groups, including MG I and MG II.

As the MG I and MG II morphologies coincide with a slow linear growth rate, it seems that the ‘mutation’ (in the broad sense) which led to the absence of synthesis of lolitrem or ergovaline is regularly accompanied by disturbances of the mycelial growth. The loss of the ability to synthetize ergovaline (or lolitrem) could be the result of a chromosome deletion which could also result in the loss of genes playing a role in mycelial growth and morphogenesis. As the deficiencies for ergovaline and lolitrem are independent and accompanied by different morphological disturbances, the two deletions probably concern different chromosomes or chromosome fragments. The loss of a minor chromosome (a ‘B chromosome’, according to Kisler & Mia (1992)) could also be an explanation: Kuldau et al. (1999) found eight chromosomes in N. lolii of which three were smaller than 3 Mb.

Figure 4 shows the frequency distribution of the concentrations of the three mycotoxins in the plants. The concentrations of lolitrem B and ergovaline are similar to those in the literature (Di Menna et al., 1992; Ball et al., 1997b; Lane et al., 1997a, 1997b). By contrast, the concentrations of peramine (median value: 6.3 µg g−1 for 53 isolates) are lower than those in the literature, which are often in the range of 10–30 µg g−1 (Ball et al., 1995a, 1995b, 1997a). However, for two particular populations, the concentrations were much higher (> 40 µg g−1). The genotype of the fungus, which controls the synthesis of a given mycotoxin, can also play a role in fixing the quantitative level of this mycotoxin within the plant. However, many other factors probably play a role, such as the host genotype, the mycelium density in the host tissues and various environmental factors; so it is difficult to assert that differences recorded in the three mycotoxins concentrations faithfully reflect genetic quantitative differences between the fungal isolates.

The isolates of MG II (which do not produce ergovaline) produced significantly more lolitrem B than the isolates of MG III. The isolates of MG III produce significantly more peramine than those of MG II. However, there was no correlation between the concentration of the three mycotoxins if all isolates are considered together.

Isolates deficient in both ergovaline and lolitrem B were not found. The predicted frequency of such strains would be the product of the individual frequencies: 6% × 37% = c. 2%. This low predicted frequency could explain why no such isolate was found. On the other hand, if it is true that both the ergovaline-free and the lolitrem B-free isolates of N. lolii are the result of genetic accidents (for instance of chromosome deletions), one would expect that the isolates carrying a double deletion would be particularly deficient and find it difficult to survive in the plant. However, screening of more than 1000 populations in New Zealand, has succeeded in finding such isolates (Fletcher & Easton, 1997).

The close link between the absence of ergovaline or lolitrem B and a specific morphology of the colonies in pure culture could, if it is confirmed on a higher number of strains from more various origins, considerably improve the process of selection of harmless strains. Such a selection could be carried out by laboratories equipped only for routine mycology and lacking the complex equipment and technical competence necessary for the analysis of mycotoxins by HPLC.

This study confirmed that the species LpTG-2, of hybrid origin (Schardl et al., 1994), is in a minority among the endophytes of L. perenne in France. Only three French isolates of this taxon were found in our study: two from the region of Marseille and one from Corsica. Four other isolates were isolated from populations from Italy and Spain. The two reference isolates obtained from AgResearch also have a Western Mediterranean origin. Thus, LpTG-2 appears to be a West-Mediterranean species.

The seven LpTG-2 isolates were homogeneous in their morphology, growth rate, isozyme pattern and mycotoxin synthesis (lolitrem B is not synthetized). The isozyme patterns MDH and PGM allow a certain identification of the species, but the morphology of the colonies on PDA may appear to be a safe method of identification.

The taxon ‘Gliocladium-like’ was also a minority among our isolates. The four isolates obtained appeared identical as concerned their isozyme pattern, macromorphology (MG VIII type), high growth rate, low isolation delay, and preference for cool temperatures (18°C). The populations concerned came from Northern France, Central France and Germany. As for LpTG-2, the identification can be easily carried out from the macromorphology alone. Neither conidiophores nor conidia were observed in planta or in culture.

In 12 cases (14%) two different endophytes were isolated from the same seed lot. As each lot had been harvested from only one plant, two different fungal isolates must have coexisted within the same plant. In one case, a Gliocladium-like fungus was found associated with an isolate of N. lolii. In three cases, the association was between LpTG-2 and N. lolii and in eight cases, between two different isolates of N. lolii. Indeed, we could only detect the situations in which the two associated isolates of N. lolii belonged to two different MGs. It is probable that other situations exist in which the two (or more) isolates are morphologically similar and could be detected only with molecular markers.

Coexistence between two endophytes in the same plant has already been reported for the associations between a clavicipitaceous and a nonclavicipitaceous endophyte (Schmidt, 1994; Siegel et al., 1995). The association between two different isolates of the same species of endophyte has already been reported and analysed by Meijer & Leuchtmann (1999) on the pair Brachypodium sylvaticum/Epichloë sylvatica. However, to our knowledge, the present work is the first mention of the presence of two different strains of Neotyphodium lolii in the same plant of Lolium perenne. Frequent coexistence between two clavicipitaveous endophytes in the same host may support parasexual hybridization in the continuum existing from parasitic Epichloë to mutualistic Neotyphodium, leading to new species.


The work was carried out with the technical collaboration of C. Astier, P. Desray, C. Huc and M. Carcelen. The authors gratefully acknowledge Mrs. Maureen Bocquet for her reading of the English manuscript.