Stolons of bermudagrass (cv. ‘Tifway’) plants were collected from the research farm at Nanjing Agricultural University in Nanjing, China, and transplanted into pots (20 cm in diameter and 20 cm in depth) filled with a mixture of soil and sand (soil: sand = 1:1, v/v). Plants were grown in a greenhouse with average temperature of 30/22°C (day/night), natural sunlight and irrigated once a week with half-strength Hoagland’s nutrient solution (Hoagland and Arnon, 1950) to establish canopy and roots for 2 months. During this period, plants were trimmed once a week to keep a canopy height of 4–5 cm. After establishment, plants were transferred to growth chambers (Xubang, Jinan, Shandong province, China) with the temperature of 30/25°C (day/night), 70% relative humidity, photosynthetically active radiation of 650 μmol⋅m^-2⋅s^-1 and a 12-h photoperiod.

The CO₂ concentrations set-up and control in growth chambers followed the same designed as described in Yu et al. (2012a,b). Each CO₂ treatment was imposed in four growth chambers on September 1, 2015. In order to evaluate the long-term effects of elevated CO₂, plants were grown under the two CO₂ concentrations for 70 days prior to the exposure to heat stress. Plants grown under either CO₂ treatment was then exposed to (45/40°C) (heat stress) or 30/25°C (non-stress control) in two growth chambers on November 9, 2015 until December 8, 2015. Plants were randomly relocated in each chamber twice per week to avoid confounding effects of environmental variation between different chambers. The CO₂ concentration inside each growth chamber was controlled by an automated, open-chamber CO₂ control system connected to a gas tank containing 100% CO₂ (Yu et al., 2015).

The experiment was arranged as factorial design with two CO₂ concentrations (ambient CO₂ concentration at 400 ± 10 μmol⋅mol^-1 and elevated CO₂ concentration at 800 ± 10 μmol⋅mol^-1) and two temperature treatments [30/25°C (day/night, optimal temperature control) and 45/40°C (day/night, heat stress)]. Each treatment was repeated in four pots of plants (four replicates).

Leaf net photosynthetic rate (P_n) was determined by inserting 4–5 individual leaves (second full-expanded from the top) collected from each pot to a 6 cm² cuvette with a portable infrared gas analyzer (Li-6400, LI-COR, Inc., Lincoln, NB, United States). Leaves were placed in a leaf chamber with a built-in red and blue light source of the Li-6400 with the light intensity of 800 μmol photon⋅m^-2⋅s^-1.

For leaf chlorophyll content (Chl), 0.2 g of fresh leaves were detached from plants and then immersed in dimethyl sulfoxide (DMSO) in dark for at least 72 h for a complete extraction of total chlorophyll. The absorbance of the Chl extract was measured at wavelengths of 663 and 645 nm, respectively, by a spectrophotometer (Ultrospec 2100 pro, Biochrom Ltd., Cambridge, England) to calculate Chla and Chlb content. Chl was determined as described by Arnon (1949). For photochemical efficiency (F_v/F_m), chlorophyll fluorescence (the ratio of variable to maximum fluorescence as F_v/F_m) was measured by a fluorescence induction monitor (Bioscientific Ltd., Herts, United Kingdom) following 30 min dark acclimation through leaf tips.

The extraction procedure was conducted following the method of Roessner et al. (2000) and Rizhsky et al. (2004). Leaf samples collected at 28 days of treatment were collected and immediately frozen in liquid nitrogen, then stored at -80°C for metabolic profiling analysis. For each sample, frozen dry leaves were ground to a fine powder with liquid nitrogen, and then 25 mg of powder was transferred into a 10 mL microcentrifuge tubes, and extracted in 1.4 mL of 80% (v/v) aqueous methanol at 23°C for 2 h. Ribitol solution of 10 μL (2 mg⋅mL^-1 water) as an internal standard was added prior to incubation. Then, extraction was performed in a water bath at 70°C for 15 min. Tubes were centrifuged for 30 min at 9660 gn and the supernatant was decanted into new tubes, 1.4 mL of water and 0.75 mL of chloroform were added. The mixture was vortexed thoroughly and centrifuged for 15 min at 5025 gn and then 1 mL of the polar phase (methanol/water) was pipetted into HPLC vials and dried in a centrifugal concentrator (Centrivap, Labconco Corporation, Kansas City, MO, United States). The dried polar phase was methoximated with 80 μL of 20 mg⋅mL^-1 methoxyamine hydrochloride at 30°C for 90 min and then was trimethylsilylated with 80 μL N-methyl-N-(trimethylsilyl) trifluoroacetamide (MSTFA) (with 1% TMCS) for 60 min at 70°C.

