Ricinosomes predict programmed cell death leading to anther dehiscence in tomato

Successful development and dehiscence of the anther and release of pollen is dependent upon the programmed cell death (PCD) of the tapetum and other sporophytic tissues. Ultrastructural examination of the developing and dehiscing anther of tomato ( Solanum lycopersicum L.) revealed that cells of the interlocular septum, the connective tissue, the middle layer / endothecium and the epidermal cells surrounding the stomium all exhibit features consistent with progression through PCD. Ricinosomes, a subset of precursor protease vesicles that are unique to some incidences of plant PCD, were also present in all of these cell types. These novel organelles are known to harbour KDEL-tailed cysteine proteinases that act in the final stages of corpse processing following cell death. Indeed a Solanum lycopersicum KDEL-tailed cysteine proteinase, SlCysEP, was identified and its gene cloned, sequenced, and characterized. SlCysEP transcript and protein were restricted to the anthers of the senescing tomato flower. Present in the interlocular septum and in the epidermal cells surrounding the stomium relatively early in development, SlCysEP accumulates later in the sporophytic tissues surrounding the locules as dehiscence ensues. At the ultrastuctural level, immunogold labelling localized SlCysEP to the ricinosomes within the cells of these tissues, but not in the tapetum. It is suggested that the accumulation of SlCysEP and appearance of ricinosomes act as very early predictors of cell death in the tomato anther. reticulum membranes. I) Stage 15. Ricinosome membranes are studded with ribosomes. J) Stage 18. Numerous ricinosomes present in the very condensed cytoplasm.


Introduction
Successful reproduction in the vast majority of angiosperms is dependent on the proper development and release of the male gametophytes, the pollen, from the anthers. A thorough understanding of the processes of pollen formation and release is useful for agricultural practices and maintenance of both agriculturally-and ecologically-important genetic banks. In those agricultural species that are normally self-crossing, the artificial induction of male sterility can facilitate cross-pollination and the production of hybrids, allowing for an increase in the pool of genetically diverse individuals. Cross-pollination also allows for the flow of genetic information between closely-related species, and is receiving a great deal of public attention with the advent and use of transgenics in agriculture. Understanding of the processes leading to pollen production and release is of great importance given the potential ecological significance of the release and transmission of transgenes from agricultural crop plants into native wild relatives (Goldberg et al., 1993;Ma, 2005).
Interestingly, the successful production of viable pollen is dependent on the death of sporophytic tissues of the anther. Microsporogenesis, microgametogenesis and the resulting formation of viable pollen within the locules of the anther are dependent on nutritive contributions from the surrounding sporophytic tissues (detailed in Ma, 2005, and references therein). As an essential part of anther development and pollen formation, cells of the tapetum are sacrificed through programmed cell death (PCD). With continued development of the microgametophytes the cellular constituents resulting from tapetal PCD provide nutrition, act in exine sculpting and are deposited as the materials characteristic of the pollen wall (Vizcay- Barrena and Wilson, 2006;Varnier et al., 2005;Wu and Cheung, 2000). PCD then extends radially to cells of the middle layer and connective tissues nearest the locular chambers, the digested cellular contents presumably providing additional nutritional resources in support of the anabolic metabolism of the microgametophytes during reserve accumulation (Wetzel and Jensen, PCD of the cells of the post-germinative castor seed endosperm and senescent Day Lily petals (Gietl et al., 1997;Schmid et al, 1999) and PCD of the nucellus during castor seed development (Greenwood et al., 2005). More recently, ricinosomes, but not the enzymes within, have been implicated in PCD of the endosperm of post-germinative tomato seed (DeBono and Greenwood, 2006). The collapse of the central vacuole essentially defines the point of death in plant cells (Jones, 2001), and results in the acidification of the cytoplasm. In those incidences of PCD where ricinosomes are involved, it is believed that the acidification of the cytosol results in the autocatalytic processing of the enzyme to its mature active form. Concomitantly, the ricinosomes swell and break open, releasing these very active enzymes which then act in the final processing of the cell corpse. It should be noted that Rogers (2006) distinguishes ricinosomes from PPVs in that the former deliver their enzymic contents directly to the cytosol following vacuolar collapse, whereas the latter fuse with the central vacuole delivering vacuolar processing enzymes responsible for the processing of a number of proteins. When present, the occurrence