The objective of the study was to assess the impact of elevated CO 2 in interaction with rhizobial strains on freezing tolerance and cold-induced molecular changes in alfalfa.
Alfalfa inoculated with two different strains of rhizobium (A2 and NRG34) was grown and cold acclimated (2 weeks at 2 °C) under either 400 (ambient) or 800 µmol mol −1 (elevated) CO 2 .
Plants acclimated under 400 µmol mol −1 CO 2 were more freezing tolerant than those maintained under 800 µmol mol −1 . Cryoprotective sugars typically linked with the acquisition of freezing tolerance such as sucrose, stachyose and raffinose increased in roots in response to low temperature but did not differ between CO 2 treatments. Similarly high CO 2 did not alter the expression of many cold-regulated (COR) genes although it significantly increased the level of transcripts encoding a COR gene homologous to glyceraldehyde-3-phosphate-dehydrogenase (GAPDH). A significant effect of rhizobial strain was observed on both freezing tolerance and gene expression. Plants of alfalfa inoculated with strain A2 were more freezing tolerant than those inoculated with strain NRG34. Transcripts of COR genes homologous to a pathogenesis-related protein (PR-10) and to a nuclear-targeted protein were markedly enhanced in roots of alfalfa inoculated with strain A2 as compared with strain NRG34. Transcripts encoding the vegetative storage proteins (VSPs) β-amylase and chitinase were more abundant in roots of non-acclimated plants inoculated with strain NRG34 than with strain A2.
Taken together, the results suggest that elevated CO 2 stimulates plant growth and reduces freezing tolerance. The acquisition of cold tolerance is also influenced by the rhizobial strain, as indicated by lower levels of expression of COR genes and sustained accumulation of VSP-encoding transcripts in alfalfa inoculated with strain NRG34 as compared with strain A2.
The atmospheric CO 2 concentration is expected to increase and perhaps even double during this century ( Intergovernmental Panel on Climate Change, 2001 ). It is well established that plants grown under elevated CO 2 exhibit enhanced photosynthesis and biomass production at least in the short term (reviewed by Drake et al. , 1997 ). However, only a few studies have focused on the impact of high CO 2 concentration on the response of perennial plants to environmental stresses, particularly freezing stress. Among these studies, contrasting effects have been reported: CO 2 enrichment led to more severe frost damage in leaves of Eucalyptus pauciflora ( Barker et al. , 2005 ) and Ginko biloba ( Terry et al. , 2000 ), and in a native temperate grassland ( Obrist et al ., 2001 ), whereas it increased frost resistance of Betula allaghanensis ( Wayne et al ., 1998 ) and Picea mariana ( Bigras and Bertrand, 2006 ) but had no effect on freezing tolerance of Picea abies ( Dalen et al ., 2001 ). There are indications that high CO 2 , by stimulating growth during cold acclimation, can predispose perennial plants to cold damage ( Margolis and Vezina, 1990 ). In contrast, the high CO 2 -driven accumulation of carbohydrate ( Poorter et al ., 1997 ) could promote superior frost hardiness through an increase in the levels of cryoprotective sugars such as stachyose, sucrose and raffinose, closely related to the development of freezing tolerance in alfalfa ( Castonguay et al ., 1995 ). Elevated CO 2 concentration not only typically increases total non-structural carbohydrate (TNC) but also reduces nitrogen (N) concentration in plant tissues ( Skinner et al ., 1999 ). Dhont et al . (2006) have shown that endogenous N pools that accumulate in taproots play an important role in the overwintering potential of alfalfa. Therefore, factors reducing N reserves such as defoliation or elevated atmospheric CO 2 could decrease the capacity of alfalfa to withstand environmental stresses during winter.
