INTRODUCTION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Embryogenesis is a critical stage of the plant life cycle because it establishes the basic body plan. Little is known about the molecular mechanisms that regulate the process (Harada, 1999). This paucity of information is partially because of the location of the embryos, which are embedded within the maternal tissue and are difficult to dissect. The generation of embryos in culture through somatic embryogenesis has become a model system for investigating factors that affect embryo growth. Somatic embryogenesis provides a large number of embryos at defined stages of development, and allows alterations of the embryonic environment through manipulations of the culture conditions.

White spruce (Picea glauca) is an economically important species in North America, utilized for pulpwood and lumber production (Hosie, 1979). Regeneration of this species via somatic embryogenesis (Hakman and Fowke, 1987; Lu and Thorpe, 1987) has represented a means of propagation and a model system for conducting physiological and biochemical studies (for review, see Stasolla et al., 2002). Generation of white spruce somatic embryos is commonly achieved by transferring embryogenic tissue onto an abscisic acid (ABA)-containing maturation medium (Lu and Thorpe, 1987). Although such embryos may appear "morphologically" mature, they do not perform well during postembryonic growth without the imposition of a drying period. Improvement of embryo quality can be achieved through the imposition of osmotic stress, which is an important factor for directing embryo development and maturation both in vivo and in vitro (Finkelstein and Crouch, 1986; Litz, 1986). In conifers, the combined application of ABA and polyethylene glycol (PEG), a non-plasmolyzing osmoticum, has become a routine method for stimulating embryo maturation (see Attree and Fowke, 1993). The effect of PEG mimics the naturally occurring water stress on seeds during late stages of maturation. Attree et al. (1991) observed a 3-fold increase in the maturation frequency of white spruce embryos after application of PEG. Such embryos closely resemble their zygotic counterparts in low moisture level and ability to tolerate desiccation (Attree et al., 1995). In addition, PEG applications resulted in increased deposition of storage proteins similar in abundance and electrophoretic mobility to those accumulated in zygotic embryos (Misra et al., 1993). After full desiccation, a large percentage of PEG-treated embryos were able to convert into plantlets (Attree et al., 1991).

To contribute to an understanding of the molecular events occurring during embryo maturation in conifers, the steady-state transcript levels of stage-specific embryos matured with ABA or ABA + PEG were analyzed using a spotted cDNA microarray, consisting of a nonredundant set of 2,178 cDNAs from loblolly pine (Pinus taeda). The utility of a pine cDNA array for studies on gene expression in spruce has been documented (van Zyl et al., 2002).

This study provides new information on global changes of gene expression during different stages of embryo development. In addition, because applications of PEG affect somatic embryogenesis by altering the morphology of the embryos produced and their ability to tolerate stress conditions and to accumulate storage products (see Stasolla et al., 2002), the molecular mechanisms regulating these responses will be discussed in detail.

	RESULTS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

White spruce somatic embryogenesis may be divided into five distinct stages of development (Fig. 1). In stage 1, the embryonic tissue was cultured for 7 d on maintenance medium. Early filamentous stage embryos were characterized by a small head of cytoplasmic cells and elongated suspensor cells. After 10 d on ABA-containing maturation medium (stage 2), the embryo proper increased in size. A well-developed shoot and root pole became visible after 20 d in culture (stage 3). After 30 d (stage 4), the embryos developed further, and a ring of cotyledons emerged from the shoot apical region. Fully mature embryos, characterized by expanded cotyledons, were visible at the end of the 40 d of culture (stage 5). Inclusion of PEG into the ABA-containing maturation medium did not alter the time course of developmental events described in Figure 1, but increased the number of fully developed embryos produced. At the optimal PEG concentration (7.5% [w/v]), 435 embryos (g fresh weight tissue¹) were produced, compared with 183 embryos (g fresh weight tissue¹) with ABA alone (control). Higher concentrations of PEG inhibited embryo growth (Table I).

View larger version (56K): [in this window] [in a new window]   Figure 1.   Micrographs of different stages of embryo development. At stage 1 (maintenance medium), early filamentous embryos were observed. After 10 d on ABA-containing medium (stage 2), the embryos increased in size. Formation of a defined shoot and root pole was only observed after 20 d in culture (stage 3). After 30 d (stage 4), a ring of cotyledons emerged from the apical pole of the embryos, which increased in size. Fully developed cotyledonary embryos (stage 5) were observed at the end of the maturation period.