The Gas Chromatography-Mass Spectrometer (GC-MS) analysis was modified from Qiu et al. (2007). The derivatized extracts were analyzed with a GC coupled with a TurboMass-Autosystem XL MS (Perkin Elmer Inc., Waltham, MA, United States). A 1 μL extracts was injected into a DB-5MS capillary column (30 m × 0.25 mm × 0.25 μm, Agilent J & W Scientific, Folsom, CA, United States). The inlet temperature was held at 260°C. After a 6.5 min solvent delay, initial GC oven temperature was maintained at 60°C; 1 min after injection, the GC oven temperature was raised to 280°C at a rate of 5°C⋅in^-1, and finally maintained at 280°C for 15 min. The injection temperature was set at 280°C and the ion source temperature was adjusted to 200°C. Helium was used as the carrier gas with a constant flow rate of 1 mL⋅min^-1. The measurements were performed through electron impact ionization (70 eV) in the full scan mode (m/z 30–550). The detected metabolites were identified with Turbomass 4.1.1 software (PerkinElmer Inc., Waltham, MA, United States). For GC/MS results, compounds were identified based on retention time (RT) and comparison with reference spectra in mass spectral libraries.

Leaf samples were collected from each tube at 28 days, immediately frozen in liquid nitrogen, then ground into fine powder and stored at -80°C until analysis. Proteins were extracted using the trichloroacetic acid (TCA)/Acetone method described from Xu and Huang (2008). Leaf powder samples (0.5 g) were homogenized on ice in precipitation solution (10% TCA and 0.07% 2-mercaptoethanol in acetone) for 10 min and then incubated at -20°C for 2 h. The protein pellet was collected and washed with cold acetone containing 0.07% 2-mercaptoethanol until the supernatant was colorless. The pellet was then vacuum-dried and suspended in resolubilization solution [8 M urea, 2 M thiourea, 2% CHAPS, 1% dithiothreitol (DTT), and 1% pharmalyte]. The suspension was centrifuged at 21000 g for 20 min and the supernatant was collected for further protein quantification. Protein content was determined using the method of Bradford (1976). A 10 μL aliquot of protein extract was mixed with 0.5 mL of a commercial color reagent (Bio-Rad Laboratories, Hercules, CA, United States) by a bovine serum albumin (BSA) standard. The absorbance was measured spectrophotometrically at 595 nm between 5 and 30 min after reaction.

An IPGPhor apparatus (GE Healthcare, Waukesha, WI, United States) was used for the first isoelectric focusing (IEF) described by Xu and Huang (2008). The extracts containing 300 μg of sample protein were used for IEF in immobilized pH gradient (IPG) strips (pH 3.0–10.0, linear gradient, 13 cm), formed by rehydrating strips for 12 h at room temperature in 250 μL of rehydration buffer (8 M urea, 2 M thiourea, 2% CHAPS, 1% DTT, 1% v/v IPG buffer, and 0.002% bromophenol blue). Following IEF, the IPG strips were equilibrated for 15 min twice at room temperature in equilibration buffer (50 mM Tris–HCl pH 8.8, 6 M urea, 30% glycerol, 2% SDS, and 1% DTT), then transferred to the same equilibration buffer containing 2.5% iodoacetamide instead of 1% DTT. The second dimension electrophoresis was run on a 12.5% SDS–polyacrylamide gel with a Hoefer SE 600 Ruby electrophoresis apparatus (GE Healthcare, Waukesha, WI, United States). The running conditions were 5 mA per strip for 30 min followed by 20 mA per strip for about 5 h. Gels were stained with Coomassie brilliant blue G-250 and scanned using a Personal Densitometer SI (63-0016-46, GE Healthcare, Waukesha, WI, United States).

Gel images analysis was performed by Progenesis software (Nonlinear Dynamics, Durham, NC, United States). Automatic default spot analysis settings were coupled with manual correction and editing of spot features. The spot volumes were normalized as a percentage of the total volume of all spots on the gel to correct the variability due to staining. Variance analysis of data was used to test the treatment effects on each transgenic line.