of ricinosomes in cells seems to infallibly predict that those cells are going to die (Gietl and Schmid, 2001) The histological details of anther development and some cytological details of cell death in the sporophytic tissues, particularly the tapetum (Papini et al., 1999;Wu and Cheung, 2000;Wang et al., 1999), have been documented for a number of species including members of the Solanaceae (Bonner and Dickinson, 1989;Koltunow et al, 1990;Goldberg et al., 1993;Sanders et al., 2005;Varnier et al., 2005). However, with limited exception (see Varnier et al, 2005;Sanders et al 2005) ultrastructural details of the progression of PCD in anther tissues other than the tapetum have been largely ignored and limited to documenting mitochondrial and nuclear changes that occur as the cells approach death. Common ultrastructural features consistent with the progression of PCD in plant cells have been provided (Gunawardena et al., 2004 and references therein;DeBono et al , 2006) and include: progressive condensation of chromatin with nuclear shrinkage, invagination and lobing; aberrant morphologies and changes in electron density of mitochondria and plastids; vesiculation and vacuolization of the cytoplasm with changes in electron density of the tonoplast and cytoplasm; shrinkage of the plasma membrane from the cell wall; and vacuolar collapse. Ricinosomes may or may not be present. epidermal cells (Fig. 1 E,F). By stage 20 the septa are completely degraded and the stomia have opened to release the pollen, completing the dehiscence process ( Fig. 1 G).

Ultrastucture of the stomial, middle layer and connective tissue cells during anther maturation to dehiscence
The ultrastructure of the epidermal cells surrounding the stomium (Fig. 2, henceforth referred to as stomial cells), and the cells of the connective tissue and middle layer / endothecium ( Fig. 3) was examined during anther maturation and dehiscence from stages 9 through 20. All cells examined over the time course exhibited ultrastructural characteristics consistent with plant PCD, with some minor variation in these characteristics observed between the cell types.
Many characteristics indicative of plant PCD, were apparent in the stomial cells ( Fig. 2 A-F). These include the vesiculation of the cytoplasm and development of autophagic vesicles ( Fig. 2 B, D), changes in plastid morphology and retraction of the cytoplasm from the cell wall abnormalities in mitochondrial structure (swelling and breakage) often associated with plant PCD, were not prevalent in stomial cells during the progression to death (Fig. 2 A-E) until late in the process. Also consistent with some occurrences of PCD in plants, organelles resembling ricinosomes were common and obvious in the epidermal cells surrounding the stomium. These organelles were evident relatively early in development of the anther, even being present in cells which had recently divided ( Fig. 2A), and persisted in the cells with progression to dehiscence ( Fig. 2 B-D, G-J). The organelles were often associated with the rough endoplasmic reticulum In comparison to the cells of the stomium, PCD in the underlying interlocular septum occurs very quickly (see Fig. 2 A-C, 3 A-C). Ricinosome-like organelles are also seen in these cells prior to death ( Fig. 2A). The accumulation of large calcium oxalate crystals in the vacuoles of the interlocular septum cells, lost during sectioning and resulting in voids, are characteristic of anthers of the Solanaceae and provide a landmark for orientation during ultrastructural studies (lower cell Fig. 2F, upper cells Fig. 3 A-C).
Features consistent with PCD as observed in the stomial cells were also seen in the cells of the connective tissue and middle layer / endothecium ( Fig. 3 A-D and E-H, respectively).
Chromatin condensation and irregularities in nuclear shape were more prevalent than that seen in the stomial cells, and cell wall integrity was increasingly compromised with progression through to dehiscence. In addition, para-crystalline structures, presumably peroxisomes, were often seen in cells of the middle layer / endothecium. Ricinosome-like organelles, however, were not easily observed in either connective tissue or middle layer / endothecial cells until later stages. Close observation revealed that ricinosome-like organelles did exist in these cells early in the progression to death, but that the organelles were small compared to those found in the stomial cells and had contrast similar to the cytoplasm (compare Fig. 3 I-L with 2 G-J, noting the differences in magnification). Size and contrast increased as death became imminent ( Fig. 3 H and M). Interestingly, ricinosome-like organelles were never observed in the tapetum at any stage through the progression to dehiscence (data not shown).
The occurrence of the ricinosome-like organelles in the cells of the stomium, middle layer /endothecium, interlocular septum and connective tissue, similar to those found in cells of castor oil seed endosperm and nucellus that are destined to die, strongly suggested that the terminal steps of PCD in these cells and tissues involved a KDEL-tailed cysteine proteinase(s).
