Microarray analysis of the abscission-related transcriptome in the tomato flower abscission zone in response to auxin depletion.

The abscission process is initiated by changes in the auxin gradient across the abscission zone (AZ) and is triggered by ethylene. Although changes in gene expression have been correlated with the ethylene-mediated execution of abscission, there is almost no information on the molecular and biochemical basis of the increased AZ sensitivity to ethylene. We examined transcriptome changes in the tomato (Solanum lycopersicum 'Shiran 1335') flower AZ during the rapid acquisition of ethylene sensitivity following flower removal, which depletes the AZ from auxin, with or without preexposure to 1-methylcyclopropene or application of indole-3-acetic acid after flower removal. Microarray analysis using the Affymetrix Tomato GeneChip revealed changes in expression, occurring prior to and during pedicel abscission, of many genes with possible regulatory functions. They included a range of auxin- and ethylene-related transcription factors, other transcription factors and regulatory genes that are transiently induced early, 2 h after flower removal, and a set of novel AZ-specific genes. All gene expressions initiated by flower removal and leading to pedicel abscission were inhibited by indole-3-acetic acid application, while 1-methylcyclopropene pretreatment inhibited only the ethylene-induced expressions, including those induced by wound-associated ethylene signals. These results confirm our hypothesis that acquisition of ethylene sensitivity in the AZ is associated with altered expression of auxin-regulated genes resulting from auxin depletion. Our results shed light on the regulatory control of abscission at the molecular level and further expand our knowledge of auxin-ethylene cross talk during the initial controlling stages of the process.


