The advantages of endophyte-infected over uninfected tall fescue in the growth and pathogen resistance are counteracted by elevated CO2

Atmospheric CO2 concentrations are predicted to double within the next century. Despite this trend, the extent and mechanisms through which elevated CO2 affects grass-endophyte symbionts remain uncertain. In the present study, the growth, chemical composition and pathogen resistance of endophyte-infected (E+) and uninfected (E−) tall fescue were compared under elevated CO2 conditions. The results showed that the effect of endophyte infection on the growth of tall fescue was significantly affected by elevated CO2. Significant advantage of E+ over E− tall fescue in tiller number, maximum net photosynthetic rate and shoot biomass occurred only under ambient CO2. With CO2 concentration elevated, the beneficial effect of endophyte infection on the growth disappeared. Similarly, endophyte infection reduced lesion number and spore concentration of Curvularia lunata only under ambient CO2. These results suggest that the beneficial effect of endophyte infection on the growth and pathogen resistance of tall fescue could be counteracted by elevated CO2. An explanation for the counteraction may be found in a change in photosynthesis and nutritive quality of leaf tissue.

C 3 plant) and purpletop grass (Tridens flavus, a C 4 plant), the growth of endophyte-infected (E+) and uninfected (E−) plants responded similarly to CO 2 enrichment. Also in perennial ryegrass, Hunt, et al. 8 reported that E+ biomass tended to be greater than E− plants only at elevated CO 2 , and they further found that E− plants had 40% lower concentrations of soluble protein under elevated CO 2 than under ambient CO 2 , but this CO 2 effect on soluble protein was absent in E+ plants. In tall fescue (Lolium arundinaceum), Newman, et al. 9 did not find interaction between CO 2 concentration and endophyte infection in the growth, but they found soluble crude protein concentration increased under elevated CO 2 for E− plants but not for E+ plants. Ryan, et al. 30 reported that endophyte-derived alkaloids decreased in response to elevated CO 2 . Taken together, the effects of endophyte infection on herbage quality as well as defensive chemistry can be affected by elevated CO 2 . Therefore, the endophyte-induced herbivore 11 and pathogen resistance 18,31 of the host are likely to be impacted by elevated CO 2 in the atmosphere.
Recently, the effect of Epichloë endophyte infection on pathogen resistance has been extensively investigated. The pioneering research by Shimanuki and Sato 32 demonstrated that timothy plants (Phleum pratense) infected by Epichloë typhina were resistant to the fungus Cladosporium phlei. In in vitro investigations, White and Cole 33 , Siegel and Latch 34 and Christensen 35 found that Epichloë isolates inhibited the growth of pathogenic fungi, only the antifungal activity of endophytes differed between the isolates. In in planta investigations, the positive effect of endophyte infection on pathogen resistance of the host grass has been observed in tall fescue 36,37 , ryegrass [38][39][40][41] and other native grasses [42][43][44] . Certainly, endophytes do not always improve disease resistance of the host. Negative 45,46 and neutral 47,48 effects have also been reported. In our previous study 44 , we found that endophyte could enhance pathogen resistance of Leymus chinensis, and this endophytic benefit was strengthened by drought treatment. These different reports suggest that the interactions between endophytes and pathogens are complex, and may be affected by species difference as well as environmental factors 31 such as elevated CO 2 concentration in the atmosphere.
In the present study, E+ and E− tall fescue were planted under contrasting CO 2 availability regimes to test the effect of the endophyte infection and CO 2 concentration on the performance in terms of growth, chemical composition and pathogen resistance of tall fescue. Specifically, we addressed the following questions: (1) does endophyte infection improve growth and pathogen resistance of the grass host? (2) does elevated CO 2 affect growth and pathogen resistance pattern of tall fescue -endophyte associate? If this is the case, (3) what is the mechanism involved might be?  Fig. 1). Under ambient CO 2 condition, tiller number of E+ was significantly more than that of E−, but under elevated CO 2 condition, no significant difference occurred.