Selected protein spots were manually excised from gels and subjected to a trypsin digestion. The peptides were identified by MALDI-TOF-MS as described by Xu and Huang (2008). Data were searched against the National Center for Biotechnology Information (NCBI) database. Proteins containing at least two peptides with a confidence interval value >95% were considered to be successfully identified (Ma et al., 2016).

Protein functional classification was performed by Mapman software (Thimm et al., 2004) in combination with the criteria proposed by Bevan et al. (1998). The identified proteins were distributed to different subcellular location by SUBA (Tanz et al., 2013). Gene ontology (GO) for biological process, molecular function and cellular component was conducted by the agrigo database¹; threshold was -log10 > 4 (Ma et al., 2016).

Data were analyzed using statistics software (SPSS 13.0; SPSS Inc., Chicago, IL, United States). Analysis of variance (ANOVA) was used to determine differences among treatment effects at a given treatment time. The means ± SE were calculated for each parameter. When a particular F-test was significant, means were tested with least significant difference (LSD) at a confidence level of 0.05.

Under normal temperature, elevated CO₂ significantly increased P_n and Chl (Figures 1A,B) while it had no significant effects on Fv/Fm (Figure 1C). Under heat stress, plants grown at elevated CO₂ had significantly higher P_n (Figure 1A), Chl (Figure 1B), and F_v/F_m (Figure 1C) than that at ambient CO₂ concentration.

A total of 53 metabolites, including 18 organic acids and phosphoric acid, 12 amino acids, 18 sugars and 4 sugar alcohols, in responsive to elevated CO₂ and heat stress were identified and quantified by GC-MS. The name, RT, derivative and mass to charge (m/z) as well as the relative expression of each metabolite was presented in Figure 2 and Table 1.

Total content of organic acids, amino acids, sugars, and sugar alcohols were presented in Figure 3. Under normal temperature, no effects of elevated CO₂ were detected on total content of organic acids, amino acids, sugars and sugar alcohols compared with ambient CO₂. Under heat stress, elevated CO₂ resulted in significant increases in the content of organic acids, amino acids, sugars, and sugar alcohols by 52%, 2.79-fold, 29% and 30%, respectively (Figure 3).

For organic acids, under heat stress, plants grown at elevated CO₂ exhibited significantly lower content of mucic acid, galacturonic acid, lactic acid, but higher content of pyruvic acid, α-ketoglutaric acid, citric acid, glyceric acid, pimelic acid, malic acid and threonic acid compared to those with ambient CO₂ (Figures 4A–C). Plants exposed to elevated CO₂ had significantly lower content of pyruvic acid and α-ketoglutaric acid than those with ambient CO₂ under normal temperature (Figure 4B).

For amino acids, under heat stress, the content of all amino acids were increased by elevated CO₂ except glycine compared with ambient CO₂ (Figure 5). The content of alanine, GABA, and serine (Figure 5C) was significantly lower and the content of aspartic acid, isoleucine, lysine and glutamic acid (Figure 5D) was significantly higher in plants exposed to elevated CO₂ compared with ambient CO₂ treatments under normal temperature.

For sugars and sugar alcohols, the content of gentiobiose, gulose, lyxose (Figure 6A), and myo-inositol (Figure 7) was decreased while that of erythrose and glucopyranose (Figure 6B) was increased by elevated CO₂ compared to plants grown at ambient CO₂ concentration under normal temperature. Under heat stress, 8 out of 18 sugars, and two sugar alcohols (mannitol and galactinol) exhibited increases in the content in plants exposed to elevated CO₂ compared with ambient CO₂ (Figures 6, 7).

Out of 53 identified metabolites, 43 were placed into the metabolic pathways associated with GABA shunt, TCA cycle, sugar and amino acid metabolism (Figure 8). These 43 metabolites included 16 organic acids, 12 amino acids, 11 sugars and 4 sugar alcohols. Under heat stressed conditions, elevated CO₂ enhanced the accumulation of metabolites associated with GABA shunt, sugar and amino metabolisms.

A total of 70 and 53 protein spots were differentially expressed in leaves of bermudagrass due to elevated CO₂ compared to those at ambient CO₂ under normal and high temperature, respectively (Figure 9 and Table 2).