View this table: [in this window] [in a new window]   Table I.   Effect of different concentrations of PEG on cotyledonary embryo production Values are expressed as no. of mature embryos (g fresh wt tissue1). Means ± SE. n = 6. Asterisks indicate treatments that are statistically different (P < 0.01) from control tissue (0% [w/v] PEG).

To analyze global changes in gene expression during the maturation process, the transcript population from stage-specific embryos (stages 1-5, Fig. 1) matured with ABA alone (control) or ABA + PEG (7.5% [w/v]; +PEG) was hybridized against a pine microarray composed of 2,178 cDNAs. All the results presented in this study were obtained from the E1 line. The other cell line (E2) was used to confirm differences in gene expression between control and PEG-treated embryos at stages 2 and 5 of development. A distribution of the 2,178 cDNAs into major functional categories is shown in Figure 2A. Changes in the transcript profile in the sample series were estimated by hybridizing adjacent samples against each other, following a loop experimental design (Kerr and Churchill, 2001). Analysis of transcript profiles revealed that 317 and 462 cDNAs were differentially expressed in PEG-treated embryos between stages 1 and 2 and 4 and 5, respectively. In control embryos, the largest difference in gene expression between neighboring stages was observed between stages 1 and 2. Only 10 cDNAs were differentially expressed between stages 4 and 5 (Fig. 2B). Major differences in transcript abundance between control and PEG-treated embryos occurred at stages 2 and 5 (see Tables I and II and supplemental material available at www.plantphysiol.org). In fully developed embryos (stage 5), more than 400 cDNAs were differentially expressed between control and PEG treatment (Fig. 2C). A list of genes with strongest difference in fold change between mature PEG-treated embryos and control embryos is shown in Table II. All these genes were up-regulated in the presence of PEG.

View larger version (31K): [in this window] [in a new window]   Figure 2.   A, Percentage distribution of the 2,178 genes present on the array into major functional categories. B, Number of cDNAs that are differentially expressed (P value of Student's t test < 0.01) between neighboring stages of embryo development in the presence or absence of PEG. C, Number of cDNAs that are differentially expressed (P value of Student's t test < 0.01) between control and PEG-treated embryos at specific stages of development. For stages of embryo development, refer to Figure 1.

Validation of the results of the microarray experiment was confirmed by RT-PCR studies of five cDNAs (NXSI_125_G03, NXSI_134_G06, NXSI_049_A01, NXSI_137_D09, and NXSI_079_C02), which were differentially expressed between control and PEG-treated cotyledonary embryos. These clones were selected because they exhibited small (<2) and large (>10) fold change differences in expression between treatments. Very similar results (up- or down-regulation) were obtained between the two hybridization techniques for all the selected clones (Table III).

Hierarchical clustering of the transcript levels of the 2,178 cDNAs allowed the identification of genes with similar expression pattern during somatic embryo development (Fig. 3). The number of genes that showed differential expression among all stages of development was determined in both control and PEG-treated embryos (Fig. 3).

View larger version (36K): [in this window] [in a new window]   Figure 3.   Hierarchical cluster analysis of the 2,178 selected genes in developing embryos cultured in the absence (control) or presence (+PEG) of PEG. The color scale on the top left indicates fold change differences in gene expression between each stage of embryo development and the neighboring previous stage. See Figure 1 for a description of embryogenic stages. Examples of major classes of genes with differential expression patterns are diagrammed on the right. Number of genes that are differentially expressed (P value of Student's t test < 0.01) among all stages of development in control and PEG-treated embryos are shown at the bottom.

To determine that the effect of PEG on gene expression was not genotype dependent, the transcript levels of control and PEG-treated embryos of the other cell line (E2) were analyzed at stages 2 and 5, where major differences in gene expression were observed (Figs. 2 and 3). PEG induced similar alterations (up- or down-regulation) for 85% and 93% of genes in both lines (E1 and E2) at stages 2 and 5, respectively.

Establishment of the Embryo Body Plan

A cluster of auxin-induced genes showed a general decline during the maturation period in both control and PEG-treated embryos, especially upon removal of 2,4-dichlorophenoxyacetic acid (2,4-D; stages 1-2). The transcript levels of two of these genes, however, significantly increased in embryos matured with PEG between stage 4 and 5 (Fig. 4). Differences in expression profiles between control and PEG-treated embryos were also observed for two ABA-responsive genes because their transcript levels decreased between stages 1 and 2 in PEG-treated embryos only. Compared with control embryos, one embryo-specific gene (NXNV_160_C12) was up-regulated in PEG-treated embryos at stages 2 and 3 of development. A similar result was also observed for one gene participating in cell division (NXSI_063_G10) at stage 4 (Fig. 4).

