In parallel to studies on the identification of the protein constituents of wine haze, work to defme proteins associated with the onset of ripening (veraison) in grapes was per- fonned. This led to the defmition of three proteins clearly synthesised at elevated rates soon after, but not before, veraison (Tattersall et al., 1997; Robinson et aI., 1997). Sur- prisingly, these proteins proved identical to those previously identified by Waters et af.
(1996) as haze proteins in wine. Cloning of cDNAs encoding the major PR5-like protein (thaumatin-like protein VVTL 1) and the PR3 proteins (chitinases) of grapes along with production of antibodies made it possible to characterise both the temporal and spatial accumulation of these proteins in V. vinifera.
In V. vinifera cv Muscat of Alexandria, both the expression of the VvtlJ gene, and the levels of the corresponding thaumatin-like protein increased dramatically immediately after the onset of berry softening (veraison) (Tattersall et al., 1997). Indeed, neither Northern nor Western blotting revealed any gene expression prior to veraison. VvtlJ con- tinued to be expressed and protein continued to accumulate throughout berry ripening.
Its expression is specific to berries with no evidence of its occurrence in other organs or tissues of V. vinifera. The deduced protein sequence of VvtlJ is identical to that of the thaumatin-like protein originally identified as a major haze fonning protein in wine by Waters et al. (1996).
A study on V. labruscana cv Concord using antibodies raised against tobacco PR- proteins (Salzman et al., 1998) also showed that thaumatin-like proteins and chitinases accumulate during berry ripening. A number of other studies have shown that chitinase activity also increases during berry development of V. vinifera (Derckel et aI., 1996;
Robinson et al., 1997; Derckel et al., 1998) and two genes have been identified in V.
vinifera cv Shiraz that encode the chitinases expressed during development (Robinson et al., 1997). The protein sequences deduced from these genes exhibit a high degree of similarity to the chitinases identified as major haze fonning proteins (Waters et al., 1996; Waters et al., 1998). The accumulation pattern observed for the smaller and probably less significant 13 kDa juice and wine protein, VVPR4a, was indistinguishable from that seen for the PR3 (chitinases) and PR5 (thaumatin-like proteins) families, at
188 D.B. TATTERSALL et al.
both the protein and the gene transcript level, except that the mRNA encoding Vvpr4a could also be detected in flowers of V. vinifera prior to berry set (Tattersall, 1999).
Southern blotting clearly revealed that all of the major berry PR proteins are encoded by genes which are members of multi gene families (Tattersall et aI., 1997; 1999; Robin- son et aI., 1997) consistent with the fact that isoforms of both the PR2 and PR3 proteins have been detected by conventional protein techniques and through cloning of highly homologous cDNAs (Loulakakis, 1997a; 1997b; Tattersall,1999).
s. THE REGULATORY ELEMENTS CONTROLLING PR-LIKE PROTEIN SYNTHESIS AT VERAISON ARE NOT KNOWN The trigger at veraison for the onset ofVVTLl and VVPR4a synthesis and accumulation throughout the remainder of berry ripening is unknown, as is indeed the trigger for any gene activity induced at veraison. In a variety of plants, a number ofPR-like proteins are induced in response to a suite of signals including pathogen attack, wounding, and sig- nalling molecules such as salicylic acid, abscisic acid and ethylene. It would be tempting to speculate that the accumulation of berry PR-like proteins is a result of pathogen attack although this now appears unlikely, due to the ubiquitous presence ofPR-like proteins in apparently healthy grapes (Tattersall et at., 1997; Pocock et at., 1999). Their homo- logues are also found in a number of other, apparently disease free, ripe fleshy fruits such as cherries, bananas and kiwifruits. Additionally, no ~-1,3-glucanase activity, a potential indicator of pathogen attack (Stintzi et at., 1993), could be detected in berry extracts in our laboratory (results not shown), and has not been detected in other ripening berries in which PR-like proteins are known to accumulate (Robinson et aI., 1997;
Derckel et at., 1998; Kraeva et aI., 1998), despite the recent cloning of a partial cDNA encoding a ~-1,3-glucanase from cultured berry cells (Kraeva et aI., 1998).
A factor which might induce berry PR-like accumulation is osmotic stress. A change in the osmotic pressure of grape berry juice occurs very rapidly at the beginning of stage III of berry development, and over the entire ripening period following veraison, there can be an increase in osmotic pressure of about 2,000 KPa (Downton and Loveys, 1978). The major solutes that contribute to this osmotic pressure change are fructose and glucose, both of which accumulate to near molar levels in grape juice. Interestingly, four tobacco PR3 protein family members (Yun et aI., 1996), and the PR5 protein family members, osmotin and NP24, which were originally purified from salt stressed tobacco and tomato cells, re- spectively (Singh et at., 1987; King et at., 1988), all accumulate in tobacco at levels di- rectly related to the degree of osmotic stress. Osmotin and NP24 are known to have a pre- dominantly intracellular location, as opposed to the presumed extracellular location of VVTLl and VVPR4a (see Tattersall, 1999). Therefore, whether VVTLl and VVPR4a have a role in osmotic stress adaptation or accumulate non-specifically in response to such stresses remains unknown. However, the gene encoding the potentially extracellular PR5 like protein, VVOSMI from grapevine, is known to be induced in cultured grapevine cells
PATHOGENESIS RELATED PROTEINS 189 in response to salt stress (Loulakakis, 1997a).