INTRODUCTION
Abscission, senescence and ripening are plant developmental processes that their timing is determined by tissue sensitivity to ethylene (Trewavas, 1986;Bleecker and Patterson, 1997;Zegzouti et al., 1999). The biological basis for this increased ethylene sensitivity is still not known, but it has been shown to be modulated also by other plant hormones. In abscission, the interplay between indole-3-acetic acid (IAA) and ethylene is well established (Abeles and Rubinstein, 1964;Sexton, 1997;Taylor and Whitelaw, 2001;Roberts et al., 2002). The generally accepted model is that a basipetal IAA flux through the abscission zone (AZ) prevents abscission by rendering the AZ insensitive to ethylene. Unlike various auxin-mediated physiological processes that are a result of transient and local changes in auxin levels (Woodward and Bartel, 2005), prevention of abscission has been found to require a continuous and constant polar supply of auxin to the AZ (Taylor and Whitelaw, 2001). If the source of IAA is removed, the AZ becomes sensitized to the action of ethylene and abscission commences (Rubinstein and Leopold, 1963;Abeles and Rubinstein, 1964;Addicott, 1982;Sexton and Roberts, 1982;Meir et al., 2003Meir et al., , 2006. Accordingly, the activities of cell wall degrading enzymes, including cellulase (Cel), polygalacturonase (PG), expansin (EXP) and xyloglucan endohydrolase endotransglycosylase (XET) have been shown to increase dramatically with the onset of abscission (Lashbrook et al., 1994;Kalaitzis et al., 1997;Agusti et al., 2008Agusti et al., , 2009Cai and Lashbrook, 2008;Roberts and Gonzalez-Carranza, 2009).
The molecular mechanisms leading to increased tissue sensitivity to ethylene in response to IAA deficiency in the AZs are still unknown. Some insights were provided by a study of abscission in Mirabilis jalapa, identifying differentially regulated genes in AZs (Meir et al., 2003, 2006. Auxin depletion led to down-regulation of several auxin-responsive genes, while application of auxin prevented their decrease. Some genes, up-regulated by auxin depletion, were homologous with known ethylene-responsive (ER) genes such as peroxidase (PER) and ER6. Since the experiments were done in the presence of the ethylene action inhibitor, 1-methylcyclopropene (1-MCP), it is likely that these effects are independent of changes in ethylene perception. Based on this study, our hypothesis postulates that acquisition of ethylene sensitivity in the AZ is associated with altered expression of auxin-regulated genes. Gonzalez-Carranza, 2009). A novel cell wall-related gene, XET-BR1, was found in our analysis to be up-regulated specifically in the AZ after flower removal (Fig. 4D) in a similar pattern as TAPG4 (Fig. 4C). This gene encodes for xyloglucan endotransglycosylase that was found before to be regulated by brassinosteroid treatment (Koka et al., 2000). The brassinosteroid receptor BRI1 is a receptor kinase that transduces steroid signals across the plasma membrane and has an extracellular domain containing 25 Leu-rich repeats (LRRs) (Wang et al., 2001).
The possible role of this domain in regulating abscission will be discussed later on.
To measure their mRNA expression levels, a semi-quantitative real-time PCR (SQ-PCR) analysis was performed for TAPG1, TAPG2, TAPG4, Cel1 and XET-BR1 (Fig. 4G) in order to compare it with the results obtained in the microarray analysis. RNA from AZ and NAZ tissues, sampled in experiments independent from those used for the microarray assays, was extracted and used for the SQ-PCR assay. The expression pattern revealed by this analysis (Fig. 4G) was in good agreement with the expression patterns obtained by the microarray experiments ( Fig. 4, A to E). Thus, these results confirm that the microarray analysis of gene expression at the AZ reflects true molecular events induced by flower removal, as detailed below. In addition, expression analyses using SQ-PCR were performed for few additional genes including: two ethylene responsive factors (ERFs) -ERF2 and ERF1; four novel AZspecific genes highly expressed in the AZ at zero time (before flower removal) -TKN4, PHANTASTICA and OVATE; and a gene which encodes for a protein phosphatase that is upregulated 4 h after flower removal in the AZ without being affected by the 1-MCP pretreatment (Supplemental Fig. S3). In this case, as well, results from the SQ-PCR expression analysis (Supplemental Fig. S3G) confirmed the expression patterns of the same genes from the microarray analysis (Supplemental Fig. S3, A to F). While the pattern of gene expression in the AZ measured by SQ-PCR generally matched nicely the expression measured by the microarray analysis, there was some discrepancy in the expression data in the NAZ obtained by the two methods. For example, the expression of Cel1 and XET-BR1 genes in the NAZ analyzed by SQ-PCR showed a slight increase at the late time points (Fig. 4G), while the microarray analysis did not indicate such an increase ( Fig. 4D and E). This difference is not significant for our analysis and may be ascribed either to biological variations or to the higher sensitivity of the SQ-PCR method.
An additional validation of the microarray analysis was performed using quantitative real time PCR (Q-RT-PCR) for seven other genes: ERT10, ERF4, TKN4, TAGL2, HB-13, Homeobox-Leu zipper and TPRP-F1. RNA was extracted from either the AZ or the NAZ tissues originating from two biologically independent experiments, and the measured expression pattern using Q-RT-PCR was compared to the microarray data analysis. Similar results were obtained by the two different methods for these seven genes (Supplemental Figs. S4 to S6), with highly significant regression coefficients obtained between the Q-RT-PCR and the microarray data (Supplemental Figs. S4, E-F; S5, G to I; S6, E-F), thereby validating the microarray analysis. The Q-RT-PCR analysis was also used to validate the microarray data obtained in the experiments examining the effects of 1-MCP pretreatment or IAA application on expression of the eight following genes in the AZ: ERT10, ERF4, Homeobox-Leu zipper, Protein phosphatase 2c, Proline transporter, Putative PK, PK7 and Ubiquitin-protein ligase (RGLG2). The results presented in Supplemental Figs. S17 to S20, and the high significant correlations obtained between the microarray and PCR expression data for these eight genes (Supplemental Figs. S17, E-F; S18, E-F; S19, E-F; S20, E-F), further support the validity of the microarray data.
Taken together, the agreement found among the expression results obtained by the different methods, including SQ-PCR, Q-RT-PCR, the published information for several genes and the microarray analyses with the high reproducibility, confirmed the reliability of this microarray-based results of the tomato AZ gene expression.
Recently, a stamen AZ transcriptome profiling study in Arabidopsis was reported, which followed the stamen abscission global gene expression during flower development (Cai and Lashbrook, 2008). This study coupled laser capture microdissection of Arabidopsis thaliana stamen AZs tissue with GeneChip Microarray profiling, to reveal the stamen AZ transcriptome responding to developmental shedding cues. This study resulted in the classification of the identified differentially expressed genes into eight clusters according to their expression. Among the identified genes, were genes with cell wall modification functions including: EXPs (three genes), extensin (EXT4), glycosyl hydrolase (three genes), pectin methylesterase (PME) (five genes), PER (11 genes), PG (three genes) and xyloglucan endotransglycosylase/hydrolase (XTH) (four genes). Differential expression of genes associated with cell wall metabolism during abscission was demonstrated also during ethyleneinduced abscission of citrus leaves (Agusti et al., 2008(Agusti et al., , 2009). These included genes for two different PGs, cellulase and two different XTHs -XTH1 and XTH2, which were over expressed in leaf AZ-enriched tissue (Agusti et al., 2008). The expression of cell wall-metabolism related genes was also studied using real-time PCR and Affymetrix GeneChip hybridization in soybean leaf AZ taken from explants exposed to ethylene (0, 1 or 2 days). This analysis showed strong up-regulation of Cel1, Cel6, Cel9, pectate lyase (PL1), PL2, PG9, PG11, EXP3, expressed sequence tags (ESTs), which are homologous to ILR3, had the same peak of expression 4 h after flower removal in both the AZ and the NAZ tissues, and 1-MCP pretreatment even further increased their expression (Fig. 5,B and C). IAA can exist in the cells either as the hormonally active free acid or in a bound form in which the carboxyl group is conjugated either to sugars via ester linkages or to amino acids or peptides via amide linkages (Cohen and Bandurski, 1982;Bartel et al., 2001;Woodgard and Bartel, 2005). IAA conjugates have a role in storage, transport and compartmentalization of IAA. The IAA conjugates have auxin activity when applied exogenously, and have a physiological activity in regulating different developmental processes such as seed germination and root elongation. This activity is mediated by the free IAA which is released following hydrolysis of IAA conjugates (Cohen and Bandurski, 1982;Meir et al., 1984;Bartel et al., 2001;Woodgard and Bartel, 2005). Using mutant screens in Arabidopsis, a family of ILRs was identified and characterized (LeClere at al., 2002;Rampey et al., 2004;Woodgard and Bartel, 2005). ILR1 protein specifically hydrolyses IAA-Leu (Bartel and Fink, 1995), while ILR3 specifically hydrolyses IAA-Ala (Davies et al., 1999). The microarray results suggest that the flower AZ tissue can sense reduction in auxin flow and react by increasing ILRs needed for the release of stored conjugated auxins. 1-MCP pretreatment had no effect on ILRs initial expression, which was low before flower removal (Fig. 5). The effect of 1-MCP pretreatment on expression of ILRs after flower removal suggests that a cross-talk mechanism between ethylene and auxin exists in the AZ. The observation showing that IAA application after flower removal prevented the increase in ILR1 and ILR3 expression (Supplemental Fig. S7) supports this suggestion. If indeed the increase in expression of ILRs after flower removal results in hydrolysis of IAAconjugates and release of active IAA, it does not seem to be sufficient to keep the AZ insensitive to ethylene and to prevent abscission. This is supported by the findings showing that exogenous IAA application on the cut end after flower removal is still required for inhibition of abscission (Fig. 2 and Roberts et al., 1984). The observed effect of flower removal on expression of other auxin-related genes supports the conclusion that auxin is decreased in the AZ after flower removal, as will be shown and discussed further on.
It should be noted that in abscising organ systems, a continuous auxin flow through the AZ is required for preventing the increase in ethylene sensitivity and abscission (Taylor and Whitelaw, 2001), which presumably also results in continuous expression of Aux/IAA genes.
Indeed, results from our microarray analysis show that seven Aux/IAA genes were downregulated following flower removal (Fig. 6, A to G). Thus, expression levels of IAA1 (Fig. 6A), 6D) and IAA10 (Fig. 6G) decreased sharply within 2 h after flower removal, and remained low. The decrease in expression of these Aux/IAA genes was similar in the AZ and NAZ tissues and was not affected by 1-MCP pretreatment (Fig. 6,A to D and G). This indicates that the decrease in the Aux/IAA gene expression as a result of IAA depletion following flower removal is neither AZ-specific nor affected by ethylene. The only two exceptions from this general pattern were observed for IAA8 (Fig. 6E) and IAA9 (Fig. 6F), the expression of which decreased more gradually within 8 h compared to the sharp and immediate decrease in the NAZ tissue (Fig. 6F). The expression of IAA9 decreased more slowly, as it remained high specifically in the AZ within 2 h after flower removal and only subsequently decreased. On the other hand, pretreatment with 1-MCP resulted in a sharp and immediate decrease of IAA9 expression within 4 h (Fig. 6F). Among the tomato Aux/IAA genes, the IAA3 is an interesting gene in functional terms, as it is thought to represent a molecular link between ethylene and auxin signaling. This hypothesis was suggested by the results showing that downregulation of IAA3 expression in tomato fruit resulted in both auxin and ethylene-related developmental defects (Chaabouni et al., 2009).
One tomato EST registered in the GenBank as an auxin-regulated protein which is expressed in roots (NCBI, AF416289.1; Zurek DM, Franke P, Rayle DL, 2001), was found in our analysis to be transiently up-regulated 2 h after flower removal in both AZ and NAZ samples (Fig. 6H). This gene is probably an auxin-repressed gene, since its expression was increased very dramatically after flower removal. This protein is probably not related to abscission.
It is now well established that the Aux/IAA proteins are actually repressors of auxininduced transcription, and auxin promotes the degradation of this large family of transcriptional regulators (TFs), leading to diverse downstream effects (Worley et al., 2000;Gray et al., 2001;Overvoorde et al., 2005). This allows Auxin Responsive Factors (ARF) proteins to bind to the Auxin Responsive Elements (ARE) within the promoters and either activate or repress expression of target genes. Rapid induction of the Aux/IAA genes is a response to the reduced levels of the Aux/IAA proteins, which ensures a tightly controlled transient response to changes in auxin concentrations via a negative feedback (Leyser, 2002;Woodward and Bartel, 2005).
Genetic evidence supporting a role for auxin in regulating Arabidopsis floral organ shedding has been elusive. Recently, functional studies of ARF2, ARF1, ARF7 and ARF19 suggested that these transcriptional regulators act with partial redundancy to promote senescence and floral abscission (Ellis et al., 2005;Okushima et al., 2005a,b). Mutations in ARF2 alone delayed the onset of floral senescence and organ shedding, which are further inhibited by loss of ARF1 activity, or by the loss of both ARF7 and ARF19 activities (Ellis et al., 2005). Changes in auxin gradients across AZs might promote abscission, and one possibility is that the activities of ARF2, ARF1, ARF7 and ARF19 might be modulated by similar gradients in floral organs (Taylor and Whitelaw, 2001;Ellis et al., 2005). Changes in these activities might also play essential roles in auxin-mediated plant development by regulating both unique and overlapping functions of ARF gene family members in Arabidopsis (Okushima et al., 2005b). Since the expression levels of ARF genes were not affected by flower removal (data not shown), it is suggested that the abscission regulation is mediated via an effect on Aux/IAA expression.