View larger version (77K): [in this window] [in a new window]   Figure 4.   Fold change differences in the expression of several genes involved in embryo development. Fold changes are estimated between each stage of embryo development and the neighboring previous stage in the absence (control) and presence (PEG) of PEG. Direct comparisons between treatments at respective stages of embryo development are also shown. Statistically significant (P value of Student's t test < 0.01) expression ratios are shaded in green (down-regulation) or in red (up-regulation). NAM, No apical meristem.

Several cDNAs present on the array were apparent homologs to Arabidopsis genes involved in establishing the pattern of embryo formation. Among the genes required for the formation and maintenance of the shoot apical meristem (SAM), ZWILLE, FIDDLEHEAD, and one KNOTTED-like gene increased in expression in PEG-treated embryos between stages 1 and 2. Comparison between treatments indicated that these genes were significantly induced in immature embryos cultured with PEG (Fig. 4). A similar result was also observed for an AGO (ARGONAUTE) gene (NXSI_050_C01) at stage 2 of development. During the last 10 d in culture (between stages 4 and 5), several genes, including ZWILLE, ERECTA, two of the three ARGONAUTE genes, and both KNOTTED-like genes were down-regulated in PEG-treated embryos (Fig. 4). Differences in expression profiles between control and PEG-treated embryos were observed for genes homologous to the Arabidopsis FUSCA, CLAVATA 1, and NO APICAL MERISTEM. Compared with control embryos, both SCARECROW genes were up-regulated in immature (stage 2) PEG-treated embryos (Fig. 4).

Stress Response Mechanisms

The expression level of the only CAT (catalase) gene present on the array was similar between control and PEG-treated embryos at all stages of development (Fig. 5). Compared with their control counterparts, PEG-treated embryos had higher expression of two superoxide dismutase genes at stage 5. Different expression profiles were observed for a cluster of ascorbate peroxidase genes present on the array. Among the genes involved in glutathione metabolism, glutathione reductase was up-regulated between stages 3 and 4 in PEG-treated embryos. At stage 4, the transcript level of this gene was higher in embryos cultured in the presence of PEG. Differences in expression levels and profiles between control and PEG-treated embryos were observed for two glutathione-S-transferase genes (Fig. 5). The transcript level of one glutathione peroxidase gene (NXNV_144_B12) increased in PEG-treated embryos between stages 1 and 2 and stages 4 and 5. At the end of the maturation period, the expression level of this gene was higher in PEG-treated embryos, compared with their control counterparts.

View larger version (77K): [in this window] [in a new window]   Figure 5.   Fold change differences in the expression of several genes involved in stress response mechanisms. Fold changes were estimated between each stage of embryo development and the neighboring previous stage in the absence (control) and presence (PEG) of PEG. Direct comparisons between treatments at respective stages of embryo development are also shown. Statistically significant (P value of Student's t test < 0.01) expression ratios are shaded in green (down-regulation) or in red (up-regulation). CAT, Catalase; SOD, superoxide dismutase; APX, ascorbate peroxidase; GR, glutathione reductase; GST, glutathione-S-transferase; GPX, glutathione peroxidase; HSPs, heat shock proteins.

Among the genes involved in drought stress mechanisms, several late embryogenic abundant (LEA) genes and genes encoding for heat shock proteins were down-regulated in control embryos between stages 4 and 5 and up-regulated in PEG-treated embryos at the same stages of development (Fig. 5). After 40 d in culture (stage 5), the transcript levels of four LEA genes and two heat shock genes were higher in PEG-treated embryos.

Carbohydrate Metabolism

Suc constitutes the main source of carbon for the growth and development of the embryos throughout the maturation period. Its utilization was investigated by analyzing the expression patterns of several enzymes involved in Suc metabolism (Fig. 6).