An additional factor bound to modify alfalfa physiology and tolerance to environmental stress is the association with its symbiont Sinorhizobium meliloti . Graham (1992) reviewed the impact of stress tolerance on legume symbiosis and concluded that soil acidity, drought, and P and N availability can affect interactions between host and rhizobia and the performance of the symbiosis. With respect to temperature stress, cold conditions delay or inhibit the establishment and the functioning of the rhizobium–legume symbiosis (reviewed by Prévost et al ., 1999 ). However, Prévost et al. (1999) showed that the selection of cold-adapted rhizobia constitutes a valuable approach to counteract the negative effect of low temperatures by improving legume productivity under these conditions. Using this approach, Prévost et al . (2003) demonstrated that regrowth of alfalfa after overwintering is influenced by the strain of S. meliloti , showing the potential of strain selection for improving plant survival after exposure to winter stresses. No information is currently available on the impact of CO 2 concentrations and the potential interaction with the strain of rhizobium on the acquisition of freezing tolerance of alfalfa.
Cold acclimation of alfalfa is characterized by significant biochemical changes including modifications in proline ( Paquin and Pelletier, 1981 ), arginine ( Dhont et al ., 2006 ), vegetative storage proteins (VSPs) ( Avice et al ., 1996 ) and concentrations of soluble sugars ( Castonguay et al ., 1995 ). Furthermore, a close relationship between freezing tolerance and gene expression in alfalfa is suggested by the differential accumulation of cold-inducible gene products in alfalfa cultivars of contrasting freezing tolerance ( Monroy et al ., 1993 ; Castonguay et al ., 1997 ). The characterization of the modification of these cold-induced molecular changes in response to CO 2 concentrations and rhizobial strains will help to unravel the impact of these factors on alfalfa freezing tolerance.
The objectives of the research project were ( a ) to assess the effect of elevated concentration of atmospheric CO 2 on the growth, the accumulation of C and N reserves, the molecular changes occurring at low temperature and the freezing tolerance of alfalfa; and ( b ) to compare those responses between strains of S. meliloti differing in their adaptation to winter stress.
MATERIALS AND METHODS
Plant material and treatments
Fifteen seeds of alfalfa, Medicago sativa L., ‘AC Caribou’ were sown in 12.5 cm diameter pots filled with non-sterile topsoil. Pots were distributed between growth chambers (Environmental Growth Chamber, model M18-SI, Chagrin Falls, OH, USA) with contrasting atmospheric CO 2 concentrations. The high CO 2 treatment was of 800 ± 30 µmol mol −1 while the near ambient CO 2 treatment was of 400 ± 20 µmol mol −1 . Seedlings were initially maintained under optimal growth conditions and were subsequently exposed to low temperature for their cold acclimation. Details of the environmental conditions, plant management and samplings performed during the experiment are provided in Table 1 . After 1 week of growth, seedlings were thinned to ten plants per pot and were inoculated with 50 mL of a fresh S. meliloti suspension containing 10 8 cells of either strain A2 or strain NRG34. These two strains are used in Canadian commercial inoculants; strain A2 is adapted to survive in soils after freezing ( Bordeleau et al ., 1977 ) and strain NRG34 is adapted for nodulation and plant yield at low temperature ( Rice et al ., 1995 ). Plants were fertilized as follows: 1 g L −1 of 8-20-30 for the first 2 weeks, 1 g L −1 of 0-52-34 plus micronutrients for the six subsequent weeks and no fertilization during cold acclimation. Plants were kept well watered throughout the experiment.
|Day||Management||Temperature (day/night) (°C)||Relative humidity (%)||PPFD (μmol m −2 s −1 )||Photoperiod (h)|
|7||Thinning and inoculation||22/17||70||600||16|
|28||First sampling and first cut||22/17||70||600||16|
|71||Third sampling and freezing test|
|Day||Management||Temperature (day/night) (°C)||Relative humidity (%)||PPFD (μmol m −2 s −1 )||Photoperiod (h)|
|7||Thinning and inoculation||22/17||70||600||16|
|28||First sampling and first cut||22/17||70||600||16|
|71||Third sampling and freezing test|
Root systems were gently washed under a stream of cold water to remove soil, and sponged. Roots and aerial parts were separated at the crown level with a cutter and weighed. Plant material was then oven dried for 48 h at 70 °C for dry matter determination and the estimation of dry weight (d. wt).