Ethylene Biosynthesis-Related Genes
Eight genes related to different steps of the ethylene biosynthetic pathway were modified for their expression following flower removal (Fig.   7 ). These genes are involved in: Met biosynthesis -homocystein S-methyltransferase (  . 7D), seemed to be more relevant for the abscission process. This suggestion is based on its expression which was induced during 8 to 14 h after flower removal, was inhibited by 1-MCP pretreatment and was highly AZ-specific (Fig. 7D). This conclusion is in accordance with the widely accepted view that Met and SAM production are not limiting steps in the ethylene biosynthesis pathway, and therefore, probably have no controlling role (Kevin et al., 2002). The late induction of this ACS gene also suggests a second increase in ethylene production, coinciding with the abscission development and execution. To the best of our knowledge, the only report on ethylene production during abscission of tomato flower explants was reported by Roberts et al. (1984). Although not showing directly this expected second increase in ethylene, these authors observed a consistent reduction of ethylene production by aminoethoxyvinylglycine (AVG) treatment that delayed abscission, but was not as effective as the 1-MCP pretreatment, which completely inhibited abscission (Fig. 2). ACO1 expression sharply increased 2 h following flower removal and then leveled off during the subsequent 4 to 14 h, showing higher expression in the AZ during this period (Fig. 7H). The initial increase in expression of ACO1 was not affected by 1-MCP pretreatment, while 1-MCP partially inhibited the later expression during 4 to 14 h, which still remained high (Fig. 7H). This observation suggests that ACO1 does not serve as a controlling factor of ethylene biosynthesis in the AZ during tomato flower abscission.

Ethylene Signal Transduction-Related Genes
Our hypothesis postulates that acquisition of ethylene sensitivity in the AZ is associated with alteration in the expression of auxin-regulated genes. Therefore, we have examined the effect of flower removal leading to auxin depletion, on expression of genes related to ethylenesignal transduction pathway. The microarray results show that out of six genes encoding for the tomato ethylene receptors (Klee, 2002(Klee, , 2004, the expression of five of them was not affected by flower removal (data not shown). Interestingly, ethylene resistant 4 (ETR4) expression increased transiently following 2 h and again at 8 to 14 h ( Fig. 8A) following flower removal, when abscission has already initiated (Fig. 2). Both the early and the late increases in ETR4 expression were inhibited by 1-MCP pretreatment (Fig. 8A). The late increase of ETR4 expression was AZ-specific ( Fig. 8A), implying that ETR4 is directly involved in the late stages of the abscission process. Expression of constitutive triple response 1 (CTR1) (Fig. 8B) was affected by flower removal and by 1-MCP pretreatment in a very similar pattern to that of ETR4 (Fig. 8A). The similar patterns of ETR4 and CTR1 expression observed following flower removal and in response to 1-MCP pretreatment might be due to the functional link which exists between the two encoded proteins. Both yeast two-hybrid system and in vitro biochemical experiments in Arabidopsis indicate that the predicted transmitter domain of ETR1 can interact directly with the regulatory domain of CTR1 (Clark et al., 1998;Gao et al., 2003;Huang et al., 2003;Binder, 2008). The late and AZ-specific increase in expression of both ETR4 and CTR1 (Fig. 8, A and B) suggests that this receptor complex is required for function in the late stages of the abscission process. However, the results presented in Fig. 2 demonstrate that 1-MCP, which was bound irreversibly to the available ethylene receptors before flower removal, prevented abscission for a relatively long period of time following flower removal. This suggests that the acquisition of ethylene sensitivity at the AZ in response to flower removal cannot be gained via modification of the ethylene receptors. It seems, therefore, that ETR4 and CTR1 are involved in the late stages of the abscission process, but do not play a regulatory role in acquisition of sensitivity to ethylene in the AZ.
Another set of genes associated with ethylene signaling are the ethylene responsive factor (ERF) which are TF genes. Analysis of the promoters of ethylene-responsive genes revealed a common cis-acting ethylene responsive element called the GCC box (Fujimoto et al., 2000). This element was shown to be necessary and sufficient for ethylene regulation in a variety of plant species. The first type of the trans-acting factors isolated from tobacco which bind to the GCC box was termed ethylene-responsive element binding proteins (EREBPs) (Ohme-Takagi and Shinshi, 1995). EREBPs play a role in plants' responses to phytohormones, pathogens attack and environmental stresses (references cited in Hu et al., 2008). Five different ERF proteins were described for Arabidopsis: AtERF1, AtERF2 and AtERF5, which function as activators of GCC box-dependent transcription, and AtERF3 and AtERF4 which act as repressors. The AtERF genes were differentially regulated by ethylene and abiotic stress conditions, via the ethylene insensitive 2 (EIN2)-dependent or EIN2-independent pathways (Solano et al., 1998;Fujimoto et al., 2000;Nakano et al., 2006). Over-expression of rice OsERF1 in Arabidopsis resulted with up-regulation of the expression of two known ER genes, plant defensin (PDF1.2) (low-molecular-weight Cys-rich 77) and b-chitinase (Hu et al., 2008).
Our microarray analysis revealed five different ERF genes, the expression of which was altered following flower removal ( Fig. 8C-8G). The homologue gene to AtERF4 repressor was down-regulated early (2 h) following flower removal, and remained at this low expression throughout the subsequent period at 4 to 14 h (Fig. 8G). This decrease in expression was not affected by 1-MCP pretreatment and seemed to be down-regulated even more in the AZ than in the NAZ tissue at 4 h (Fig. 8G). The other ERFs tended to transiently increase in expression after flower removal: ERF2 increased early (2 h) and transiently and was not affected by 1-MCP pretreatment (Fig. 8E); ERF3 increased early (2 h) and late (14 h) and its later increase was affected significantly by 1-MCP pretreatment (Fig. 8F); ERF1b increased early (2 h) and transiently and was affected by 1-MCP pretreatment (Fig. 8C); ERF1c expression increased early (2 h) and transiently and increased again at 8 to 14 h following flower removal; this increase was highly AZ-specific and was not inhibited by 1-MCP pretreatment (Fig. 8D).
Based on its expression pattern, which was highly AZ-specific at 14 h (Fig. 8D), and the prevention of its increased expression by IAA application (Supplemental Fig. S11B), ERF1c can be considered as a good candidate for encoding an ERF involved in abscission regulation.
The linkage between ERFs and auxin signaling is further supported by published results obtained in peaches, showing that the active ethylene and auxin signaling cross-talk throughout fruit development and ripening is mediated by ERFs and the Aux/IAA genes (Trainotti et al., 2007). Additionally, LeERF1 was reported to mediate the ethylene signals in tomato, as it was positively related with ethylene triple response, plant development and fruit ripening and softening (Li et al., 2007).