Those 70 proteins up- or down-regulated by elevated CO₂ under normal temperature were found in plastid (67.1%), cytosol (18.6%), mitochondrion (5.7%), peroxisome (4.3%), cytosol and plasma membrane (1.4%) and unknown locations (2.9%) (Figure 10). GO category enrichment showed that 70 proteins participated in various biological processes (metabolic process, response to stress, generation of precursor metabolites and energy, photosynthesis, carbohydrate metabolic process, glycolysis, protein folding, and carbohydrate biosynthetic process), molecular functions (catalytic activity and oxidoreductase activity) and cellular components (intracellular, cell, cytoplasm, organelle, plastid, chloroplast, membrane, protein complex, thylakoid, mitochondrion, envelope, plastoglobule, cytosol, stromule, and photosystem) (Figure 11A).

The majority of these 53 proteins up- or down-regulated by elevated CO₂ under heat stress mainly distributed in plastid (62.3%) followed by cytosol (17%) (Figure 10). GO category enrichment indicated that the biological processes regulated by CO₂ included cellular metabolic process, responses to stress, response to abiotic stimulus, generation of precursor metabolites and energy, photosynthesis, response to inorganic substance, electron transport chain, carbohydrate metabolism, molecular functions (catalytic activity and oxidoreductase activity) and cellular components (cytoplasm, intracellular, cell, organelle, plastid, chloroplast, membrane, mitochondrion, thylakoid, protein complex, envelope, photosystem, and stromule) (Figure 11B).

Based on the Venn analysis, 18 and 19 differential proteins were up-regulated by elevated CO₂ only under either normal temperature or heat stress, respectively (Figure 12A). Elevated CO₂ caused 12 proteins to be up-regulated regardless of temperature (Figure 12A). A total of 40 proteins were down-regulated by elevated CO₂ under normal and high temperature (Figure 12B). There were 27 proteins down-regulated under normal temperature and 7 proteins down-regulated under heat stress alone due to elevated CO₂ treatment. The differentially expressed proteins in responses to elevated CO₂ and heat stress were classified into different functional categories (Figure 13).

Under normal temperature condition, the differential proteins caused by elevated CO₂ compared with ambient CO₂ were involved in photosynthesis (55.7%), followed by amino acid metabolism (8.6%), glycolysis (7.1%), protein synthesis and degradation (4.3%), stress defense (8.6%), nucleotide metabolism (4.3%), and the remaining (11.4%) including those unknown functions (Figure 13A). For proteins related to photosynthesis, significant increases in the relative fold change were found in ATP synthase subunit (ATPA, n476) by 1.5-fold, rubisco large subunit (RBCL, n551, n561, n574) by 1.1- to 2.6-fold, glyceraldehyde-3-phosphate dehydrogenase (GAPDH, n660, n762, n770, n795) by 1.1- to 1.4-fold, ATP synthase subunit gamma (ATPC, n769) by 1.6-fold, fructose-bisphosphate aldolase (FBA, n781) by 1.4-fold, ferredoxin-NADP reductase (FNR, n865) by 1.3-fold, oxygen-evolving enhancer protein (OEE, n941) by 1.6-fold, chlorophyll a-b binding protein (LHC, n1066, n1076, n1127, n1195) by 1.3- to 2.4-fold, cytochrome b6-f complex iron-sulfur subunit (PGR, n1364) by 2.0-fold, rubisco small chain (RBCS, n1650) by 1.5-fold (Figure 14A). Other 21 proteins involved in photosynthesis [(chaperonin-60 alpha, CPN60A), n411 by 1.14-fold; (cpn60 chaperonin family protein, CPN60B), n417 by 1.52-fold; n486, ATPA by 1.53-fold; n525, n530 and n537, ATPB by 1.17- to 1.23-fold; n666, GAPDH by 1.23-fold; (Phosphoglycerate kinase, PGK), n673, n675 and n678 by 1.17- to 1.21-fold; (Sedoheptulose-1,7-bisphosphatase, SBPase), n705 by 1.18-fold; n793 and n794, FBA by 1.22 – 1.35; n942, OEE by 1.37-fold; n1073, TIM by 1.36-fold; (Photosystem I reaction center subunit, Psa), PsaD, n1403 and n1410 by 1.47- to 2.11-fold; n1473, PsaE by 1.23-fold; n 1507, ATPE by 1.3-fold; n1511, PsbQ by 2.02-fold; n1802, PsaH by 1.89-fold] were significantly down-regulated by elevated CO₂ under normal temperature compared with ambient CO₂ (Figure 14A).