Further research was conducted to substantiate this.

Tomato has a gene encoding a KDEL-tailed cysteine proteinase
Ricinus communis CysEP encodes a KDEL-tailed cysteine proteinase that accumulates specifically in ricinosomes and is known to be a predictor of some incidences of PCD in that species (Schmid et al., 1999). A Basic Logical Algorithm Search Tool (BLAST) search of the NCBI database was conducted using the CysEP mRNA sequence (GenBank accession number AF050756) and a highly homologous, uncharacterized cDNA from tomato fruit was identified In silico translation of the open reading frame of SlCysEP predicts a 360 amino acid papain-like pre-pro cysteine proteinase containing a hydrophobic N-terminal signal peptide and having a C-terminal KDEL ER-retrieval motif, among other features (Fig. 4). Alignment of the predicted SlCysEP amino acid sequence to those of CysEP (Ricinus communis), SH-EP (Vigna mungo), TPE4A (Pisum sativum) and SEN11 (Hemerocallis spp) demonstrates that there is a significant amount of amino acid identity between SlCysEP and these senescence-related proteinases (Fig. 4). CysEP and the SlCysEP translation product both contain 360 amino acids with 76% identity, and this comparison allowed a prediction of the SlCysEP cleavage sites based on the known cleavage sites within CysEP (Fig. 4). The signal peptide is likely cleaved between amino acids 20 and 21, and the propeptide between amino acids 125 and 126 producing a 235 amino acid mature enzyme (amino acids 126 to 360). The predicted molecular weights (MW) of unprocessed SlCysEP, the pro-protein and mature enzyme are 40.6, 38.3 and 25.5 kDa respectively.
Southern analysis, using a DIG-labeled cDNA probe complementary to 479 base pairs of exon 1, against restriction digested genomic DNA of tomato suggests that SlCysEP is a present as a single copy in the tomato genome (Fig. 5).
addition of DTT but was completely abolished by the cysteine proteinase-specific inhibitor, E64 ( Fig. 6B). Human cathepsin B, a well characterized cysteine proteinase, behaved similarly, although specific activity against azocasein was much lower than that of SlCysEP (Fig. 6B). Recombinant pro-SlCysEP undergoes self-hydrolysis at pH 4.8 with complete conversion from an isoform of about 45 kDa to an isoform of about 28 kDa isoform occurring within 12 minutes; no processing occurs at pH 7.0 ( Fig. 6C) . At pH 4.8, an intermediate cleavage product of 43 kDa was observed immediately upon the addition of the acid, and two intermediate cleavage products of 32 and 30 kDa could be seen after 3 minutes. Only the 28kDa isoform was present after 12 minutes of incubation (Fig. 6C).

SlCysEP is restricted to specific tissues of the anthers during floral development and senescence
Northern blot analyses determined that SlCysEP transcripts were abundant in stage 13 to 15 flowers and that expression in these flowers was restricted to the stamen (Fig. 7A). Western immunoblot analyses confirmed that SlCysEP was only minimally expressed in stage 1 to 12 flowers (Fig. 7B). Three peptides were detected in stage 13 to 15 flowers; a 44 kDa peptide, slightly larger than the predicted molecular weight of the entire SlCysEP protein (40.6 kDa), a smaller peptide of approximately 43 kDa that could be a processed form lacking the signal peptide (predicted at 38.3 kDa), and a 29 kDa peptide similar to the predicted molecular weight of the mature enzyme (25.5 kDa) (Fig. 7B). The 29 kDa peptide alone was detected in stage 18 to 20 flowers. Accumulation of the proteinase was restricted to the stamens (Fig. 7B).
To address the possibility that the affinity-purified anti-SlCysEP antibodies were cross reacting with other cysteine proteinases, an immunoprecipitation experiment was performed using total protein extracted from stage 18-20 anthers. A single peptide corresponding to the 29 kDa peptide in Fig. 7B was pulled down (see Fig. S1). That the antibodies did not detect any proteins in other senescencing floral tissues, at the time when the anthers are becoming competent to dehisce (see Fig. 7B), is also good evidence that there is minimal cross-reactivity with other cysteine proteinases.
