Novel Insight into Vascular, Stress, and Auxin-Dependent and -Independent Gene Expression Programs in Strawberry, a Non-Climacteric Fruit

Abstract

Using cDNA microarrays, a comprehensive investigation of gene expression was carried out in strawberry (Fragaria × ananassa) fruit to understand the flow of events associated with its maturation and non-climacteric ripening. We detected key processes and novel genes not previously associated with fruit development and ripening, related to vascular development, oxidative stress, and auxin response. Microarray analysis during fruit development and in receptacle and seed (achene) tissues established an interesting parallelism in gene expression between the transdifferentiation of tracheary elements in Zinnia elegans and strawberry. One of the genes, CAD, common to both systems and encoding the lignin-related protein cinnamyl alcohol dehydrogenase, was immunolocalized to immature xylem cells of the vascular bundles in the strawberry receptacle. To examine the importance of oxidative stress in ripening, gene expression was compared between fruit treated on-vine with a free radical generator and non-treated fruit. Of 46 genes induced, 20 were also ripening regulated. This might suggest that active gene expression is induced to cope with oxidative stress conditions during ripening or that the strawberry ripening transcriptional program is an oxidative stress-induced process. To gain insight into the hormonal control of non-climacteric fruit ripening, an additional microarray experiment was conducted comparing gene expression in fruit treated exogenously with auxin and control fruit. Novel auxin-dependent genes and processes were identified in addition to transcriptional programs acting independent of auxin mainly related to cell wall metabolism and stress response.

The attractive characteristics of strawberry (Fragaria × ananassa) fruit are not only aroma, taste, color, and texture, but also their essential nutrient, mineral, vitamin content, and antioxidant properties. Their antioxidant properties, coupled with high dietary fiber content, have been medically recognized as having positive influences on protecting against the risk of many diseases (Brownleader et al., 1999). To date, we still lack valuable information on the molecular events that control strawberry fruit development, ripening, and adaptation to environmental cues that are all complex biological processes involving the coordinated regulation of genes and biochemical pathways.

Unlike fruit botanically defined as arising from the expansion of the ovary, strawberry is actually the swollen base of the flower (receptacle) with one seeded fruit (termed achenes) located on the outer surface (Perkins-Veazie, 1995). Vascular bundles supply nutrients that move acropetally to the developing embryos in the achenes and surrounding cells of the receptacle (Hancock, 1999). In the ripening stage of strawberry fruit development, the vascular tissue comprises long fibers composed of cellulose, protein, pectin, and lignin (Suutarinen et al., 1998). Because strawberry fruit is composed of approximately 90% water and 10% total soluble solids, it is not inconceivable that the vascular system beginning from the achenes and connecting to the pith plays an important role in the texture and structural integrity of the ripe fruit (Jewell et al., 1973; Suutarinen et al., 1998). To date, studies have neglected to explore the role of the vascular system in the development and ripening of strawberry fruit and instead have focused on the remainder of the receptacle tissue.

In plants, tracheary element (TE) differentiation/xylogenesis has been extensively studied using the Zinnia elegans mesophyll cell system. It commences with rearrangements of the microtubules in a cortical banding pattern that reflects the position of future secondary thickenings (Fukuda, 1997). Subsequently, cellulose is deposited in the initial thickenings, followed by lignification and cell death (Domingo et al., 1998). Programmed cell death (PCD) is an active process that occurs in plants during development and in response to environmental cues. Cell death occurring during differentiation of procambium into TE is one such example (Greenberg, 1996). During the PCD process, TEs degrade their cellular contents and become hollow corpses serving as a water conducting system.

Organ senescence is an example of a PCD process occurring in plants. Senescence is a dynamic and tightly regulated developmental process that involves an array of changes at both physiological and biochemical levels including gene expression. Fruit ripening is considered by some to be a specialized form of senescence (Seymour et al., 1993). A large number of biotic and abiotic factors accelerate the process. In fruit, external environmental factors such as heat (Cheng et al., 1988; Kagan-Zur et al., 1995), cold (Masia, 1998), salt (Avsian-Kretchmer et al., 1999), and ozone (Kirtikara and Talbot, 1996) have been proven to induce oxidative stress. Ripening itself, however, may impose stress conditions on the fruit. In grape (Vitis vinifera), the accumulation of 10 cDNAs encoding putative stress response proteins upon ripening was recently reported (Davies and Robinson, 2000). To date, no studies testing the hypothesis that a transcriptional program related to stress, and in particular oxidative stress, exists in ripening fruit have been reported.

Concurrent with the supply of nutrients to the achenes (described above), the hormone auxin is translocated basipetally through the phloem of the vascular bundles from the achenes to the peduncle (Perkins-Veazie, 1995). It has been unequivocally demonstrated that growth and early fruit development of strawberry is stimulated by auxin originating in the achenes (Nitsch, 1950). Later in fruit development (middle green stage) before ripening, auxin levels decline in the receptacle, possibly due to the cessation of auxin transport from the achenes, and this invokes the ripening process (Given et al., 1988). Ripening triggered by reduced auxin levels is accompanied by de novo synthesis of specific mRNAs, which encode proteins responsible for the dramatic changes in fruit such as pigmentation and texture (Manning, 1994, 1998). In climacteric fruit such as tomato (Lycopersicon esculentum), banana (Musa spp.), apple (Malus domestica), and melon (Cucumis melo), ethylene is the hormonal signal that triggers ripening; however, not all ripening processes are ethylene dependent (Lelièvre et al., 1997). Although it has been well documented that exogenous ethylene has no effect on the ripening process in non-climacteric fruit, it appears that in strawberry and other fruit such as citrus and pineapple (Ananas comosus) it may play a role (Goldschmidt et al., 1993; Alonso et al., 1995; Cazzonelli et al., 1998). Strawberry exhibits a low and slightly elevated level of ethylene production during the late stage of ripening (Perkins-Veazie et al., 1996). Thus, it would seem that in both climacteric and non-climacteric fruit, not all ripening processes are affected by the same hormone.