Assessment of freezing tolerance
Freezing tests were performed in programmed freezers following an 8 h equilibration period at −2 °C according to a procedure described previously ( Castonguay et al ., 1993 ). The temperature was lowered by 2 °C during a 30 min period followed by a 60 min plateau at each test temperature. Two simultaneous freezing tests were performed in separate freezers as replicates. Plants were tested between −8 and −18 °C. At the end of each temperature plateau, four pots for each CO 2 treatment and strain were withdrawn from the freezers and thawed at 2 °C for 24 h. Plants were then transferred to regrowth conditions (21/17 °C day/night, 16 h photoperiod, ambient CO 2 concentration). After 3 weeks, survival counts were taken and the 50 % lethal temperature (LT 50 ) was computed using the SAS Probit procedure (SAS Institute, Cary, NC, USA, 1999–2001). Freezing tolerance was assessed once, after the cold acclimation period.
Carbohydrate and amino acids analyses
Plants were sampled three times during the experiment, after 1 and 2 months of growth and after 2 weeks of cold acclimation (Table 1 ). The following measurements were made on four pots (ten plants per pot) for each combination of CO 2 treatments and strain inoculant: growth parameters, biochemical analysis and gene expression characterization.
For leaf analysis, all plant leaves were thoroughly mixed to obtain a homogenous sample. For root analysis, the root system was separated from the stem and cut in approx. 0·3 cm sections with a chopper to obtain a homogenous sample. For each pot, approx. 1 g of fresh weight (f. wt) of a pooled sample of the ten plants was collected for both leaves and roots. Samples were ground to a fine powder in liquid nitrogen and extracted in methanol: chloroform: water (12 : 5 : 3, v/v/v) in a 1 : 6 (w/v) ratio, heated for 30 min at 65 °C to stop enzymatic activity and stored at −80 °C until extraction. Tubes were subsequently centrifuged for 10 min at 1500 g and the supernatant was collected. For phase separation, 0.25 mL of water was added to a 1 mL sub-sample of supernatant, tubes were shaken and centrifuged for 3 min at 21 000 g and the upper phase was collected. A 1 mL sub-sample was evaporated to dryness on a Savant Speed Vac (SC210A) vacuum concentrator/centrifugal evaporator system (Fisher Scientific, Ottawa, ON, Canada), and then solubilized in 1 mL of EDTA (Na + , Ca 2+ , 50 mg L −1 ) and kept frozen at −80 °C. The non-soluble residues left after extraction were washed twice with 10 mL of methanol and used for starch determination. For each pot, a sub-sample of leaves and roots was oven dried for 48 h at 70 °C for determination of the percentage of dry matter.
Mono-, di-, tri- and tetrasaccharides were separated and quantified by high-performance liquid chromatography (HPLC). Samples were centrifuged for 3 min at 21 000 g prior to analysis. The HPLC analytical system was controlled by WATERS Millennium 32 software (WATERS, Milford, MA, USA) and was composed of a Model 515 pump, a Model 717 plus autosampler and a Model 2410 refractive index detector. Sugars were separated on a WATERS Sugar-Pak column eluted isocratically at 85 °C at a flow rate of 0·5 mL min −1 with EDTA (Na + , Ca 2+ , 50 mg L −1 ) and detected on a refractive index detector. Peak identity and sugar quantity were determined by comparison with standards.
Starch was quantified as glucose equivalent with the p -hydroxybenzoic acid hydrazide method of Blakeney and Mutton (1980) after gelatinization at 100 °C and digestion for 90 min with amyloglucosidase (Sigma A7255, Sigma Chemical Co., St Louis, MO, USA). The amounts of starch were determined spectrophotometrically by reference to a glucose standard curve. Results from carbohydrate analyses were expressed on a concentration basis (mg g −1 d. wt).
Amino acids, with the exception of proline, were analysed by HPLC according to a procedure described previously in Bertrand and Bigras (2006) , using pre-column derivatization of aqueous extracts with o -phtalaldehyde in 0·5 m K Borate buffer, pH 10. For proline determination, 500 µL of the resuspended aqueous extract was mixed with 300 µL of a ninhydrin solution [0·125 g of ninhydrin dissolved in 5 ml of H 3 PO 4 (6 m ): glacial CH 3 COOH, 2 : 3, v/v) and 200 µL of glacial CH 3 COOH. Each sample was thoroughly mixed before being heated at 100 °C for 45 min. Samples were then cooled before adding 800 µL of toluene. After 45 min, the optical density (OD) of the upper phase (toluene) was assessed by spectrophotometry at 515 nm. The proline content was calculated from the regression curve of OD obtained from standard solutions of pure proline, ranging from 0 to 10 µg.
