Introduction

Freezing tolerance is an extremely complex trait involving many physiological and morphological processes and is usually induced or increased by exposure to non-lethal low temperatures in cold hardy plants (cold acclimation). To understand the mechanisms, alterations during cold acclimation have been extensively studied at physiological, cellular and molecular levels (Thomashow 1999). Many of the processes and molecular mechanisms, however, still remained ambiguous. One instance is the role of abscisic acid (ABA).

Physiological studies have shown that endogenous ABA levels increase in plants during cold acclimation (Daie and Campbell 1981; Chen et al. 1983; Lalk and Dörffling 1985; Lång et al. 1994), which allowed to hypothesize that ABA may work as a second messenger in cold acclimation process. Exogenous application of ABA increased freezing tolerance in some plants (Lång et al. 1989; Mäntylä et al. 1995; Veisz et al. 1996) and suspension or callus cultures (Chen and Gusta 1983; Keith and McKersie 1986; Orr et al. 1985; Ishikawa et al. 1990; Galiba et al. 1993; Dallaire et al. 1994).

In accordance with this hypothesis, ABA-deficient mutant (aba1) and ABA-insensitive mutant (abi1) of Arabidopsis plants following exposure to cold-acclimating conditions were found to be less cold hardy compared to wild type plants (Heino et al. 1990; Gilmour and Thomashow 1991; Mäntylä et al. 1995). But the results have to be interpreted with caution as these mutants have much less vigor than wild-type plants, which may result in lower capability to cold-acclimate (Thomashow 1999). Comparison of COR gene expression in these mutants and wild-type plants revealed that some COR genes were highly responsive to exogenous ABA but their expression by low temperature was not necessarily mediated by ABA (Gilmour and Thomashow 1991; Nordin et al. 1991). More recently, molecular analyses of low temperature-responsive genes have demonstrated that there are ABA-dependent and ABA-independent transcriptional pathways (Yamaguchi-Shinozaki and Shinozaki 1994) and even cross-talks between these pathways (Ishitani et al. 1997). The role of ABA in activation of low temperature responses is considered to be minor than it was thought (Thomashow 1999). Yet, question still remains unanswered as to how simple exogenous application of ABA alone can induce high levels of cold hardiness in some plant systems and how it should be interpreted.

Physiological characterization has revealed that in many systems, ABA best induces freezing tolerance at an ambient temperature and does not require lengthy exposure to low temperatures and does not necessarily accompany growth cessation. In suspension cultures of bromegrass (Bromus inermis Leyss), which has been extensively studied, high levels of ABA-induced freezing tolerance were best achieved at 25–30 °C and only marginally at 5–15 °C (Ishikawa et al. 1990). Interestingly, cultures incubated with ABA showed high growth rates equivalent to the control cultures at 25–30 °C. The levels of ABA-induced freezing tolerance were linearly correlated with the amount of culture growth attained in the presence of ABA at 5–30 °C. This is contrary to cold acclimation process in cold hardy plants where the growth is usually suppressed during induction of freezing tolerance. When ABA-treated cells were transferred to an ABA-minus medium and incubated at 4 °C, they quickly lost the attained freezing tolerance (Ishikawa et al. 1990). These observations allowed us to consider that freezing tolerance induction by ABA is fairly different from the one by cold. ABA-induced cold hardiness has attracted attention as a novel system where mechanisms directly involved in freezing tolerance could be discriminated from those involved in metabolic adaptation to low temperatures.

In bromegrass cell cultures, freezing tolerance can also be induced by exposure to low temperatures (Ishikawa et al. 1990). Comparison of cold-induced freezing tolerance and ABA-induced freezing tolerance has been considered to provide a unique approach to understand cold hardiness mechanisms. Protein analysis revealed that ABA induced accumulation of specific, unique polypeptides (Robertson et al. 1987, 1988, 1995; Wilen et al. 1996), most of which were not induced or increased by cold stress. From physiological and morphological analyses, Ishikawa et al. (1990) considered that behavior of bromegrass cells during ABA-induced cold-hardening was similar to that of seed formation process. Several genes responsive to ABA (Lee and Chen 1993; Robertson et al. 1994; Wu et al. 2004) and to cold (Nakamura et al. 2008) were isolated from bromegrass. However, these genes represented only a part of gene expression induced by ABA or cold and it remains unknown how they contribute to freezing tolerance induced and whether ABA-induced cold hardiness involves similar mechanisms as low temperature-induced cold hardiness or entirely different mechanisms. To understand these, more global approach on gene expression analysis during cold and ABA-induced cold-hardening is required.