In this study, our goal was to better understand the processes underlying strawberry fruit maturation and non-climacteric ripening. By using DNA microarray technology, we were able to perform large-scale and simultaneous investigation of gene expression during fruit development, in different tissues and after exposure to stress (oxidative stress) and hormonal treatments. The results highlighted two key processes active during fruit development and ripening relating to vascular development and oxidative stress. They also showed that not all the processes associated with strawberry ripening are under the same genetic control, and are probably a collection of processes regulated in a discrete manner.

RESULTS AND DISCUSSION

First and Second Generation Microarrays

In this paper, we refer to strawberry “fruit” as the receptacle including the seeds (achenes). The main stages of strawberry fruit development are depicted in Figure 1A. The time course from anthesis (full petal opening) to medium green, large white, turning, and red (ripe) stages of fruit development is approximately 10, 21, 24, and 30 d, respectively. We used microarrays comprising 1,701 strawberry cDNAs (probes) derived from a red fruit cDNA library to perform four first generation microarray experiments (Fig. 1B). The focal point of these four experiments was to identify ripening-related genes and processes not previously disclosed. The first three experiments (hybridizations) compared green versus red (I), white versus red (II), and turning versus red (III) stages of fruit development. A fourth microarray experiment was performed comparing achene versus receptacle (IV) to differentiate between genes expressed in either of these two fruit tissues.

Figure 1.

Using cDNA microarrays to follow gene expression patterns in strawberry fruit during development, in different tissues, under oxidative stress conditions and hormonal treatment. A, Strawberry fruit developmental stages. 1, Small flower bud; 2, large flower bud; 3, anthesis; 4, small green; 5, medium green; 6, large green; 7, small white; 8, large white; 9, turning; 10, red. Stages 5 and 8–10 were used for microarray experiments (referred to as G, W, T, and R, respectively). B, Experimental schematic diagram of first (I–IV) and second (V and VI) generation microarray experiments.

Combining the results from all four experiments, a total of 537 unique cDNAs were identified as differentially expressed at least once (Fig. 2A). Two hundred fifty-nine cDNA clones (48%) showed higher expression in the achenes (AchA, Fig. 2A) and 182 (34%) showed higher expression in the receptacle (RecA, Fig. 2A; 441 in total expressed in either achene or receptacle). Eighty-eight percent of the achene cDNAs (228) and 56% of the receptacle cDNAs (102) were not developmentally regulated. Ninety-six cDNAs (18%) were equally expressed in either tissue type but were differentially expressed during development. A large number (42%) of the 537 cDNA clones identified were “unknown” or “novel.”

Figure 2.

Differential gene expression in strawberry fruit. The diagrams show the numbers of overlapping and nonoverlapping genes differentially expressed during fruit development (Dev), receptacle associated (RecA), and achene associated (AchA) as detected in experiments I through IV (A). Those genes are auxin repressed (AuxRep), ripening regulated (RipR), receptacle associated (RecA), and stress induced (StrInd), detected in experiments V and VI and based on experiments I through IV (B).

Based on these data, a second generation microarray was prepared comprising 384 probes. This new microarray allowed us to focus our analysis primarily on ripening-regulated receptacle-associated cDNAs. Array elements included: (a) those showing elevated expression in the receptacle tissue; (b) those that were differentially expressed during development, including ripening-regulated cDNAs (112 individual cDNAs were identified as ripening regulated and 80 of them [RipR, Fig. 2B] were arrayed on the second generation microarray); (c) cDNAs identified in our original expressed sequence tag collection that did not show differential expression in the first four experiments; and (d) appropriate controls (for a detailed description of array elements, see “Materials and Methods”).

Two additional microarray experiments were performed, an oxidative stress experiment (V) and an auxin experiment (VI). The first experiment was designed to identify ripening-regulated cDNAs that might be also oxidative stress induced (in the receptacle). The second experiment was performed to detect auxin-dependent and independent ripening-related cDNAs and processes. A schematic diagram showing the experimental outline of the first and second generation microarray experiments is shown in Figure 1B.

The capability and reproducibility of our microarray experiments in scoring differential gene expression was described in detail in a previous paper (Aharoni et al., 2000). Each of the microarray experiments described in this study was performed twice with the dyes reversed between the two replicates. A statistical analysis of variance model was used to evaluate the data and to determine a threshold value, which indicates a significant up or down regulation of gene expression (see “Materials and Methods”). We further demonstrated the quality and reliability of our microarray experiments by comparing gene expression results from RNA gel-blot analyses with expression ratios originating from microarray data (Table I). It is beyond the scope of this report to describe in detail all the genes that were differentially expressed and novel processes identified. Several of our main discoveries are provided below.

Table I.

Confirmation of differential expression detected by microarrays with RNA gel-blot analysis

^a Definition and accession of nucleotide sequence of the first BLAST X homolog. 

^b Expression ratio detected by microarray experiment (I) comparing red (R) and green (G) fruit stages of development. 

^c Gel blots loaded with total RNA derived from leaf (L), green fruit stage (G), and red fruit stage (R). 

The Importance of the Vascular System and Lignification in the Developing Receptacle

Gene expression analysis during strawberry fruit development (experiments I–III) revealed a group of 112 unique (i.e. distinct) ripening-regulated genes. Ripening-regulated genes were only those which showed higher levels of expression in the red stage compared with either the green, white, or turning stages (in one or more of the three cases). The majority of the cDNA clones were either receptacle associated (64 of 112) or did not show any difference in expression between the achene and receptacle tissues (41) and are possibly expressed in both tissues (as deduced from experiment IV). Several of them were previously associated with a specific ripening process (e.g. pigmentation and cell wall), whereas others had no previous recognized role in any or a certain aspect of fruit ripening. One such group comprised cDNAs encoding putative cinnamoyl-CoA reductases (CCRs: JB116 and JB196) and CADs (F193, F138, and F122), enzymes performing the last committed steps in the biosynthesis of lignin (Chapple and Carpita, 1998).