Among proteins associated with the function of protein synthesis and degradation, cytokinin inducible protease (CLPC, n196) had a 1.9-fold up-regulation and the other two [(elongation factor, EF), n598 and n616] with 1.2- to 3.0-fold down-regulation compared with ambient CO₂ under normal temperature. Transketolase (n324) involved in oxidative pentose phosphate showed a 1.8-fold increase in response to elevated CO₂ (Figure 14A). Proteins involved in amino acid metabolism exhibited increases in methionine synthase protein (MS, n248, n250) by 1.4- to 1.7-fold and cysteine synthase C1 (CSase, n814) by 1.6-fold as well as decreases in aspartate aminotransferase (ASP, n736, n760) by 1.3- to 1.5-fold and S-adenosylmethionine synthase (SAMS, n619) by 1.3-fold in plants grown at elevated CO₂ compared with ambient CO₂. GAPDH associated with glycolysis were all down-regulated by 1.4- to 1.9-fold (n382, n776, n784) under elevated CO₂ except n808, which was increased by 2.4-fold under elevated CO₂ concentration. There were six proteins associating with stress defense of which four proteins [Heat shock protein (Hsp) 90, n238; Hsp 70, n300; Catalase, n447; Ascorbate peroxidase, n1019] were up-regulated by 1.2- to 1.4-fold and tow proteins (Hsp 70, n303; 2-cys peroxiredoxin BAS1, n1173) were down-regulated by 1.1- to 1.2-fold under elevated CO₂ concentration (Figure 14B).

Under heat stress, elevated CO₂-regulated proteins were classified the into following functional categories: photosynthesis (52.8%), TCA cycle (11.3%), stress defense (9.4%), glycolysis (7.5%), amino acid metabolism (5.7%), nucleotide metabolism (3.8%), metal handling (1.9%), major CHO metabolism (1.9%), protein synthesis (1.9%), cell cycle (1.9%) and transport (1.9%) (Figure 13B). Proteins associated with photosynthesis were mainly up-regulated by elevated CO₂ under heat stress, including CPN60A (h712) by 1.2-fold, CPN60B (h730, h736) by 1.2- to 1.3-fold, ATP synthase subunit beta (ATPB, n826) by 1.4-fold, RBCL (h843) by 1.5-fold, PGK, (h1067) by 1.3-fold, GAPDH (h1150, h1162, h1162, h1192) by 1.3-fold, SBPase (h1154, h1157) by 1.3-fold, ATPC (h1176) by 1.3-fold, FBA (h1211, h1233) by 1.3- to 1.5-fold, OEE (h1398) by 2.1-fold, LHC (h1655) by 1.6-fold, PGR (h1900, h1930) by 1.2- to 1.7-fold, PsaD (h1905) by 2.5-fold, photosystem I reaction center subunit N (PsaN, h2563) by 1.5-fold (Figure 15A). One protein associated with protein synthesis was upregulated by elevated CO₂ under heat stress (Figure 15A). All proteins involved in amino acid metabolism and nucleotide metabolism were up-regulated by elevated CO₂ under heat stress, including MS (1.6-fold), SAMS (1.5-fold), CSase (1.4-fold), pyrophosphorylase 6 (PPa6, 1.7-fold), adenylate kinase (ADK, 1.5-fold). Most TCA cycle related proteins [(Malic enzyme, ME), h716 and h724] by 1.4- to 1.5-fold; [(Malate dehydrogenase, MDH), h1237, h1242 and h1258] by 1.2- to 1.5-fold)] were up-regulated by elevated CO₂ compared with ambient CO₂ during heat stress (Figure 15B).