Immunohistochemistry revealed that SlCysEP is restricted to the interlocular septum in anthers from stage 11 flowers (Fig. 8A), but later accumulates in the cells of the stomium, the middle layer / endothecium and the connective tissues (Fig. 8B). With progression to dehiscence, SlCysEP becomes restricted to the cells of the stomium (Fig. 8C), with some remaining in the epidermal cells that were surrounding the stomium even after dehiscence is complete (Fig. 8D). No signal was seen in controls using pre-immune serum (Fig. 8, E-H).

SlCysEP localizes to ricinosomes in cells undergoing PCD
ImmunoGold labelling using rabbit anti-SlCysEP localizes SlCysEP to the ricinosomelike organelles in the stomial, interlocular septum and middle layer / endothecium cells during development and dehiscence ( Fig. 9), confirming that the organelles are indeed ricinosomes.
Accumulation of the enzyme begins as early as stage 9 in the stomial and interlocular septum cells ( Fig. 9 A and E) and by approximately stage 13 in the cells of the middle layer / endothecium as tapetal degeneration proceeds (Fig. 9J). Labelling intensity over the ricinosomes increased as the ricinosomes became more electron dense with increased maturtity of the cells and tissues ( Fig. 9 B, F, G, K and L). At death the ricinosomes rupture, in some cases evidently in conjunction with the loss of tonoplast integrity (Fig. 9C), releasing SlCysEP into to the cytoplasm (astrices with arrows, Fig. 9 C, D, I and M) which has shrunk away from the cell wall

Multiple cysteine proteinase genes are expressed throughout anther development
Given that multiple cysteine proteinases are often expressed during PCD (reviewed in Trobacher et al., 2006), there is a strong possibility that other cysteine proteinases act either before, or concurrently, with SlCysEP. To investigate this possibility, gene-specific primer sets were designed to amplify fragments of transcripts from several tomato cysteine proteinase genes, or unigenes. Reverse transcription-PCR performed on total RNA extracted from anthers at stages 1-13, 13-18, and 18-20 confirmed that multiple cysteine proteinase genes are expressed in tomato anthers (Fig. 10). SlCysEP expression in stages 1-13, and 13-18 is similar, but decreases in stages 18-20. TDI-65, encoding a drought-inducible 65 kDa cysteine proteinase found in the nucleus, chloroplasts, and some cytoplasmic regions of mesophyll cells (Tabaeizadeh et al., 1995) is expressed evenly in all stages examined. Expression of SlCYSPRO, a member of the C1A family of cysteine proteinases, is similar to TDI-65. SGN-U321596, a unigene predicted to encode a 42.7 kDa C1A family cysteine proteinase similar to Arabidopsis XBCP3 (Zhao et al., 2000), is also evenly expressed in anthers at all stages examined. Transcipts encoding a putative 40.3 kDa metacaspase, SlMCA2, belonging to the C14B family of cysteine peptidases, are detectable in stages 1-13, however, expression increases through stages 13-18 and 18-20.
Expression of another putative C1A family member-unigene, SGN-U321072, could not be detected despite EST data suggesting it is expressed in flower buds (data not shown).

Discussion
The progression of programmed cell death in the non-tapetal sporophytic tissues of the tomato anther, from mid-development leading up to dehiscence, has been documented using light and electron microscopy. Our observations of the morphological and cytological features of anther development and dehiscence are consistent with those observed previously for Solanum lycopersicum (Bonner and Dickinson, 1989, and references therein) and tobacco, another member of the Solanaceae (Koltunow et al.. 1990;Goldberg et al., 1993;Sanders et al., 2005).
The general features of anther development, as described by Goldberg et al. (1993) apply very well to tomato, with minor exceptions. Unlike tobacco, where the stomium consists of multiple cell layers (Sanders et al., 2005) and the endothecium is easily identified, the stomium in tomato is only one cell layer thick, and the endothecium is not well-defined and may be absent in large portions of the mature anther (Bonner and Dickinson, 1989).
The majority of previous cytological studies of anther development fail to detail the changes in cellular ultrastructure that accompany the progression to dehiscence. Although a number have provided ultrastructural details of tapetal PCD during pollen development (Brighigna and Papini, 1993;Papini et al., 1999), these tend to ignore other anther tissues that play a fundamental role both in providing additional nutrition for the pollen and in the dehiscence process itself. By documenting changes in nuclear and mitochondrial structure, The timing of the appearance of these characteristics, however, is dependent on cell type.