Microarray technology has emerged as a powerful tool that allows simultaneous monitoring of the expression levels of numerous genes and is useful for analyzing genome-wide regulatory networks of a stress response. Arabidopsis microarrays were also applied for gene expression analysis during salt stress in salt cress (Thellungiella halophila), a halophyte closely related to Arabidopsis (Taji et al. 2004). Rice cDNA microarrays were used to analyze mRNA expression in Fe-deficient barley roots (Negishi et al. 2002) while barley cDNA-macroarrays were successfully used to study cold-induced transcriptome in wheat (Kocsy et al. 2010). In heterologous microarray analyses, highly conserved gene sequences between the related probe and target plant species allow the heterologous probes (microarray spots) to hybridize the target mRNAs to identify and analyze the expression of orthologs in the target plant. This has been shown to be a valid approach for cross-species transcriptional profiling of rather distantly related species: a Pinus array for Picea (van Zyl et al. 2002) and Arrabidopsis arrays for poplar and oat (Horvath et al. 2003). Since bromegrass and rice are phylogenically related in the recent molecular analysis of Poaceae (Kellogg, 2001), rice microarrays seemed useful for analyzing gene expressions in bromegrass.

The objective of this study is to roughly characterize and compare transcriptome profiles involved in two physiologically different systems of freezing tolerance induction, stimulated by cold and ABA using bromegrass cell cultures, which may provide a unique approach. For this purpose, the use of rice microarrays for bromegrass mRNA (heterologous system) was likely a reasonable compromise. Our focus was on long-term expressed genes that may be involved in freezing tolerance mechanisms rather than transiently expressed genes. Therefore microarray analyses were done with cells exposed to cold or ABA for 7 days. We attempted to identify up-regulated genes that were cold-specific, ABA-specific, or common to both treatments and characterize these systems. Furthermore, several cold- and ABA-responsive genes were isolated from bromegrass suspension cells.

Materials and methods

Plant material and culture conditions

A non-embryogenic cell suspension culture of smooth bromegrass (Bromus inermis Leyss cv. Manchar) was used in this work and was subcultured biweekly in ER medium (pH 5.8, 0.5 mg/mL 2,4-D) at 25 °C as described previously (Ishikawa et al. 1990; Nakamura and Ishikawa 2006).

Cold treatment was conducted by exposure of 3 day-old cultures (initiated from 1.5 g of cell inoculum in 50 mL of ER medium and grown at 25 °C) to 15 °C for a day and then to 4 °C for 6 days. Bromegrass cells (1.5 g) were inoculated in 50 mL of ER medium in the presence (ABA treatment) and absence (control) of 50 μM ABA for 7 days at 25 °C.

Freezing tolerance determination

Following each treatment, the cells were aseptically harvested by filtering on an 80 μm mesh and washed with 250 mL of sterile water to remove extracellular sugars that may affect freeze survival. About 0.26 g fresh weight of cells (control, ABA- or cold-treated) were placed in plastic centrifuge tubes (15 mL, IWAKI) with 0.2 mL of sterile water. After equilibration at −3 °C for 30 min, freezing was initiated by touching the tubes with a liquid nitrogen-cooled steel rod. The cells were held at −3 °C overnight, then cooled at 2 °C/h to −12 °C and then 5 °C/h to −35 °C. The cells were removed at designated temperatures and thawed at 4 °C before being incubated in 15 mL of ER medium at 25 °C for 7 days to assess the viability by regrowth capacity. After 7 days of reculture, the cells were harvested and washed with distilled water prior to determination of dry weight by oven-drying at 70 °C for 2 days. Viability of cells was calculated from the regrowth data (average of three determinations) and freezing tolerance was expressed as LT50, the temperature at which 50 % of the cells were killed as described previously (Ishikawa et al. 1995).

Isolation of mRNA

Total RNA extraction from cells grown under various conditions was described previously (Nakamura et al. 1997). Poly(A)+ RNA was purified using oligo(dT) latex beads (Oligotex dT30 Super, Takara) according to the method specified by the supplier.