In our present study, 67 out of 123 proteins (54.5%) associated with photosynthetic pathways, including proteins involved in electron transport chain and Calvin cycle, were responsive to elevated CO₂ concentration under normal and high temperatures, as shown by the decrease or increase in their abundance (Figure 16). Under heat stress, the majority of proteins involved in photosynthesis exhibited accumulation in response to elevated CO₂ concentration, such as ATP synthase subunit (h826, h1176), photosystem I reaction center subunit [(PsaD, h1905) and (PsaN, h2563)] in light reactions of photosynthesis, and fructose-bisphosphate aldolase (FBA, h781, h1211, h1233), phosphoglycerate kinase (PGK, h1067) and sedoheptulose-1,7-bisphosphatase (SBPase, h1154, h1157) in Calvin cycle (Figure 15). FBA is a primary enzyme involved in the sixth reaction of Calvin cycle to convert fructose 1,6-bisphosphate into glyceraldehyde-3-phosphate (G3P) and dihydroxyacetone phosphate as well as ATP (Abbasi and Komatsu, 2004). FBA content at the level of protein significantly increased under elevated CO₂ in C3 tall fescue under heat stress (Yu et al., 2014) and creeping bentgrass (Agrostis stolonifera) under both well-water and drought stress (Burgess and Huang, 2016). SBPase functions as a bisphosphatase enzyme catalyzing sedoheptulose 1,7-bisphosphate dephosphorylation to sedoheptulose-7-phosphate during the regeneration phase of Calvin cycle (Raines et al., 1999). Overexpression of SBPase in C₃ tobacco (Nicotiana tabacum) had higher photosynthesis at elevated CO₂ compared with that at ambient CO₂ under field conditions (Rosenthal et al., 2011). The benefits of SBPase on the stimulation of photosynthesis depended on light intensity (Lefebvre et al., 2005; Rosenthal et al., 2011). Therefore, our study and another case in creeping bentgrass, conducted in light saturated growth chambers, the abundance of SBPase was enhanced by elevated CO₂ under abiotic stresses (Burgess and Huang, 2016). FBA is a primary enzyme involved in the sixth reaction of Calvin cycle to convert fructose 1,6-bisphosphate into glyceraldehyde-3-phosphate (G3P), dihydroxyacetone phosphate and ATP (Abbasi and Komatsu, 2004). In addition, FBA could directly affect ribulose-1,5-bisphosphate (RuBP) regeneration which actions as substrate of carbon fixation (Taiz and Zeiger, 2010). FBA content at the level of protein showed the grater accumulation in elevated CO₂ in C₃ tall fescue under heat stress (Yu et al., 2014) and creeping bentgrass (Agrostis stolonifera) under both well-water and drought stress (Burgess and Huang, 2016). PGK is a major enzyme catalyzing the phosphorylation of 3-phosphoglycerate to produce 1, 3-bisphosphoglycerate and ADP which is one of vital steps regenerating RuBP during Calvin cycle (Bernstein et al., 1997). The regulation of FBA and PGK induced by elevated CO₂ indicated that elevated CO₂ availability in atmosphere could be helpful for sustaining ATP supply and RuBP regeneration for plant growth under heat stress. ATP synthase is a critical enzyme for creating energy storage molecule ATP. Under high CO₂ availability, ATP synthase was found to decline in wheat grain (Högy et al., 2009). To our knowledge, our case is the first report on the abundance of ATP synthase and PGK in response to elevated CO₂ were found in C₄ plant species grown under heat stress. Our previous study in C₃ plant species found differential responses of photosynthesis-related proteins to elevated CO₂ different from found in bermudagrass in our study. In tall fescue, the abundance of ATP synthase subunit and PGK did not change in response to elevated CO₂ under heat stress (Yu et al., 2014). The increase in abundance and activity of a single or some enzyme(s) during photosynthesis could enhance carbon assimilation (Rosenthal et al., 2011). Taken together, the enhanced accumulation of proteins involved in photosynthesis by elevated CO₂ under heat stress in bermudagrass suggested that elevated CO₂ could help to maintain photosynthesis to withstand the adverse environments as various proteins are involved in the light harvesting, electron transport, and carbohydrate assimilation processes of photosynthesis.

Other proteins related to photosynthesis such as GAPDH, OEE, PGR exhibited the enhanced expression in plants grown at elevated CO₂ concentration under both temperatures in our study. GAPDH could convert G3P to D-glycerate 1,3-bisphosphate as well as mediating the formation of NADH and ATP (Tristan et al., 2011). It has multiple functions, such as two chloroplastic forms playing photosynthetic function locating in chloroplast and one cytosolic form participating in glycolysis in higher plants (Sparla et al., 2005; Tarze et al., 2007). In chloroplasts, GAPDH catalyzes a reaction of NADPH-consuming which is regulated by light utilizing thioredoxins and metabolites during Calvin cycle (Sparla et al., 2005). Various stresses caused the decline in chloroplastic GAPDH whereas stress-tolerant species exhibited higher GAPDH abundance than stress-sensitive plants, such as creeping bentgrass under heat stress (Xu and Huang, 2010a; Merewitz et al., 2011), salinity stress (Xu and Huang, 2010b) and drought stress (Burgess and Huang, 2016). Plants with lower GAPDH abundance were generally associated with decreased photosynthetic capacity resulted from reduced RuBP regeneration rate, followed with the decline in accumulation of photosynthetic products (Price et al., 1995; Burgess and Huang, 2016). However, elevated CO₂ had no effects on chloroplastic GAPDH abundance under heat stressed condition but caused significant decrease under non-stressed control plants (Yu et al., 2014). Overall, our study suggested that enhanced abundance of photosynthesis-related proteins could contribute to the improved photosynthetic activities by elevated CO₂, particularly under heat stress, which could be reflected with improved P_n and increased content of sugars, such as fructose, glucose, sucrose, erythrose, and glucopyranose.

