Origin, Divergence, and Phylogeny of Asexual Epichloë Endophyte in Elymus Species from Western China

Asexual Epichloë species are likely derived directly from sexual Epichloë species that then lost their capacity for sexual reproduction or lost sexual reproduction because of interspecific hybridization between distinct lineages of sexual Epichloë and/or asexual Epichloë species. In this study we isolated asexual Epichloë endophytes from Elymus species in western China and sequenced intron-rich regions in the genes encoding β-tubulin (tubB) and translation elongation factor 1-α (tefA). Our results showed that there are no gene copies of tubB and tefA in any of the isolates. Phylogenetic analysis showed that sequences in this study formed a single clade with asexual Epichloë bromicola from Hordeum brevisubulatum, which implies asexual Epichloë endophytes that are symbionts in a western Chinese Elymus species likely share a common ancestor with asexual E. bromicola from European H. brevisubulatum. In addition, our results revealed that asexual E. bromicola isolates that are symbionts in a western Chinese Elymus species and sexual Epichloë species that are symbionts in a North American Elymus species have a different origin. Further analysis found that Epichloë species likely originated in Eurasia. In addition, the results support the hypothesis that migratory birds or humans might have aided the dispersal of these fungal endophytes to other continents.


Introduction
Fungi species in the genus Epichloë (Clavicipitaceae, Ascomycota) and closely related asexual Epichloë species are common endophytes of cool-season grasses in the subfamily Pooideae [1,2]. Epichloë species often provide numerous benefits to their hosts, such as increased tolerance to drought [3][4][5], disease-resistance [6], resistance to herbivory and parasitism [7,8], and enhanced above-ground and below-ground vegetative and reproductive growth [9]. Previous studies have confirmed that certain alkaloids in Epichloë species play a crucial role in pasture persistence. For example, lolines and peramine are toxic and confer significant deterrent activity against insect pests [10,11].
Sichuan, Xinjiang and Tibet. We examined endophyte-infected grasses using the aniline blue coloring (0.1% aqueous) method [33] and isolated fungal endophytes on potato dextrose agar, incubated in darkness at 25°C for four weeks [34].

DNA extraction, amplification and sequencing
Endophyte total genomic DNA was extracted from fresh mycelia using the HP fungal DNA kit (OMEGA, Beijing, China). Intron-rich portions of the housekeeping genes β-tubulin (tubB) and translation elongation factor 1-α (tefA) were amplified by polymerase chain reaction (PCR) using universal primers according to the previous study of Moon et al. [15]. The primers in this study were as follows: tub2-exon 1d-1: GAGAAAATGCGTGAGATTGT and tub2-exon 4u-2: GTTTCGTCCGAGTTCTCGAC; and tef1-exon 1d-1: GGGTAAGGACGAAAAGACTCA and tef1-exon 5u-1: CGGCAGCGATAATCAGGATAG. The PCR standard reaction was carried out with 0.5 μ l of genomic DNA, 1 μ l of each primer (10 pmol), 12.5 μ l 2 × taq master-mix and RNAse-free water added to a total of 25 μ l. The tubB PCR cycling program was as follows: 94°C for 5 min, followed by 35 cycles of 94°C for 30 s, 45°C for 45 s and 72°C for 2 min, followed by a final extension at 72°C for 10 min. The tefA PCR cycling program was as follows: 94°C for 5 min, followed by 35 cycles of 94°C for 30 s, 55°C for 45 s and 72°C for 2 min, followed by a final extension at 72°C for 10 min. PCR products were cloned into the pMD 18-T vector (TaKaRa, Dalian, China) based on the manual. Five positive clones per genes were sequenced by TaKaRa Biotechnology Co. Ltd (Dalian, China).