Other Ethylene-Responsive Genes
The expression of six more ethylene-responsive genes was modified following flower removal ( Fig. 9). ER1 and ER49 are tomato ripening-related genes with yet unclear functions.
ER49 was suggested to function as a post-transcriptional regulator (Zegzouti et al., 1999). Our microarray data show that ER1 (Fig. 9A), and ER49 ( Fig. 9C) were up-regulated and downregulated, respectively, after flower removal. This effect on expression was maintained throughout the abscission process, was not affected by 1-MCP pretreatment and the expression was not AZ-specific (Fig. 9, A and C). ER5 is a tomato ripening-associated gene, and its expression increases in mature green and breaker fruit development stages as the fruit becomes sensitive to ethylene, and is also activated by ethylene treatment (Zegzouti et al., 1999). ER5 expression increased early and transiently after flower removal, and this increase in expression was not affected by 1-MCP pretreatment, and was three-fold higher in the AZ tissue compared to the NAZ (Fig. 9B). Another ripening-related gene, ERT10, was up-regulated transiently at 2 h and later at 8 and 14 h after flower removal (Fig. 9D). These early and late increases in expression were inhibited by 1-MCP pretreatment and IAA application (Supplemental Fig.   S12D), and the second increase was highly AZ-specific ( Fig. 9D), similar to the pattern observed for ACS gene (Fig. 7D). This suggests that it can serve as a good marker for ethylene response that is regulated by IAA levels in the AZ. It is interesting to note that the competence of tomato fruit to ripen and to respond to ethylene while undergoing the transition phases from a green fruit (which does not respond to ethylene) to a mature-green fruit (which is ethyleneresponsive), is very similar to the abscission process. Therefore, genes associated with tomato ripening and which are modified upon transition between these two ripening stages, such as ER5 and ERT10, may be significant to the general phenomenon of acquisition of ethylene sensitivity manifested in these two systems.
Chitinases are pathogenesis-related (PR) proteins that had been shown to be transcriptionally regulated by ethylene, and very often their induction is considered as a marker for ethylene activity (Broglie et al., 1989;Díaz et al., 2002;Hall and Bleecker, 2003;Taira et al., 2005), including in abscission systems (Butenko et al., 2006). Two chitinase genes were up-regulated in the AZ 2 h after flower removal and remained highly expressed during 14 h, while in the NAZ their observed increase of expression was only transient and peaked at 2 h after flower removal (Fig. 9, E and F). The increase in expression of the gene coding for basic endochitinase ( Fig. 9E) was inhibited by 1-MCP pretreatment at all time points. On the other hand, for the gene encoding chitinase class II, 1-MCP pretreatment inhibited only the late (8 and 14 h) high expression (Fig. 9F). The late high expression of both chitinase genes was AZspecific. The results suggest that the early (2 h) increase in the chitinase expression which is not AZ-specific, is a wounding response which is transient only in the NAZ.

