INTRODUCTION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION CONCLUDING REMARKS MATERIALS AND METHODS LITERATURE CITED

Polygalacturonase (PG) (EC 3.2.1.15) hydrolyzes pectin in the cell wall and middle lamella of plant cells. Pectin polysaccharides make up approximately 30% of the dry mass of the primary cell wall in flowering plants and a much greater proportion of the middle lamella (Carpita and Gibeaut, 1993). PG expression increases during a number of developmental processes thought to involve cell wall breakdown. These include fruit ripening (Fischer and Bennett, 1991), lateral root emergence (Peretto et al., 1992), root cap cell detachment (Hawes and Lin, 1990), organ separation (abscission) (Tucker et al., 1984), pod dehiscence (Jenkins et al., 1996), and pith autolysis (Huberman et al., 1993). In addition, wounding (Bergey et al., 1999) and attack by plant pathogens (Ryan and Farmer, 1991; Cote and Hahn, 1994) also correlate with an increase in PG expression. Despite a great deal of work to elucidate the role of PG in these processes, its function is still a matter of speculation (Hadfield and Bennett, 1998).

In tomato (Lycopersicon esculentum cv Ailsa Craig) there are currently nine different PGs recorded in the public nucleotide databases. One of the nine PGs is specific to fruit ripening (Della Penna et al., 1986). Several are expressed in weakening abscission zones and pistils (Kalaitzis et al., 1997; Hong and Tucker, 1998, 2000), and another is induced by wounding and systemin (Bergey et al., 1999). The expression patterns for two of the nine PG genes have not been determined (Hong and Tucker, 1998).

Our primary interest in the PG genes is their use as tools to study molecular mechanisms that control the abscission process. Abscission and the synthesis of PG in abscission zones in dicotyledonous plants are stimulated by ethylene and suppressed by auxin (Sexton and Roberts, 1982; Sexton, 1995). The balance between these two hormones may determine the timing of abscission (Sexton, 1995). Despite the significance of abscission to agricultural productivity, the molecular mechanisms underlying tissue differentiation and hormonal control are not currently understood.

The weakening of tomato abscission (TA) zones and subsequent organ separation is typical of most abscission systems that have been studied (Sexton, 1995). Abscission of the flower pedicel of tomato involves the breakdown of the connective middle lamella and partial degradation of the primary wall in a four- to 10-cell-wide separation layer (Roberts et al., 1984; Tucker et al., 1984). In common with gene expression examined in abscission zones of other plants, the weakening of tomato zones is correlated with the localized synthesis of both endoglucanases (Lashbrook et al., 1994; del Campillo and Bennett, 1996) and PGs (Roberts et al., 1984; Kalaitzis et al., 1997).

Several of the tomato PG genes expressed during abscission are also expressed in mature pistils (Kalaitzis et al., 1997). The breakdown of the middle lamella between the longitudinal files of cells composing the transmitting tract in pistils resembles the cell separation that occurs during abscission (Cresti et al., 1976; Dumas et al., 1978).

In addition to a role for PGs in the digestion of cell walls, they are also implicated in the elicitation of defense responses and signaling events in growth and development (Ryan and Farmer, 1991; Cote and Hahn, 1994). Oligogalacturonides released from the plant cell wall during infection have been shown to induce the accumulation of several pathogenesis-related (PR) genes (Cote and Hahn, 1994). Moreover, Bergey et al. (1999) recently described the cloning of a PG cDNA induced by wounding. They suggested that PG gene expression might potentiate a defense response elicited by wounding. Defense genes are expressed in ripening fruit (Dopico et al., 1993), mature pistils (del Campillo and Lewis, 1992; Harikrishna et al., 1996), and abscission zones (del Campillo and Lewis, 1992). Expression of defense genes in these tissues may be important in the prevention of opportunistic infection of cells made vulnerable by the loss of a protective cell wall structure (del Campillo and Lewis, 1992).

We previously identified and characterized cDNA and genomic clones encoding four tomato PGs expressed in weakening abscission zones, i.e. TAPGs 1, 2, 4, and 5 (Kalaitzis et al., 1995, 1997; Hong and Tucker, 1998). TAPGs 1, 2, and 4 were also expressed in the upper portion of pistils that included the stigma and style (Kalaitzis et al., 1997). No transcripts for these genes were detected in stems, petioles, or fruit. Expression of multiple PGs in the same tissue suggests that each may have a different enzymatic function and/or different temporal or spatial expression pattern. For simplicity abscission zones are often visualized as identical cells acting in synchrony. However, in reality they contain different classes of cells with different wall chemistries and different temporal patterns of wall degradation (Sexton and Roberts, 1982; Sexton, 1995). It is possible that each PG gene product has a discrete role in this complex process.