View larger version (33K): [in this window] [in a new window]   Figure 6.   Metabolic pathways related to Suc metabolism. Arrows represent enzymatic reactions. Colored bars next to arrows indicate relative fold changes of the corresponding gene between each stage of embryo development and the neighboring previous stage in the absence (C) or presence (P) of PEG. For expression scale see Figure 3. Genes included are: Suc synthase (NXSI_007_H12), cellulose synthase (NXNV_148_H06, NXNV_065_E11, NXSI_007_B11, NXSI_108_H05, NXSI_024_H01, and NXSI_087_D09), 1,3--glucan synthase (NXSI_134_B11), Glc-6-phosphate isomerase (NXSI_021_D06), fructokinase (NXCI_157_B10 and NXNV_079_G08), 6-phosphofructokinase (NXCI_034_B04 and NXSI_082_A04), Fru-bisphosphate aldolase (NXCI_026_G09, NXNV_044_C09, and ST_07_E05), triose-phosphate isomerase (NXSI_105_D03 and NXNV_124_C02), glyceraldehyde-3-P-dehydrogenase (NXNV_117_F02, NXSI_069_G09), phosphoglycerate kinase (NXCI_115_A02), pyruvate kinase (NXSI_126_D02 and NXSI_143_H06), pyruvate dehydrogenase (NXNV_074_H11, NXCI_094_G11 and NXCI_150_E08), citrate synthase (ST_29_A09), aconitase (NXSI_023_H11 and ST_02_E09), succinyl-CoA-synthetase (ST_33_D11, NXSI_039_A11), fumarase (NXCI_106_D10), malate dehydrogenase (NXSI_048_D06, NXNV_076_E08, and NXCI_032_G05). Numbers close to color bars represent the BLASTX score for that particular gene. The + or  signs in boxes indicate significant (P value of Student's t test < 0.01) up-regulation (+) or down-regulation () of the gene between control and PEG-treated embryos at that particular stage of development.

The expression of Suc synthase decreased in control embryos during the five stages of development, whereas it increased significantly in PEG-treated embryos between stages 3 and 4. The transcript levels of the six cellulose synthase genes present on the array were generally similar in both control and PEG-treated embryos (Fig. 6). Differences between treatments were only observed for one cellulose synthase gene (NXNV_065_E11) at stages 3 and 4 of development. At the end of the maturation period (stage 5), the transcript levels of many enzymes involved in glycolysis and the tricarboxylic acid (TCA) cycle were lower in embryos cultured with PEG. These included two Fru-bisphosphate aldolase genes, one triose-phosphate isomerase gene, one glyceraldehyde-3-P-dehydrogenase gene, one pyruvate kinase gene, citrate synthetase, aconitase, succinyl-CoA-synthetase, and one malate dehydrogenase gene (Fig. 6).

Nitrogen Metabolism

In fully mature somatic embryos, a large proportion of the storage products are proteins. Thus, nitrogen metabolism was investigated in both control and PEG-treated embryos (Fig. 7). Assimilation of NH⁴⁺ in the tissue occurs through the activities of two enzymes participating in the GS/GOGAT cycle: Gln synthase and Glu synthase. At the end of the maturation period (stage 5), the transcript levels of one Gln synthase gene and one Glu synthase gene were higher in PEG-treated embryos, compared with their control counterparts (Fig. 7). No significant differences in transcript levels between control and PEG-treated embryos were observed for several genes encoding Asp transaminase, Arg decarboxylase, Orn transaminase, and S-adenosyl-Met dedecarboxylase. At the end of the maturation period (stage 5), ACC oxidase was up-regulated in PEG-treated embryos, whereas two of the four AK genes present on the array had lower expression in PEG-treated embryos, compared with their control counterparts (Fig. 7).

View larger version (34K): [in this window] [in a new window]   Figure 7.   Metabolic pathway related to nitrogen metabolism. Arrows represent enzymatic reactions. Colored bars next to arrows indicate relative fold changes of the corresponding gene between each stage of embryo development and the neighboring previous stage in the absence (C) or presence (P) of PEG. For expression scale, see Figure 3. Genes included are: Gln synthase (NXCI_147_D06 and ST_20_C09), Glu synthase (NXCI_075_C07 and NXNV_063_H07), Asp transaminase (NXNV_136_H04, NXNV_125_E12, NXCI_124_H09 and ST_40_G07), Asn synthase (NXNV_096_C09), Orn transaminase (NXSI_104_E11), Arg decarboxylase (NXCI_127_G06, ST_22_E07, and NXCI_150_A07), adenosine kinase (AK; NXSI_059_G09, ST_22_G10, NXCI_037_F08, and NXSI_116_A09), S-adenosyl-Met decarboxylase (NXSI_099_F10), S-adenosyl-Met synthase (NXNV_090_A12, NXSI_060_E02, ST_08_F07, NXCI_050_B07, and NXCI_031_E05), and ACC oxidase (NXCI_125_G03). Numbers close to color bars represent the BLASTX score for that particular gene. The + or  signs in boxes indicate significant (P value of Student's t test < 0.01) up-regulation (+) or down-regulation () of the gene between control and PEG-treated embryos at that particular stage of development.