Effect of Flower Removal on Expression of Transcription Factor Genes
The regulation of gene expression at the transcription level has a profound role in the control of many biological processes. TFs are acting as major switches of regulatory cascades during development, and alterations in the expression of such genes may affect various developmental processes (Riechmann et al., 2000). Recently, the developments in identifying and assigning roles to various TFs involved in regulation of organ abscission and dehiscence processes were reviewed (Nath et al., 2007). Also reviewed recently is the association of TFs The results from our tomato flower AZ microarray show that the expression of different TF genes was affected in different ways by flower removal (Figs. 10 and 11). Two genes belonging to the KNOX family TFs were sharply down-regulated in the AZ 2 h after flower removal (Fig. 10, A and B). The gene showing homology to class I knotted-like homeodomain gene was expressed similarly in the AZ and the NAZ, was down-regulated in a similar rate in both tissues and was not affected by 1-MCP pretreatment (Fig. 10A). On the other hand, TKN4 was expressed initially and before flower removal three-fold higher in the AZ compared to the NAZ tissue, and 1-MCP pretreatment slowed moderately the rate of its expression reduction in the AZ (Fig. 10B). The AZ cells differ considerably from NAZ cells since they are nondifferentiated cells, suggesting that cell growth and differentiation is arrested at an early stage in the AZ (Van Nocker, 2009). It was shown that cells in the shoot apical meristem (SAM) are prevented from differentiation through the activity of the KNOX family of TFs. For example, the closest Arabidopsis KNOX TF (At1g62360; Fig.10A) which is SHOOTMERISTEMLESS (STM), was shown to be required for SAM formation during embryogenesis (Long et al., 1996), It was speculated by Van Nocker (2009) in his review that the apparent lack of development and differentiation of AZ cells may result from the persistent expression of the KNOX gene in this region. Our results support this speculation, as one KNOX gene, knotted TKN4, was preferentially expressed in the AZ, and this expression decreased 2 h after auxin depletion in the AZ from flower removal (Fig. 10B). The fact that IAA application after flower removal prevented the decrease in the expression of both knotted genes (Supplemental Fig.   S13, A and B) suggests that these increases resulted from auxin depletion. A previous study showed a connection between the expression of class I knotted-like gene, (STM) and auxin transport in the SAM (Heisler et al., 2005). It was shown that cycles of auxin build-up and depletion, caused by rapid reversal in the polarity of the auxin efflux carrier PIN1, accompanied and may directed different stages of primordium development (Heisler et al., 2005). On the other hand, the possibility that STM may act upstream to PIN1 to influence its behavior, was also suggested (Heisler et al., 2005).
The expression of three Homeobox-Leu zipper TF genes was down-regulated within 2 h after flower removal and remained low later on ( Fig. 10, C, D and E). 1-MCP pretreatment delayed the reduction of one of these genes (Fig. 10E), and the reduction of one was more AZspecific ( Fig. 10D). Three TGA-type basic Leu zipper TFs were suggested to be involved in abscission and to regulate the expression of abscission-related genes, as indicated by their binding to bean abscission cellulase promoter (Tucker et al., 2002). The promoter of this cellulase gene includes a cis DNA element that can function both in negative or positive regulation of the gene (ERF). Based on the observed reduction in their expression following flower removal found in our analysis, it is possible that the three Homeobox-Leu zipper TF encoding genes ( Another TF gene that was highly expressed in the AZ before flower removal and was sharply down-regulated in the AZ after flower removal is a basic helix-loop-helix (bHLH) (Fig.   10F). The closest homologous gene in Arabidopsis (At3g26744; Fig. 10F) is SCREAM/ICE1, which was reported to be involved in regulation of freezing tolerance and stomata differentiation in the epidermis (Kanaoka et al., 2008). A myc/bHLH TF ALCATRAZ (ALC), expressed in the valve-replum margin of Arabidopsis siliques, was found to have an important role in dehiscence, as indicated by the consequence of its inactivation with a disruption of dehiscence and the separation of valve cells from the replum (Rajani and Sundaresan, 2001).
An additional TF gene, bZIP, had a different pattern of expression, showing initially a transient down-regulation until 4 h after flower removal, followed by a continuously increased expression later on. This expression was AZ-specific and was inhibited by the 1-MCP pretreatment (Fig. 11A).
The expression of a gene encoding for a TF containing APETALA2 (AP2) domain was transiently up-regulated specifically in AZ without any effect of the 1-MCP pretreatment (Fig.   11B). AP2 plays an important role in the control of Arabidopsis flower and seed development, and encodes a putative TF that is distinguished by a novel DNA binding motif referred to as the AP2 domain (Okamuro et al., 1997). It has also been reported that expression of ERF genes, including Arabidopsis thaliana ethylene-responsive element binding protein (AtEBP), was regulated by the activity of AP2, a floral homeotic factor (ERF). Over-expression of AtEBP caused up-regulation of AP2 expression in leaves. AP2 expression was affected by EIN2, but was not regulated by ethylene treatment (Ogawa et al., 2007). Actually ERF2 also contains a conserved AP2 domain. The role of this gene in the regulation of flower abscission remains to be examined.  et al., 2000). STK (SEEDSTICK), that encodes a MADS domain TF known to be required for seed abscission (Pinyopich et al., 2003), is closely related to an AGAMOUS SHATTERPROOF -SHP1 and SHP2 which are required for silique dehiscence (Liljegren et al., 2000, Pinyopich et al., 2003. The WRKY is a super family of TF proteins with up to 100 representatives in Arabidopsis, which are highly divergent and are categorized into distinct groups possibly reflecting their different functions (Eulgem et al., 2000;Eulgem and Somssich, 2007). WRKY factors hold central positions mediating fast, positive and negative regulation of disease resistance. Two WRKY genes were found in our analysis to be up-regulated transiently at 2 h after flower removal, followed by an additional increase at 8 and 14 h (Fig. 11, E and F), which coincides with abscission development (Fig. 2). The first increase of both WRKY1 and WRKY IId-1 genes was not AZ-specific and was not affected by the 1-MCP pretreatment. However, the second rise of both genes was inhibited by the 1-MCP pretreatment, while only WRKY lld-1 expression was AZ-specific (Fig. 11,E and F). This suggests that the first transient increase of these WRKY genes is a wounding response, while the second increase, mainly that of WRKY lld-1, may be involved in regulation of a pathogen defense response in the separation layer. In accordance with our observation, specific expression of AtWRKY33, which shows high homology to WRKY1 and AtWRKY6 encoding genes, was described to occur in Arabidopsis at the flower base around the AZs of petals, sepals and stamens (Robatzek and Somssich, 2001;Lippok at al., 2007).
Overall, our results show differential expression of genes coding for TFs belonging to different families, including ARF, Aux/IAA, KNOX, Homeobox Leu zipper, bHLH, AP2, NAC (AY498713, Supplemental Table S2.4C), AGL and WRKY in tomato pedicel AZ, which exhibit different patterns of expression after flower removal. Different members of all these TF families were shown to be expressed in plants AZ or DZ, and were suggested to participate in different sub-processes of abscission or dehiscence, such as the development of the AZ, the execution of AZ separation, and regulation of defense-related processes in the abscission layer.
In some of these TF encoding genes the changes in transcript levels are observed early, such as 2 h after auxin depletion due to flower removal. This quick response may indicate their involvement in the early regulatory events associated with the development of ethylene sensitivity in the AZ. The exact roles of these regulatory factors in pedicel abscission responding to auxin depletion due to flower removal remain to be established.