Detailed analysis of CCR and CAD expression in different strawberry tissues and during fruit development and maturation using RNA gel blots confirmed the microarray data and showed elevated levels of both transcripts in the red stage (Fig. 3A). Although the expression of CCR gradually increased during ripening, CAD expression decreased after the green stage (in the white and turning stages) before increasing again at the red stage (Fig. 3A). Expression of both genes could be detected in achene and receptacle (fruit with no achenes), petioles, leaves, and flowers. Because these genes were strongly expressed in the ripening receptacle tissue, we suspected that some of them might be actively expressed in the vascular bundles and associated with their lignification (Fig. 3B). To localize where active lignification is occurring in the fruit, we performed histochemical staining on sections from the four different stages of fruit development (green, white, turning, and red) using the Weisner reagent (phloroglucinol-HCl). This reacts with aldehyde groups (cinnamaldehydes and benzaldehydes) in the lignin, giving characteristic deep reddish-purple coloration in the xylem of the vascular bundles (Clifford, 1974). Strong staining indicating the presence of lignin was detected in all stages of development in immature xylem cells of the fibrovascular strands of the receptacle (Fig. 3, C and D).

Figure 3.

The vascular system and lignin-associated gene expression and protein localization in strawberry fruit. A, RNA gel-blot analysis of strawberry CCR and CAD expression in various strawberry tissues and during fruit development. 1, Petiole; 2, leaf; 3, flower; 4, green fruit; 5, white fruit; 6, turning fruit; 7, red fruit; 8, red fruit without achenes; 9, achenes; 10, overripe fruit. The entire strawberry CCR and CAD cDNAs were used as probes for hybridizations. B, Section of a green and red ripe strawberry fruit showing fibrovascular strands (vb, vascular bundles) connecting the achenes (a) to the interior of the receptacle (p, pith). C and D, Presence of lignin in the vascular system (xylem vessels) in the receptacle, visualized after staining with phloroglucinol, E and F, Cross sections of the receptacle stained by immunolocalization of CAD in the lignified vascular tissue (immature xylem) with the strawberry anti-CAD (F193) antiserum. G and H, Receptacle stained by immunolocalization with pre-immune antiserum (negative controls). Sections C, E, and G and D, F, and H are green and red stage strawberry receptacle, respectively. In C and F, bar = 12 μm; in D and E, bar = 6 μm; in H and G, bar = 7 μm.

Expression of a CAD cDNA homolog (F193, GenBank accession no. U63534) in yeast (Saccharomyces cerevisiae) cells and enzymatic activity assays demonstrated a CAD activity of the recombinant enzyme (i.e. only a clear activity was found using sinapylaldehyde and coniferaldehyde as substrates; R. Blanco-Portales, N. Medina-Escobar, J.A. López-Ráez, J.A. Gónzalez-Reyes, J.M. Villalba, E. Moyano, J.L. Caballero, and J. Muñoz-Blanco, unpublished data). Immunological detection of the strawberry CAD (U63534) protein in the receptacle using its corresponding primary antistrawberry CAD polyclonal antiserum showed that this particular CAD protein was present during all stages of fruit growth and development and localized specifically to immature xylem cells undergoing active lignification (Fig. 3, E–H). At this stage, it cannot be ruled out that CAD enzyme activity in the receptacle might also be associated with wound response or with the biosynthesis of flavor compounds as suggested in an earlier study by Mitchell and Jelenkovic (1995). The authors (Mitchell and Jelenkovic, 1995) reported on a ripening-regulated and receptacle-specific CAD enzyme activity and correlated it with the interconversion of aldehydes and alcohols implicated in flavor of ripe strawberry fruit. However, substrate specificity of the recombinant CAD including the immunolocalization data presented here clearly suggest a role for this particular CAD in the lignification of vascular elements in the receptacle.

Analogy in Gene Expression between Strawberry Fruit Development and TE Differentiation in Z. Elegans

Apart from genes associated with lignification, the expression pattern and putative identity of other clones suggested that a large set of genes detected in this study might be related to processes occurring in the vascular tissue. From our first three microarray experiments (I–III), we could deduce that 31 of 112 (28%) distinct genes identified as ripening regulated show similarity to Z. elegans genes expressed during the process of TE differentiation. In Table II, we show the previously reported Z. elegans genes associated with TE differentiation together with their strawberry ripening counterparts. It is feasible that the strawberry genes (or other members of a particular gene family) suggested here as vascular associated might function in other strawberry fruit ripening processes and tissues as well.

Table II.

Parallels in gene expression between tracheary element differentiation in Z. elegans and strawberry development and ripening

