A PLDδ-knockout mutant was previously isolated from Arabidopsis Wassilewskija ecotype (WS). The loss of PLDδ was confirmed by the absence of its transcript, protein and activity (Li et al., 2008).

Two Arabidopsis genotypes were grown in water in a controlled growth chamber at 23°C (day) and 19°C (night) and 60% relative humidity under a 12 h photoperiod, with fluorescent lighting at 120 μmol m^-2 sec^-1. Fully expanded leaves of the same age were collected from approximately 6-week-old plants of the two genotypes of Arabidopsis; the detached leaves were rinsed briefly with sterile water and placed with the adaxial side up in Petri dishes containing 50 μM ethephon (Sigma, C0143). We chose ethephon rather than ethylene for incubation of the detached leaves because it is easier to control, but has an identical effect on the detached leaves. The leaves were incubated at 23°C under a 12 h photoperiod and light at 120 μmol m^-2 sec^-1.

Chlorophyll was extracted incubation of leaves after incubation with ethephon in, 80% acetone. Chlorophyll content was determined spectrophotometrically at 663 and 646 nm as described previously (Craftsbrandner et al., 1984). Chlorophyll fluorescence was analyzed using an imaging chlorophyll fluorometer, MAXI-Imaging Pulse-Amplitude (PAM; Walz, Germany; Bonfig et al., 2006). The maximal quantum yield of photosystem II (PS II; F_v/F_m) photochemistry was measured after adaptation to complete darkness for 20 min.

Cell death, indicated by loss of plasma membrane integrity, was quantified spectrophotometrically by Evans blue staining of detached leaves, using a previously described method with minor modifications (Rea et al., 2004). Briefly, detached leaves were incubated with 0.1% (w/v) Evans blue for 2 h with shaking, and then washed extensively to remove unbound dye. The leaves were ground into powder in liquid nitrogen. The tissue powder was incubated with 50% (v/v) methanol and 1% (w/v) SDS at 60°C for 30 min, and then centrifuged. For a control measurement of 100% cell death, the leaves were heated at 100°C for 5 min. Absorbance was measured at 600 nm.

The processes of lipid extraction, ESI-MS/MS analysis and quantification were performed in accordance with a protocol from Welti et al. (2002). Data processing was performed as previously described. The lipids in each class were quantified by comparison with two internal standards of the class. Five replicates of each treatment for each genotype were analyzed. The Q test was performed on the total amount of lipid in each head-group class, and data from discordant samples was removed (Devaiah et al., 2006).

Statistical analysis was performed using Origin 7.0 (Origin Lab Corporation, Northampton, MA, USA). For all quantitative measurements in this study, five replicates from each sampling time were analyzed. The data were subjected to one-way ANOVA analysis (duncan’s multiple range test) of variance with SPSS 16.0.

Leaves detached from WS plants started yellowing 1 day after treatment and turned almost completely yellow 5 days after incubation in 50 μM ethephon under light. In contrast, most parts of the PLDδ-KO leaves were still green after the 5-day ethylene treatment, which indicated a much slower senescence process in the PLDδ-deficient leaves (Figure 1A, top). Consistent with the visible yellowing, the photochemical quantum efficiency of the photosystem II (PS II) reaction center (F_v/F_m) in ethylene-treated leaves was much lower in WS plants than in PLDδ-KO mutants (Figure 1A, bottom). Measurements of chlorophyll content showed that chlorophyll was lost more quickly from WS leaves, diminishing by 37% after 5 days, whereas PLDδ-KO mutant leaves lost just 20% of their chlorophyll content upon treatment with ethylene (Figure 1B, top). Data on cell viability showed that the rate of cell death was significantly higher in the WS leaves during the ethylene-promoted senescence process, as measured by Evans blue staining (Figure 1B, bottom). These results indicate that the suppression of PLDδ retarded, to some extent, ethylene-promoted senescence.