Effect of Flower Removal on Expression of Some Other Regulatory Genes
The expression of a gene coding for a Leu-rich repeat trans-membrane receptor-like kinase (LRR-RLK) was found to be down-regulated specifically in the AZ 8 and 14 h after flower removal (Fig. 12A). HAESA (HAE) and HAESA-Like 2 (HLS2) that serve as receptors for the inflorescence deficient in abscission (IDA/IDL), were identified as receptor-like kinases (RLKs) (Cho et al., 2008;Stenvik et al., 2008). RLKs are components of signal transduction pathways that elicit cellular responses to extracellular information. In plants, the RLKs have been implicated in prevention of self-pollination, pathogen response, hormone perception and signaling and plant development and defense responses (Becraft, 1998;Lease et al., 1998). A Ser/Thr protein kinase (PK) encoding gene, PK7, was up-regulated specifically in the AZ at 4 and 14 h after flower removal, and 1-MCP inhibited this increase (Fig. 12B). This pattern of induction follows the progress of the abscission process (Fig. 2) and matches the pattern of expression induction found for the cell wall modifying genes: TAPG1, TAPG2, Cel1 and Cel5 (Fig. 4, A, B, E and F, respectively). These results suggest the possibility that PK7 is involved in abscission regulation, similar to our earlier observation regarding the MADS-box protein encoding gene, TAGL12 (Fig. 11C). Similar to PK7, some other kinases were also found to be involved in the abscission process. This is based on the data showing that the expression of a gene (AF332960) coding for the auxin-regulated dual specificity cytosolic kinase was found to be up-regulated in the AZ 2 h after flower removal (Supplemental Table   S3.4C). The expression of another gene (BM410830) coding for the PK/peptidoglycan-binding LysM domain-containing protein was also up-regulated in the AZ following flower removal (Supplemental Table S3.4C), as well as that of a Putative PK (Supplemental The expression of the AGO1 gene, encoding for argonaute-like protein, was rapidly down-regulated within 2 h after flower removal, and remained low later on. This decrease in AGO1 expression was not AZ-specific, was not affected by IAA application (Supplemental The gene encoding for a Pro-rich protein, TPRP-F1, was found to be specifically expressed in the AZ tissue at a high level before abscission induction, but was dramatically inhibited after flower removal (Fig. 12D). Pretreatment with 1-MCP reduced to some extent the TPRP-F1 transcript initial level in the AZ, but did not have any effect on its decrease once the flower was removed (Fig. 12D). The TPRP-F1 was originally identified as a gene encoding for a Pro-rich protein preferentially expressed in young tomato fruit (Salts et al., 1991). While the specific functions of TPRP-F1 and related Pro proteins in plants are not yet clear, studies focusing on various members of this plant gene family indicate functions related to different developmental aspects or responses to environmental factors (Goodwin et al., 1996;Holk et al., 2002;Battaglia et al., 2007). In accordance with our observations, a gene encoding for a Prorich protein was previously identified to be up-regulated specifically in the DZ of Brassica napus pods during dehiscence (Coupe et al., 1994).

Effects of IAA Application After Flower Removal on Expression of Genes Modified by Flower Removal
Application of IAA to the cut surface of the remaining tissue after flower removal nullified pedicel abscission during 38 h after flower removal (Fig. 2). Indeed, IAA application clearly inhibited during the late (8-14 h) time points after flower removal the increased expression of genes encoding for cell wall modifying enzymes (Fig. 13), which are known to be induced in the AZ following induction of the abscission process, Thus, IAA supplementation completely inhibited the expression of TAPG1 (Fig. 13A), TAPG2 (Fig. 13B), Cel1 (Fig. 13E), and Cel5 (Fig. 13F) at all time points after flower removal, and prevented the tremendous increase in their expression induced by flower removal, which is AZ-specific (Fig.   4). This further confirms the role of IAA in preventing organ abscission. Similarly, this increase in expression of these genes, induced by flower removal, was also prevented by 1-MCP pretreatment (Fig. 4A, B, E, F). On the other hand, IAA treatment had no effect on the early (2-4 h) increase in TAPG4 expression in the AZ, but it reduced TAPG4 expression during the 8-14 h after flower removal (Fig. 13C). This lack of IAA effect on TAPG4 expression during the early phase after flower removal does not contradict the role of IAA in inhibiting abscission, as TAPG4 induction was detected much earlier than TAPG1 and TAPG2 mRNAs (Kalaitzis et al., 1997) or Cel1 (Lashbrook et al., 1994) during both leaf and flower abscission in tomato, and as demonstrated in Fig. 4. The increase in XET-BR1 induced by flower removal in the AZ was prevented by IAA application at all time points after flower removal (Fig. 13D).
To the best of our knowledge, this is the first report showing an AZ-specific increase in XET-BR1 expression (Fig. 4D), which is also inhibited by IAA treatment (Fig. 13D). The role of XET-BR1 in the abscission process remains to be determined.
As demonstrated for the cell wall modifying enzymes, the effect of IAA application after flower removal on gene expression can help us to clarify which genes are likely to be regulated by IAA (Table II). Genes that were down-regulated following flower removal, with IAA application preventing this reduction, are probably genes positively induced by IAA.
Genes that were up-regulated by flower removal, with IAA application preventing this induction, are probably genes that are repressed by IAA. Genes whose expression is modified, either induced or inhibited, by flower removal but IAA application does not affect their expression are probably not regulated by IAA. Therefore, any observed modification in the expression of such genes could stem from events not directly related to the abscission process.
Such events may include reduced levels of signals originating from the flower, or events resulting from the wounding effect due to flower removal.
Among these two abscission-related sub-groups, sub-group 2II includes genes which are also ethylene-regulated downstream to IAA, and sub-group 2III includes genes which are specifically IAA-regulated in the AZ.
Group 3: Includes genes whose expression was transiently up-regulated early after flower removal, and was followed by a continuous second increase from 8 to 14 h, with IAA application preventing these changes. Therefore, genes in this group can be considered as IAArepressed genes, whose expression increased due to IAA depletion following flower removal.
The second increased expression of three of these genes, ACS6, ERT10 and WRKY lld-1, was prevented by the 1-MCP pretreatment (Figs. 7F, 9D and 11F), and was AZ-specific for three of them, ERF1c, ERT10 and WRKY lld-1 (Figs. 8D, 9D and 11F). This suggests that all these genes are involved in abscission regulation.

