DELLA factors promote high ABA accumulation in imbibed seeds
Dry seeds contain endogenous ABA, which plays an essential role during embryogenesis to establish seed osmotolerance and dormancy. Normal germination conditions, such as seed imbibition under white light, trigger a decrease in endogenous ABA levels within the first 12 hours upon seed imbibition, reaching the lowest ABA concentration between 12 and 24 hours after imbibition (Ali-Rachedi et al., 2004; Piskurewicz et al., 2008). However, seeds that cannot synthesize GA maintain high ABA levels (and further increase them) upon seed imbibition and
do not germinate in a DELLA-dependent manner (Piskurewicz et al., 2008; Piskurewicz et al., 2009). Indeed, under white light conditions ga1/rgl2 fails to sustain high ABA levels upon seed imbibition, similar to ga1/rgl2/gai/rga that under both white light and far-red conditions does not maintain high ABA levels (Piskurewicz et al., 2009). Lower ABA levels in ga1/rgl2 and ga1/rgl2/gai/rga mutants upon seed imbibition might result from lower ABA accumulation in dry seeds or from the failure to sustain ABA production upon imbibition. The first possibility seems unlikely because ga1 dry seeds contain roughly the same ABA amounts as ga1/rgl2 and ga1/rgl2/gai/rga (Piskurewicz et al., 2009). This suggests that DELLA factors are necessary to sustain high ABA levels upon seed imbibition and not during seed development or maturation, consistent with the observation that their expression is induced upon imbibition. Indeed, RGL2, GAI and RGA mRNA and protein levels are almost undetectable in dry seeds and rise upon seed imbibition (Piskurewicz et al., 2008). Under white light conditions, where RGL2 plays a major role to repress germination, the levels of ABA are sustained only under low GA conditions, i.e.
when the RGL2 protein highly accumulates but not under normal germination conditions when it is expressed transiently and does not persist further than 48 hours after seed imbibition (Piskurewicz et al., 2008). Thus, the pattern of RGL2, GAI and RGA expression is consistent with the genetic evidence for DELLA-dependent ABA accumulation upon seed imbibition. However, it was shown that after-ripened sly1 mutant seeds can germinate despite persistently high RGL2 accumulation (Ariizumi, 2007). We have confirmed this observation and have shown that sly1 germination was correlated with ABI5 protein disappearance but not that of RGL2 (Piskurewicz et al., 2008). This suggests that RGL2 activity to promote ABA accumulation can be lost during the process of after-ripening (Ariizumi, 2007; Piskurewicz et al., 2008).
How DELLA factors promote accumulation of ABA is not known. It was proposed that under normal germination conditions, the drop in ABA levels is a result of CYP707A2 activity (Kushiro et al., 2004). Thus, DELLA factors might sustain high ABA levels under low GA conditions by preventing the expression or activity of this enzyme. However, this seems unlikely.
Indeed, ABA levels drop about 20 fold, in both wild type seeds and seeds unable to synthesize GA, within the first 12 hours after dry seed imbibition (Piskurewicz et al., 2008). However, when GA synthesis is prevented, the endogenous ABA levels increase reaching 10 fold higher levels relative to seeds able to synthesize GA in a DELLA-dependent manner (Piskurewicz et al., 2008;
Piskurewicz et al., 2009). Thus, DELLA factors must act to stimulate de novo ABA synthesis in
addition to the possibility that they may prevent ABA catabolism. They could act to stimulate the expression or activity of any of ABA biosynthetic enzymes or else prevent that of AtBG1, coding for the enzyme that hydrolyses the inactive ABA-GE form to the bioactive one (see Introduction p.38). Moreover, they could stimulate the rise in bioactive ABA concentration by stimulating XERICO expression, XERICO being the only DELLA specific target reported so far (discussed below).