The coding regions of the four TAPGs expressed during abscission share 72% nucleotide sequence identity. The 5'-upstream promoters of the TAPGs are composed of two domains, a 300-bp proximal domain, which is partially conserved in all four TAPGs, and an upstream distal domain, which is widely divergent with the exception of TAPG1 and 2 that share 66% identity in the first 1,140 bp of upstream sequence (Hong and Tucker, 1998). The TAPG4 transcript accumulates much earlier in abscission zones than that of TAPG1 and 2. Although all three genes are expressed in abscission zones, the details of their spatial expression patterns have not been determined.

To better understand processes that regulate hydrolase expression in abscission and pistil development, we prepared chimeric gene fusions between the 5'- and 3'-flanking regions of TAPG1 and 4 with a -glucuronidase (GUS) reporter gene. In addition to these full-length promoter constructs, we prepared reduced-length and deletion constructs of the TAPG1 promoter to better define potentially important cis-acting elements in this gene. Transgenic tomato plants containing these chimeric gene fusions were surveyed for GUS expression, and the temporal and spatial distribution of GUS activity was determined in abscission zones and pistils. In addition, the response of the full-length and deletion transgenes to inhibitory concentrations of auxin and silver thiosulfate (STS), an ethylene action inhibitor, was investigated.

	RESULTS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION CONCLUDING REMARKS MATERIALS AND METHODS LITERATURE CITED

Tissue-Specific and Temporal Expression Patterns of GUS Activity in Transgenic Plants

Kalaitzis et al. (1997) demonstrated that TAPG1, TAPG2, and TAPG4 were expressed in weakening leaf abscission zones (LAZ), floral abscission zones (FAZ), and pistils. The temporal expression patterns for TAPG1 and 2 were very similar. TAPG4 transcript, however, appeared much earlier than TAPG1 and 2 in both abscission zones and pistils. The coordinated expression of TAPG1 and 2 is reflected in 66% sequence identity in their 5'-upstream regions between 1 and 1,140 bp (Hong and Tucker, 1998). The same region of the TAPG4 gene is more divergent (Hong and Tucker, 1998). Therefore we have focused our studies on TAPG1 and TAPG4.

The 5'-upstream sequences from TAPG1 and 4 (2.1 and 2.4 kb, respectively) and 3'-downstream sequences (0.4 and 0.8 kb, respectively) were fused to GUS reporter genes to generate constructs p1-1 and p4-4, respectively (Fig. 1). Putative kanamycin-resistant transformants were screened by PCR analysis followed by histochemical staining to confirm GUS activity in ethylene-induced abscission zones. Three transgenic lines for 1-1 and five for 4-4 were selected for further study.

View larger version (29K): [in this window] [in a new window]   Figure 1.   Gene constructs used for Agrobacterium transformation of tomato. E, EcoRI; K, KpnI; B, BamHI; X, XbaI; H, HindIII; Sm, SmaI; S, SalI; St, SstI; P, PstI; PG1 5', TAPG1 5'-flanking region; PG1 3', TAPG1 3'-flanking region; PG4 3', TAPG4 3'-flanking region.

Histochemical localization and quantitative fluorometric assays of GUS were carried out on the pistils as well as the corolla abscission zone, FAZ, and LAZ from single-copy homozygous second and third generation plants (Fig. 2). There was considerable variation in the magnitude of GUS activity between independent transformants of both constructs, which is reflected in the large SEs for the means (Fig. 2). Although the amount of the GUS activity varied greatly, the temporal and spatial patterns of GUS expression were very consistent from one transformant to another.

View larger version (83K): [in this window] [in a new window]   Figure 2.   Time course and histochemical staining for TAPG1:GUS and TAPG4:GUS expression in pistils and selected abscission zones from 4-4 and 1-1 transformants. Expression of GUS activity (left side) was measured in protein extracts from pistils (ovaries removed) and corolla abscission zones taken from flower buds with opened sepals but unopened petals (SO), fully open but non-senescent flowers (FO), and senescent flowers (SE). Flower parts used for GUS activity measurements were not exposed to ethylene. GUS activity was measured in LAZ and FAZ, respectively, in explants treated with 25 µL/L ethylene at 25°C for the indicated times at the bottom of each graph. Error bars indicate the SD of the means. The numbers in parentheses to the immediate right of the symbol definitions in each graph indicate the number of independent transformants used to derive the means for each construct. Histochemical staining of pistils from 4-4 and 1-1 plants was performed at the indicated stages described above. In addition, a pistil from a very young (VY) unopened flower bud was also stained for GUS activity. Explants that included the flower corolla abscission zone (AZ), pedicel FAZ, and LAZ were exposed to ethylene for the indicated time intervals (in parentheses under the labels) and histochemically stained for GUS activity.