Results
Maximum net photosynthetic rate and biomass. Maximum net photosynthetic rate and shoot biomass were significantly affected by three-way interaction among CO 2 concentration, N supply and endophyte infection (Table 1). Only under ambient CO 2 and high N condition, both maximum net photosynthetic rate and shoot biomass were greater in E+ than in E− plants (Fig. 2).
Leaf carbon, nitrogen and C:N ratio. Leaf C concentration was significantly affected by N supply and endophyte infection (Table 1). Leaf N concentration was significantly affected by three-way interaction among CO 2 concentration, N supply and endophyte infection. Leaf N concentration of E+ plants was lower than that of E− plants only under ambient CO 2 and high N condition (Fig. 3a). Both elevated CO 2 concentration and endophyte infection significantly improved leaf C:N ratio (Fig. 3b,c). Lesion number and spore concentration of the pathogen. Both lesion number and pathogen spore concentration were significantly affected by the interaction between CO 2 concentration and endophyte infection ( Table 2). Under ambient CO 2 concentration, endophyte infection reduced lesion number and pathogen spore concentration of the host leaves when exposed to Curvularia lunata. Elevated CO 2 significantly improved pathogen resistance of both E+ and E− plants. However, no difference occurred in either lesion number or pathogen spore concentration between E+ and E− plants under elevated CO 2 (Fig. 4). That is to say, the advantage in pathogen resistance of E+ over E− plants was alleviated by elevated CO 2 .
Soluble sugar and amino acids. Soluble sugar concentration was significantly affected by CO 2 concentration and endophyte infection ( Table 2). Elevated CO 2 significantly increased soluble sugar concentration while endophyte infection significantly decreased soluble sugar concentration of tall fescue (Fig. 5).
Because the responses of the 17 amino acids that were measured were not independent, after measurement, we used a PCA to reduce the number of amino acid response variables to a new set of composite variables. To facilitate interpretation of the principal components, we subjected the first four principal components to factor  rotation and retained four rotated factors (RF1, RF2, RF3, and RF4, which accounted for 82.61% of the total variance) (Fig. 6). As the values of the rotated factor increased, the variables that load heavily and positively (loading ≥ +0.5) also increased, while the variables that load heavily but negatively (loading ≤ −0.5) decreased. The standardized univariate responses of these variables are shown in Fig. 7 to facilitate the interpretation of the multivariate responses and to allow a closer inspection of the variables loading heavily onto RF1, RF2, RF3, and RF4.
Lignin accumulation. Lignin concentration was significantly affected by interaction between CO 2 concentration and pathogen inoculation ( Table 2). Pathogen inoculation resulted in lignin accumulation in the leaf of tall fescue under ambient CO 2 concentration, and this trend was further strengthened by elevated CO 2 concentration (Fig. 8). Lignin concentration was significantly affected by interactions among CO 2 concentration, endophyte infection and pathogen inoculation. Only under ambient CO 2 and pathogen inoculation condition, lignin concentration of the leaf was greater in E+ than in E− plants (Fig. 8).