	DISCUSSION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

The beneficial effect of PEG in increasing embryo number and quality in coniferous species has been well documented (Attree and Fowke, 1993; Attree et al., 1995; Stasolla et al., 2002). In this study, applications of PEG in the maturation medium induced two major changes in gene expression: the first in immature embryos (stage 2) and the second in fully developed embryos (stage 5, Figs. 2 and 3). The first alteration in global gene expression observed after transferring the tissue on ABA-containing medium (stage 2) occurs in conjunction with the initiation of embryo development. The differences in transcript levels and profiles of several auxin- and ABA-responsive genes observed in embryos cultured in the absence or presence of PEG (Fig. 4) suggest that this compound may affect these initial phases of development, possibly by altering the responsiveness of the tissue to growth regulators. Responsiveness of the tissue to ABA is fundamental for the establishment of the embryo body plan, which occurs through the coordination of an apical-basal and radial growth.

Two important morphological events occurring during the initial phases of embryogenesis in both angiosperms (see Laux and Jurgens, 1997) and gymnosperms (Yeung et al., 1998) are the formation of a SAM and a root apical meristem. Although some of the genetic processes controlling these events have been elucidated in flowering plants (Scheres et al., 1994; Fletcher and Meyerowitz, 2000), no information is currently available for conifers.

In angiosperms, some of the molecular components required for the formation and maintenance of the SAM have been identified, and they require a complex network of interactions among genes, including ZLL (ZWILLE), AGO (ARGONAUTE), ERECTA, NAM (NO APICAL MERISTEM), CLV1 (CLAVATA 1), and FIDDLEHEAD (Clark et al., 1996; Souer et al., 1996; Lolle et al., 1997; Bohmert at al., 1998; Moussian et al., 1998; Yokoyama et al., 1998). The existence of several pine cDNAs, apparent homologs to these genes, suggests that similar mechanisms govern SAM activity in angiosperms and gymnosperms. In our study, the changes in transcript levels of some of these genes appear to be developmentally regulated. Of particular interest is the expression of ZLL, which, during the initial stages of embryo development (stage 2), is higher in PEG-treated embryos compared with their control counterparts (Fig. 4). In Arabidopsis, the role of ZLL is to maintain stem cells within the SAM in an undifferentiated state, thus preserving meristematic identity (Moussian et al., 1998). Mutation of the ZLL gene, which is expressed in the shoot pole during embryogenesis, results in the differentiation of the stem cells in the SAM and leads to meristem abortion at germination (Moussian et al., 1998). Thus, the up-regulation of the ZLL gene in the presence of PEG may contribute to proper SAM formation by conferring stem cell identity to the apical cells of developing spruce embryos. The maintenance of a large group of undifferentiated cells within the SAM might: (a) enhance the rate of embryo formation because meristematic cells may be more responsive to ABA, and (b) increase the postembryonic performance of the embryos at germination, when through the resumption of mitotic activity new leaf primordia are produced. The poor postembryonic performance of spruce somatic embryos is often the result of abnormal SAM formation at maturation (Kong and Yeung, 1992). In addition to ZLL, PEG may control SAM activity through the expression of other genes, including FIDDLEHEAD, AGO, and KNOTTED-like genes, which are up-regulated in immature embryos cultured with PEG (Fig. 4). These genes are implicated in the control of cell division and differentiation. Mutations in AGO (Bohmert et al., 1998), as well as in a member of the KNOTTED gene family, SHOOTMERISTEMLESS (Barton and Poethig, 1993), result in meristem abortion.

In spruce embryos, formation of the root apical meristem is also an early event, which occurs at the filamentous stage of embryo development (stages 2 and 3) (Yeung et al., 1998). Molecular analyses in Arabidopsis embryonic roots have demonstrated the importance of the SCR (SCARECROW) gene in delineating the radial pattern of growth. Expression of SCR is needed very early during embryogenesis for conferring endodermis identity because mutation in this gene results in the loss of one ground tissue layer (Di Laurenzio et al., 1996). Two SCR genes present on the array are up-regulated in the presence of PEG at stage 2 of embryo development (Fig. 4). Thus, PEG may be important for controlling the radial pattern of growth in embryonic roots. The development of a functional root during embryogenesis is critical for successful conversion at germination.