Expression in the Pistil

Expression in the Corolla Abscission Zone

Expression in the FAZ

View larger version (97K): [in this window] [in a new window]   Figure 3.   TAPG4-driven GUS activity (construct 4-4) appears first in the vicinity of the vascular bundle and its expression is independent of expression in other vascular bundles in the abscission zone. A, Pedicel FAZ from a senescent flower not exposed to ethylene (0 h). B, Distal FAZ showing initiation of localized GUS expression after 12 h of exposure to 25 µL/L ethylene at 25°C. C, Longitudinal half-section of FAZ showing vascular localization of GUS expression after 16 h of exposure to ethylene. D, 0.1 mM IAA was applied to the distil end of the petiole in a lanolin paste and the petiole sliced half-way through 5 mm distal to the LAZ. The LAZ explant was then exposed to ethylene for 84 h and the LAZ was stained for GUS activity.

Expression in the Fruit Calyx Abscission Zone

Expression in the LAZ

View larger version (74K): [in this window] [in a new window]   Figure 4.   TAPG:GUS expression for the full-length constructs 1-1 and 4-4 but not the minimal construct 1M-N is detected in vascular bundles 1 mm distal to the separation layer of the LAZ. Explants were exposed to 25 µL/L ethylene at 25°C for 72 h and the distal petiole sliced into 1-mm sections. In addition to the fracture surfaces (proximal and distal) the faces of the sections taken 1 mm distal to the fracture are shown (distal + 1 mm).

Other Organs and Tissues

Hormonal Regulation and Role of the 3' End and Conserved TAPIR Element in the TAPG1 and 4 Gene Promoters

To ascertain if the 3' end of the TAPG1 gene contained information that was essential for tissue specificity or hormonal control of TAPG1 gene expression, a construct was prepared in which the nopaline synthase termination sequence (NOS 3') replaced the TAPG1 3' termination sequence, construct p1-N (Fig. 1). Moreover, sequence comparison of the 5'-upstream sequences for the TAPG genes identified a conserved 300-bp inverted repeat (TAPIR) that contained several potential hormone response elements (Hong and Tucker, 1998). The 300 bp containing this inverted repeat was deleted from the p1-N promoter construct to generate construct p1D-N (Fig. 1). Each of these constructs was transformed into tomato. Histochemical staining for GUS activity in transformants containing these constructs, p1-N and p1D-N, were indistinguishable from transformants containing the more complete p1-1 construct. In addition to histochemical staining, GUS activity was quantified in the FAZ of several primary transformants for each of these constructs (Fig. 5). The temporal expression patterns in FAZ for these deletion constructs in transgenic tomato were not significantly different from those for the complete construct, p1-1, in transgenic tomato (Fig. 5).

View larger version (20K): [in this window] [in a new window]   Figure 5.   GUS activity in abscission zones and pistils from transgenic tomato containing the TAPG1 promoter constructs described in Figure 1. Ethylene exposure and stages of flower development are as described in Figure 2. The numbers in parentheses to the immediate right of the symbol definitions in each graph indicate the number of independent transformants used to derive the means displayed.

Treatment of leaf explants with 0.1 mM indole-3-acetic acid (IAA) or 0.5 mM STS (an ethylene action inhibitor) prior to exposure of explants to 25 µL/L ethylene for 72 h inhibits petiole abscission and TAPG gene expression (Kalaitzis et al., 1995; P. Kalaitzis and M.L. Tucker, unpublished data). Similarly, treatment of flower explants with 0.1 mM IAA inhibits pedicel abscission (del Campillo and Bennett, 1996). GUS activity in IAA-treated FAZ and IAA- or STS-treated LAZ from 1-1, 1-N, 1D-N, and 4-4 transgenic explants was reduced by greater than 90% relative to the GUS activity in abscission zones of explants not pretreated with IAA or STS (Fig. 6). There was no significant difference in the inhibitory response of the different promoters in these constructs (Fig. 6).

View larger version (26K): [in this window] [in a new window]   Figure 6.   Relative GUS activity in FAZ and LAZ of explants from transgenic tomato not treated (control) or pretreated with 0.1 mM IAA or 2.0 mM then 0.5 mM STS. All FAZ and LAZ explants were exposed to 25 µL/L ethylene at 25°C for 24 or 72 h, respectively. TAPG1 and TAPG4 promoter constructs are as described in Figure 1. ND, Not determined. The numbers in parentheses immediately below the construct name indicate the number of independent transformants used to derive the means displayed.

Although explants treated with IAA show no sign of weakening in the abscission zones after 72 h of exposure to ethylene, abscission zone sections did contain sporadic islands of staining around the vascular tissue. This would happen if the IAA applied onto the petiolar stump does not reach all parts of the abscission zone equally. If IAA is added to the petiole and one side of the petiole cut half-way through to prevent movement of IAA down that side, GUS staining develops in the abscission zone predominantly on the same side as the cut where IAA concentration should be lowest (Fig. 3D).