DELLAs may sustain high ABA levels by stimulating XERICO expression
We have shown that repression of seed germination under low GA conditions is correlated with increased accumulation of XERICO mRNA (Piskurewicz et al., 2008;
Piskurewicz et al., 2009). XERICO is a zinc-finger protein that contains a RING-H2 motif (Ko et al., 2006). When compared to wild type Arabidopsis plants, transgenic plants over-expressing XERICO exhibited hypersensitivity to salt and osmotic stress during germination and early seedling growth and accumulated higher endogenous ABA levels (Ko et al., 2006). Moreover, xerico mutants accumulate lower ABA levels in dry seeds (Zentella et al., 2007) suggesting that XERICO is involved in promoting ABA accumulation. Interestingly, although the over-expression lines accumulated substantially higher levels of ABA than wild type plants even without stress conditions, the GeneChip analysis showed that the expression of genes involved in ABA biosynthesis was not substantially changed (Ko et al., 2006). This suggests that XERICO might be involved in the post-transcriptional control of factors involved in ABA homeostasis. On the other hand, a yeast-two-hybrid screen indicated that XERICO interacts with an E2 ubiquitin-conjugating enzyme (AtUBC8) and ASK1-interacting F-box protein (AtTLP9), which is a positive regulator of ABA signaling during seed germination (Lai et al., 2004; Ko et al., 2006).
Thus, XERICO might function on ABA homeostasis through ubiquitin/proteasome-dependent substrate-specific degradation of ABA synthesis inhibitors by interacting with AtTLP9 (Ko et al., 2006).
Our experiments showed a sharp increase in XERICO mRNA accumulation under low GA conditions (Piskurewicz et al., 2008; Piskurewicz et al., 2009). Interestingly, XERICO expression was not stimulated by exogenous ABA (Piskurewicz et al., 2008). Thus, XERICO is the only GA-dependent and DELLA-specific target reported so far (Piskurewicz et al., 2008).
How DELLA factors, especially RGL2, promote XERICO mRNA accumulation during seed germination and how XERICO stimulates endogenous ABA production under low GA conditions is not known. Chromatin immunoprecipitation assays showed that the DELLA factor, RGA, is associated with XERICO’s promoter sequences (Zentella et al., 2007). However, the enrichment in XERICO sequences was subtle suggesting that RGA-DNA interaction may be indirect and may involve other transcription factor(s). It is tempting to speculate that RGL2 also directly or in combination with other unknown transcription factors stimulates XERICO expression in order to modulate ABA biosynthesis.
The steady-state levels of bioactive ABA in the seed result from the balance between the synthesis and degradation processes. Elevation in ABA concentration might result from increased expression of any gene encoding enzymes involved in ABA biosynthesis or from decreased expression of enzymes encoding ABA catabolic enzymes. Alternatively, increases in ABA levels might result from increased activity of ABA biosynthetic enzymes or decreased activity of ABA catabolic enzymes. XERICO could be involved in the regulation of any of those processes. Germination of the ga1/rgl2 mutant under white light conditions is associated with low NCED6 levels while the non-germinating ga1 mutant accumulates high levels of NCED6 mRNA (Piskurewicz et al., 2009). Similarly, germination of the ga1/rgl2/gai/rga quadruple mutant under far-red light conditions was associated with low NCED6 levels while the ga1/rgl2 mutant that cannot germinate under those conditions accumulated high NCED6 (Piskurewicz et al., 2009). However, it is hard to conclude if ABA biosynthetic genes like NCED6 are direct targets of DELLA factors or their expression reflects only the ability of ga1/rgl2 and ga1/rgl2/gai/rga to germinate. In this respect, the XERICO expression pattern is unique.
XERICO expression is not stimulated by exogenous ABA but it is specifically up-regulated by low GA levels and in a DELLA-dependent manner (Piskurewicz et al., 2008). Thus, XERICO is the only DELLA specific target during seed germination reported so far.