Microarray analysis

A DNA chip with 8987 randomly selected ESTs prepared by the microarray project in Japan (Kishimoto et al. 2002), with a system developed by Amersham Bioscience, was used for microarray analysis. The identity and the accession number of each of the EST clones are listed at http://microarray.rice.dna.affrc.go.jp/. Preparation of DNA chips, and microarray analysis methods (experimental procedures and data analysis) were previously described by Yazaki et al. (2000, 2003). Briefly, 8987 cDNA clones from japonica rice (cv. Nipponbare) were printed on an aluminum-coated, DMSO-optimized glass slide in duplicate. Target mRNAs purified from bromegrass samples were labeled by reverse transcription with Cy5 and hybridized with the microarrays using Atlas Glass Fluorescent Labeling Kit (Clontech). Images of the microarrays scanned were analyzed using Fujifilm ArrayGauge software. Following an appropriate background subtraction and global normalization, the signal intensities of different samples (cold or ABA-treated vs. untreated control) were compared. Data presented are the mean of three determinations, each of which was the average of duplicated spots. Genes that were up-regulated more than twofold (the means with SD) after 7 days of cold or ABA treatments compared to the untreated control were shown in Tables 1, 2, 3 and 4.

Table 1 Numbers of the functional categories of transcripts in bromegrass cultured cells increased by cold and/or ABA treatments
Full size table
Table 2 Cold-inducible genes identified by microarray analyses
Full size table
Table 3 ABA-inducible genes identified by microarray analyses
Full size table
Table 4 Cold- and ABA-inducible genes identified by microarray analyses
Full size table

Preparation of a cDNA library and screening procedure

Complementary DNA libraries of cold-stressed and ABA-treated bromegrass cells were constructed using Lambda ZAP II XR Library Construction kit (Stratagene) according to the method specified by the supplier. For screening the cDNA library, rice cDNAs listed in Table 5 were obtained from the Rice Genome Research Program (RGP: http://rgp.dna.affrc.go.jp/) and used as probes.

Table 5 List of genes used for Northern blot analyses as probes
Full size table

Northern blot analysis

Northern blot analysis was performed as described previously (Nakamura et al. 1997). Rice full-length cDNAs and gene specific regions of bromegrass cDNAs were used as probes for Northern blot analysis.

Results

Effects of low temperature and ABA treatments on the freezing tolerance of bromegrass cells

Freezing tolerance was measured for bromegrass cells grown at low temperature (15 °C for 1 day and 4 °C for 6 days without ABA) and for cells grown at 25 °C for 7 days with 50 μM ABA or without ABA (control). Fig. 1 shows the viability of cells frozen from −3 to −35 °C as determined by the regrowth method. LT50 for control, cold and ABA-treated cells were −3.9, −8.1 and −12.3 °C, respectively. Low temperature and ABA treatments increased freezing tolerance by more than 4 and 8 °C, respectively, as compared to the control.

Fig. 1
figure 1

Effect of freezing temperatures on the viability of untreated control (circle), cold-acclimated (triangle) and ABA-treated (square) bromegrass cells. Cold treatment was conducted by incubating 3 day-old bromegrass cell cultures at 15 °C for a day and then at 4 °C for 6 days. Bromegrass cell cultures were incubated in culture medium in the presence or absence of 50 μM ABA for 7 days at 25 °C. Data were the mean ± SE of three determinations (SE smaller than the symbols were not shown)

Full size image

Although the experimental conditions were not exactly the same (e.g., the size of cell culture inoculum), essentially similar levels of increase in freezing tolerance were obtained with cold- and ABA-treated bromegrass cells in our previous studies (Ishikawa et al. 1990, 1995, 2006).