Discussion
Plant growth response. The effects of elevated CO 2 on growth of plants, especially C 3 plants, have been widely studied, but most published papers on plant response to elevated CO 2 fail to even state the endophyte status of their plant material. In the pioneering study, Marks and Clay 29 found no significant interactions between CO 2 enrichment and endophyte infection on the growth of perennial ryegrass. Similar results have been reported by Newman, et al. 9 in tall fescue. In contrast, in the present study, we found a significant endophyte infection × CO 2 interaction for tiller number, maximum net photosynthetic rate and shoot biomass. We found that growth advantage of E+ over E− plants occurred only under high N and ambient CO 2 conditions. Under high N conditions, elevated CO 2 improved shoot growth of both E+ and E− plants, but the growth advantage of E+ disappeared under elevated CO 2 . That is to say, elevated CO 2 counteracted the beneficial effect of endophyte infection on the growth of the host. Although significant endophyte infection × N supply × CO 2 interaction for growth response has not been reported, this result is consistent with most published reports in that growth advantage of E+ plants occurred under high N conditions 9,23,30,49 , and consist with the results in tall fescue that no significant difference appeared in growth between E+ and E− plants under elevated CO 2 9, 50 . In tall fescue, Brosi, et al. 51 found that endophyte infection frequency was significantly higher under elevated CO 2 compared to ambient; and Ryan, et al. 30 found that endophyte concentration increased under elevated CO 2 . If fungal concentration was correlated with vegetative vigor of the host plant directly 28 , elevated CO 2 may promote the plant-fungal endophyte mutualism. In the present study, we did find elevated CO 2 improve the growth of E+ plants, but elevated CO 2 improve the growth of E− plants in a higher degree, and thus a significant growth difference between E+ and E− plants did not exist anymore under elevated CO 2 . This phenomenon might be related to photosynthetic ability of tall fescue. Tall fescue belongs to a C 3 grass. Because of the lack of CO 2 concentrating ability, at ambient CO 2 , its carboxylation function of Rubisco is thought to be limited by CO 2 . With CO 2 concentration in the air increasing, its photosynthetic rate will increase 3,52,53 . Under ambient CO 2 , photosynthetic ability of grasses can be improved by endophyte infection 9, 54-56 . Under elevated CO 2 , the carboxylation function  of Rubisco in tall fescue might be near saturation, and the effect of endophyte infection on photosynthesis might be negligible. That is to say, elevated CO 2 might counteract the beneficial effect of endophyte infection in photosynthesis and thus biomass to the host plants.
C and N metabolism. Independent of endophyte infection, elevated CO 2 altered tall fescue tissue chemistry in some expected ways 3, 57-59 , such as increasing carbohydrates (here soluble sugar concentration), decreasing N concentration and thus increasing C:N ratio. As for amino acid concentrations, studies have reported both positive 60, 61 and negative 30, 62 effects of elevated CO 2 on amino acids. In the present study, elevated CO 2 tended to enhance the concentration of 12 out of 17 amino acids tested. From an herbivore perspective, increased concentrations of soluble sugar and amino acids would increase palatability 63, 64 . However, the subsequent decrease in the percentage of N and the increase in C:N ratio under elevated CO 2 could offset this impact 62,65 . Endophyte infection has been described to result in a reduction of nitrogenous compounds in tall fescue 66, 67 and ryegrass 8 . In the present study, we found that endophyte infection significantly decreased the soluble sugar concentration, leaf N and increased C:N ratio of the host grass. Endophyte infection had no effect on most amino acids tested, except decreased RF4. Although we found no interaction between CO 2 and endophyte on soluble sugar, amino acids concentration and C:N ratio, similar to the results from Ryan, et al. 30 , we did find significant interaction between CO 2 and endophyte infection on leaf N concentration. Under ambient CO 2 and high N conditions, E+ plants had smaller leaf N concentration than E− plants. With CO 2 elevated, however, no difference between E+ and E− plants occurred. Here, both elevated CO 2 and endophyte infection can decrease leaf N concentration, but the decreasing degree resulted from CO 2 was even larger. Under high N conditions, elevated CO 2 resulted in 54.5% less while endophyte infection resulted in 20.8% less in leaf N concentration. Alkaloids are considered to contribute to defense. Although we did not measure alkaloids in the present study, both Ryan, et al. 30 and Brosi, et al. 51 in tall fescue found that alkaloid production decreased with CO 2 concentration elevated. Ryan, et al. 30 further suggested that plants where the C:N ratio was highest would have the lowest alkaloid per unit endophyte concentrations. All these results suggest that CO 2 enrichment might buffer the effect of endophyte infection on the N-metabolism of host plants. Pańka, et al. 37 observed stronger susceptibility of E− tall fescue to Rhizoctonia zeae than E+ counterparts. A significant increase in resistance to dollar spot disease, caused by Sclerotinia homoeocarpa, has also been observed in Festuca rubra 42 . In the present study, we found that endophyte infection improved pathogen resistance of tall fescue, but the significant effect occurred only under ambient CO 2 concentration.
In studies examining plant response to fungal disease under elevated CO 2 , disease incidence and severity are variable, from decreased 69, 70 , unchanged 71, 72 to increased 73 . When endophyte infection was considered, up to now, we found no report on its contribution to the host grasses under elevated CO 2 . In the present study, we found that disease severity of both E+ and E− plants decreased under elevated CO 2 .
The interesting result in the present study is that the advantage of E+ over E− plants in pathogen resistance under ambient CO 2 disappeared with CO 2 elevated. One possible explanation might be that the nutritive quality of leaves is responsible for pathogen development. Thompson and Drake 74 found positive correlations existed between plant N concentration and disease severity. And this correlation has been proved by Mcelrone, et al. 75 and Plessl, et al. 70 . In the present study, the main effects of elevated CO 2 and endophyte infection were similar on reducing leaf N concentration and decreasing pathogen severity, but the contribution of elevated CO 2 was even bigger. So it might be larger degree reduction of N concentration resulted from elevated CO 2 that cover up the role of endophyte infection on N concentration and thus pathogen resistance. In the present study, we further found that both elevated CO 2 and endophyte infection resulted in lignin accumulation in tall fescue after pathogen inoculation. Lignin is one of important phenolic compounds, whose deposition is believed to play a crucial role in barricading the pathogen from invading the plant through physical exclusion 76 . In the present study, lignin accumulation went along with a decrease in susceptibility and might be a factor contributing to pathogen resistance 77,78 .
Our results shed some light of the effects of elevated CO 2 on the mutualistic relationship between a grass and a fungus. Besides CO 2 concentrations, other factors such as temperature and water availability are likely to be altered in coming years 1 . Therefore, the response of grass-endophyte symbiosis to pathogens will be more complex and depend largely on the specific environmental conditions encountered. Given the extensive acreage of tall fescue worldwide and the fact that the ecological effects of this grass-fungal endophyte symbiosis have been observed at population, community, and ecosystem-scales 79 , understanding the response of tall fescue and its endophytic fungi to climate change may be important in predicting not only the responses of pathogens, but also grazing herbivores and ecological processes such as litter decomposition and nutrient cycling.