It remains unclear whether XERICO accumulation is sufficient for a rise in ABA concentration so as to inhibit seed germination under low GA conditions. The levels of ABA in xerico mutant seeds germinating under low GA levels were not measured. The characterization of xerico mutant germination showed its weak insensitivity to PAC treatment (Piskurewicz et al., 2008). Thus, it is unsurprising that xerico mutants maintained higher ABI5 levels than rgl2 mutant under low GA conditions (Piskurewicz et al., 2008). This indicates that the xerico mutant
accumulates high ABA levels under low GA conditions and that other RGL2-dependent factors exist, playing a role to elevate endogenous ABA. On the other hand, low levels of xerico insensitivity to PAC might be partially due to the fact that it accumulates wild type-like RGL2 protein levels, which strongly inhibit testa rupture, thus preventing germination by mechanically blocking endosperm rupture (Piskurewicz et al., 2008). Furthermore, xerico mutant seeds used in our experiments were not a null allele; they accumulate approximately ten fold lower XERICO transcript levels than wild type (Zentella et al., 2007). Thus, germination of other xerico mutant alleles under low GA conditions might help to determine whether XERICO is necessary to promote higher ABA levels and inhibit seed germination under low GA conditions.
How does ABA inhibit endosperm rupture?
Until now I have discussed the fact that DELLA factors are necessary to inhibit testa rupture and to sustain elevated ABA concentrations under low GA conditions. In turn, high ABA levels in imbibed seeds efficiently inhibit endosperm rupture. Genetic experiments have shown that two transcription factors, ABI3 and ABI5, are necessary to inhibit endosperm rupture in response to ABA (Giraudat et al., 1992; Finkelstein and Lynch, 2000; Lopez-Molina L, 2000).
Indeed, abi3 and abi5 mutants can rupture the endosperm in the presence of exogenously applied ABA and, as expected, under low GA conditions (Giraudat et al., 1992; Finkelstein and Lynch, 2000; Lopez-Molina L, 2000; Piskurewicz et al., 2008; Piskurewicz et al., 2009). However, the mechanism by which those two factors inhibit endosperm rupture is unclear.
It was proposed that prior to radicle protrusion through the seed coat, the endosperm undergoes a process called weakening, during which cell wall modifying enzymes (mostly hydrolases) trigger loosening of endosperm cell adhesions. In the Solanaceae family endo- β-mannanases and β-1,3-glucanases were proposed to be involved in the endosperm weakening process. Moreover, β-1,3-glucanase mRNA expression was shown to be repressed by ABA (Leubner-Metzger et al., 1995). Thus, a possible mechanism for the ABA-imposed and ABI3- and/or ABI5- dependent endosperm rupture inhibition is repression of enzymes that are involved in cell wall hydrolysis. However, it is difficult to address this hypothesis because enzymes involved in endosperm weakening in Arabidopsis have not yet been identified. A recent report showed that ABI5 positively regulates the expression of PGIP1 and PGIP2, which encode
polygalacturonase-inhibiting proteins (Kanai et al., 2010). It has been reported that these two PGIP proteins act redundantly to inhibit polygalacturonases, which are known to degrade pectin in the cell wall (Ferrari et al., 2003; Ferrari et al., 2007). Pectin is known to be associated with cell-wall adhesion and strength (Willats et al., 2001). Consistent with their role in protecting pectin in the cell-wall from degradation, PGIP1 and PGIP2 were localized in the cell wall and in the apoplast (Boudart et al., 2005; Irshad et al., 2008). However, the ABI5-dependent regulation of PGIPs seems to be indirect since Kanai et al. could not identify any ABI5 consensus binding sequences within the 5kb upstream region of either PGIP1 or PGIP2 (Kanai et al., 2010).
ABI3 direct binding to DNA was not reported and ABI3 is proposed to activate its targets by interacting with other transcription factors, such as ABI5 (Nakamura et al., 2001). Thus, in response to ABA, ABI3 might recruit ABI5 to promote PGIP1 and PGIP2 accumulation in the endosperm. Consistent with this hypothesis, both ABI3 and ABI5 are expressed in the endosperm layer in ABA and PAC treated seeds (Penfield et al., 2006).