Data analysis
Endophyte tubB and tefA sequences were aligned using the MAFFT 7.0 program [35] and alignments were adjusted by eye. Maximum parsimony (MP) trees were constructed in the PAUP 4.0b10 package [36]. MP trees were estimated using a heuristic search with tree bisection-reconnection (TBR) branch swapping and 100 random addition replicates. Alignment gaps were treated as missing information. Nucleotide substitutions were unordered and unweighted. Maximum likelihood (ML) trees were constructed in the MEGA 6.0 program [37]. The optimal model of nucleotide evolution was HKY+G for tubB and tefA, according to MrModeltest 2.3 [38] and this model was used in the ML analysis. ML heuristic searches were performed with 100 random addition sequence replications and TBR branch swapping algorithm. Bootstrap support values were calculated with 1000 replicates.
Nucleotide diversity was calculated using Tajima's π [39] and Watterson's θ [40] statistics. The Tajima's π measure quantifies the mean percentage of nucleotide differences among all pairwise comparisons for a set of sequences; whereas, Watterson's θ is simply an index of the number of segregating (polymorphic) sites. To tests the neutral evolution, Tajima's D and Fu and Li's D statistics were inferred as described by Tajima [39] and Fu and Li [41]. The software program DnaSP 5.0 [42] was used to perform the above calculations.
Median-joining (MJ) network analysis was applied to display the genealogical relationships between taxa [43,44]. Previously there has been very little information published that related to the phylogenetic network of the endophytes. In the present study, we determined the haplotype of sequences excluding the outgroup based on the DnaSP 5.0 program [42] and constructed a network of the endophytes using the Network 4.1 program (Fluxus Technology Ltd, Clare, Suffolk, UK).

Sequence variation
Sixteen asexual Epichloë endophytes were isolated from western Chinese Elymus species (S1 Table). The fragment sequence of the Epichloë tubB sequence in this study includes four exons and three introns, while the fragment sequence of the Epichloë tefA sequences contains three exons and three introns. Furthermore, no gene copies of tubB and tefA were observed in any of the isolates. In addition, tubB and tefA sequence of isolates NI_201207 and NI_201209 could not be obtained (S1 Table). Analyses of the Epichloë tubB and tefA sequences found that the exon sequences are more conserved than the intron sequences (data not shown). More importantly, previous studies [15,45] used intron sequences to provide insight into the phylogenetic relationships of endophytes. Therefore we also used the intron sequences.
The length of the tubB intron sequences in this study varied from 244 to 403 bp. The length of the alignment of the tubB intron sequences was 495 bp, including 62 invariable sites and 115 variable sites, 55 of which were parsimony informative sites. The length of the tefA intron sequences varied from 348 to 551 bp. The length of the alignment of the tefA intron sequences was 732 bp, including 90 invariable sites and 136 variable sites, 77 of which were parsimony informative sites.