It has been widely known that glycolysis and TCA cycle are vital pathways for energy supply, amino acid synthesis and various other biological processes in plants (Fernie et al., 2004; Ma et al., 2016). As substrate of photosynthesis for carboxylation, plants grown at elevated CO₂ tended to accumulate the larger amount of non-structural carbohydrates (Yu et al., 2012a; Song et al., 2014). Most monosaccharides (glucose, fructose, galactose, etc.) as substrate or intermediates play vital roles during glycolysis. Glycolysis pathway could convert glucose into pyruvate via a series of intermediate metabolites and cytosolic GAPDH is one of essential enzymes catalyzing the sixth step of respiratory glycolysis to convert G3P to 1, 3-bisphosphateglycerate (1, 3-BPG) which is one of the most important reactions during the glycolytic pathway (Mijeong et al., 2000). The increase of pyruvic acid (pyruvate) as the product of glycolysis, followed by the enhanced content of valine, isoleucine and alanine, was partly due to elevated CO₂-caused accumulation of glucose under heat stress in our study, since those metabolites are all derived from glucose. In C₃ tall fescue, we also observed the significant increases in valine and alanine but not for isoleucine resulted from elevated CO₂ under heat stress (Yu et al., 2012a). During the pathway of glycolysis, the abundance of GAPDH in cytosol (n776, n784, n816 except n808) and phosphoglycerate mutase (PGAM, n382) exhibited the down-regulation in response to elevated CO₂ rather than ambient CO₂ under normal temperature, while under heat stressed conditions elevated CO₂ caused up-regulation in GAPDH (h845, h1164, h1167 except h1182) in bermudagrass (Figure 16). GAPDH might serve as a provider of additional energy for plant growth and development under stressed conditions and stress tolerance could be enhanced by improved abundance of GAPDH to cope with environmental stresses (Mijeong et al., 2000; Bertrand et al., 2007). In C₃ plants, no consistent changes were found due to variations in plant species. For example, in tall fescue and creeping bentgrass, the abundance of cytosolic GAPDH exhibited either no changes or decrease under elevated CO₂ and heat stressed condition (Yu et al., 2014; Burgess and Huang, 2016). Kappachery et al. (2015) found that GAPDH gene-silenced lines showed more sensitive traits to drought stress than non-silenced lines in potato (Solanum tuberosum). By contrast, the higher shoot length and weight were detected in GAPDH overexpression transgenic plants compared with wild-type plants (Kappachery et al., 2015). In the level of transcription in potato, cytosolic GAPDH RNA accumulation was also increased under biological stress (Laxalt et al., 1996). Therefore, in our study, the higher abundance of GAPDH caused by elevated CO₂ was beneficial for energy supply to support plant growth under heat stress.

Malate dehydrogenase (MDH) acts as an enzyme to catalyze the oxidation of malate to oxaloacetate via the reduction of NAD⁺ to NADH in mitochondrial matrix during TCA cycle (Musrati et al., 1998). Environmental stresses including drought (Burgess and Huang, 2016), heat (Xu and Huang, 2010a), salinity (Xu et al., 2010) and Al-stress (Ramírez-Benítez et al., 2008) have been shown to decrease the level of MDH in various plant species. However, limited studies about MDH were found in plants grown at elevated CO₂ concentrations, especially under stressed conditions (Burgess and Huang, 2016). In this study, elevated CO₂-responsive MDH (n827, h1237, h1242, h1258) involved in TCA cycle exhibited up-regulated expression regardless of temperatures, suggesting that CO₂ inhibited the heat-induced reduction in MDH to catalyze the enhanced malate (malic acid) to oxaloacetate (oxaloacetic acid) during malate metabolism.