The cells of the middle layer and connective tissue show a steady progression of those features commonly associated with PCD, with nuclear invaginations and some condensation of chromatin occurring early in the process. Although present, ricinosomes are not well-developed until late in the death program in these cells. In contrast, modifications to nuclear structure in the cells immediately surrounding the stomium occur relatively late in the process, but ricinosomes are well-developed very early, potentially being present even in dividing cells. The cells surrounding the stomium perform a number of functions. They form the first line of defence against pathogen invasion and insect attack during pollen development and maturation, yet present the most vulnerable point of entry as the anther wall in that area is only one cell thick after the loss of the interporangial septum. The stomium must remain impenetrable until the pollen is mature, but is also the site of dehiscence, which itself requires a weakening of cell wall adhesion at the point of the stomium. This weakening of the adhesion between adjacent cells is a late event and is dependent on the activity of wall degrading enzymes produced by the cells themselves (Jenkins et al., 1999). The cells surrounding the stomium are fated to die very early, but must remain alive and be actively synthesizing and exporting wall degrading enzymes, thus they would likely have to maintain functional transcriptional and translational machinery. Our observation that there is little evidence of nuclear invagination and chromatin condensation in these cells until relatively late in the dehiscence process supports this notion.
Perhaps the most novel finding of the current study was that ricinosomes were common to all of the incidences of PCD in the sporophytic tissues followed during tomato anther development and dehiscence, with the exception of that occurring in the tapetum. In agreement with the notion of Rogers (2006), the term "ricinosome" is preferred as the organelles are operationally different from precursor protease vesicles. Unlike the latter, ricinosomes do not fuse with the central vacuole; instead, they lyse upon the collapse of the central vacuole at death, releasing their contents. The appearance of ricinosomes has been correlated with the occurrence of PCD in the nucellus of developing, and endosperm of post-germinative, castor oil seed (Ricinus communis) (Schmid et al, 1999;Geitl and Schmid, 2001;Greenwood et al., 2005), and the term ricinosome is indeed derived from the genus name (Mollenhauer and Totten, 1970).
Outside of these, ricinosomes have been observed and linked to the PCD of endosperm cells in post-germinative tomato seed (DeBono and Greenwood, 2006) and, perhaps, with the senescence of Day Lily, flower petals (Schmid et al., 1999). No other instances of ricinosome involvement in plant PCD have been documented until the present study.
In castor oil seed, ricinosomes harbour CysEP, an inactive pro-cysteine proteinase having a KDEL ER-retrieval motif. The enzyme functions in the final stages of corpse processing during PCD of the nucellar and endospermic cells during development and following germination, respectively (Schmid et al, 1999;Geitl and Schmid, 2001;Greenwood et al., 2005) and may be involved additionally in the digestion of extensins in the walls of the dead cells (Helm et al, 2008). The tomato genome contains a single copy of SlCysEP, a gene encoding an enzyme homologous to CysEP that localizes to the same organelle, and our results suggest that it has a similar function. Interestingly, the developmental accumulation of SlCysEP mimicked almost precisely the pattern of expression seen for the TA56 transcript in tobacco anthers (Koltunow et al., 1990), which encodes a cysteine proteinase, recently shown to be KDEL-tailed (accession number EU429306). According to the tomato flower development schedule used throughout this research (refer to Brukhin et al., 2003;see also Polowick and Sawhney, 1993), tapetum degeneration begins at stage 13 and is complete by stage 18, which coincides very well with SlCysEP transcript and protein accumulation. As the tapetum is known to undergo PCD, and ricinosomes and their marker proteinase predict PCD in some instances, we fully expected to find SlCysEP in this tissue. Surprisingly, the enzyme was not seen to accumulate in the tapetum but was found in the other sporophytic tissues, including those immediately adjacent to the tapetum. The specificity of localization in this case would have been lost had we relied solely on whole tissue extraction for analysis. We have not yet determined the substrate characteristics necessary for SlCysEP action, but Ricinus communis CysEP is known to have a broad substrate specificity, as would be expected for a corpse processing enzyme (Than et al., 2004). The tapetum, however, not only contributes to the nutrition of the developing pollen grains but also provides materials responsible for sculpting the surface and that can be important in pollen recognition events (Vizcay- Barrena and Wilson, 2006). Having a broad spectrum proteinase involved in the PCD of the tapetum would release the enzyme to the locular chamber, where it could potentially act in damaging or destroying these important pollen surface-located proteins (Wang et al., 2003).