Conclusions
Our experiments provided evidence that endophyte infection improved the growth of tall fescue, but this benefit was affected by elevated CO 2 and N supply. Only under ambient CO 2 and high N conditions, both maximum net photosynthetic rate and shoot biomass were greater in E+ than in E− plants. With CO 2 concentration elevated, the beneficial effect of endophyte infection on the growth disappeared. Similarly, endophyte infection can enhance resistance of tall fescue towards Curvularia lunata only under ambient CO 2 . Elevated CO 2 counteracted the beneficial effect of endophyte infection on the growth and pathogen resistance of the host grass.

Materials and Methods
Plant material. Endophyte-infected (E+) seeds of tall fescue (Lolium arundinaceum Darbyshire ex. Schreb., KY-31) were naturally infected with Epichloë coenophialum 80,81 , and uninfected (E−) seeds were acquired by eliminating the endophyte through the long-term storage of E+ seeds at room temperature. This procedure reduces the viability of the endophyte but not the seeds 82 . E+ and E− seeds were originally obtained from Professor Keith Clay at Indiana University, USA. The seeds used in this experiment were several generations distant from the storage treatment and came from freely cross-pollinated field-grown parents. To re-isolate the endophyte, 30 E+ and 30 E− plant individuals were randomly sampled, and the method described by Latch & Christensen 83 was used with a slight modification that the time for sodium hypochlorite treatment was 8-10 min, and the petri plates containing potato dextrose agar (PDA) were incubated in the dark at 25 °C. Up to 4 weeks' examination, only one species of endophyte, E. coenophialum, was isolated from E+ seedlings while no endophyte was found in E− seedlings. Meantime, seed germination rates for E+ and E− seeds were compared before the experiment. No significant differences were found between them, with regard to the number of days to first seedling emergence and germinations rates. Four weeks later, seven equally sized seedlings were transferred into each plastic pot (15 cm × 13.5 cm) filled with 1.4 kg of sterilized sand. After a week's growth, they were differently treated and were placed into two separate growth chambers set at 400 or 800 ppm CO 2 . Plants were maintained at 30000 lux and a 12/12 h light/dark cycle at 25/20 °C, respectively. Endophyte status of the plants was checked both immediately before and after the experiment by microscopic examination from leaf sheaths stained with aniline blue described by Latch & Christensen 83 . We found that seedlings from E+ seeds were all infected (100%) while no seedling from E− seeds was infected (0%). Experiment design. The present study included two experiments. In the first experiment, we addressed the questions: does endophyte improve growth of the grass host under elevated CO 2 concentration? If this is the case, how does nitrogen (N) availability affect the symbiosis-dependent benefits? In the second experiment, we addressed the question: does elevated CO 2 affect pathogen resistance of grass-endophyte symbiont? From the first experiment, we found that endophyte-associated benefit only occurred in high N condition. So in the second experiment, test was performed only in high N level.