Role of a 320-bp Proximal TAPG1 Sequence in the Regulation of Abscission-Specific Expression

Optimal sequence alignment of the 5'-upstream sequences for the four PG genes known to be expressed during abscission indicated that the first 300 bp is partially conserved in all four PG genes (Hong and Tucker, 1998). Accordingly, to test if this proximal region is sufficient for regulated expression in TA zones, 247 bp of 5'-upstream sequence and 73 bp immediately downstream from the start of transcription of the TAPG1 gene were fused to a GUS reporter gene to generate construct p1M-N (Fig. 1).

GUS activity was quantified in primary transformants of 1M-N plants and single-copy first or second generation 1-1 plants (Fig. 5). The temporal expression patterns for GUS activity in 1M-N plants containing the minimal promoter construct were similar to the patterns found for 1-1 plants containing the full-length constructs (Fig. 5). However, the level of GUS expression in FAZ and LAZ from the minimal promoter plants was generally less than that measured in plants containing the full-length construct (Fig. 5). Histochemical staining for GUS activity in four independent transformants (1M-N) containing the minimal construct was generally very faint. However, tissues that stained for GUS activity in the minimal TAPG1:GUS transformants (1M-N) were essentially the same as those that stained in the full-length TAPG1:GUS transformants (1-1), i.e. LAZ, FAZ, fruit abscission zones, and pistils. The most striking difference in the histochemical staining pattern for the full-length TAPG1:GUS transformant (1-1) compared with the minimal construct transformant (1M-N) was that in the minimal transformant very little or no stain was observed in the distal surface of the separation layer after 72 h of exposure to ethylene (Fig. 4). However, the lack of stain in the distal portion may not indicate a total loss in a cell-specific response in the minimal construct but rather a lower rate of accumulation of GUS in the distal abscission zone, which was below the threshold necessary for GUS staining.

The temporal expression patterns for GUS activity in pistils and corolla abscission zones were very similar between the minimal and full-length TAPG1:GUS transformants (Fig. 5). However, the GUS activity for the minimal promoter (1M-N) in pistils and corollas abscission zones was greater than that for the full-length promoter (Fig. 5). Because of the large amount of plant material required and the difficulty in collecting it, GUS activity in the staged flower parts was measured in only one 1M-N transformant. The 1M-N transformant selected had the highest GUS activity in LAZ and FAZ of three 1M-N transformants. The selection process might explain why 1M-N flower parts had higher activity than did those from the full-length 1-1 transformant.

Another difference between the full-length construct (p1-1) and the minimal construct (p1M-N) was that the relative GUS activity in 1M-N transgenic plants treated with IAA was inhibited by only 60% in FAZ and 80% in LAZ (Fig. 6). This is in contrast to approximately 95% inhibition in plants containing the full-length construct (Fig. 6). However, treatment with the ethylene action inhibitor STS inhibited the accumulation of GUS activity by greater than 95% in LAZ from both 1M-N and 1-1 transgenic plants (Fig. 6).

	DISCUSSION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION CONCLUDING REMARKS MATERIALS AND METHODS LITERATURE CITED

Correlation of TAPG-Promoted GUS Expression in Transgenic Plants with TAPG Transcript Accumulation in Wild-Type Plants

Analysis of the spatial and temporal expression of the GUS reporter gene in TAPG1:GUS and TAPG4:GUS transgenic cv Ailsa Craig tomato plants revealed virtually identical patterns to those of the native genes in the tomato cv Rutgers. Kalaitzis et al. (1997) demonstrated using an RNase protection assay that the accumulation of both TAPG1 and TAPG4 transcript in cv Rutgers tomato correlated with the weakening of FAZ and LAZ and was spatially delimited to the abscission zones. In addition, abscission and accumulation of TAPG1 transcript in LAZ were induced by ethylene and inhibited by auxin. The hormonal regulation of GUS expression from the TAPG1 and 4 promoter constructs was retained in the transgenic plants (Fig. 6). In addition to expression in abscission zones, the promoter constructs retained the ability to promote GUS expression in the mature stigma and upper style as was demonstrated for the native genes (Kalaitzis et al., 1997; Hong and Tucker, 2000). Accumulation of GUS in transgenic cv Ailsa Craig plants, however, is slightly slower than that for the PG transcripts in the native cv Rutgers. This may simply be a difference between transcript and protein accumulation or it may reflect different rates of abscission between the two cultivars of tomato.