Z. elegans Gene	Referencea	Strawberry (PRI Clone)	Homolog Definition (GenBank Accession No.) (E Value)b	Ratioc
				R:G	R:W	R:T
Activation of protein synthesis apparatus and dedifferentiation
ZeRPS	Fukuda (1997)	F187	40S Ribosomal protein S13 (AB043974) (3e-54)	0.9	1.5	2.6
ZeRPS3a		C23	60S Ribosomal protein L13E (AJ132537) (6e-60)	10.6	2.1	2.9
ZeRPPO		H88	QM-like protein (AF295636) (1e-52)	2.7	1.8	2.2
ZeEF1β	Fukuda (1997)	C197	Elongation factor 1α (1) (AJ223969) (3e-69)	4.0	1.5	1.7
TED4	Fukuda (2000)	B90	Lipid transfer protein (2) (X79604) (4e-18)	1.3	3.8	1.7
TED 63.30	L.T. Koonce (1995d)	B105	Nuclear protein (AL357612) (2e-11)	0.7	1.5	2.4
Changes in cytoskeleton including actin reorganization
ZeTub1,2	Yoshimura et al. (1996)	G84	Beta-tubulin (1) (D63138) (3e-80)	12.5	6.5	1.5
		H54	Alpha-tubulin (2) (U12589) (2e-66)	3.4	2.3	1.8
Ngr		C122	Profilin (AF129427) (4e-49)	13.9	8.2	1.8
Ngr		H78	Annexin (AF188832) (3e-59)	2.5	3.5	1.2
Primary and secondary cell wall metabolism
ZePel	Domingo et al. (1998)	D86	Pectate lyase (1) (U63550) (3e-78)	4.8	0.8	0.6
ZeExp1,2,3	Im et al. (2000)	F39	Expansin (1) (AF159563) (3e-28)	23.3	1.6	1.2
Ngr	Fukuda (1997e)	D56	Extensin (2) (AF026382) (3e-43)	6.5	0.5	0.6
ZePG1	Y. Ohdaira et al. (2001d)	D15	Polygalacturonase (2) (U20431) (1e-19)	4.0	1.2	0.6
Phenylpropanoid pathway and lignification
ZCAD1	Fukuda (2000)	F122	Cinnamyl ADH (1) (U63534) (8e-57)	4.3	5.2	3.5
		F138	Cinnamyl ADH (2) (D13991) (3e-65)	4.2	3.5	1.5
		F193	Cinnamyl ADH (3) (U63534) (2e-53)	16.0	8.2	1.8
Ngr		JB116	CCR (1) (AJ295838) (3e-49)	3.5	1.5	0.8
		JB196	CCR (2) (X79566) (1e-06)	4.0	1.6	0.9
CAOMT	Ye and Varner (1995)	F102	O-methyl transferase (2) (AF220491) (1e-18)	2.6	1.0	1.3
Programmed cell death
ZCP-4	Fukuda (1997)	C56	Cys proteinase (3) (Z14028) (8e-42)	4.1	1.3	1.0
P48h-17	Ye and Varner (1996)	JB202	Cys proteinase (7) (Z99954) (3e-68)	4.7	1.6	0.6
TED2	Fukuda (2000)	E149	Quinone reductase-like protein (AL16372) (3e-36)	3.3	3.7	1.8
Ngr	Woffenden et al. (1988e)	JB75	Ubiquitin (1) (U29159) (2e-81)	4.9	0.9	1.5
		D53	Ubiquitin extension protein (AJ223329) (7e-48)	2.9	1.3	1.4

^a Report on expression of the corresponding gene in Z. elegans (depicted under Z. elegans Gene) during TE differentiation. 

^b Definition, accession no. (nucleotide sequence), and E value (see Altschul et al., 1990) of the first BLAST X homolog. Not always the described strawberry cDNA shows first homology to the corresponding Z. elegans gene in the BLAST search result. The no. in parentheses following the putative definitions represents the no. of the sequence contig in the case when more than one sequence showed similar BLAST result but did not align in the sequence alignment. 

^c Expression ratios obtained in first three microarray experiments comparing red versus green (R:G), red versus white (R:W), and red versus turning (R:T) stages of fruit development. Values depicted in bold represent significant changes in expression (see “Materials and Methods”). The processes depicted above each gene set (in bold) are known to occur during TE differentiation in Z. elegans.  

^d Unpublished data (only sequence is currently available). 

^e Cases in which protein activity has been reported or suggested but not expression of the corresponding gene (ngr, no gene reported). If no report on protein exists, the association with TE differentiation was suggested in this study. 

Genes involved in the first phase of Z. elegans TE differentiation are dominated by components of RNA and protein turnover machinery, such as ribosomal proteins and elongation factors type 1 (Table II; Fukuda, 1997). During strawberry ripening, a dramatic induction of genes related to DNA/RNA/protein turnover, such as those encoding elongation factors (types 1 and 2) and ribosomal proteins were also observed in the receptacle. A putative strawberry lipid transfer protein, similar to the Z. elegans TED4 gene, was also identified as ripening regulated. The TED4 protein is secreted into the apoplastic space and associated with morphological changes of TEs (Endo et al., 2001). TED4 is suggested to act as an inhibitor of the proteasome that induces TE differentiation and the progression of TE program in committed cells. By inhibiting the proteasome, TED4 protects healthy cells from injury due to proteolytic activities exudated from dying TEs.

During the second phase of Z. elegans TE development, before secondary wall thickening, the cytoskeleton undergoes dynamic changes reflected by the accumulation of transcripts encoding tubulins (Fukuda, 1997). Tubulin synthesis increases the amount of microtubules, facilitating the regulation of secondary cell wall formation in subsequent stages of TE development. The identification in strawberry of Z. elegans tubulin homologs (ZeTub1 and ZeTub2) never previously associated with ripening of soft fruit, showing a dramatic increase in expression during fruit development, provided strong supporting evidence to the analogy between the two systems.

Along with the increase in tubulin synthesis, changes in actin organization occur, in which actin filaments form thick cables functioning in cytoplasmatic streaming (Fukuda, 1997). An important added value for gene expression analysis using microarrays is the association of genes with an unclear role in a certain process to those already identified showing similar expression profile (i.e. “guilt by association”). Using a similar approach, we suggest that the dramatic accumulation of profilin in ripening (14-fold increase in expression between the green to red stages), combined with its specificity to the receptacle tissue, is possibly related to its role in vascular bundle development. Profilin is an actin-binding protein and therefore affects the structure of the cytoskeleton by regulating the organization of actin filaments (at high concentrations, profilin prevents the polymerization of actin, whereas it enhances it at low concentrations; Kovar et al., 2000).

Z. elegans genes known to be involved in primary and secondary cell wall metabolism (before cell wall thickening) include pectate lyase (ZePel), expansins (ZeEXP 1, 2, and 3), polygalacturonase (ZePG1), caffeic acid 3-O-methyltransferase (CAOMT), and CADs (ZCAD1). Apart from extensins, which are specifically associated with secondary walls of TEs (Fukuda, 1997), the expression of expansin genes was recently correlated with primary cell wall expansion and secondary cell wall thickening during Z. elegans TE development in vitro (Im et al., 2000). It is possible that the pectate lyase and expansin enzymes previously identified in strawberry as ripening regulated and associated with cell wall metabolism in the receptacle cells (Medina-Escobar et al., 1997; Civello et al., 1999) might be involved in remodeling the cell wall during the development of the vascular system.