Microarray analyses

To reveal global gene expression changes during these increases in freezing tolerance, we isolated mRNA from low temperature-treated (7 days), ABA-treated (25 °C for 7 days) and untreated (25 °C for 7 days) bromegrass cells and used for analysis with rice cDNA microarrays containing 8987 cDNAs of rice EST. In preliminary experiments to check the hybridization efficiency of bromegrass mRNA, we observed only marginal differences in the signal intensities between bromegrass mRNA and rice mRNA following hybridization with the rice cDNA microarrays (data not shown). This allowed us to conclude that rice microarrays were applicable to bromegrass gene expression analysis. Yet, down-regulated genes might not possibly be properly evaluated due to lower hybridized signals. These genes might include not only down-regulated genes but also orthologous genes with low homology. These considerations allowed us to agree with the idea that expression data obtained with heterologous microarray analysis can be used for identifying up-regulation of genes (Taji et al. 2004). Therefore, we focused on only up-regulated genes. Table 1 summarizes functional categories of the genes up-regulated more than twofold by 7 days of cold and ABA treatments compared to the untreated control. The functions of 79 % (237/300) of the cold-responsive genes were known, whereas those of 61.6 % (295/479) were known for ABA-responsive genes.

Cold-inducible genes

The genes up-regulated more than twofold by 7 days of cold acclimation as compared to the untreated control are listed in Tables 2 and 4. A total of 300 genes were identified as long-term or later-stage cold-inducible genes, out of which 149 genes were also increased by 7 days of ABA treatment. Homologues of several genes in Tables 2 and 4 were reported to be cold-inducible in plants, such as zinc finger protein, phospholipase, calcium-dependent protein kinase, chitinase, heat shock protein, alternative oxidase, superoxide dismutase, elongation factor and ribosomal protein (Kim et al. 2001; Thomashow 1999; Yeh et al. 2000; Ito et al. 1997; Seki et al. 2002b). These results show that the rice cDNA microarray system functioned properly to detect cold-inducible genes in bromegrass.

ABA-inducible genes

Tables 3 and 4 show the genes that were up-regulated by ABA treatment. The transcripts of 479 genes increased after ABA treatment, with inducibilities of more than twofold compared to those of control. The lists (Tables 3, 4) include many genes whose homologues were reported to be ABA-inducible (Skriver and Mundy 1990; Giraudat et al. 1994; Busk and Pages 1998; Seki et al. 2002a). Several genes have been previously reported as responsive to ABA in bromegrass (Lee and Chen 1993). Some of these genes such as germin, aldose reductase and embryo globulin were also shown to be ABA responsive in our microarray results.

Expression analyses of selected genes

To evaluate the validity of the microarray data, the expression patterns of 8 selected clones (listed in Table 5) were further evaluated by Northern blot analysis using heterologous probes. In Fig. 2, the results of microarray analyses were compared with those of Northern blot analyses. For each of 8 clones, the expression patterns under cold and ABA treatments were fairly consistent between the two methods, showing that microarray and Northern blot analyses gave similar quantitative results (Fig. 2). Through this process, we cloned 6 bromegrass genes, three of which were cold-responsive (BiSDB1, BiGP1, BiCHT1) and three of which were ABA-responsive (BiPOR1, BiLEA1, BiWSI1) (Table 5).

Fig. 2
figure 2

Comparison between microarray and Northern blot analyses showing the expression patterns of 8 selected genes. The graphs show the relative amounts of mRNA observed in the microarray analysis. In Northern blot analyses, each lane was loaded with 15 μg of total RNA isolated from bromegrass cells treated with cold (L) and ABA (A), and untreated bromegrass cells (C). Microarray results show the expression of element No. 187 (a), 505 (b), 1236 (c), 1818 (d), 1952 (e), 2379 (f), 2388 (g) and 6993 (h) listed in Table 5. Rice full-length cDNAs corresponding to the element numbers were used as probes for Northern blot analyses

Full size image

Expression of cold-responsive genes during the periods of cold acclimation

To investigate expression patterns of the cold-responsive genes, mRNA accumulation of 3 selected clones during the course of cold acclimation was examined. Northern blot analyses of total RNA extracted from cells cold-acclimated for 0, 1, 1.5, 2, 6 and 13 days were performed using bromegrass cDNAs as probes (Fig. 3). BiSDB1 mRNA, encoding a putative SAR DNA binding protein, accumulated after 1 day of exposure to 15 °C and continued to increase with 5 days of treatment at 4 °C, then slightly decreased after 13 days. RNA of BiGP1, encoding a putative glycoprotein, was detected in the control cells and was expressed at elevated levels during cold acclimation. BiCHT1 transcript, encoding a putative chitinase, accumulated within 1 day of cold treatment and continued to increase during the course of cold acclimation. This cold-inducible bromegrass chitinase had a high homology (89 %) at the protein level with rye chitinase (CHT9), which has been reported to be an antifreeze protein (Yeh et al. 2000). The product of BiCHT1 had chitinase activity but no antifreeze activity under any conditions tested (Nakamura et al. 2008).