In addition to LAZ, FAZ, and pistils, which had been previously demonstrated to accumulate TAPG1 and 4 transcript, we examined fruit pedicel, fruit calyx, and corolla abscission zones. GUS accumulated in all of these abscission zones in explants from both the TAPG1:GUS (1-1) and TAPG4:GUS (4-4) transgenic plants. Peretto et al. (1992) measured PG activity associated with root initials in Allium porrum. In a few thin sections from 4-4 transgenic plants a narrow band of cells immediately above an adventitious root initial stained faintly for GUS activity. This staining pattern was difficult to reproduce but may suggest transient expression of the TAPG4 promoter during adventitious root initiation. Histochemically detectable GUS activity was not observed in fruit, anthers, roots, or root caps. Other genes from the tomato PG family may be expressed in these organs and tissues.

Deletion analysis of the TAPG1 promoter indicates that the conserved TAPIR element and 3' end of the TAPG1 gene, which included multiple auxin and ethylene response elements (Hong and Tucker, 1998), were not essential to the hormonal or tissue-specific regulation of TAPG1-promoted GUS expression in abscission (histochemical observations and Figs. 5 and 6). Tissue-specific and hormone-responsive elements in the TAPG1 gene must reside within the first 247 bp of the promoter or 73 bp of the 5'-upstream untranslated region of the transcript, because the minimal promoter construct (p1M-N) retained these major control characteristics.

Nevertheless, organization of gene promoters can be quite complex. A pertinent example in the context of our studies are those for the tomato fruit PG (TFPG) (Montgomery et al., 1993). GUS expression from a minimal 231 TFPG promoter construct demonstrated expression of GUS in the outer pericarp of ripening fruit very similar to that of the longer 1.4-kb promoter construct, which faithfully reflected that of the native PG transcript. However, when TFPG deletion constructs of 443 and 806 were examined for GUS expression, GUS stain was observed in both the inner and outer fruit pericarp indicating the presence of a positive regulatory element between 231 and 1,150 that controlled expression in the inner pericarp. Montgomery et al. (1993) further concluded that a negative regulatory element between 1,150 and 1,411 inhibited GUS expression in the inner pericarp re-establishing an expression pattern more like that of the native PG transcript.

Clearly, we cannot be certain that there are no additional regulatory elements between 247 bp and 2.1 kb in the TAPG1 gene promoter. Nevertheless, the minimal promoter in p1M-N must include elements that can control abscission specificity and hormonal regulation. Examination of this region of the TAPG1 gene promoter for regulatory motifs did not identify any motifs of particular interest (Hong and Tucker, 1998). Regions upstream of 247 must at the very least play an ancillary role since the overall strength of the TAPG1 minimal promoter in abscission zones was significantly reduced compared with the longer promoter construct (Fig. 5) and the IAA inhibition of expression less absolute (Fig. 6). This is not unexpected since enhancer elements commonly reside considerable distances from the start of transcription and regulatory elements may be repeated several times (Nicholass et al., 1995; Singh, 1998).

Significance of the Spatial Expression of GUS as Abscission Progresses

In the p1-1 and p4-4 gene constructs we have attempted to include as much of the native gene regulation as possible to create a reporter gene that can be used to study the biology of abscission in tomato. The p1-1 and p4-4 constructs include 2.1 and 2.4 kb, respectively, of native PG sequence 5' to the start of translation of PG and 0.4 and 0.8 kb, respectively, of 3' sequence below the stop codon for the end of PG translation (Fig. 1). Sequence analysis of the three introns in each of the PG genes did not highlight any notable motifs or conserved sequences (Hong and Tucker, 1998). Therefore, we did not include any intron sequences in the gene constructs. Observations described above indicate that expression of the GUS reporter gene from the respective constructs faithfully reflects that of the native PG genes.

The localization of GUS presented above allows a detailed insight into the progressive patterns of expression of an abscission-specific gene. GUS expression in the FAZ is initiated in isolated vascular bundles and surrounding parenchyma cells. Examples were found where only one or two bundles were stained and others where a crescent or horseshoe shaped group of bundles stained blue. As weakening progressed in FAZ the stain spread and intensified in a doughnut-shaped ring of parenchyma surrounding the circle of vascular bundles. This ring of smaller parenchyma cells was described by Biain de Elizalde (1980) who suggested it was the result of intense meristematic activity in this region. It has been proposed that these extra cell divisions might be necessary for the programming of cells to respond to the abscission-inducing signal whereas neighboring cells do not (Sexton and Roberts, 1982). The ring of intensely stained cells surround the larger cells of the central pith on the inside that only become stained late in the development of the separation layer. In addition, the FAZ includes five to seven rows of outer cortex cells that do not stain for GUS (Fig. 2).