Group 4:
Includes two genes, ETR5 and ETR6, encoding for ethylene receptors, whose expression was not affected by flower removal, but IAA application induced their expression (Supplemental Fig. S10, B and C). It is suggested that these receptors are not involved in abscission regulation, but they may contribute to the IAA effect expressed in reducing the sensitivity of the AZ to ethylene. It is widely accepted that the level of the ethylene receptor proteins is negatively correlated with sensitivity to ethylene (Binder, 2008;Kevany et al. 2007).
Group 5: Includes genes whose expression was modified early during the 2-4 h period after flower removal, either transiently or continuously. IAA application did not affect this early gene modification, but affected the expression at the later period of 8-14 h after flower removal. Therefore, it is suggested that the modification in expression of these genes does not result from IAA depletion but rather from the wounding and/or other signals omitted due to flower removal. Indeed, this group includes genes whose products are involved in the regulation of the systemic signaling during wound response such as lipoxygenase and JA2 (Leόn et al. 2002, Howe and Schilmiller 2002, Schilmiller and Howe 2005, Wasternack et al. 2006, and genes associated with ethylene biosynthesis and signaling involved in wound ethylene responses (Saltveit and Dilley 1978, Boller and Kende 1980, Dourtoglou et al. 2000, Wasternack et al. 2006. The pattern of ethylene evolution in tomato AZ explants was characterized by a sharp peak at the AZ 2 h following flower removal, which then decreased to the basal level within the subsequent hour (Roberts et al., 1984). This initial burst of ethylene evolution induced by flower removal probably exhibits a typical wounding response. Our microarray results support the occurrence of such a wounding response, since numerous wound-related genes, such as lypoxygenase (LOX), wound-induced proteinase inhibitor I, jasmonic acid 2, protease inhibitor II, osmotin-like protein and wound-inducible carboxypeptidase, were up-regulated in our tomato system within 2 h after flower removal (cluster groups 1 and 4 in Supplemental Table S3 and Table S4; Supplemental Fig. S16).
The effects of IAA on the expression of some genes modified at the late phase of 8-14 h after flower removal, may operate via affecting ethylene sensitivity of the AZ. This possible explanation is supported by the findings showing that 1-MCP pretreatment also inhibited or prevented the modified expression for some of the genes classified in this group, such as: TAPG4 (Fig. 4C), Auxin-regulated protein (Fig. 6H), SAM synthase (Fig. 7B), and ACS (Fig.   7D). The observation showing that IAA application after flower removal did not affect the increase in Chitinase genes expression after 2 h in the AZ (Supplemental Fig. S12, E and F), strongly suggests that alteration in the expression of chitinases may not result from IAA depletion due to flower removal. On the other hand, the late (8 to 14 h) high gene expression of Chitinase which is AZ-specific (Fig. 9, E and F), indicates possible participation of these chitinases in the defense against microorganism occurring in the defense layer formed after tissue separation, which is an ethylene-dependent process and was inhibited by IAA treatment (Supplemental Fig. S12, E and F).
Group 6: Includes genes whose expression was modified early during 2-4 h after flower removal, either transiently or continuously, but IAA application did not affect their expression at any time. It is suggested that the modification in expression of these genes after flower removal does not result from IAA depletion, but rather from wounding and/or other signals omitted due to flower removal. Indeed, most of the genes classified in this group are associated with ethylene biosynthesis, signaling and response. Also the AP2 domaincontaining TF that belongs to the genes included in Group 6, can be associated with the wounding response (Okamuro et al., 1997).

CONCLUSIONS
The aim of this research was to further explore the molecular changes occurring during acquisition of abscission competence in the AZ following auxin depletion, by using the Affymetrix Tomato GeneChip. Application of IAA after flower removal, that prevented the abscission process, enabled us to differentiate between genes whose expression was affected by IAA due to flower removal, which are the interest of this research, from genes whose expression was modified by flower removal and were not affected by re-supplement of IAA.
Based on the kinetics of pedicel abscission, the identity and kinetics of expression of the genes affected by flower removal, the effects of IAA application and of 1-MCP pretreatment, we can separate the sequence of events which occur during tomato flower abscission into two phases: Early events (0 to 4 h after flower removal) that probably lead to acquisition of ethylene sensitivity and abscission competence, and late events (8 to 14 h after flower removal) when processes leading to the execution of pedicel abscission and development of the defense layer occur (Fig. 15). The late events, which are ethylene-induced, are inhibited by 1-MCP pretreatment, while the early events are not necessarily so. On the other hand, IAA application immediately after flower removal inhibited all the cascade of abscission events.
The sequence of molecular events occurring after flower removal is summarized in Fig.   15. Genes showing early modified expression might be involved in mediating auxin regulation of ethylene sensitivity in the AZ. These include three sets of genes (1, 2 and 3 in Fig. 15): Set 1 includes genes that are directly regulated by auxin and are therefore down-regulated early-on after IAA depletion, such as: Aux/IAA genes such as IAA1,3,4,7,8 and some of the TFs whose expression is early down-regulated after flower removal, such as the knotted, Homeobox-Leu zipper genes and bHLH. Set 2 includes genes that are directly IAA-repressed which were upregulated early-on after IAA depletion, such as PK7, ERF1c, WRKY lld-1, Protein phosphatase. Set 3 includes other TF and/or post transcriptional regulators, which are probably regulated by the modified IAA-related genes as their expression is modified at a relatively late stage of the process (Groups 1 and 2 in Table II), such as LRR-RLK, PK7, TPRP-F1, Phantastica, and Ovate.
As the AZ becomes sensitive to ethylene, the basic level of ethylene production together with its autocatalytic increase is mediated specifically in the AZ by specific expression of ethylene biosynthesis-related genes (e.g. ACS -M34289, Fig. 7D). This induction of ethylene levels leads in turn to activation of AZ-specific genes involved in the late stage of abscission and its excution after 4 h (Set 4 in Fig. 15; Groups 1-4 in Table II). The genes included in Set 4 can be classified into three sub-groups, based on their putative functions: I -TF genes or genes belonging to ethylene signal transduction or abscission regulators such as: ETR4, CTR1, ERF1c, TAGL12, LRR-RLK, and PK7; II -Genes encoding cell wall modifying proteins; III-Genes involved in the PR defense and development of the defense layer such as WRKY TFs, ERT10, Chitinase, and Peroxidases.
The analysis of the microarray results for the flower AZ allowed us to establish a clear sequence of events occurring during acquisition of tissue sensitivity to ethylene, and to confirm our hypothesis that acquisition of ethylene sensitivity in the AZ is associated with altered expression of auxin-regulated genes. These results shed light on the mechanism of increased sensitivity of the AZ to ethylene, and further expanded our knowledge of auxin-ethylene cross talk during the abscission process.
The present study has established a powerful platform for further analysis of possible regulatory abscission-related genes involved in acquisition of ethylene sensitivity at the AZ.
Based on this study, microarray experiments, aimed to examine the effects of IAA and 1-MCP on gene expression in the leaf AZ after leaf deblading, are in progress for comparing the two types the AZs. In parallel, we have initiated functional analysis of selected candidate genes and some function in abscission was strongly suggested, based on the phenotypic consequences of modifying their expression. These genes are currently specifically inhibited using RNAi in stably transformed tomato plants, as regulated by an abscission-specific promoter. This functional analysis will enable us to further reveal the role of key regulators in the early events of the abscission process.