Continuation of embryo growth is accompanied by an active aerobic metabolism, which results in the production of active oxygen species. As suggested by Stasolla and Yeung (2001), the ability of the embryos to develop an efficient antioxidant system may be the key for successful development and postembryonic growth. The transcript levels of several antioxidant enzymes are higher in some stages of PEG-treated embryos, compared with their control counterparts. The up-regulation of superoxide dismutase and glutathione peroxidase at stage 5, as well as glutathione reductase at stage 4 in PEG-treated embryos, may represent a protective mechanism against reactive oxygen species, which is reduced in control embryos (Fig. 5). The glutathione system was found to be the major detoxifying mechanism in sunflower (Helianthus annuus) seeds, and seed viability was strictly dependent upon glutathione metabolism because high GR activity was observed only in those seeds with high germinability (Torres et al., 1997). A positive correlation between activity of some antioxidant enzymes and embryogenic capability was also demonstrated in white spruce cultured cells (Stasolla and Yeung, 2001).

Major changes in gene expression were observed in PEG-treated embryos between stages 4 and 5 (Figs. 2 and 3). Such changes, not observed in control embryos, may be required for completing all those metabolic events that take place during the late stages of maturation in zygotic embryos, before the desiccation period. Imposition of a drying period, which represents a developmental transition between maturation and germination (see Kermode, 1995), is necessary for increasing the germination frequency of white spruce somatic embryos (for review, see Stasolla et al., 2002). Unlike embryos matured in ABA alone, PEG-treated embryos can survive severe water deficit and resume growth at germination (Attree et al., 1991, 1995). Thus, tolerance to water stress is critical for successful postembryonic growth.

One of the key metabolic events associated with dehydration tolerance is the synthesis of LEA proteins (for review, see Kermode, 1995). The transcript levels of many of these hydrophilic proteins, which protect cellular components from severe dehydration, are increased in the presence of PEG during the late stages of embryo maturation, whereas they are decreased in control embryos. A less pronounced but similar trend was also observed for the heat shock proteins, which may play a protective role during drought conditions (Lindquist and Craig, 1988; Fig. 5). Therefore, the increased transcript levels of these two classes of proteins in response to PEG, also observed by Dong and Dunstan (1996a, 1996b) may enhance the desiccation tolerance of the embryos, thus improving postembryonic growth.

Two other important changes in metabolism occurring during the late stages of embryo maturation are: (a) the overall reduction of carbohydrate catabolism (see Foley, 1996), and (b) the increase in protein biosynthesis (see Kermode, 1995). In apple (Malus domestica) seeds, a decreased rate of glycolysis was observed during desiccation (Bogatek and Lewak, 1988). In our study, the transcript levels of many enzymes involved in Suc catabolism, including the glycolytic pathway and the TCA cycle, were down-regulated in PEG-treated embryos (Fig. 6). The high expression level of several catabolic enzymes observed in mature control embryos may be related to lack of physiological maturity and may lead to precocious germination. A fully operative glycolytic pathway and TCA cycle, in fact, are important at the onset of germination for the mobilization of storage products. Therefore, the inhibitory effect of PEG on precocious germination of spruce somatic embryos (Attree et al., 1991) may be the result of altered carbohydrate metabolism.