RNA extraction and analysis
Total RNA was extracted as described by De Vries et al . (1988) from 1 g of homogenous sub-sample of roots collected at the end of the growth period (day 56) and after 2 weeks of cold acclimation (day 71). Extracted RNA was dissolved in Tris-EDTA buffer (TE; 10 m m Tris–HCl, pH 7·4, 1 m m EDTA) and quantified by UV absorption at 260 nm.
The relative level of expression of cold-regulated (COR)- and VSP-encoding genes was assessed using dot blot analysis of total RNA (5 µg per well). Total RNA from each replicate sample was transferred by vacuum blotting to a nylon membrane (Immobilon- NY + , Millipore, Billerica, MA, USA) using a 96-well BIO-DOT microfiltration apparatus (Bio-Rad Laboratories Inc., Mississauga, ON, Canada). The cDNA of the COR genes msa CIB ( Monroy et al ., 1993 ), msa CIC ( Castonguay et al ., 1994 ), msa CID and msa CIE ( Castonguay et al ., 1997 ), and msa CIG, homologous to cas17 ( Wolfraim and Dhindsa, 1993 ), as well as the cDNA of alfalfa VSP, chitinase ( Meuriot et al ., 2004 ) and β-amylase ( Gana et al ., 1998 ) were radiolabelled using the Rediprime II random prime labelling system (Amersham Biosciences, Piscataway, NJ, USA) using [α- 32 P]dCTP. Hybridization was performed at 68 °C in 2 × SSC (0·6 m NaCl, 300 m m Na citrate), containing 0·25 (w/v) low-fat powder milk. Membranes were washed free from probe with 2 × SSC. The dots were then cut out and placed in Ready Safe scintillation cocktail (Beckman Coulter, Fullerton, CA, USA) in order to determinate the relative transcript abundance of each gene.
The experimental design was a split–split plot with main plots arranged in four blocks. Each block consisted of two growth chambers containing the main plots; each growth chamber was assigned a CO 2 level. The sub-plots consisted of the two rhizobial strains randomized within the main plot. The dates represented the sub-sub-plots randomized within rhizobial strains. The experiment included a total of 288 pots with ten plants in each: four blocks × three sampling dates × two treatments (ambient CO 2 or high CO 2 ) × two strains (A2 or NRG34) × one pot for biochemical analysis, and the remaining pots were used for the determination of LT 50 at the end of the experiment. Data were subjected to an analysis of variance including sampling dates, CO 2 treatments and rhizobial strains as main factors, and by using the GLM procedure of the SAS software package (SAS Institute, Cary, NC, USA, 1999–2001).
Plant dry weight increased during the experiment ( P < 0·001) and was higher under high than under ambient CO 2 ( P < 0·001). Although under ambient CO 2 plants had similar growth responses regardless of the strain of rhizobium that was used, dry matter accumulation markedly differed between the two inoculants under high CO 2 . Plants inoculated with strain A2 had significantly higher dry matter accumulation on day 56 than those inoculated with strain NRG34. However, after a 2 week exposure at low temperature, both groups had similar dry weight as a result of continued growth of plants inoculated with NRG34 and a near cessation of dry weight accumulation for plants in symbiosis with strain A2 (Fig. 1 ).
Freezing tolerance was significantly lower for plants that were grown and cold acclimated under high CO 2 than for those kept under ambient CO 2 ( P < 0·001). Plants of alfalfa inoculated with strain A2 were more freezing tolerant than alfalfa inoculated with strain NRG34 under both CO 2 concentrations ( P < 0·001, Fig. 2 ).