Experiment 1.
A three factors randomized block design was used in this experiment. The first factor was two CO 2 concentrations with two levels: ambient CO 2 (400 ppm, AC) and elevated CO 2 (800 ppm, EC). The second factor was N availability with two levels: high N (HN) and low N (LN). The third factor was endophyte infection status: endophyte-infected (E+) and uninfected (E−). Each treatment was replicated five times, totally 40 pots.
The nutrients were supplied by the addition of a modified Hoagland nutrient solution. O, and pH 6.0 ± 0.1. Nitrogen was added in the form of NH 4 NO 3 , which was delivered as 1 mM N (LN) or 10 mM N (HN), respectively. During the experiment, 100 ml of nutrient solution was added once a week to each pot, a total of 9 times. Plants were watered as necessary with deionized water. In each block, the positions of the pots were randomly rotated each week to minimize location effects. The experiment lasted for 63 days.

Experiment 2.
A three factors randomized block design was used in this experiment. The first factor was two CO 2 concentrations with two levels: ambient CO 2 (400 ppm, AC) and elevated CO 2 (800 ppm, EC). The second factor was pathogen inoculation with two levels: uninoculated control (P−) and inoculated by Curvularia lunata (P+). The third factor was endophyte infection status: E+ and E−. Each treatment was replicated five times, totally 40 pots. Pathogen inoculation was performed after 8 weeks' growing in the growth chamber with different CO 2 concentrations. All treatments were sampled at the 6th day after pathogen inoculation.
Response variables in Experiment 1. Photosynthesis parameters. At the end of experiment 1, gas exchange measurements were made on the youngest fully expanded attached leaf in a pot with a LI-COR 6400 infrared gas analyzer (LI-Cor, Lincoln, NE, USA). Under 400 μmol mol −1 or 800 μmol mol −1 CO 2 , net photosynthetic rate (Pn) was measured at 1,500, 1,200, 1,000, 800, 600, 400, 200, 150, 100, 50, 20 and 0 μmolm −2 s −1 PPFD (photosynthetic photon flux density). According to Pn-PPFD curve, Pmax were determined. Growth and biomass. At the end of experiment 1, regular measurement of tiller number, leaf number, and shoot height of the longest tiller were made on all ramets. Then, the shoot and the root were harvested separately. The harvested material was ven-dried at 80 °C for biomass measurement and C and N analyses.
Carbon (C) and nitrogen (N) concentration. C and N concentrations were determined using the dry combustion method with an Elemental Analyser (Vario EL/micro cube, Elementar, Hanau, Germany).

Response variables in Experiment 2. Pathogen inoculation and lesion index recorded. C. lunata was
obtained from Grassland Protection Institute, Lanzhou University, China. It was originally isolated from Poa pratensis. For inoculum, the pathogen was cultured on PDA at 25 °C for 2 weeks. Spores were washed with sterile distilled water and filtered through two-layer sterile gauze. A haemocytometer was used to count the spores, and the spore concentration was 13.44 × 10 5 /ml. Plants were inoculated by spraying the spore suspensions using a sprayer until small droplets were seen on the leaves 84 , and the control was sprayed with sterile distilled water. After inoculation, plants were immediately covered with a plastic bag for 36 h to maintain humidity.
Ten fully expanded mature leaves per pot were chosen for measuring the number and length of disease lesions. After measurement, pathogen spore concentration on the leaves was decided according to Nan & Li 84 .
Soluble sugar, amino acid and lignin. Soluble sugar content was analyzed using the phenol-sulphuric acid method according to Buysse and Merckx 85 . Amino acids were analyzed by reverse-phase high-performance liquid chromatography (HPLC, Waters 1500-series) with pre-column derivatization using dinitroflurobenzene (DNBF) according to Li and Sun 86 . Lignin measurement was according to the procedure of Reddy, et al. 87 .
Statistical analyses. For the amino acids, we performed a principal components analysis (PCA) on the correlations among the 17 response variables and then performed factor rotation using the varimax method 63, 88 . After varimax rotation, we retained four rotated factors (RF). The RF variables and all other indexes were subjected to three-way analyses of variance (ANOVA). Differences between the means were compared using Duncan's multiple-range tests at P < 0.05. All statistical analyses were performed using SPSS 21.0 software.