We estimated the haplotypes and nucleotide polymorphisms of the Epichloë species in North American and western Chinese Elymus species. In the tubB sequences, the number of haplotypes (6) in the asexual Epichloë endophytes from western Chinese Elymus species was higher than the number of haplotypes (4) in the Epichloë species from North American Elymus species. The number of polymorphic sites (10) in the asexual Epichloë endophytes from western Chinese Elymus species is much lower than the number of polymorphic sites (29) in the Epichloë species from North American Elymus species (Table 1). The nucleotide diversity Tajima's π and Watterson's θ values in the asexual Epichloë endophytes from western Chinese Elymus species were 0.0038 and 0.00881, respectively, while in the Epichloë species from North American Elymus species they were 0.0326 and 0.0332, respectively ( Table 1), indicating that the nucleotide diversity in Epichloë species from North American Elymus species is higher than the nucleotide diversity in asexual Epichloë endophytes from western Chinese Elymus species. The Tajima's D and Fu and Li's D values of asexual Epichloë endophytes from western Chinese Elymus species were -2.0146 (p < 0.05) and -2.4555 (p < 0.05), respectively, while for the Epichloë species from North American Elymus species they were -0.2916 (p > 0.1) and -0.1983 (p > 0.1), respectively, indicating that the asexual Epichloë endophytes from western Chinese Elymus species had a significant departure from neutrality. In the tefA sequences, the number of haplotypes (10) in the asexual Epichloë endophytes from western Chinese Elymus species was much higher than the number of haplotypes (5) in the Epichloë species from North American Elymus species (Table 1). The number of polymorphic sites (12) in the asexual Epichloë endophytes from western Chinese Elymus species was much lower than the polymorphic sites (48) in the Epichloë species from North American Elymus species. The nucleotide diversity or Tajima's π value of the asexual Epichloë endophytes (0.0052) from western Chinese Elymus species is lower than the Tajima's π value in the Epichloë species (0.0450) from North American Elymus species. The nucleotide diversity or Watterson's θ value of the asexual Epichloë endophytes (0.0075) from western Chinese Elymus species is lower than the Watterson's θ value in the Epichloë species (0.0402) from North American Elymus species ( Table 1). The Tajima's D value was -1.1814 (p > 0.1) for the asexual Epichloë endophytes from western Chinese Elymus species and it was 0.4240 (p > 0.1) for the Epichloë species from North American Elymus species. The Fu and Li's D value was -1.8640 (p > 0.1) for the asexual Epichloë endophytes from western Chinese Elymus species, and it was 0.7607 (p > 0.1) for the Epichloë species from North American Elymus species, indicating that different selection pressures affected the asexual Epichloë endophytes from western Chinese Elymus species and Epichloë species from North American Elymus species.