The LAZ showed a similar progression of GUS expression. Staining begins around isolated vascular bundles and then spreads to the entire U-shaped stele. Eventually GUS activity spreads laterally to the outer cortex where it is restricted to a fracture plane several cell layers deep. A similar spread of cellulase activity from the stele to the cortex has been described in the abscission zones of bean (Sexton et al., 1981; del Campillo et al., 1990).

By surgically removing the stele from the LAZ of bean, Thompson and Osborne (1994) demonstrated that the vascular tissue was essential to initiate -1,4-glucanase (EGase) and abscission in the surrounding cortical cells. If they delayed removal of the stele for several hours or removed the stele and then immediately replaced it, EGase was expressed in the cortical cells, and cell separation progressed across the outer cortical parenchyma. They concluded that a signal emanates from the stele that then elicits a program of abscission-related gene expression in the cortical cells. They further speculated that this signal could be a diffusible cell wall oligosaccharide arising from hydrolysis of cell walls in the stele. The observation that GUS is first detected in the vascular traces of FAZ and LAZ supports the first tenet of the model, which predicts that the target cells for the initial inductive stimulus are in the stele. The subsequent spread of GUS staining to the cortex is also consistent with the lateral diffusion of a secondary signal.

GUS expression was detected in the fracture plane and in vascular tissues proximal and distal to the separation layer (Fig. 4). A similar distribution was observed for an EGase expressed in bean LAZ (del Campillo et al., 1990; Tucker et al., 1991). The enhanced expression of cell wall hydrolases in vascular bundles has not only been implicated in the separation of the small, tightly packed cells but has also been implicated in cell wall breakdown necessary for the formation of tyloses (Sexton et al., 1981). Tyloses are cells that penetrate and block the xylem vessels preventing both water loss and microbial access after fracture. Close examination of thin sections of vascular tissue from TAPG4:GUS transgenic plants revealed GUS stain in all the cells of the vascular bundle including those next to the vessels that are producing tylose outgrowths (data not shown).

As abscission progresses GUS activity is detected progressively farther away from the separation layer primarily distal to the fracture and is restricted to the stele (Fig. 4). Progressive GUS expression along the vascular traces is not dependent on a secondary signal produced initially in the abscission zone. If the petiole is sliced into sections prior to exposure to ethylene in a manner similar to that shown in Figure 4, GUS stain develops after 72 h of exposure to ethylene in vascular traces of slices closest to where the abscission zone was attached (data not shown). In a similar experiment with bean, cellulase activity increased in petiole sections distal to the LAZ even though there was no contact with the abscission zone (Sexton et al., 1981).

TAPG:GUS Expression in Pistils

Recently, Hong and Tucker (2000) showed that TAPG4 transcript accumulated only in mature pistils and was limited to the upper one-third of the pistil. This pattern of GUS accumulation was faithfully reproduced in pistils of 4-4 transformants (Fig. 2). TAPG4:GUS and TAPG1:GUS are not expressed in pistils of young or unopened buds (Fig. 2). GUS expression begins in pistils of 4-4 transformants as the flower opens and continues to increase as the flower senesces (Fig. 2). In pistils from 1-1 transformants, GUS activity was very low until flowers began to senesce, and faint GUS staining was observed only at the stigma surface of pistils from senescent flowers (Fig. 2).

Tomato has what is described as a solid style with a wet stigma (Kadej et al., 1985). GUS staining in the stigma and upper style is present in regions corresponding to the outer papillae, the underlying stigmatic zone, and the upper transmitting tract (Dumas et al., 1978). Cell walls in these tissues are degraded to produce longitudinal canals in which large amounts of a stigmatic exudate accumulate and through which the pollen tubes grow (Dumas et al., 1978; Janson et al., 1994). The loss of cohesion between stigma cells anatomically resembles that in abscission zones and could involve PG in a similar role. Although GUS staining was very intense in the stigmatic region and in the upper 500 µm of the style below it, GUS staining was not present in the lower style or ovary. This indicates that TAPG1 and 4 are not involved in the formation of the lower transmitting track, which passes through the style to the ovary (Cresti et al., 1976).

It is interesting that transcript for another tomato PG gene, TPG7, was found to be abundantly expressed in both mature and immature pistils (Hong and Tucker, 2000). In mature pistils, transcript accumulation for TPG7 was also limited to the upper one-third of the pistil. It appears that PG is important to several stages of pistil development.