In the third phase of Z. elegans TE development, the deposition of secondary cell wall components (secondary cell wall thickening and lignification) in conductive tissues consisting of dead TEs is tightly coupled to PCD (Fukuda, 2000). Hydrolytic activities of enzymes, such as Cys proteases, and of the ubiquitin and proteasome systems have been implicated in the PCD process during organ senescence and TE differentiation, acting both as mediators of signal transduction and as effectors of PCD (Groover and Jones, 1999). A strawberry homolog of the TED2 gene (E149) was identified as ripening regulated. The TED2 gene previously shown to be expressed in developing vasculature has homology to crystallin, a quinone oxidoreductase (Demura and Fukuda, 1994). It was also previously shown that plant γ-cystallins play a distinct role in plant oxidative stress tolerance (Babiychuk et al., 1995).

Ripening-Regulated Genes in the Receptacle Are Induced by Oxidative Stress

Results from the first four experiments (I–IV) revealed more than a dozen putative stress related ripening-regulated cDNAs that were preferentially expressed in the receptacle. This prompted us to initiate a stress experiment to identify whether ripening-regulated genes identified through our microarray study could form part of a transcriptional program responsive to oxidative stress conditions, which could arise in the receptacle during the fruit maturation process. Oxidative stress conditions develop from reactive oxygen species that can be generated as a result of uncontrolled respiration and damaged electron flow in mitochondria, leading to the induction of stress- and detoxification-related gene expression (Leprince et al., 2000). Reactive oxygen species are natural by products of metabolism, and often result from exposure to free radical generating compounds such as natural quinones, xenobiotics, and pollutants (Babiychuk et al., 1995).

To impose oxidative stress conditions, white stage fruit were treated on-vine with the free radical generating compound 2,2′-azobis (2-amidinopropane) dihydrochloride (AAPH), a water-soluble substance that decomposes thermally, yielding two carbon-centered radicals, which subsequently react with oxygen to form peroxy radicals (Henkow et al., 1996). A concentration of 100 mm was selected for the AAPH treatment after performing two independent RNA gel-blot analyses using the strawberry ferritin cDNA as a probe (Fig. 4A), which is known to be induced under stress conditions (Deak et al., 1999). Gene expression in AAPH-treated fruit and those treated with buffer only was compared using the second generation microarrays (experiment V). The strong induction of ferritin transcript by oxidative stress was confirmed by this experiment, which revealed an additional 45 significantly induced cDNAs.

Figure 4.

Identification of ripening-regulated genes induced by oxidative stress. A, Ferritin gene expression in strawberry fruit after treatment with the free radical generator AAPH. Two RNA gel-blot experiments were performed: I, comparing expression in non-treated fruit (Ntr), fruit treated with buffer only (buffer), and treated with buffer containing 100 mm AAPH (100 mm); II, comparing expression in fruit treated with 10 and 100 mm AAPH (depicted as 10 and 100 mm, respectively). Gene expression in fruit treated with 100 mm AAPH and control fruit treated with buffer only was compared using the second generation microarray (experiment V). B, Graph representing cDNAs shown to be ripening regulated from microarray experiments (I–III) and oxidative stress induced from microarray experiment V. Gray bars represent genes showing differential and elevated expression in receptacle tissue compared with achenes, whereas black bars represent genes showing no differential expression between the two tissues as deduced from experiment IV. The number in parentheses after the putative definitions represent the number of the sequence contig in the case when more than one sequence showed a similar BLAST result but did not align in the sequence alignment. Genes marked with an asterisk did not show any change in expression upon application of exogenous auxin (as detected by experiment VI).

Of the 46 induced cDNA clones, 20 were detected in earlier microarray experiments (I–III) as ripening regulated (StrInd, Fig. 2B) and 17 of them showing significant homology to other genes in the public databases are represented in Figure 4B. Results derived from experiment (IV) showed that nine cDNAs (of the 20) displayed elevated expression in receptacle tissue compared with achene tissue, whereas the rest (11 cDNAs) did not show differential expression between the two tissue types. As depicted in Figure 4B, the developing fruit appears to respond to the oxidative stress treatment with an increase in the production of detoxifying enzymes (glutathione S-transferases, glutaredoxin, and quinone reductase-like protein [TED2]), protective enzymes (ferritin and 60S ribosomal protein L13E), and pathogenesis related proteins (harpin-induced protein and CAD). Overexpression of ferritin was previously shown to confer resistance to free radical toxicity in tobacco (Nicotiana tabacum) plants (Deak et al., 1999). Interestingly, ferritin, which was the most responsive cDNA clone in the oxidative stress experiment (9-fold induction), contains an electrophile response element with sequence similarity to electrophile response element motifs found in antioxidant response genes, such as glutathione S-transferases and quinone reductase (Tsuji et al., 2000), also identified in this study as ripening regulated and oxidative stress induced. The strawberry glutaredoxin (2.3-fold increase) is a homolog of a glutaredoxin from Ricinus communis, an abundant sieve tube exudate protein that was previously shown to prevent oxidative damage and to regulate the redox status of other sieve tube proteins (Szederkenyi et al., 1997).

As in other non-climacteric fruit, ethylene levels in strawberry fruit are very low compared with climacteric fruit such as tomato and banana. However, it has been observed that the levels of ethylene in strawberry decrease from the green to the white stage and then rise again in the red ripening stage (Perkins-Veazie et al., 1996). A strawberry 1-aminocyclopropane-1-carboxylic acid (ACC) oxidase putatively encoding the enzyme catalyzing the terminal step in the biosynthesis of ethylene was identified in this study. It showed 3-fold higher expression in the red stage compared with the turning stage and a strong induction due to the oxidative stress treatment (8-fold). This could suggest an alternative function for ethylene in strawberry as a potential regulator of a stress response induced by ripening rather than triggering ripening itself. Because it was previously reported that ethylene might play a role in the induction or progression of Z. elegans TE differentiation (Fukuda, 1997), we tentatively suggest its possible involvement in a stress response associated with TE differentiation in strawberry (such as, for example, PCD in the vascular bundles). However, the correlation between ACC oxidase gene expression and ethylene formation in strawberry remains to be clarified. It has been previously demonstrated that the activity of the preceding enzyme (ACC synthase) in the ethylene biosynthetic pathway is the key step in ethylene biosynthesis (Cazzonelli et al., 1998). In this study, we did not identify a strawberry cDNA encoding ACC synthase.