Fig. 3
figure 3

Time course of mRNA accumulation during cold acclimation as detected by RNA blot analysis. Cold treatment was conducted by incubating 3 day-old bromegrass cell cultures at 15 °C for a day and then at 4 °C for up to 12 days. Gene specific regions of bromegrass cDNAs, listed in Table 5, were used as probes. Ribosomal RNA was visualized by ethidium bromide staining

Full size image

Discussion

The development of microarray technology for monitoring the expression of thousands of genes at a time has enabled global analysis of the responses of gene expression to abiotic stresses and to plant hormones. This has been successfully applied to analyses of Arabidopsis responding to cold stress (Seki et al. 2001, 2002b; Fowler and Thomashow 2002; Hannah et al. 2005; Oono et al. 2006; Kilian et al. 2007) and ABA (Seki et al. 2002a; Takahashi et al. 2004). More recently, it has been applied to cold acclimation process of agriculturally important grasses such as barley and wheat (Svensson et al. 2006; Monroy et al. 2007; Winfield et al. 2010; Laudencia-Chingcuanco et al. 2011).

In this study, we analyzed gene expressions during induction of cold hardiness in bromegrass cell cultures exposed to cold and ABA using rice cDNA microarrays (heterologous system). In preliminary experiments, we found that bromegrass mRNA had a hybridization efficiency with the rice cDNA microarrays almost comparable to rice mRNA. This may arise from the tendency that genes in cereal crops are highly conserved at the DNA level (Devos and Gale 2000). Bromegrass has a genome 30 times larger than rice (Tuna et al. 2001) and probably contains more genes (perhaps with more gene families) than rice in the genome. In this heterologous analysis, our rice microarrays may not necessarily distinguish cross hybridization of gene families of bromegrass. Yet, individual hybridization data of cold or ABA treatment were always presented as the ratio compared to hybridization of mRNA from control cells. Each data set may contain possible cross hybridization of gene families. Thus, our analyses likely represent the overall responses (compared to the control) of target genes plus their gene families (if there are any) or approximate estimation of gene expression tendencies rather than collection of pinpoint expression of individual genes. Eventually, northern blot analyses of 8 selected genes (listed in Table 5) using heterologous probes revealed that the expression patterns of these genes under cold and ABA treatments were quantitatively consistent between the microarray and Northern blot analyses (Fig. 2). This indicates that our microarray analyses worked properly. The up-regulation of three genes was further confirmed by Northern blots using gene specific sequences of bromegrass cDNA (orthologous probes) (Fig. 3). This also indicated the highly conserved sequences in the genes of the grass family (Table 5). These things allowed us to consider that rice microarrays were useful for approximate estimation of gene expression tendencies in bromegrass. Similar successful heterologous microarray analyses have been reported between distantly related species (van Zyl et al. 2002; Horvath et al. 2003) and between species differing in the genome size: barley and rice (10 times larger) or wheat and barley (4 times larger) (Negishi et al. 2002; Schweizer 2008: Kocsy et al. 2010). And yet, for the reasons detailed in the microarray section of the Results, we only focused on up-regulated genes as heterologous microarrays were used (Taji et al. 2004).

Many of the genes detected to be up-regulated by the rice microarrays in response to ABA or cold (Tables 2, 3, 4) have been previously reported to be ABA or cold inducible in Arabidopsis (e.g., Fowler and Thomashow 2002; Seki et al. 2002a, b; Takahashi et al. 2004; Oono et al. 2006) as briefly summarized in the Results and detailed below. Commonalities of our results with Arabidopsis studies also imply that rice cDNA microarray is useful for roughly characterizing cold hardiness induction systems by two different stimuli in bromegrass, which is the purpose of this study. Yet, a problem in this approach may be that rice is not a freezing-tolerant plant and the rice microarray may not contain freezing tolerance related genes. Such cautions have to be paid in interpreting the results that bromegrass-specific genes which have no homologous genes in rice or on the used microarray and genes expressed at low levels are not readily detectable in this system. With these limitations (discussed in the present and previous paragraph) in mind, we compared gene expression profiles of cold and ABA-treated bromegrass cell cultures to characterize the two cold hardiness induction systems.