High levels of free hexoses have now been shown to repress the synthesis of chloro- phyll alb-binding proteins and induce the synthesis of a number of defence-related genes, such as proteinase inhibitor II (Johnson and Ryan, 1990), chalcone synthase (Tsukaya et a!., 1991) and four PR proteins (PAR-I, PR-lb, PR-Q and SAR 8.2) in the leaves of transgenic tobacco plants (Herbers et al., 1995). Such photo assimilate induced gene expression occurs only above a certain threshold hexose concentration and the situation is thus reminiscent of that seen in grape berries, when the onset of stage III (veraison) occurs. Therefore, it is suspected that the temporally and spatially defined induction of not only Vvtll and Vvp4a, but perhaps of a majority of ripening related genes, directly or indirectly is caused by the onset of sugar accumulation, possibly due to the presence of regulatory DNA sequences (so called "sugar boxes") (Tsukaya et al., 1991), within the promoters of these genes. The mechanism by which such hypothetical
"sugar sensing" and intermediate signals operate remains unclear. It has recently been demonstrated that the presence of an invertase in the vacuole in tobacco leaves leads to systemic acquired resistance (Herbers et a!., 1996a) and that hexose sensing therefore appears to take place in the secretory pathway. In this context, it is important to note that vacuolar grape invertases are synthesised well in advance of veraison (Davies and Rob- inson, 1996) with no apparent Vvtll and Vvpr4a gene expression as a result, possibly because the free hexoses generated do not reach critical threshold levels at early stages of berry development, due to a multitude of competing metabolic demands. Preliminary experiments performed in our laboratory to detect an effect of high sugar concentration (1 M glucose or I M fructose) on PR-like protein gene expression in detached and halved pre-veraison berries were inconclusive (Tattersall, 1999).
Another consideration in determining what triggers Vvtll, Vvpr4a and indeed class IV chitinase (Robinson et al., 1997) gene expression at veraison is that of the role of signal- ling molecules. For example, salicylic acid could playa role in gene induction as it is a product of the phenylpropanoid pathway, which is induced in grape berries at veraison (Boss et a!., 1996). The concentration of salicylic acid during berry ripening remains unknown, however, exogenous application of salicylic acid is known to induce chitinase activity in wounded grapevine leaves (Derckel et aI., 1996) and berries (Derckel et al., 1998). Conversely, it has now been demonstrated that sugar-dependent induction of PR proteins discussed previously, can occur independent of salicylic acid (Herbers et aI., 1996b). Abscisic acid (ABA) is a plant hormone which is known to accumulate to high levels during grape berry sugar accumulation (Coombe and Hale, 1973), and to be capa- ble of inducing a PR protein encoding gene in cultured grapevine cells (Loulakakis,
1997a). ABA is known to be upregulated during osmotic adaptation of plant cells (Skriver and Mundy, 1990). Furthermore, the treatment of berries with a synthetic auxin, benzothiazole-2-oxyacetic acid (BTOA), delayed the onset of ripening, ABA production, and the induction of four different ripening related genes in parallel (Davies et al., 1997).
Alongside the potential role of ABA, that of indole acetic acid (IAA) must also be con- sidered. IAA, which is present at high levels during early berry development (Cawthorn
190 D.B. TATTERSALL et ai.
and Morris, 1982), may repress berry PR-like protein gene expression until after verai- son, when IAA levels become very low and ABA levels peak. As the grapevine berry is a non-climacteric fruit (Kanellis and Roubelakis-Angelakis, 1993), it is unlikely that ethylene, which is an inducer of PR protein encoding genes in many plants, has a major role in berry PR-Iike protein accumulation.
The isolation and characterisation of the VVTLl and VVPR4a encoding genes and their sequencing (Tattersall, 1999) potentially provides the materials needed to elucidate which factors trigger the accumulation of PR-Iike proteins at the onset of ripening. Char- acterisation of the Vvtll and Vvpr4a genes in our laboratory (Tattersall, 1999) revealed the existence of two putative regulatory elements, which are present in promoter regions of both genes. This suggests that VVTLl and VVPR4a may accumulate in the berry as a result of a specific developmental program coordinated by the action of common tran- scription factors. The sequence of one element is highly similar to silencer elements pre- sent in the promoter of the gene encoding bean chalcone synthase (chs15) (Lawton et al., 1991). These silencer elements bind a protein factor, SBF -I (Harrison et al., 1991), which has been suggested to repress transcription of chs15 in hypocotyls, yet enhance it (transcription of chs15) in young cotyledons (Lawton et aI., 1991). The other putative regulatory element is near identical to a core sequence required to bind the myb-Iike transcription factor P in maize (Grotewold et al., 1994). P binds to and activates tran- scription of the al gene, which is required for the formation of pigments in the pericarp of the cob (Grotewold et al., 1991). A similar protein may control transcription of the Vvtll and Vvpr4a genes, which are highly expressed in the ripening pericarp tissue of the grape. The involvement of promoter elements, homologous to the fore-mentioned, may indeed drive a berry specific transcription, unique to the ripening phase. Unfortunately the non-coding DNA sequences of other ripening induced grape berry PR-like protein genes have yet to be defined.