Correlating microarray expression data from the five experiments (I–V) has provided us with preliminary data to support our hypothesis that strawberry fruit may contain a transcriptional program responsive to stress (more particularly oxidative stress), induced during ripening.

Auxin and Gene Expression in Strawberry Development and Ripening

Early in strawberry research, it was demonstrated that the decline in auxin levels supplied from the achenes to the receptacle tissue during fruit development was associated with the onset of strawberry fruit ripening (Given et al., 1988) and triggered ripening by inducing expression of ripening-related genes (Manning, 1994, 1998; Medina-Escobar et al., 1997; Harpster et al., 1998; Moyano et al., 1998). Therefore, it is expected that by artificially treating green strawberry fruit on the vine with exogenous auxin, one would suppress the transcription of ripening-related genes. In this manner, auxin-repressed ripening-related cDNAs as deduced from the microarray experiment (VI) are those normally up-regulated in the receptacle during ripening.

We used the second generation microarray to perform a comprehensive examination of auxin action on gene expression and ripening processes in strawberry (experiment VI). By doing so, we could discriminate between auxin-dependent (repressed or induced) and -independent ripening genes and processes. Strawberry fruit on the vine were covered with paste with or without the auxin (1-naphthaleneacetic acid [NAA]). Auxin-treated fruit were morphologically similar to the non-treated fruit; however, they did not accumulate anthocyanins, indicated by lack of red coloration typical of ripe strawberries. Samples generated from treated and non-treated fruit were used for comparative hybridization on the second generation microarray.

Thirty ripening-regulated cDNAs were repressed by the auxin treatment (of 45 repressed in total) and 28 of them are depicted in Table III (cDNAs classified as “no hit” are not presented). Fruit ripening processes that appear to be auxin dependent include pathways related to pigmentation, stress/defense, cell wall metabolism, cell structure, fatty acid metabolism, and flavor/aroma synthesis (aldehyde, ester, and possibly terpene biosynthesis). Fourteen of the genes reported in this study as ripening regulated and auxin repressed were previously reported by Manning (1998) and others (e.g. Medina-Escobar et al., 1997; Harpster et al., 1998) and this provides an additional evidence to the quality and reliability of the microarray hybridization data obtained. A cDNA encoding a dioxygenase-like protein (H142) with unknown function showed the strongest repression (8.7-fold) by the auxin treatment. As expected, many cDNAs related to flavonoid metabolism and pigmentation were relatively strongly repressed by the auxin treatment (A104, A135, C179, F157, H61, and JB77). Interestingly, expression of the two receptacle- and cell structure-associated genes profilin (C122) and tubulin (G84) was strongly reduced by the auxin treatment (6.5 and 7.2-fold, respectively).

Twenty-five cDNAs were induced by the auxin treatment. None of them was ripening regulated and 19 did not show any change in expression during development, as deduced from microarray experiments (I–III). However, the remaining six were both induced by auxin and showed elevated gene expression in early to mid-strawberry fruit development compared with the red stage (green to white stage, Table III). Among them, we identified a pectin esterase-like protein (H163), which may be involved in early cell wall metabolism and fruit softening related to expansion, and another two cDNAs related to Met biosynthesis (5-methyltetrahydropteroyltri-Glu-homo-Cys methyltransferase [H117] and S-adenosyl-Met synthetase [JB67]).

Of the 80 individual ripening-regulated cDNA clones identified in the microarray experiments (I–III) and arrayed on the second generation microarray (RipR, Fig. 2B), 48 cDNAs (61.5%) did not show repression or induction by auxin and thus represent auxin-independent ripening processes. Nineteen selected cDNAs of the 48 are listed in Table IV and another 12 are depicted in Figure 4, as induced by the oxidative stress treatment (indicated by an asterisk). Although we have identified ripening-regulated and auxin-repressed cDNAs associated with certain metabolic processes such as fatty acid metabolism, cell wall, and stress, other cDNA clones related to the same processes appeared to be auxin independent. For example, ripening-regulated and cell wall-related cDNAs such as expansin (F22), extensin-like/Pro-rich protein (F93), and polygalacturonase (D15) did not show any change in expression as a result of the auxin treatment (Table IV), whereas pectate lyase (E30) and endo-1,4-β-glucanase (E80) were repressed (Table III). Our observation that not all ripening-regulated cell wall-related genes in strawberry are auxin dependent is supported by a previous study on the strawberry expansin gene FaExp2, which was reported to be auxin insensitive (Civello et al., 1999). Interestingly, FaExp2 expression was also not affected by ethylene treatment.

Table IV.

Identification of auxin-independent and ripening-regulated genes and processes in strawberry

PRI Clone	R:Aa	Homolog Definition (GenBank Accession No.) (E Value)b	Putative Process
A145	n.d.	Unknown (AL138641) (7e-49)	Unknown
B1	6.0	Aldo keto reductase (AC002292) (5e-14)	Unknown
B73	n.d.	Protein kinase (AC009325) (1e-33)	Signal transduction
B85	n.d.	Myo-inositol-1-phosphate synthase (U38920) (4e-60)	Stress
B109	13.1	Adenosine kinase (AJ012281) (3e-79)	Primary metabolism
C54	6.9	Stearoyl-acyl carrier protein desaturase (X56508) (9e-49)	Fatty acid metabolism
C94	n.d.	HMGR (L10390) (3e-59)	Primary metabolism
C112	6.0	Ascorbate peroxidase (AF022213) (2e-51)	Stress
D5	2.4	Dehydrin (U69633) (2e-13)	Stress
D15	4.9	Polygalacturonase precursor (U20431) (1e-19)	Cell wall
D120	14.1	Globulin-like protein (AF206627) (9e-24)	Unknown
D135	2.6	Unknown (AP002031) (3e-35)	Unknown
F22	12.5	Expansin (AF159563) (4e-54)	Cell wall
F93	3.7	Pro-rich protein (AF026382) (2e-34)	Cell wall
G128	4.7	Inorganic pyrophosphatase (AF093629) (3e-25)	Primary metabolism
G144	n.d.	Short chain alcohol dehydrogenase (AC009273) (9e-29)	Stress
H78	11.0	Annexin (AF188832) (3e-59)	Stress
H168	2.4	Unknown (AF325033) (9e-20)	Unknown
H182	6.2	Ubiquitin (X98063) (3e-51)	Stress

^a Expression ratio in microarray experiment (IV) comparing expression in receptacle (R) versus achenes (A). n.d., No difference in expression between achene and receptacle. 