Cold-inducible genes

In this study where the gene expression profiles of bromegrass cells that were cold-hardened for 7 days were analyzed, 300 genes were identified as long-term or later-stage cold-inducible (Tables 2, 4) and whose products may confer freezing tolerance (LT50: −8.1 °C as compared to −3.9 °C of control cells in Fig. 1). Several genes were common when the expression profiles of cold-hardened bromegrass were compared with those of cold-hardened Arabidopsis (Fowler and Thomashow 2002) which had LT50 of −6 to −8 °C after 7 days exposure to 4 °C compared to −4.5 °C of non-hardened control (Gilmour et al. 2000). These genes included pyruvate decarboxylase, zinc finger protein, xyloglucan endotransglycosylase, peroxidase, myb protein, extensin, SNF1 like protein kinase, cysteine proteinase, fibrillarin and chitinase. Some of other genes such as SAR DNA binding protein, nucleoside diphosphate kinase and histone H2A were newly identified as cold-inducible genes in our microarray results (Tables 2, 4).

One newly found cold-inducible gene is BiSDB1, encoding a putative SAR DNA binding protein. Scaffold attachment regions (SARs) are AT-rich sequences and are known to bind specifically to components of the nuclear scaffold (Hall et al. 1991). The function of SARs is to enhance transgene expression and normalize it by protection against surrounding chromatin, called position effects (Breyne et al. 1992). In plants, a typical SAR DNA binding protein was purified and whose cDNAs (MARBP-1 and MARBP-2) were isolated from pea (Hatton and Gray 1999). MARBP-1/MARBP-2 genes were expressed in all the tissues, and the function of these products was predicted as housekeeping. Expression of BiSDB1 was up-regulated after 1 day of exposure to 15 °C and gradually increased throughout the cold acclimation period (Fig. 3). This implies that BiSDB1 may have a housekeeping role in the normalization of gene expression under cold conditions and/or that BiSDB1 may possibly regulate cold-inducible genes at the chromatin level.

When bromegrass suspension cells were grown at 4 °C for 0–3 weeks, they showed slow steady growth in the second and third weeks: the fresh weight of cultured cells increased to 118 and 136 % of the initial cell weight, respectively. In contrast, they showed only marginal culture growth (104 % of the initial weight) in the first 7 days at 4 °C (Shinkawa et al. manuscript in preparation), which was the target condition of this study. During this period, bromegrass cells most likely spent most of the energy to adjust their metabolism to cold conditions (a part of housekeeping) and to scrap/rebuild many enzymes and protein machinery, which may have resulted in a limited culture growth.

In agreement with this, numerous genes involved proteolysis, RNA stabilization and protein synthesis were up-regulated after 7 days at 4 °C as compared to untreated control and ABA treatment (Table 2). For instance, expression of protein synthesis related genes such as elongation factors, translation initiation factors, and ribosomal proteins, were increased in cold-acclimated bromegrass (Tables 2, 4). Protein synthesis is necessary for cold response and the accurate translation machinery is an important factor for properly processing cold acclimation. In Arabidopsis, the LOS1 gene encodes translation elongation factor 2 and the los1-1 mutant is impaired in its ability to cold-acclimate and defective in protein synthesis in the cold (Guo et al. 2002). Three ribosomal protein genes induced by low temperatures in soybean show long-term up-regulated expressions (Kim et al. 2004). In yeast, ribosomal proteins stimulate cell growth (Warner 1999). Thus ribosomal proteins expressed at low temperatures may be required for bromegrass cells to grow under cold stress.

Cold up-regulated genes involved in proteolysis included proteasome related genes and various proteases whilst those involved in RNA function included various t-RNA synthetases, RNA binding proteins, RNA helicase, etc. (Tables 2, 4). Some RNA binding proteins are known to be cold responsive and work as RNA chaperones whilst RNA helicases are considered to be involved in the removal of inhibitory secondary structures of mRNA formed at low temperatures and may contribute to maintaining protein synthesis and increased survival under cold conditions (Kwak et al. 2011; Vashisht and Tuteja 2006).

Increased expression of some genes by cold such as plasma membrane ATPase and alternate oxidase (AOX) were consistent with literatures on physiological and molecular studies on alterations during cold acclimation (e.g., Ishikawa and Yoshida 1985; Umbach et al. 2005).