Metabolism of membrane lipids is one of several biochemical manifestations of cellular senescence (Thompson et al., 1998). Since the above results suggest that PLDδ is involved in ethylene-promoted senescence, combined with the participation of PLDδ in the metabolism of membrane lipids, we performed ESI-MS/MS analysis to determine whether the degradation of membrane lipids is affected by the suppression of PLDδ during ethylene-promoted senescence. ESI-MS/MS allowed the identification of alterations in >120 diverse polar glycerolipids, including six head-group classes of phospholipids [PC, phosphatidylethanolamine (PE), phosphatidylinositol (PI), phosphatidylserine (PS), PA, and PG] and two head-group classes of galactolipids (MGDG and DGDG; Table 1). Each molecular species was identified in terms of the total numbers of acyl carbon atoms and double bonds (Welti et al., 2002).

As an overview, most lipid species changed dramatically in terms of both their level (absolute value; Figure 2, left) and the composition (relative value; Figure 2, right) during ethylene-promoted senescence in the leaves of both genotypes of Arabidopsis. The levels of most lipids declined in both WS and PLDδ-KO leaves, although there were some differences in the profiles of membrane lipids between plants of the two genotypes during ethylene-promoted senescence (Figure 2). Clustering of the lipid contents of leaves in ethylene-promoted senescence suggested that the ethylene treatment was the main factor inducing the degradation of membrane lipids. The differences between WS and PLDδ-KO leaves subjected to ethylene treatment were greater than those between WS and PLDδ-KO leaves without such treatment (Figure 2). These results suggest that ethylene treatment affected lipid degradation, and that PLDδ participated in this lipid degradation during ethylene-promoted senescence.

During ethylene-promoted senescence in WS leaves, the levels of leaf membrane lipids decreased significantly (Table 1). After ethylene treatment for 5 days, we found that the level of total lipids decreased by 68.7%, from 302.20 nmol/mg (non-senescent leaves, NS) to 94.53 nmol/mg (leaves treated with ethylene for 5 days). The levels of PG, PC, PE, DGDG, and MGDG all decreased significantly, but the abundances of PA, PI, and PS remained unchanged. As shown in Table 1, after ethylene-promoted senescence for 3 days, the levels of MGDG and DGDG, two main classes of plastidic lipid, decreased significantly compared with those in untreated leaves, whereas the levels of the main extraplastidic lipids (PC and PE) decreased significantly only after ethylene treatment for 5 days. These findings imply that plastidic lipids might be degraded before extraplastidic lipids.

To compare lipid degradation between plastidic and extraplastidic membranes further, the changes in the levels of molecular species of PG were analyzed. In Arabidopsis, PG includes four molecular species, namely, PG 34:1 (total carbon number:double bond number), 34:2, 34:3, and 34:4 (Welti et al., 2002). PG 34:4, which contains a 16:1 acyl chain at the sn-2 position, is part of the plastidic membrane, whereas both PG 34:1 and 34:2 are extraplastidic lipids. Of the two molecules that correspond to PG 34:3, one contains a 16:1 acyl chain which is part of the plastidic membrane, whereas the other is extraplastidic (Marechal et al., 1997). During ethylene-promoted senescence in WS leaves, the level of PG 34:4 had decreased by 33.2% at day 3, whereas the levels of the other three PG molecular species, PG 34:3, 34:2, and 34:1, showed no significant decline after ethylene treatment for 3 days (Table 2). These results indicate that the degradation of plastidic membrane lipids occurred before that of extraplastidic lipids during ethylene-promoted senescence.