^b Definition, accession no. (nucleotide sequence), and E value (see Altschul et al., 1990) of the first BLAST X homolog. 

CONCLUSION

In this paper, we have employed microarray technology to provide a comprehensive view of gene expression patterns during strawberry fruit development, in different tissues, under oxidative stress and auxin treatment conditions. The broad picture of gene expression obtained by our microarray analysis enabled new biological insights, which could not have been identified using conventional single-observation methodologies. Combining the expression data from six different microarray experiments resulted in three major findings in relation to: (a) a novel yet uncharacterized ripening process in strawberry namely the development of the vascular system, (b) the association between ripening-related gene expression and oxidative stress response, and (c) hormone (auxin)-dependent and -independent processes.

One of the intriguing outcomes emerging from our microarray data analysis was the parallelism in gene expression patterns between TE differentiation in Z. elegans and strawberry. Based on the experimental data, although at this stage mainly correlative, we would like to put forward the hypothesis that the development of the vascular system is a significant event coupled to fruit maturation. Whether vascular development in strawberry proceeds in the same way as in Z. elegans still remains to be established. This finding on the importance of gene expression in the vascular tissue of strawberry receptacle has some important implications concerning the possible function of genes identified earlier as ripening regulated in strawberry. Part of the genes might play a specific or additional role in the developing vasculature rather than only in the receptacle tissue itself.

The identification of 20 ripening-regulated cDNAs induced by an oxidative stress treatment implies that oxidative stress could be part of certain strawberry ripening processes. The genes identified in this study might be triggered to actively respond to the stress conditions and/or play a role in different ripening processes induced by stress. Stress may arise in the fruit during ripening from changes in osmotic potential due to the accumulation and storage of osmotically active substances (e.g. hexoses), or from abiotic or biotic factors. A potential source of stress could also be (possibly in addition to other sources) the lignifying vasculature. Recently, a basic peroxidase isozyme was located in the concentric array of the vascular bundles and in the vascular connections with the achenes in strawberry (Lopez-Serrano and Barcelo, 2001). Peroxidases are involved in the oxidation of phenolic compounds in cell walls, polymerization of lignin and suberin, and several other oxidation processes. Whether the activity of this peroxidase could contribute to oxidative stress conditions in the receptacle remains to be established.

Finally, we have identified novel ripening-induced genes that were either repressed by auxin or not affected by the auxin treatment, suggesting that another signal molecule(s) besides auxin may regulate the developmental ripening process in strawberry. A set of 25 genes was induced by the hormone; none of them were ripening regulated, whereas six showed high expression levels in early to mid-fruit development. High auxin levels are known to promote early fruit growth in strawberry (Nitsch, 1950) and our data provide supporting evidence at the level of gene expression to this early observation. Auxin influence on gene expression in early fruit development was also reported in grapes, another non-climacteric fruit (Davies et al., 1997).

The results presented demonstrate the complexity of the hormonal control of non-climacteric fruit ripening, and indicates that the ripening process is likely to be a collection of subprocesses differentially regulated yet coordinated into a general ripening program. Further experiments to examine gene expression and protein localization in the vascular bundles compared with cortical tissue, and in fruit treated with other phytohormones (e.g. ethylene, abscisic acid, gibberellins, and cytokinins), will provide additional valuable data on the genetic controls governing ripening in strawberry.

MATERIALS AND METHODS

Plant Material and Preparation of mRNA

For developmental microarray experiments, medium-size green fruit, white fruit with no sign of pigmentation, turning (fruit are partially pigmented), and red ripe stage fruit obtained from the domesticated strawberry (Fragaria × ananassa cv Elsanta) were used. For the comparison of receptacle and achenes, 5 kg of red ripe fruits were blended with water and the achenes sinking to the bottom of the beaker were collected and used for RNA isolation. Achenes were removed manually from frozen red ripe fruits, and the remaining receptacle tissue was used for the comparison with achene tissue. Total RNA was prepared as described by Schultz et al. (1994). For mRNA preparation, the mRNA purification kit (Amersham-Pharmacia Biotech, Uppsala) was used.

Production of First Generation Microarrays

The first generation microarrays were produced as described previously (Aharoni et al., 2000). In brief, the source of the probes arrayed was a red ripe strawberry fruit tissue cDNA library including the achenes. The library was constructed in the UNI–XR vector (Stratagene, La Jolla, CA). After mass excision, plasmid DNA from 1,701 strawberries picked randomly was extracted using the BioROBOT 9600 (Qiagen, Chatsworth, CA). The cDNAs were amplified by PCR using the T₃ and T₇ universal primers using the GeneAmp PCR system 9600 (Perkin Elmer, Foster City, CA). The primers contained a six-carbon amino modification (Isogen Bioscience BV, Maarssen, The Netherlands). PCR products were purified using the QIAquick PCR purification kit (Qiagen) and eluted in 100 μL of 0.1× Tris-EDTA, pH 8.0. Samples were dried to completion, resuspended in 7.5 μL of 5× SSC (approximately 1 mg mL⁻¹), and transferred to a 384-format plate to be subsequently used for spotting. Amplified cDNAs were spotted in duplicate onto silylated microscope slides (CEL Associates, Houston) using a 16-pin print head and a custom-built arraying robot. After arraying, the slides were air dried and stored in the dark.