Carbohydrate and nitrogen reserves
The concentration of carbohydrate and nitrogen reserves did not differ in response to rhizobial strains. In both leaves and roots, concentrations of total soluble sugars combining sucrose, raffinose and stachyose increased throughout the experiment ( P < 0·001) but did not differ between CO 2 treatments (Fig. 3 A, D). In contrast, starch concentration in leaves continuously declined from day 28 until day 71 and was slightly higher (Fig. 3 B, P = 0·06) in plants grown under high than under ambient CO 2 . Starch concentration in roots initially increased from day 28 to 56, and subsequently declined upon transfer to low temperature concomitant with a sharp increase in soluble sugars (Fig. 3 E). There were no differences in starch concentration measured in roots of plants growing under either ambient or high CO 2 . The soluble sugar pinitol was more abundant in leaves (Fig. 3 C, P = 0·007) and in roots (Fig. 3 F, P = 0·04) of plants kept under ambient than under high CO 2 .
Total amino acids increased in leaves with time and were slightly less abundant under elevated CO 2 ( P = 0·07) (Fig. 4 A). The proline concentration in leaves differed significantly ( P = 0·01) in response to the CO 2 treatment and was more abundant under ambient than under elevated CO 2 (Fig. 4 B). Tyrosine was the only amino acid that showed a significant difference between strains ( P = 0·04), with a higher level in leaves with strain A2 (3·24 ± 0·97 nmol g −1 d. wt) than in those with strain NRG34 (1·74 ± 0·32 nmol g −1 d. wt). Concentrations of total amino acid in roots were similar for both CO 2 treatments; they initially increased during plant growth and slightly declined during cold acclimation (Fig. 4 C). Arginine concentrations increased in roots throughout the experiment but more markedly after the temperature was lowered (Fig. 4 D). In both leaves and roots, the most abundant amino acids were asparagine and proline. The mean asparagine concentrations for all treatments and sampling dates were of 7·6 ± 2·8 nmol g −1 d. wt in leaves and of 79·6 ± 14·2 nmol g −1 d. wt in roots.
As expected, the levels of transcripts encoded by COR genes from alfalfa were much higher in cold-acclimated than in non-acclimated roots ( P < 0·001). Relative levels of expression varied between COR genes, with the dehydrin homologue msa CIG having approx. 5-fold higher transcript abundance than the msa CIB, CIC and CIE genes (Fig. 5 E). The gene msa CIE, homologous to glyceraldehyde-3-phosphate-dehydrogenase (GAPDH), was the only COR gene whose transcripts were found to be more abundant in roots of plants cold-acclimated under high than under ambient CO 2 (Fig. 5 A, P = 0·01). However, the level of expression of the msa CIB (nuclear targeted, P = 0·002), msa CIC (bi-modular protein, P = 0·01) and msa CID [pathogenesis (PR) protein, P = 0·01] genes was significantly higher in roots of plants inoculated with strain A2 than in those noculated with NRG34 (Fig. 5 B, C and D, respectively).
Transcripts encoding the VSP β-amylase and chitinase were more abundant in non-acclimated than in cold-acclimated roots (Fig. 6 A, B). The gene expression of β-amylase differed significantly between rhizobial strains in non-acclimated plants ( P = 0·01), with more abundant transcripts being detected in plants inoculated with strain NRG34 as compared with strain A2. The transcript level of chitinase remained higher in acclimated alfalfa inoculated with strain A2 and was at its lowest in plants inoculated with strain NRG34 under high CO 2 ( P = 0·05).