A second important event occurring during spruce embryo maturation is the accumulation of storage proteins (Joy et al., 1991, 1997; Misra et al., 1993). Misra et al. (1993) reported quantitative and qualitative differences in storage proteins of embryos cultured in the presence or absence of PEG. Applications of PEG increased deposition of storage proteins, which were similar in profile and structure to those found in seed embryos (Misra et al., 1993). The transcript levels of one Gln synthase gene and one Glu synthase gene were higher in mature embryos treated with PEG (Fig. 7). The up-regulation of these two enzymes that play a key role in nitrogen assimilation may be required for sustaining Glu synthesis in maturing embryos. Glu, together with Arg, is the most abundant amino acid found in storage proteins of conifers (Newton et al., 1992; Leal and Misra et al., 1993). A positive correlation between Glu synthase activity and amino acid production was also documented by Joy et al. (1997). Another important class of nitrogen compounds, which are synthesized during embryo development, includes polyamines. These small molecules are preferentially accumulated in PEG-treated embryos (Kong et al., 1998) and play an important role during embryogenesis (see Thorpe and Stasolla, 2001). Synthesis of spermidine and spermine requires decarboxyl-S-adenosyl-Met, which is produced from S-adenosyl-Met (Fig. 7). Thus, availability of S-adenosyl-Met, which is also utilized as an intermediate by pathways involved in ethylene and AMP production, affects the rate of polyamine synthesis. It appears that the activities of these two pathways, which divert S-adenosyl-Met from polyamine synthesis, may be differentially regulated in the presence or absence of PEG. Compared with mature control embryos, two AK genes, involved in the production of AMP, were down-regulated in PEG-treated embryos, whereas the reverse was observed for ACC oxidase, the last enzyme of ethylene biosynthesis (Fig. 7). An increased accumulation of ethylene in PEG-treated embryos is predictable because this volatile hormone is often produced in culture under stress conditions (for review, see Gaspar et al., 1996).

In conclusion, studies on transcript accumulation during the embryogenic process have revealed the presence of many genes that are developmentally regulated during the different stages of embryo development. Furthermore, besides elucidating the mechanism of action of PEG during somatic embryogenesis, these findings may have important implications in the identification of target genes or metabolic products for improving somatic embryo quality in conifers, through genetic engineering or modification of media during development.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Plant Material

White spruce (Picea glauca [Moench] Voss) embryogenic tissue was generated from zygotic embryos (Lu and Thorpe, 1987). Two cell lines (E1 and E2) were generated from open pollinated seeds (lot nos. 7431580.1 and 7231587.2, respectively) provided by the National Tree Seed Center (Fredericton, NB, Canada). Seeds were sterilized in 20% (v/v) commercial bleach for 20 min and rinsed three times with sterile water. Dissected embryos were placed on induction (von Arnold and Eriksson [AE]) medium (von Arnold and Eriksson, 1981) containing 10 µmol L¹ 2,4-D, 5 µmol L¹ N⁶-benzyladenine, 5% (w/v) Suc, and 0.8% (w/v) Bacto-agar, pH 5.8 (DIFCO Laboratories, Detroit). The stock culture was maintained in the dark at 26°C for 4 to 5 weeks. Embryogenic tissue was transferred onto a maintenance medium (AE medium containing 10 µmol L¹ 2,4-D, 2 µmol L¹ N⁶-benzyladenine, and 3% [w/v] Suc) and was subcultured every 7 d.

Somatic embryo development was initiated by transferring the embryogenic tissue onto solid maturation medium (AE medium containing 50 µmol L¹ ABA and 5% [w/v] Suc; Kong and Yeung, 1992). To test the effect of osmoticum, the following concentrations of PEG were included into the maturation medium: 0%, 2.5%, 5%, 7.5%, and 10% (w/v). At the end of the maturation period (40 d in culture), the number of normal-looking cotyledonary embryos was scored. For statistical analysis, the Student Newman-Keuls test (Zar, 1999) was utilized. All the results presented in this study were obtained from the E1 line. The other line (E2) was used to confirm the observed differences in gene expression between control and PEG-treated embryos at stages 2 and 5 of development.

Microarray Procedure

The 2,178 cDNAs were selected from 55,000 expressed sequence tags grouped in 9,000 contigs. These expressed sequence tags were obtained from five different cDNA libraries: NXNV (xylem normal wood vertical), NXCI (xylem compression wood inclined), NXSI (xylem side-wood inclined), ST (shoot tip), and PC (pollen cone; http://web.ahc.umn.edu/biodata/nsfpine/contig_dir6). The cDNAs were selected closest to the 3' end of the respective contig and were run on BLASTX against the Arabidopsis database (ftp://ftpmips.gsf.de/cress/arabiprot/). The best hit from the BLAST search was utilized for grouping the cDNAs into functional categories, as proposed for Arabidopsis (http://pedant.gsf.de). The selected cDNAs were transformed into Escherichia coli XL-1 blue competent cells and the plasmids were isolated using Qiagen kits (Qiagen USA, Valencia, CA).