Phylogenetic analyses of tubB and tefA sequences
To reveal the phylogenetic relationships between asexual Epichloë endophytes from western Chinese Elymus species and other Epichloë species, we constructed phylogenetic trees with maximum parsimony (MP) and maximum likelihood (ML) methods using tubB and tefA se- The MP method produced a tree topology nearly identical to the ML trees. We only showed MP tree in S1 and S2 Figs, respectively.
In the tubB tree the asexual Epichloë endophytes from western Chinese Elymus resolved in subclade VI-tub (Fig 1 and S1 Fig). Interestingly, sexual Epichloë species from North American Elymus species were not in subclade VI-tub, but instead, resolved in subclade III-tub and subclade IV-tub. Although a previous study [27] showed that asexual Epichloë species originated from sexual Epichloë species and then lost the ability to sexually reproduce, our phylogenetic results found that isolates of western Chinese (asexual endophytes) and North American (sexual endophytes) Elymus species have different origins, indicating that they are different species. Subclade VI-tub contained 16 tubB sequences, including 15 asexual Epichloë endophytes from 15 western Chinese Elymus species and one asexual Epichloë bromicola from European Hordeum brevisubulatum. This suggests that the isolates from western Chinese Elymus species likely share a common ancestor with the asexual E. bromicola from European H. brevisubulatum. In addition, subclade III-tub contained three sexual E. elymi from North American Elymus canadensis, El. villosus and El. virginicus, one asexual E. canadensis from North American El. canadensis and one asexual E. funkii from North American Achnatherum robustum. Subclade IV-tub contained three sexual E. amarillans from North American Agrostis hyemalis, El. virginicus and Sphenopholis obtusata respectively, one asexual E. canadensis from North American El. canadensis and one asexual E. chisosa from North American Ac. eminens.
The topology of the tefA tree is consistent with that of the tubB tree. Isolates from western Chinese (asexual endophytes) and North American (sexual endophytes) Elymus species were grouped in different subclades: subclade III-tef, subclade IV-tef and subclade VI-tef, respectively (Fig 2 and S2 Fig). Subclade VI-tef contained 21 tefA sequences, including 15 asexual Epichloë endophytes from western Chinese Elymus species, three asexual E. sinica from Chinese Roegneria spp., two sexual E. liyangensis from Chinese Poa pratensis ssp. pratensis and one asexual E. bromicola from European H. brevisubulatum. E. sinica and E. liyangensis appear  to be hybrids. The results of the tefA tree confirmed that the asexual Epichloë endophytes from western Chinese Elymus species are likely derived from the same ancestor with the asexual E. bromicola from European H. brevisubulatum.
Subclade III-tef contained three sexual E. elymi from North American El. canadensis, El. villosus and El. virginicus, one asexual E. canadensis from North American El. canadensis and one asexual E. funkii from North American Ac. robustum. Subclade IV-tef contained three sexual E. amarillans from North American S. obtusata, Ag. hyemalis and El. virginicus, one asexual E. canadensis from North American El. canadensis and one asexual E. chisosa from North American Ac. eminens. These results suggest that asexual Epichloë endophytes from western China and sexual Epichloë species from North American Elymus species have different origins. related, while Htub 7, 17 and 44 were in different clades (Fig 3). The results also confirm that asexual Epichloë endophytes from western Chinese Elymus species and sexual Epichloë species from North American Elymus species have different origins, and the genetic diversity is higher in sexual Epichloë species from North American Elymus species than in asexual Epichloë endophytes from western Chinese Elymus species. Htub 65 contained only one tubB sequence from asexual Epichloë endophytes in western Chinese Elymus species, but Htub 20 contained 18 tubB sequences from Asia and Europe, including one sexual Epichloë species and 17 asexual Epichloë endophytes. Among the 17 asexual Epichloë endophytes, one asexual E. bromicola was isolated from the European species, H. brevisubulatum, and 16 asexual Epichloë endophytes were isolated from Asian Leymus chinensis (2) and western Chinese Elymus species (14). There were 14 asexual Epichloë endophytes from western Chinese Elymus species, and two asexual E. bromicola from Asian L. chinensis. In addition, the asexual E. hordelymi was isolated from European Hordelymus europaeus. Htub 36 contained a sexual E. bromicola tubB sequence from Asian El. repens. Htub 36 and 20 were distributed in one clade, indicating closer relationships between these haplotypes. Htub 7 included three sexual E. elymi and one asexual E. canadensis. The three sexual E. elymi were isolated from North American El. canadensis, El. villosus and El. virginicus. The asexual E. canadensis was isolated from North American El. canadensis. Htub 17 and 44 included sequences from sexual E. amarillas and asexual E. canadensis. Sexual E. amarillas was isolated from North American El. virginicus, and asexual E. canadensis was isolated from North American El. canadensis.
The sexual Epichloë species from North American Elymus species shared four haplotypes: Htef 11, 27, 56 and 57. Htef 27, 56 and 57 only contained one tefA sequence, respectively. Htef 27 was from the North American sexual E. amarillas (El. virginicus), and Htef 56 and 57 were from the North American asexual E. canadensis (El. canadensis). HapF 11 was from the sexual E. elymi, whose host plants include El. canadensis, El. villosus and El. virginicus in North America.