ABA-inducible genes

ABA-treated cells acquired higher freezing tolerance than cold-acclimated cells in 7 days of treatments (Fig. 1). In bromegrass cultured cells, ABA treatment for 7 days induced many stress-related genes such as germin, LEA protein, water stress protein, peroxiredoxin, desaturase and superoxide dismutase (Table 3). These genes were also induced by salt and drought in Arabidopsis (Seki et al. 2001, 2002b; Takahashi et al. 2004) and rice (Kawasaki et al. 2001; Rabbani et al. 2003), but not induced by cold in bromegrass (Table 2). ABA-inducible genes may increase freezing tolerance as a part of ABA-mediated cross-adaptation to various stresses (Ishikawa et al. 1995). Microarray results indicated that freezing tolerance induced by cold or ABA accompanied up-regulation of considerably different sets of genes, and that cold-specific or ABA-specific genes may represent different freezing tolerance mechanisms involved.

ABA has been known to play important roles in seed formation and maturation. In agreement with this, many genes related to storage proteins (e.g., globulin, lectin, allergen), lipid body protein (e.g., oleosin), cell wall proteins (e.g., extension, expansin) were up-regulated in response to 7 days of ABA treatment (Tables 3, 4). The list of ABA-up-regulated genes indicated intensified stress and defense related responses: antioxidation (e.g., peroxidase, peroxiredoxin, superoxide dismutase), biosynthesis of proteins responsive to pathogen, heat, water stress, salt, wound, heavy metals and UV (e.g., heat shock proteins, various types of LEA, embryo specific proteins, water stress protein, osr40c1, germin, chitinase, etc.) and biosynthesis of secondary metabolites such as jasmonic acid (allene oxide cyclase), flavonoid (e.g., phenylalanine ammonia-lyase, 4-coumarate-CoA ligase), lignin (e.g., cinnamyl alcohol dehydrogenase, cinnamoyl-CoA reductase) and phytoalexin (e.g., hydroxyanthranilate hydroxycinnamoyltransferase). Activation of glucose, sucrose and starch biosynthesis was also implied (e.g., fructose biphosphate aldolase, phosphoglycerate kinase, malate dehydrogenase, glucose-6-phosphate isomerase and sucrose synthase).

Since ABA treatment of bromegrass cultures was done at 25 °C, adjustment of metabolism and protein machinery to low temperatures is not required (thus circumventing the process of proteolysis/synthesis/RNA maintenance) and functional stress-related genes may have been expressed rather directly and constitutively, which may partly attribute to the differences in transcriptome profiles between ABA-treated cells and cold-treated cells.

Genes up-regulated by cold and ABA treatments

It is noted that 149 clones were induced by both cold and ABA treatments (Table 4). These genes may possibly be more important for freezing tolerance. Functions of these genes include regulation of transcription, stress and defense related proteins (e.g., heat shock proteins, chitinase, thaumatin-like protein, glucanase, ankyrin repeat containing protein), antioxidants (e.g., various peroxidases), cell wall biogenesis and alteration (e.g., monosaccharide transporter, xyloglucan endotransglycosylase, remorin, extensin, cinnamoyl-CoA reductase), lipid synthesis (e.g., erg1, choline phosphate cytidyltransferase, acyl-CoA synthase), etc. Some transcription factors detected, such as zinc finger protein, may possibly regulate the expressions of downstream genes induced by cold- and ABA. Further understanding of these interactions will become feasible with the availability of regulatory sequences of the target genes. It is likely that common regulatory networks and/or signaling intermediates govern the expression of the genes up-regulated by both treatments. Several of the downstream responses shared by cold and ABA may play important roles in freezing tolerance mechanisms.

One interesting observation is that key enzymes in ethylene biosynthesis (1-aminocyclopropane-1-carboxylate oxidase and 1-aminocyclopropane-1-carboxylate synthase) and ethylene response factor were up-regulated by both cold and ABA in coordination with many disease resistance related genes and a key enzyme in polyamine biosynthesis (S-adenosylmethionine decarboxylase).