This study on the effects of elevated atmospheric CO 2 on growth of alfalfa confirms predictions that elevated CO 2 will stimulate dry matter production in N-fixing legumes ( Fuhrer, 2003 ). However, a more detailed analysis of indirect effects of an increase in CO 2 on the capacity of alfalfa to withstand sub-freezing temperatures and potential interactions with rhizobial strains reveal that more complex responses are bound to occur at the agroecosystem level. In northern climates, the persistence of stands of alfalfa is to a large extent determined by the capacity of the plants to withstand exposure to low sub-freezing temperatures during the winter months ( Castonguay et al ., 2006 ). It was observed that plants of alfalfa grown and acclimated to low temperature under elevated CO 2 were more sensitive to freezing temperatures than those maintained under ambient CO 2 . Similar increases in frost sensitivity of evergreens ( Dalen and Johnsen, 2004 ; Barker et al ., 2005 ; Loveys et al ., 2006 ) and herbaceous perennials ( Obrist et al ., 2001 ) grown under elevated CO 2 have been reported in recent years. On the other hand, the acceleration of phenological development leading to earlier acquisition of autumn hardiness in conifers exposed to high CO 2 ( Bertrand and Bigras, 2006 ) might help explain reports of positive effects ( Tinus et al ., 1995 ) or absence ( Wiemken et al ., 1996 ) of effects of high CO 2 on freezing tolerance. However, a report by Dalen et al . (2001) that elevated CO 2 did not affect the timing of bud set or freezing tolerance in Norway spruce indicates that developmental acceleration might not be a general response to an increase in atmospheric CO 2 .
A significant impact of the rhizobium strain on the capacity of alfalfa to withstand sub-freezing temperatures was also observed. The results support the suggestion by Prévost et al . (1999) of the use of strains adapted to cold to improve symbiotic nitrogen fixation in temperate regions and plant freezing tolerance. Although CO 2 reduces freezing tolerance with both strains, the results show, however, that it is possible to select a better inoculant strain, helping to mitigate this reduction.
The underlying causes of the effects of elevated CO 2 and strains of rhizobium on freezing tolerance are unknown and probably involved multiple interactions between environmental factors and both physiological and biochemical processes ( Loveys et al ., 2006 ). In the present study, an in-depth investigation on alfalfa physiology and genetic expression allows a better determination of the influence of the rhizobial strain on the mechanisms of cold resistance of alfalfa.
Low temperature-induced hydrolysis of starch reserves in taproots and the concomittant accumulation of soluble sugars, mainly sucrose, raffinose and stachyose, have been positively associated with the acquisition of freezing tolerance and winterhardiness potential in alfalfa ( Castonguay et al ., 1995 ; Cunningham et al ., 2003 ). In the current study, concentrations of sucrose, raffinose and stachyose did not differ between CO 2 treatments and rhizobial strains and were not related to the observed differences in freezing tolerance between these treatments. It was observed, however, that concentrations of the sugar alcohol pinitol were superior in the more freezing-tolerant plants grown under near ambient CO 2 than in those maintained under elevated CO 2 . Streeter et al . (2001) found higher pinitol concentrations in leaves of water-stressed soybean than in those of well-watered plants, and suggested that because of its slow turnover and limited reactivity, pinitol has good potential as an osmoprotectant. Orthen and Popp (2000) observed that cyclitols, including pinitol, decreased membrane damage by freezing and thawing to a similar degree to known cryoprotectants such as sucrose and trehalose. The present observation of higher concentrations of pinitol in roots and leaves before cold acclimation suggests that this compound may contribute to the acquisition of freezing tolerance by preparing cellular constituents for upcoming freezing events. Although the accumulation of stachyose and raffinose did not differ in taproots of alfalfa exposed to different concentrations of CO 2 in the current study, it is interesting to note that pinitol is an intermediate in one of the metabolic pathways leading to the synthesis of these members of the raffinose family oligosaccharides ( Peterbauer and Richter, 2001 ) and could contribute through alternative protective roles to the acquisition of freezing tolerance in alfalfa. The level of carbohydrates was not affected by the rhizobial strain, suggesting that their effect on freezing tolerance is not mediated through a modification of plant carbohydrate composition.