Thus the gradual dismantling of the tapetum through PCD must employ mechanisms distinct from those used in the remaining sporophytic cell types.
While the data presented above clearly demonstrate that SlCysEP is present in ricinosomes within the stomium and the interlocular septum, they do not preclude the possibility that other enzymes may be present as well. Our data demonstrate that several other cysteine proteinase genes are in fact expressed in anthers. These, however, are unlikely to be ricinosomal-resident candidates. SlCYSPRO has a signal peptide suggesting it targets through the ER, but lacks an ER-retrieval signal, a feature that seems to be requisite for residence in ricinosomes. Further, in western analyses, antibodies directed against rSlCysEP do not detect SlCYSPRO, and those against rSLCYSPRO do not detect SlCysEP in total protein extracts of multiple tissues (Senatore, 2006). TDI-65 is reported to localize to the nucleus, chloroplasts and cytosol (Tabaeizadeh et al., 1995), while both SGN-U321596, and SlMCA2 lack a signal peptide and must therefore localize outside of the endomembrane system. Since multiple cysteine proteinases are expressed during instances of PCD we fully expect that other enzymes are present and play a co-ordinated role in the PCD of sporophytic tissues in tomato (Trobacher et al., 2006).
Although not present in all incidences of PCD in plants, when they do occur, ricinosomes and the marker proteinase are only seen in cells that are going to die, and their occurrence predicts the cell's death (Geitl andSchmid, 2001, Greenwood et al., 2005). The enzyme is involved in corpse processing, active only after the collapse of the central vacuole and cell death, and thus should only be expressed in cells that are fated to die. Ricinosomes accumulate within the stomial cells and those of the interlocular septum very early in the development of the anther.
If the premise of Geitl and Schmid (2001) holds, then these cells have been fated to die at the same time. Both cell types lie immediately adjacent to each other, but the cells of the interlocular septum are among the first to undergo PCD, whereas those surrounding the stomium are among the last. This suggests that the initiation of PCD, as identified by the formation of ricinosomes, may be uncoupled from whatever may be acting as an effector to complete the process. In this case, one cell type may be competent and responding to some additional death signal that the other is incapable of responding to until very late. One of the recurring issues that arises in studies of plant PCD is that we do not have any firm understanding as to how the process is initiated and then regulated. Given that the expression of SlCysEP, and concomitant formation of ricinosomes, can occur well prior to death and that these seem to invariably predict that the cells are going to die, examining the factors controlling the expression of SlCysEP may provide some insight as to how PCD in plants is initiated.

Plant materials
Seeds of Solanum lycopersicum L. cv. 'Glamour' were purchased from Stokes Seeds (St. Catharines, ON, CAN). Plants were grown in 3:1 peat:turface in a growth chamber set to 14 hour, 25ºC days : 10 hour, 22ºC nights (Brukhin et al., 2003). Lights were maintained at an intensity of 205 µM m -2 s -1 . Plants were watered three times per week, and were supplemented once a month with 20-20-20 fertilizer, 1 g L -1 , 500 mL per plant. The S. lycopersicum flower development schedule used for this research was proposed by Brukhin et al., (2003) and is based on morphological landmarks in development. Entire flowers were directly ground in liquid nitrogen for developmental stage protein and RNA extraction; flowers were dissected into pedicel, sepal, petal, stamen, and carpel prior to grinding for floral tissue protein and RNA extraction.

Bright Field Microscopy
Stamens were fixed and embedded in paraffin according to de Almeida Engler et al., (2001).  (Table I) (Table I) and SlCysEP R1 are complementary to the 5' and 3' extremes of the BTO14302 mRNA sequence and were used to PCR-amplify the SlCysEP cDNA from the RT sample. The cDNA was subsequently cloned into pGEM®-T vector (Promega, Fisher Scientific, Nepean, Canada) and sequenced.