Proposed Roles for PG

The patterns of PG expression are consistent with a proposed role in cell separation, which occurs in both of the abscission zones, in the stigma, and in the style. However, these same cells also share the characteristic that they are unprotected by mechanical barriers and therefore are susceptible to microbial attack. The antimicrobial enzymes, chitinase, -1,3-glucanase, and several other PR proteins accumulate abundantly in abscission zones at the time of shedding (Gomez et al., 1987; Weiss and Bevan, 1991; del Campillo and Lewis, 1992). Moreover, PR gene expression is evoked in mature pistils in a similar manner to that observed in abscission zones (del Campillo and Lewis, 1992; Atkinson et al., 1993; Harikrishna et al., 1996). Oligosaccharides and specifically oligogalacturonides have been shown to activate plant defensive genes (Baydoun and Fry, 1985; Cote and Hahn, 1994; Doares et al., 1995). Oligosaccharides move only very small distances and therefore are not part of a long-distance signal to mount a systemic defense response in tomato (Baydoun and Fry, 1985). As a result PG may play a secondary role in abscission zones and pistils by releasing oligogalacturonides from cell walls to mount a local defense response.

	CONCLUDING REMARKS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION CONCLUDING REMARKS MATERIALS AND METHODS LITERATURE CITED

Unfortunately, we have not been able to produce reliable in situ hybridization data to confirm the transgenic results. In addition to typical difficulties with in situ hybridization, expression of a family of at least four related PG genes in abscission (Hong and Tucker, 1998) complicates probe selection for gene-specific hybridization. However, the ease of identifying and following the expression of the GUS reporter gene has generated a level of resolution concerning its topographical and temporal expression that we have not achieved for the native gene transcripts.

TAPG1:GUS and TAPG4:GUS expression were observed in all the abscission zones examined, i.e. leaf petiole, flower and fruit pedicel, fruit calyx, and corolla abscission zones. Our model for PG expression, and abscission in general, is that an abscission signal is perceived first in a target cell localized somewhere in the vascular bundle. The target cell then evokes an abscission program that includes TAPG4 gene expression. As proposed by Thompson and Osborne (1994), the abscission signal is then further propagated to surrounding cortical and pith cells in the separation layer through another secondary signal that regulates the expression of both TAPG1 and TAPG4. Each vascular bundle works independently of each other to propagate a localized abscission signal across the separation layer. In addition to propagation of the signal transversely across the separation layer, an abscission signal is also perceived in the vascular bundle a short distance proximal and distal to the separation layer. Perception of the signal distal to the separation layer is not dependent on a secondary signal, because transverse sections cut across the petiole 1 to 2 mm distal to the separation layer can be induced by ethylene to express TAPG4:GUS in the vascular bundles without contact with the separation layer cells. The target cells in the distal vascular bundles must be different from those in the separation layer or, as suggested by Thompson and Osborne (1994), the cortical cells in the separation layer must already be predetermined to respond to a secondary signal emanating from the target cells in the vascular bundles. Although the above model fits the current data, the precise location of the target cells in the vascular bundles and the role of a secondary signal in abscission zones need to be rigorously examined with further experimentation.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION CONCLUDING REMARKS MATERIALS AND METHODS LITERATURE CITED

Construction of Gene Fusions

The 5'-flanking sequences of TAPG1 and 4 (accession nos. AF001000 and AF001002, respectively) were contained in a 3-kb EcoRI fragment of the genomic clone PG1-2 and a 7-kb EcoRI fragment of the genomic clone PG3-3, respectively (Hong and Tucker, 1998). The 2.1- and 2.4-kb sequences upstream from ATG start codons of TAPG1 and 4 were amplified by long PCR, respectively, using Taq extender PCR additive (Stratagene, La Jolla, CA) and oligonucleotide primers with restriction sites at their 5' ends. The resulting fragments were purified, cleaved, and cloned into HindIII/XbaI sites of the vector pBI221 to make transcriptional fusions with the coding region of the GUS gene with the 3'-terminator sequence from the NOS 3'. The HindIII/EcoRI fragments from the pBI221 containing the TAPG:GUS:NOS fusions were then subcloned into the pBluescript SK⁺ vector (Stratagene). Finally, the KpnI/PstI fragments of the pBluescript SK⁺ vector containing TAPG:GUS:NOS fusions were inserted into the binary vector pCGN1547 (McBride and Summerfelt, 1990). The resulting binary constructs containing TAPG1:GUS:NOS and TAPG4:GUS:NOS fusions were called p1-N and p4-N, respectively (Fig. 1).

PCR was used to generate a deletion construct in which the TAPIR element (Hong and Tucker, 1998) was deleted from an otherwise normal 2.0-kb 5'-upstream sequence from TAPG1 and named p1D-N (Fig. 1). In addition, a minimal construct was prepared for the TAPG1 gene, p1M-N, that contained 247 bp of sequence 5' upstream and 73 bp downstream of the start of transcription for the TAPG1 transcript (Fig. 1). This 320-bp TAPG1 fragment was sequenced and determined not to contain any errors introduced by the PCR amplification. Next, the NOS 3'-terminator sequence of p1-N and p4-N was replaced with 350 and 800 bp of the native 3'-downstream sequences of TAPG1 and 4, respectively. The TAPG1 and TAPG4 3' sequence include 38 and 12 bp, respectively, of 3'-translated sequence upstream from the respective stop codons. However, these TAPG translated sequences are downstream of the native stop codon for the GUS open reading frame. These two new constructs, which now contain both 5'- and 3'-flanking sequences of TAPG1 and 4, were named p1-1 and p4-4, respectively (Fig. 1).