In addition to carbohydrate, free amino acids accumulate in roots of alfalfa during autumn hardening, and are thought to constitute an important source of organic nitrogen for spring regrowth ( Dhont et al ., 2006 ) or to act as osmoprotectants ( Rai, 2002 ). Dhont et al . (2003) noted significant increases in proline, arginine and histidine in winter-hardened alfalfa. Even though a continuous increase in total leaf amino acid concentrations throughout plant growth and cold acclimation was observed, root levels tended to decrease in plants transferred to low temperature. In the present study, arginine was the only amino acid that increased markedly in roots of plants exposed to 2 °C. Dhont et al . (2006) observed a positive association between arginine concentration in autumn and spring yield of alfalfa. The high N to C ratio of arginine makes it a good storage compound for N that could be readily incorporated into other nitrogenous compounds for active regrowth ( Sagisaka, 1974 ). Proline commonly accumulates in plants in response to environmental stresses, and has been related to cold tolerance in perennial weeds ( Cyr et al ., 1990 ), Arabidopsis ( Nanjo et al ., 1999 ) as well as alfalfa ( McKenzie et al ., 1988 ). Even though conclusive evidence on a role for proline in cold tolerance of alfalfa cannot be derived from the present study, the higher accumulation of proline that occurred in leaves under ambient CO 2 is noteworthy and could potentially be associated with the acquisition of the superior freezing tolerance of plants exposed to this treatment as compared with high CO 2 . Interestingly, Bertrand and Paquin (1991) also noted that the accumulation of proline preceded the acquisition of freezing tolerance during cold acclimation of alfalfa. In contrast to the effect found with carbohydrate, there was a rhizobial strain effect on the accumulation of the amino acid tyrosine. The higher accumulation of tyrosine in roots of alfalfa inoculated with A2 as compared with NRG34 could also contribute to the superior freezing tolerance of these plants. Tyrosine is a precursor of phenolic compounds that could act as antioxidants with mechanisms involving both free radical scavenging and metal chelation ( Rice-Evans et al ., 1997 ). The activation of antioxidant mechanisms during plant cold acclimation and their positive contribution to winter survival of alfalfa have been demonstrated ( McKersie et al ., 2000 ).
Many COR genes have been isolated and shown to be strongly upregulated in alfalfa ( Castonguay et al ., 2006 ). In spite of correlative observations between the expression of COR genes and variations in either freezing tolerance ( Mohapatra et al ., 1989 ; Monroy et al ., 1993 ) or winter survival ( Cunningham et al ., 2001 ), no causal relationship between the levels of COR transcripts and freezing tolerance of alfalfa has thus far been established. In that perspective, it is not entirely unexpected that the expression of all the COR genes tested was not significantly higher in cold-tolerant alfalfa grown and acclimated under ambient levels of CO 2 than in those exposed to elevated CO 2 . However, transcripts of the COR gene msa CIE encoding a protein homologous to GAPDH in alfalfa ( Castonguay et al ., 1997 ) accumulated to significantly higher levels in plants under high CO 2 . Velasco (1994) observed an increase in mRNA levels and enzyme activity of a cytosolic GAPDH in response to drought stress and abscisic acid (ABA) treatment, and concluded that enhanced glycolysis is a typical response of plants coping with environmental stress. In addition to its upregulation by low temperature, the increased expression of msa CIE that was observed in plants under elevated CO 2 could be reflective of higher metabolic rates at low temperature attributable to the availability of more substrates for glycolysis.
In response to the rhizobial strain, three COR genes ( msa CIB, msa CIC and msa CID) showed significantly higher cold-induced accumulation of their transcripts in roots of the more cold-tolerant alfalfa inoculated with strain A2 than in the less tolerant plants inoculated with strain NRG34. The superior accumulation of msa CID transcripts encoding a homologue of the pathogenesis-related PR-10 protein in plants inoculated with strain A2 could be directly linked with the higher level of freezing tolerance of these plants. PR proteins with antifreeze properties have been shown to accumulate in winter rye during cold acclimation, allowing winter cereals to be more resistant to both fungal diseases and freezing stress ( Griffith et al ., 1997 ). Differential accumulation of msa CID could alternatively be due to differences in a cell attachment motif that has been characterized in lupin PR-10 ( Bantignies et al ., 2000 ). This sequence, also conserved in alfalfa ( Truesdell and Dickman, 1997 ), was found in peas to mediate the first step in attachment of rhizobium cells to plant root hair tips ( Swart et al ., 1994 ) and could