Plant Material and Treatments
Flower bunches of cherry tomato (Solanum lycopersicum Mill, cv. 'Shiran' 1335, Hazera Genetics Ltd., Israel) were harvested between 10 to 12 AM from a commercial greenhouse in Israel. Bunches containing at least 2 to 4 fresh open flowers, were brought to the laboratory under high humidity conditions. Senesced flowers and young flower buds (unopened) were removed and the stem ends were trimmed. Groups of 3 to 4 bunch explants ( Fig. 1, A and B) were placed in a vial containing 10 mL of organic chlorine (50 µl L -1 TOG-6, Milchan Bros, Ltd., Israel) in water to prevent microorganism development. The bunch explants in vials were kept in a covered box containing a moistened paper to maintain high humidity, and were divided into two groups: one was incubated at 20°C (control), and the second group was exposed to the ethylene action inhibitor, 1-MCP (0.4 µl L -1 ) in a sealed 200-L chamber at 20°C for 12 h, before flower removal. IAA (10 -3 M) was applied in lanolin paste immediately after flower removal to the cut surface of the remaining tissues.
Flowers were removed with a sharp razor blade (Fig. 1C, b), and pedicel abscission was monitored in control, 1-MCP-pretreated and IAA-treated explants at various time intervals after flower removal up to 60 h. Groups of 15 vials (containing about 50 explants with ~120 flowers) were used for each treatment.
Tissue samples for RNA extraction were taken from the AZ (100 segments of less than 1 mm thickness for each time point, excised less than 0.5 mm from each side of the visible AZ), and from the NAZ (20 segments of 3 mm for each time point) (Fig. 1C). The AZ tissue was sampled at five time points (0, 2, 4, 8 and 14 h), and the NAZ tissue was sampled at four time points (0, 2, 4 and 14 h) following flower removal (Fig. 1D). Since no major abscissionrelated changes were expected to occur in the NAZ with or without 1-MCP, less time points were sampled from these tissues. Samples for time zero were taken from entire explants without flower removal (Fig. 1C, a). Tissues were sampled from control, 1-MCP-pretreated and IAA-treated explants. All samples were frozen in liquid nitrogen, and stored at -80°C until RNA extraction.

RNA Extraction
RNA was extracted from tissue segments collected as described above. The tissue was cDNA was synthesized using the Reverse Transcription System (PROMEGA, Madison, WI, USA) using 2 µg of total RNA from each sample.

Microarray Analysis
The microarray analysis was employed to measure global gene expression in the AZ and NAZ tissues of tomato flower pedicel, sampled at various time intervals after flower removal (Fig. 1D). The samples were taken from control explants, from explants pretreated with 1-MCP before flower removal, to block any direct effects of ethylene, and from explants treated with IAA immediately after flower removal. The use of 1-MCP should allow us to distinguish between IAA-related genes that affect ethylene sensitivity of the AZ, and ethylenerelated genes that induce the abscission process. IAA application after flower removal should allow us to distinguish between changes in gene expression resulting from auxin depletion or from wounding or other non-related signals.

For microarray analysis we have used the Affymetrix GeneChip® Tomato Genome
Array which is designed specifically to monitor gene expression in tomato. The comprehensive array consists of over 10,000 tomato probe sets to interrogate over 9,200 transcripts. We have used RNA extracted from biological duplicates of two independent experiments performed in a 3-week interval for the 1-MCP pretreatment and three for the other treatments. All procedures for probe preparation, hybridization, washing, staining and scanning of the GeneChip® Tomato arrays, as well as data collection were performed at the Microarray Core Facility, Department of Biological Services, The Weizmann Institute of Science, Rehovot, Israel. We have used the Affymetrix GeneChip Exp 3' One-Cycle kit according to the relevant Affymetrix GeneChip® Expression Analysis Technical Manual (No. 701021 Rev. 5) and Data Analysis Fundamentals booklet (P/N 701190) manual. cDNA was prepared using the two-cycle target labeling procedure and was used for further synthesis of biotin-labeled target cRNA by in vitro transcription as described in the Affymetrix GeneChip® Manual. The cRNA was fragmented before hybridization and hybridized to the probe array for 16 h at 45°C. Specific experimental information was defined using Affymetrix® GeneChip Operating Software (GCOS) on a PCcompatible workstation. Immediately following hybridization, the probe array went an automated washing and staining protocol on the fluidics station using the fluorescent molecule streptavidin-phycoerythrin that binds to biotin, and for signal amplification anti-streptavidin and biotinylated goat IgG antibodies. Probe Array scan was also controlled by the GCOS software to define the probe cells and to compute intensity for each cell. The Data image was analyzed for probe intensities as described in the Data Analysis booklet.
Initially, probe signal summarization, normalization, and background subtraction were performed using robust multichip analysis (RMA;Irizarry et al., 2003) in the 'affy' package with default parameters. The statistical test for differentially expressed genes was performed using the linear models for microarray (LIMMA) package (Smyth, 2004), which allows a better variance estimation by calculating the moderated t-statistic using empirical Bayesian techniques. These moderated t-statistics were calculated separately for each of the following In order to enable a gene ontology analysis, we had to assign each tomato gene with its nearest homologous gene from Arabidopsis, as the FatiGo Gene Ontology analysis tools (http://fatigo.bioinfo.cipf.es/) we have used are available only for Arabidopsis. This was done by using the BLAST tool in order to match the tomato transcripts to the best Arabidopsis homologues. The criterion for finding the best Arabidopsis homologue was chosen as genes with less than E value of 1e -5 .

Validation of Microarray Analysis of the Tomato Flower AZ by SQ-PCR
Validation of the microarray expression results was performed for few genes which exhibited abscission-specific type of expression in the tomato flower AZ. Beside validation of the microarray results for newly discovered abscission-specific genes, we have followed the expression of genes encoding for cell wall hydrolases known to be associated with abscission. to the length of the PCR products, which were documented using an Image Master VDS system.

PCR (Q-RT-PCR)
Flower samples for AZ and NAZ tissues were collected as described above.  (609267), SL-Actin (U60481/ Q96483) and GAPDH (U97257) genes were used as internal controls, and relative expression levels of these genes were computed by the 2 -ddCt method of the relative quantification (Livak and Schmittgen, 2001). All experiments were carried with Non template control and Negative control (RNA sample), were repeated at least three times and yielded similar results. Linear regression analyses between the microarray and the Q-RT-PCR expression data were performed, using the statistical program SigmaStat (Jandel Scientific).

SUPPLEMENTAL DATA
The following materials are available in the online version of this article: Supplemental Figure S1. Kinetics of changes in array-measured expression of genes encoding cell wall-related enzymes following flower removal.   Fig. 3 The symbols of the cluster types are detailed in the legend of Fig. 3.

h) or late (8-14 h) phases after flower removal
The table shows how IAA application affected the modifications in gene expression induced by flower removal. The sign (-) indicates no effect. The genes listed in Group 2 were classified into three sub-groups, I, II and III.  Proline transporter increase increase prevention S19A

Gene name
Protein phosphatase 2c increase increase -prevention S18C Putative PK increase increase prevention S20C Chitinase class II increase --decrease S12F Basic endochitinase increase --similar but lower S12E