Tomato Transformation and Propagation

The chimeric TAPG:GUS fusions in the binary vectors p1-1, p1-N, p1D-N, p1M-N, and p4-4 were transformed into Agrobacterium tumefaciens (strain EH105) by the freeze-thaw method of Holsters et al. (1978). Transgenic tomato (Lycopersicon esculentum cv Ailsa Craig) plants were generated by the method of McCormick et al. (1986). Transformed shoots were selected on kanamycin (100 µg/mL). After rooting of shoots, the plantlets were transferred to sterile potting soils and gradually acclimated before transfer to the greenhouse. The integrity and copy number of the introduced genes was checked by PCR and Southern-blot analysis. GUS activity was quantified fluorometrically in the FAZ of all the selected primary transformants, some of which had several copies of the transgenes. Because of the large amount of plant material required for GUS quantification, single-copy transformants were selected for 1-1 and 4-4 transformants and propagated by seed.

Plant Tissue Preparation

Ethylene, STS, and IAA treatments of tomato explants were performed as described previously (Koehler et al., 1996). For STS treatment, tomato explants with two to three attached leaves were pretreated for 3 h in 2 mM STS. Leaf blades were then removed, leaving the subtending petiole attached, and the explants placed upright in beakers containing 0.5 mM STS and then treated with 25 µL/L of ethylene at 25°C for an additional 72 h.

Petal wilt is an indication that abscission may have already been induced. Floral explants harvested for the time course experiments had fully open and bright yellow petals with no sign of flower senescence or petal wilt. IAA treatment of FAZ and LAZ was done by applying 0.1 mM IAA in lanolin to the cut ends of the pedicels of fully open flowers or petioles of expanded leaves, respectively. After application of lanolin the explants were allowed to stand in air for 4 h before exposure to 25 µL/L ethylene at 25°C. At least 20 FAZ and 10 LAZ were collected for each time point by cutting approximately 2-mm sections that included the abscission zone.

Assay for GUS Activity

Quantitative GUS enzyme assays were performed as described by Jefferson et al. (1987). Pooled tissues were ground in GUS extraction buffer (100 mM sodium phosphate, pH 7.0, 10 mM EDTA, 0.1% [v/v] Triton X-100, 0.1% [w/v] SDS, and 10 mM -mercaptoethanol) followed by centrifugation. Protein concentrations were estimated using the colorimetric method of Bradford (1976). To reduce endogenous GUS-like activity in tomato, sample extracts were heated to 55°C for 30 min. The volume equivalent to 20 µg of protein was incubated with 4-methylumbelliferyl glucuronide solution for 60 min at 37°C, and the reaction stopped by adding 40 µL of the reaction mixture to 1.0 mL of 0.2 M sodium carbonate. Fluorescence was measured with a Bio-Rad VersaFluor fluorometer (Bio-Rad Laboratories, Hercules, CA). The instrument was calibrated with a solution of 1.0 µM 4-methylumbelliferone in 0.2 M sodium carbonate. Histochemical staining was conducted by incubating hand-cut sections for 4 h at 37°C in a buffer (100 mM NaH₂PO₄, pH 8.0, 10 mM EDTA, 0.1% [v/v] Triton X-100, and 0.5 mM or 3.0 mM potassium ferro- and ferricyanide) containing 500 µg/mL of X-gluc (5-bromo-4-chloro-3-indolyl--glucuronic acid) (Stomp, 1992). Sections were cleared of chlorophyll with several changes of a mixture of ethanol:acetic acid at a 3:1 ratio. Samples were stored in the same 3:1 mixture of ethanol:acetic acid.

GUS histochemical staining can be susceptible to artifacts caused by the diffusion of the reaction product before precipitation. This problem can be overcome by increasing the ferricyanide concentration in the reaction mixture (Guivarc'h et al., 1996). In preliminary experiments we directly compared the two halves of bisected abscising pedicel and leaf base zones incubated in media with either 0.5 or 3.0 mM ferricyanide. The higher concentration markedly inhibited staining, although staining was more discrete at higher concentration. It appeared that the risk of obtaining "false-negative" results when using the higher ferricyanide concentration was greater than the risk of false distributions. As a result, we adopted the strategy that all initial investigative incubations were carried out in 0.5 mM ferricyanide and any observations of significance were checked at a higher concentration.