Discussion
Asexual Epichloë species are thought to have derived from asexual and/or sexual Epichloë species by hybridization [27]. Among the recognized asexual Epichloë species, hybrid endophytes (19 of 26 taxa) outnumber non-hybrid endophytes [12]. Moreover, hybrid endophytes are abundant in wild grass populations in temperate areas across the world [47][48][49]. However, we did not find any gene copies of tubB and tefA through PCR and sequencing. The tubB and tefA phylogenetic estimates do not suggest different origins of asexual Epichloë endophytes from western Chinese Elymus species. Our results reveal that asexual Epichloë endophytes from western Chinese Elymus species do not hybridize, which indicates that these asexual Epichloë species are derived from the same ancestor with E. bromicola (S1 and S2 Figs).

Endophyte diversity in North American and western Chinese Elymus species
Our tubB and tefA phylogenetic trees suggest that asexual Epichloë endophytes from western Chinese Elymus species and sexual Epichloë species from North American Elymus species have different origins (Figs 1, 2, 3 and 4). Several scenarios would lead to this kind of pattern. Transmission of sexual Epichloë species to new hosts occurs through horizontal transmission through the stigmata [14]. In contrast, transmission of asexual Epichloë species to the next generation of grass generally occurs in the seed, when hyphae penetrate the developing embryo [15,16]. However, researchers have found that asexual Epichloë species could be horizontally spread from plant to plant through conidia produced on the leaf surface [19,24,50]. We found that the asexual Epichloë endophytes from western Chinese Elymus were closely related to asexual E. bromicola from European H. brevisubulatum, asexual E. sinica from Asian Roegneria spp. and sexual E. liyangensis from Asian P. pratensis ssp. pratensis. This set of relationships indicates that a horizontal transmission mechanism probably exists. In addition, Moon et al. [27] found that transmission of endophytes occurred within the same tribe, but also between different tribes. Elymus, Hordeum and Roegneria are members of the Triticeae tribe, but P. pratensis ssp. pratensis is a member of the Poeae tribe. Although there is only a little information available about horizontal transmission between Triticeae and Poeae, the data presented here and the results described by Moon et al. [27] are sufficient to support horizontal transmission between Triticeae and Poeae.

Origin and spread of endophytes
The MJ network reflects the ancestral and derived relationships based on haplotype data, but also reveals a clear geographical pattern of distribution to the new or old world [44]. North American (new world) haplotypes of Epichloë are nested within old world samples of Epichloë species, whereas Asian (old world) haplotypes are grouped within the new world haplotypes (Figs 3 and 4). This pattern indicates that Epichloë gene-flow between the new and old world is common. Note that most haplotypes from European and Asian endophytes (old world) are located in MJ network (Figs 3 and 4). These results suggest that European and Asian endophytes have high haplotype diversity. In addition, Europe (19) and Asia (11) had more Epichloë speices (70%, 30/43) than any other continent and E. festucae var. lolii, E. typhina, E. coenophiala and E. occultans were introduced from Europe to other places [12]. This finding is consistent with this location being the center of origin for the genus [51] as this is where the greatest species diversity is to be found. In phylogenetic tree, European and Asian endophytes are placed near the root of the clades (S1 and S2 Figs), indicating they are diverged early in the phylogeny. Because of most of Europe and Asia belong to Eurasia from the perspective of geography. So we speculated that Epichloë species likely originated in Eurasia.
The tubB and tefA networks suggest that gene-flow among continents is common. We found that Eurasian endophytes are widely distributed across the globe and that this indicates that Eurasian endophytes "bridge" new and old world endophyte diversity (Figs 3 and 4). The spread of endophytes could be caused by, but is not limited to, the following processes. European endophytes are closely related to American endophytes (Figs 3 and 4) and it is possible that European animals carried endophyte-infected grass plants or seeds while they crossed the Bering land bridge to North America [52]. Another hypothesis is that the spread of endophytes from North America to South America may have been caused by the transport of endophyte-infected grass plants or seeds by rafting or carriage by migratory birds [15]. Endophytes could have spread more easily throughout Eurasia and Africa by human or birdmediated dispersal. More interestingly, the Epichloë endophytes from Oceania are closely related to Epichloë from America and Europe (Figs 3 and 4). The most likely distribution scenario to the Oceania is European or American migration to the islands, accompanied by endophyte-infected seeds or plants. This is perhaps not surprising given that asexual E. festucae var. lolii was first introduced to New Zealand in the 1800s, in seeds brought to New Zealand by British immigrants [53]. Furthermore, haplotype Htub 22 and Htef 21 are from African and Oceania endophytes, indicating gene-flow exchange between these two areas. The most likely reason for this pattern is European migration to Oceania, with stop-overs in Africa where there were food and water supplies, and then a selection of endophyte-infected seeds or plants left Africa and were transferred to Oceania.