In addition to function in TCA cycle, MDH also participates in the process of amino acid synthesis due to the relations among malate, oxaloacetate and aspartate (Musrati et al., 1998; Wen et al., 2015). Several amino acids including aspartate (aspartic acid), methionine, threonine, isoleucine, lysine derived from oxaloacetate and aspartate is the precursor of methionine, threonine, isoleucine and lysine (Muehlbauer et al., 1994). Along with the significant increase in malic acid and aspartic acid, the content of threonine, isoleucine and lysine were stimulated by elevated CO₂ during heat stress. Furthermore, the content of alanine, valine and serine were also enhanced by elevated CO₂ compared with ambient CO₂ under heat stress. Alanine, valine and serine are used for synthesis of several proteins and associated with many metabolic processes (Bourguignon et al., 1999). The stimulation of elevated CO₂ concentration on the content of alanine, valine and serine was found in other species under abiotic stresses, as previously reported in C₃ grass species under heat stress (Yu et al., 2012a) and tree seedlings under drought stress (Tschaplinski et al., 1995). The increase in synthesis of both alanine and valine in present study is directly associated with the higher content of pyruvate (pyruvate acid) which is the final product of glycolysis (Schulzesiebert et al., 1984). Superior stress tolerance has been reported with the higher content of alanine, valine and serine as well as other amino acids such as GABA, glutamic acid, proline and 5-oxoproline involved in the GABA shunt pathway in plant species, including perennial grasses (Merewitz et al., 2012; Xu et al., 2013; Shi et al., 2014; Li et al., 2016a,b). The GABA shunt was considered to be a part of the TCA cycle during respiration besides its central role in primary carbon and nitrogen metabolism (Fait et al., 2008). In tall fescue, the content of GABA was significantly decreased by elevated CO₂ under high temperature (Yu et al., 2014). While, in bermudagrass of this case, GABA and glutamic acid exhibited the opposite response to elevated CO₂ under heat stress. Increased GABA caused the enhanced content of alanine and pyruvate which was turned into TCA cycle and proline metabolism. The content of all amino acids except arginine during GABA shunt was increased by elevated CO₂ under heat stress suggesting a predominant role of elevated CO₂ in carbon and nitrogen metabolism in C₄ bermudagrass.

Proteins including methionine synthase (MS), cysteine synthase (CSase) and S-adenosylmethionine synthase (SAMS) associated with amino acid metabolism were up-regulated by 1.4- to 1.6-fold by elevated CO₂ under heat stress. MS and SAMS serve as regulators in the synthesis and degradative pathways of various amino acids (Bohnert and Jensen, 1996). It was detected by the same proteomic analysis that many proteins involved in amino acid metabolism accumulated more or degraded less in stress-tolerant plants, such as MS and SAMS (Merewitz et al., 2011). CSase functions in the final strep in cysteine synthesis in plants. Plants with overexpressing CSase gene displayed high tolerance to toxic environmental pollutants, such as sulfur dioxide and sulfite (Noji et al., 2001), cadmium toxicity (Harada et al., 2001) in tobacco and aluminum toxicity in rice (Yang et al., 2007). The accumulation of many amino acids as well as proteins involved in amino acid metabolism in this study could contribute to elevated CO₂-improved heat tolerance.

In summary, elevated CO₂ concentration suppressed heat-induced damages in bermudagrass, as shown by the increased P_n, Chl and F_v/F_m. The improvement of heat tolerance under elevated CO₂ could be associated with some important metabolic pathways during which proteins and metabolites were up-regulated, including proteins, sugars and/or amino acids involved in light reaction (ATP synthase subunit and photosystem I reaction center subunit) and carbon fixation of photosynthesis (GAPDH, FBA, PGK, SBPase and sugars), glycolysis (GAPDH, glucose, fructose and galactose) and TCA cycle (pyruvic acid, malic acid and MDH) of respiration, amino acid metabolism (aspartic acid, methionine, threonine, isoleucine, lysine, valine, alanine and isoleucine) as well as the GABA shunt (GABA, glutamic acid, alanine, proline and 5-oxoproline). The molecular factors and mechanisms underlying the metabolic changes caused by elevated CO₂ during plant responses to heat stress require further investigation.