Production of Second Generation Microarrays

The second generation array contained 384 probes: 356 strawberry cDNAs, 16 peach (Prunus persica) fruit cDNAs, one petunia (Petunia hybrida) cDNA, one NPTII gene, and 10 controls (five fragments of the firefly luciferase gene, two mouse (Mus musculus) cDNAs, and three probes not containing DNA). Amplification, purification, and arraying of the probes were performed as described for first generation microarray with a few modifications. Samples were resuspended in 10 μL of 5× SSC before arraying using the PixSys 5500 (Chartesian Technologies, Irvine, CA) including the ArrayIt, ChipMaker 3 microspotting device and pins (TeleChem, Sunnyvale, CA). Arraying was performed on amino-silane-coated slides (Corning, Corning, NY). Each array was printed a second time at the opposite side of the same slide. After printing, the microarrays were rehydrated above a beaker containing hot water for 5 s and then snap dried for 2 s on a hot plate (100°C). The DNA was then UV cross-linked to the surface by subjecting the slides to 20 mJ energy (Stratalinker, Stratagene).

Hybridization and Data Analysis

For first and second generation microarrays, hybridization, scanning and data acquisition, and statistical analysis were performed as described previously (Aharoni et al., 2000). Each of the microarray experiments was performed in a duplicate with the dyes reversed. For the first generation arrays, we used two separate slides and for the second generation arrays, the duplicate experiment was performed on the same slide using the duplicated arrays and hybridizing under separated coverslips. For the first three experiments, green/red, white/red, and turning/red, the threshold ratio for detection (minimum ratio for differential expression) was 2.60, 3.32, and 2.24, respectively. For the microarray experiments comparing achene and receptacle tissues, oxidative stress and auxin, the threshold ratio of detection was 1.97, 1.63, and 1.73, respectively (in all experiments significant at single test, P < 0.05).

RNA Gel-Blot Analysis

For RNA gel-blot analyses, 10 μg of glyoxal (1.5 m)-denaturated total RNA was electrophorized and blotted onto Hybond N⁺ (Amersham, Buckinghamshire, UK). After fixation (2 h at 80°C), blots were hybridized as described by Angenent et al. (1992). The hybridization probes were made by random-labeling oligonucleotide priming (Feinberg and Vogelstein, 1984) of the entire cDNAs. Blots were washed two times for half an hour each in 0.1× SSC and 0.1% (w/v) SDS at 65°C.

Immunolocalization of the Strawberry CAD Protein and Structural Studies

For the cytolocalization of the CAD polypeptide, tissues were fixed in ethanol:acetic acid (3:1, v/v), dehydrated through an ethanol-tertiary butanol series, and embedded in Paraplast Plus (Sherwood Med. Co., St. Louis). Five-micrometer microtome sections were mounted on slides covered with gelatin, deparaffinized in xylene, and rehydrated through an ethanol series, and blocked with 2% (w/v) nonfat dried milk in Tris-buffered saline. Immunological detection was performed using a primary anti-strawberry (clone F193) polyclonal antiserum (1:25), and a secondary anti-rabbit alkaline phosphatase-conjugated antibody (Sigma, St. Louis) (1:250). The reaction of alkaline phosphatase was developed with nitroblue tetrazolium and 5-bromo-4-chloro-3 indolyl-phosphate for 15 to 30 min. The sections were dehydrated through graded ethanols, cleared in xylene, and mounted in Entellan New (Merck, Rahway, NJ). An AH-2 photomicroscope (Olympus, Tokyo) was utilized for sample visualization and photography. Lignified structures were visualized by performing the Weisner reaction using phloroglucinol-HCl (Clifford, 1974).

Oxidative Stress Treatment

For the oxidative stress experiment, white stage fruits (attached to the plants, approximately 20 fruit of eight plants per treatment) were submerged in a solution containing 100 mm AAPH (Polysciences, Warrington, PA) dissolved in 10 mm MES buffer, pH 6.0, and 0.05% (w/v) Tween 20. Control fruits were submerged in the same buffer lacking AAPH. Fruit were submerged two times for 30 min in the solutions with a 17-h gap between treatments. Six hours after the second treatment, fruits were picked, immediately frozen in liquid nitrogen, and stored at −80°C until mRNA isolation. The AAPH concentration used in the microarray experiment was chosen from preliminary RNA gel-blot experiments, using RNA from fruit treated with buffer only and 10 and 100 mm AAPH. Four known stress-related genes served as probes in these experiments.

Auxin Treatment

For hormone treatment, we used NAA at a concentration of 0.5 mm in a lanolin paste containing 1% (v/v) dimethylsulfoxide. The paste was applied gently using a spatula to the entire fruit, on the vine, at the middle green stage of development (fruit grown on 20 plants were used). Seven days after the treatment, berries were picked, wiped clean of lanolin, and used for total RNA isolation (25 berries for each sample). Control fruit were treated in a similar manner except for the absence of NAA in the paste.

Sequence Analysis

cDNAs (1,100) of a total of 1,700 cDNAs were partially sequenced from the 5′ end before performing the microarray experiments. Other non-sequenced cDNAs, which showed differential expression in the microarray experiments, were sequenced using the Applied Biosystems (Foster City, CA) dye terminator cycle sequencing Ready Reaction kit and the 310 DNA sequencer. Comparison analysis of the sequences was conducted with the advanced basic local alignment search tool BLAST X (version 2.2.1) server (Altschul et al., 1990) and the National Center for Biotechnological Information (http://www.ncbi.nlm.nih.gov) nonredundant protein database. DNA and protein analyses was performed using Geneworks (IntelliGenetics, Oxford) and DNASTAR (DNASTAR Inc., Madison, WI). cDNA clones showing differential expression that could not be classified to any functional category were annotated as unknown or novel. The “unknown” category included sequences showing significant homology to genes with unknown function. The “novel” category included sequences showing no homology at all (no hit) or low homology to database sequences.