Another surmise is that some physiological or metabolic processes are shared in cold- and ABA-treated cell cultures, which results in the expression of common genes, which means that the results in Table 4 have to be interpreted with caution. For instance, phytosulfokine-alpha, a sulfated peptide growth factor, is known to be involved in cell proliferation of cultured rice cells (Yang et al. 1999; Lorbiecke and Sauter 2002). A bromegrass homologue of OsPSK4 that encodes a precursor of phytosulfokine-alpha was up-regulated both by cold and ABA (Table 4). Interestingly, bromegrass cell cultures showed cell growth in the cold or in the presence of ABA (Ishikawa et al. 1990; Shinkawa et al. manuscript in preparation).

Taken together, bromegrass cell cultures provided a unique system for studying cold hardiness mechanism, where freezing tolerance were independently induced by two different stimuli, cold and exogenous ABA at 25 °C. Microarray analyses of long-term up-regulated genes during freezing tolerance induction by these stimuli revealed that cold stress was characterized by triggering of numerous genes involved in protein degradation/synthesis and RNA maintenance in addition to cold stress-related genes whilst ABA treatment was characterized by induction of numerous genes involved in seed formation and functional genes related to biotic and abiotic stresses. The results were in good agreement with physiological and protein studies of the two systems: proteins induced by cold and ABA treatments were fairly different with more newly induced bands by ABA whilst ABA-treated cells had thicker cell walls, denser cytoplasm with more lipid and protein bodies and starch accumulation in plastids compared to cold-treated cells (Robertson et al. 1987, 1988; Ishikawa et al. 1990). Physiologically, ABA conferred bromegrass cells tolerance to freezing, heat, salt and osmotic stresses simultaneously (cross-adaptation) (Ishikawa et al. 1995). The results may not be conclusive but these two systems had fairly different transcriptome profiles and likely involve different mechanisms of cold hardiness induction. The genes commonly induced or increased by cold and ABA are likely more directly involved in freezing tolerance mechanism and especially transcription factors are of interest for further studies.

More recently, we have found that exogenous ABA can induce freezing tolerance in suspension cells of rice, which is a chilling sensitive plant (Shinkawa et al. manuscript in preparation). Proteomic studies revealed that ABA-treated rice cells had a protein profile similar to that of seed embryos, which shared a common implication with the present study.

The data obtained with heterologous microarrays have to be interpreted with cautions as detailed in the first part of Discussion. A bromegrass gene ROB5, encoding a highly ABA-responsive LEA protein (42 kD) (Robertson et al. 1994), was not detected as ABA-inducible by the rice microarray. This gene has been claimed to confer various stress tolerance to transformed crops without affecting the yield (Robertson et al. 2008). Meanwhile, bromegrass COR (cold-regulated) genes (Thomashow 1999) were not detected as cold-inducible by the rice microarray. These cold hardiness genes most likely specific to bromegrass were not detectable with the rice microarrays. Interestingly, bromegrass transcriptome profiles obtained using such rice microarrays that may lack bromegrass specific freezing tolerance genes still had numerous commonalities with those of Arabidopsis treated with cold or ABA. Moreover, the rice microarray, in spite of being an approximate probe, seems to have successfully characterized two cold hardiness induction systems of bromegrass cells without having bias on well known freezing tolerance genes. This is most likely due to the highly conserved sequences in the genes of the grass family.

In this study, several genes were newly found to be cold-inducible. The expression of three cold-inducible genes (BiSDB1, BiGP1, BiCHT1) was confirmed by Northern blots using the corresponding bromegrass cDNA. We also isolated three ABA-inducible genes (BiPOR1, BiLEA1, BiWSI1) from bromegrass (Table 5). These genes were introduced into rice plants by transformation to see any changes in stress tolerance and performance.

The present study may help understand the nature of ABA-induced freezing tolerance. Yet numerous questions still remain to be unraveled. Why can freezing tolerance induction by ABA be achieved in only limited number of systems? In Arabidopsis suspension cells, for example, ABA treatment only marginally induces freezing tolerance. Why can ABA induce higher levels of cold hardiness in cultured cells than cold treatment (Ishikawa et al. 1990), sometimes irrespective of the frost sensitivity of the examined genotypes (Galiba et al. 1993)? Why can cultured cells achieve only limited levels of freezing tolerance even after prolonged cold treatment compared to the high levels of freezing tolerance acquired in their derived plants after cold acclimation (Ishikawa et al. 1990)?