participate in the determination of the specificity of rhizobial strains and, hence, the observed differential levels of expression. The levels of transcripts of msa CIB (homologous to cas 15; Monroy et al ., 1993 ) encoding a nuclear-targeted protein were also higher in cold-tolerant alfalfa inoculated with strain A2. Monroy et al . (1993) observed previously that the level of cas 15 transcripts correlates with the degree of freezing tolerance developed during cold acclimation of alfalfa. Though of unknown function, the encoded protein shares common features (hydrophilic, simple amino acid composition, repeated amino acid motifs) with Arabidopsis COR15a , wheat wcs 120 and barley HVA1 (for a review, see Thomashow, 1999) and may play a role in stabilizing membranes against freezing damage. Primary structure analysis of the putative bipartite protein encoded by msa CIC reveals distinct proline-rich and hydrophobic domains that also suggest a possible association with membranes ( Castonguay et al ., 1994 ). Such an association would be consistent with earlier results by Mohapatra et al . (1988) which showed the occurrence of significant changes in membrane protein composition at low temperature. Changes in the expression of these COR genes, linked to the level of plant freezing tolerance, hint at a major role for the encoded polypeptides for low-temperature survival. How rhizobium modulates variations in COR gene expression and the adaptive value of these effects are questions that certainly deserve further investigation. The effect of strains on differential accumulation of transcripts could occur at the plant transcriptional or post-transcriptional level. The transcriptional activation of alfalfa genes has been observed during the earliest stages of rhizobium–legume interaction ( Journet et al ., 1994 ).
The superior freezing tolerance of plants inoculated with strain A2 under both CO 2 conditions as compared with strain NRG34 differs from previous observations by Prévost et al . (2003) of superior regrowth potential of overwintering alfalfa inoculated with strain NRG34. Thus, the better adaptation for nodulation and N fixation at low temperature of strain NRG34 does not show benefits for freezing tolerance of alfalfa. This is indicative that strains of rhizobium interact with many traits that can affect spring regrowth of alfalfa, including cold tolerance, organic reserve accumulation, autumn dormancy and disease resistance. For instance, we observed that transcripts encoding the alfalfa VSPs chitinase and amylase were more abundant in unacclimated alfalfa inoculated with strain NRG34 than those inoculated with strain A2. Root VSPs are preferentially mobilized during alfalfa shoot growth in the spring or its regrowth after defoliation in the summer ( Dhont et al ., 2006 ). Lower VSP transcript levels in acclimated plants confirm the report by Noquet et al . (2001) of a decline in alfalfa VSP-encoding transcripts at low temperature. It is noteworthy, however, that low temperature repression of VSP was significantly less pronounced for the chitinase than for the β-amylase, especially in the case of the more cold-tolerant plants inoculated with strain A2. We observed the most intense repression of chitinase in the least cold-tolerant plants that had been inoculated with strain NRG34 and cold acclimated under high CO 2 . Although the main function of VSP is the accumulation of N reserves during autumn and winter to support spring regrowth, there are indications that they may, in some cases, also confer protection against abiotic and biotic stresses (for a review, see Avice et al. , 2003 ). Chitinase has been shown to have dual roles: storage and PR activity ( Meuriot et al ., 2004 ), as well as properties typical of antifreeze proteins ( Yeh et al ., 2000 ).
Elevated atmospheric CO 2 has an impact on multiple aspects of alfalfa physiology with potential consequences on long-term persistence through a reduced freezing tolerance. The delay in cold acclimation could be due to the higher metabolic rates observed under high CO 2 . It is possible to select or identify rhizobial strains to improve alfalfa performance under high CO 2 . Freezing tolerance as well as the expression of key overwintering genes of alfalfa can be altered by the strain of rhizobium. Taken together, the results of gene expression suggest that the acquisition of cold tolerance could be delayed in alfalfa inoculated with strain NRG34, as shown by a lower level of COR genes and enhanced accumulation of VSP transcripts. The study of the combined effects of rhizobial strain and high CO 2 is a promising research avenue to develop alternatives to lower the dependence on N fertilizer in agriculture and to mitigate the impact of global change.
The technical assistance of Lucette Chouinard, Pierre Lechasseur, Josée Bourassa and Carole Gauvin from Agriculture and Agri-Food Canada (Québec) and of Yves Dubuc from Natural Resources Canada (Quebec) is gratefully acknowledged.