Probe Preparation and Printing

The cDNAs were PCR amplified in 50-µL reactions in 96-well reaction plates. Each 50-µL reaction contained 39.1 µL of distilled, deionized water; 0.5 µL of PCR reaction buffer containing 15 mM MgCl₂, 1 µL of dNTPs, 1 µL of forward- and reverse-specific primers (10 µM), respectively; 0.4 µL of Taq polymerase (5units µL¹); and 2.5 µL of 100-fold diluted plasmid stock. Amplifications were carried out in thermocyclers (MJ Research, Waltham, MA) with the following conditions: denaturation at 94°C for 30 s, annealing at 57°C for 1 min, and elongation at 72°C for 4 min. After 35 cycles, the final chain elongation was performed at 72°C for 10 min. The PCR products were purified on Multiscreen filter plates (Millipore Corp., Bedford, MA) and analyzed on ethidium bromide agarose gels. The purified DNA was denatured in 50% (w/v) dimethyl sulfoxide and spotted in four replicates onto Corning microarray technology-gamma amino propyl silane aminosilane-coated glass microscope slides (Corning, NY) by a 417 Arrayer (Affymetrix, Woburn, MA). Identity of the clones discussed in this study was confirmed by resequencing for 92% of the cDNAs.

Target Preparation

For each developmental stage of control and PEG-treated embryos, tissue (1 g fresh weight) was pooled from three separate petri dishes, combined, and utilized for RNA extraction as described by Chang et al. (1993). cDNA probes were labeled using the aminoallyl procedure developed by J.L. De Risi (University of California, San Francisco; http://cmgm.stanford.edu/pbrown/protocols/index.html). RNA from each sample was labeled with Cy3 and Cy5 and used for reciprocal hybridizations. Hybridization and stringency washes were performed using the recent protocol from The Institute of Genomic Research protocol (Hegde et al., 2000). The slides were scanned using a ScanArray 4000 Microarray Analysis System (GSI Lumonics, Ottawa). Raw, non-normalized intensity values were collected with QUANTARRAY software (GSI Lumonics). Using the quantification option, spots were visually inspected for spot morphology and background. Only a very few spots were flagged as bad and excluded from further analysis.

Experiment Design and Statistical Analysis

A fully balanced, incomplete loop experimental design was used in our experiment, as proposed by Kerr and Churchill (2001). Gene significance was then estimated using the "mixed model system" developed by Wolfinger et al. (2001) and Jin et al. (2001). This model is highly sensitive and shows that changes in gene expression less than 2-fold can be statistically significant (Jin et al., 2001). In brief, the log₂ transformed data (y_ijk) were subjected to a normalization model: y_ijk = µ + A D_j + (A × D)_ij + _ijk, where µ is the sample mean, A is the effect of the array, D_j is the effect of the dye, (A × D)_ij is the effect of the array-dye interaction, and _ijk is the stochastic error. The residual values from this model were then fit into gene-specific model in the form of r_ijk = µ + A + T_j + N_k + _ijk, where T_j corresponds to the jth treatment (control and PEG), and N_k is the effect of the clone position on the array. Both models were implemented using PROC MIXED in SAS (SAS/STAT Software version 8, SAS Institute Inc., Cary, NC). The least square means (probability value: P < 0.01) and the differences in least square means between treatments were calculated from the gene-specific model and utilized for calculating fold changes (two differences in least square means). Fold changes were imported into GENESPRING, version 4.1 (Silicon Genetics, Redwood City, CA) and the "Make Tree" function was utilized to perform hierarchical clustering of the genes.

RT-PCR

The transcript levels of five cDNAs (NXCI_125_G03, NXSI_134_G06, NXSI_049_A01, NXSI_137_D09, and NXSI_079_C02), which appeared differentially expressed between control and PEG-treated cotyledonary embryos in the microarray experiments, were confirmed by RT-PCR. Total RNA was extracted using the cetyl-trimethyl-ammonium bromide extraction procedure published by Chang et al. (1993). The concentration of total RNA was measured using the RiboGreen RNA Quantitation Reagent and Kit (Molecular Probes, Eugene, OR). First strand cDNA was reverse transcribed from 300 ng of total RNA using Taq-Man Reverse Transcription Reagents (PE-Applied Biosystems, Foster City, CA) according to the manufacturer's instructions. Gene-specific primers were designed by the Primer Express 1.0 (PE-Applied Biosystems). The relative transcript abundance was monitored on an Applied Biosystems 7700 Sequencer using SYBR Green PCR Master Mix. The 18S amplicon was used as an internal control for normalization.