DNA was extracted according to Richards et al. (2001). Primers SlCysEP F1 and SlCysEP R1 (Table I) were used to PCR-amplify the genomic DNA complementary to the BT014302 mRNA sequence. Sequences upstream and downstream of the BT014302 coding sequence were obtained using GenomeWalker™ (Clontech, Mountainview USA) according to manufacturer's instructions. Libraries were prepared using genomic DNA digested with EcoRV, NaCl, 0.1% (v/v) TWEEN-20) containing 5% (w/v) skim milk powder, then cut into smaller squares. Three ml of anti-SlCysEP antiserum was added to 7 mL of TTBS and 0.1 g skim milk and the antibody solution incubated with the rSlCysEP squares for 2 hours at RT, washed twice with TTBS for 15 minutes, and then transferred to small trays. Purified antibodies were eluted using 3.0 mL of elution buffer (25 mM glycine pH 2.3, 0.5 M NaCl, 0.5% (v/v)  repeatedly pipetted over each set of squares for 1 minute. The antibody containing supernatants were transferred to separate 15 mL centrifuge tubes each containing 3 mL of neutralization buffer (200 mM Tris pH 7.8). Tubes were vortexed, sodium azide was added to 0.02% (w/v), and affinity-purified antibodies were stored at 4ºC.

Recombinant SlCysEP proteolytic activity
To determine pH optimum of rSlCysEP, citrate-phosphate buffers in a range from pH 3.0 to 8.0 were prepared by combining the appropriate amounts of 0.1 M citric acid pH 2.0 and 0.2 M Na 2 HPO 4 pH 8.4 in 1.5 mL centrifuge tubes (Ruzin, 1999). To each tube, 600 µL of appropriate pH buffer was added, along with 90 µL of 5% (w/v) azocasein solution, 150 µL ddH 2 O, and 10.7 µg of recombinant protein, (Charney and Tomarelli, 1947). The samples were incubated in the dark at 37ºC O/N. The following day 150 µL of ddH 2 O and 150 µL of 50% (w/v) trichloroacetic acid (TCA) were added. Tubes were vortexed and centrifuged at 13 000 x g for 10 minutes at 4ºC to pellet undigested azocasein. Control samples were prepared similarly, with TCA being added prior to the addition of the enzyme. 174 µL aliquots from each sample were pipetted into a microwell plate and 26 µL of 10 M NaOH was added to each well. The plates were read at 490 nm using the THERMO max microplate reader (Molecular Devices, Sunnyvale, USA).
Citrate-phosphate buffers were prepared (Ruzin, 1999) at pH 4.8 and 5.6 in order to accommodate for the respective pH optima of SlCysEP and human cathepsin B (Azaryan et al., 1985). To 450 µL of the appropriate citrate-phosphate buffer was added 100 µL of 2.5% (w/v) only with saturated aqueous uranyl acetate, and imaged on the CM-10 TEM as above. Controls grids were treated identically using either pre-immune rabbit antiserum or rabbit-anti SlCysEP IgG at approximately 28 µg•ml -1 pre-absorbed with 200 µg•ml -1 rSlCysEP protein.
Sequence data from this article can be found in the GenBank/EMBL data libraries under accession number EU122386 (Solanum lycopersicum KDEL-tailed cysteine endopeptidase [CysEP] gene, complete cds).

Reverse Transcription-PCR
Total RNA was extracted from dissected anthers using TRI REAGENT™ (Sigma) according to the manufacturer. Each sample was treated with TURBO DNAfree™ DNase (Ambion) according to the manufacturer and used immediately for cDNA synthesis. First-strand         Gene specific primers were used to amplify regions of cysteine proteinase transcripts via reverse transcription-PCR on total RNA from stage 1-13, 13-18, and 18-20 anthers. Actin is included as a loading control.  Ricinosomes may be more numerous and the cytoplasm becomes somewhat flocculent in appearance (D).
The interlocular septal cell is dead by this stage (lower right, in C), evidenced by the loss of cytoplasm (see also    ) Effect of DTT and E 64 on proteolysis by rSlCysEP and human cathepsin B at optimal pH at 37ºC. DTT (white bars) increases activity of both enzymes over that seen in control reaction (gray bars). E64 (black bars) completely abolishes activity. Error bars represent the standard deviation between triplicate independent assays. C) Western analysis of rSlCysEP self-hydrolysis in acidic pH. Self-hydrolysis occurs within three minutes at pH 4.8, but does not occur not at pH 7.0. Values above each lane indicate reaction time in minutes.
A) pH optimum for rSlCysEP activity at 37ºC using azocasein as substrate. Error bars represent the standard deviation between triplicate absorbance values for each pH.