During ethylene-promoted senescence, the amount of total lipids declined by 50.7% in PLDδ-KO plants (Table 1). Most of this decrease could be attributed to a decrease in plastidic lipids. For example, the level of MGDG in PLDδ-KO leaves decreased by 52.4% (from 217.78 to 103.67 nmol/mg), the level of DGDG decreased by 48.4% (from 31.85 to 16.43 nmol/mg), and the level of PG decreased by 64.0% (from 12.67 to 4.56 nmol/mg). The levels of both total lipids and the main plastidic lipids (MGDG, DGDG) were significantly higher in PLDδ-KO leaves than in WS leaves, whereas no differences in the levels of PC, PE, PI, PA, and PS were detected between the plants with different genotypes (Table 1). Upon further analysis of the content of molecular species of each membrane lipid, we found that the contents of molecular species MGDG 34:2, 34:5, 34:6, and 36:6 as well as DGDG 36:6 in PLDδ-KO detached leaves were much higher than that in WS detached leaves (Table 3). Furthermore, the levels of the plastidic lipids PG 34:4 and 34:3 were higher in PLDδ-KO leaves than that in WS leaves. The extraplastidic lipids PG 34:2 and 34:1 in PLDδ-KO leaves showed no clear difference compared with those in WS leaves (Table 2). The lipids that are largely synthesized and localized in plastids, PG 34:4, MGDG, and DGDG, were previously shown to be the most abundant in leaves (Devaiah et al., 2006). Plastidic lipids have also been shown to have a direct role in photosynthesis (Dormann and Benning, 2002). We also found that the main MGDG species 34:6 and 36:6 significantly decreased at day 3 of ethylene treatment, whereas the level of DGDG 34:6 increased obviously. These results indicated that these head group remodeling of galactolipids might contribute to maintenance of photosynthesis at early stage of leaf senescence. Combining these findings with our results, it is conceivable that the attenuated degradation of plastidic lipids might have contributed to higher photosynthetic activity in PLDδ-KO leaves during ethylene-promoted senescence, and the retardation of ethylene-promoted senescence by the suppression of PLDδ was a consequence of qualitative changes in plastidic lipids.

For the analysis of the relative contents of membrane lipids, for which the data are expressed as mol% lipids, we found that the most important changes concerned the two galactolipids in WS plants after ethylene treatment for 5 days. The MGDG percentage decreased from 77.55% (NS) to 63.70% (leaves treated with ethylene for 5 days). In contrast, the DGDG percentage increased from 9.70% (NS) to 14.22% (leaves treated with ethylene for 5 days). The PG percentage showed a slight decrease in WS plants. In addition, the relative percentages of the non-chloroplastic PE, PC, PI, and PS, which are mainly located in the membranes of non-photosynthetic organelles such as the plasma membrane, endoplasmic reticulum, and mitochondria (Singh and Privett, 1970), increased during senescence, in keeping with the preferential destruction of chloroplast membranes. The ratio of galactolipids/phospholipids decreased from 11.02 to 5.41, which represent a decrease of 50.9% (Table 4). This decrease might have resulted from the degradation of 67.4% of the MGDG, which initially constituted 77.6% of total lipids (Tables 1 and 4). The ratio of DGDG/MGDG increased in both WS and PLDδ-KO leaves during ethylene-promoted senescence. The relative content of MGDG was clearly higher in PLDδ-KO plants than in WS plants, whereas the relative content of DGDG in PLDδ-KO plants resembled that in WS plants, which resulted in the ratio of DGDG/MGDG in PLDδ-KO detached leaves being significantly lower than that in WS plants after ethylene treatment for 5 days, namely, 0.19 and 0.22, respectively (Table 4). The higher relative content of MGDG in PLDδ-KO leaves might have contributed to stabilizing the ultrastructure, fluidity and permeability of the chloroplast membrane, which resulted in higher photosynthetic activity.

To investigate how PLDδ functions in ethylene-promoted senescence, we analyzed the changes in the absolute level and relative content of PA under ethylene treatment in the two genotypes plants leaves. During ethylene-promoted senescence, no significant changes were detected in absolute levels of PA in either WS or PLDδ-KO plants. Upon analysis of the relative content of membrane lipids, we found that the relative content of PA increased 3.5-fold (from 0.02 to 0.07%) in WS plants, but remained unchanged in PLDδ-KO plants, which resulted in the relative content of PA in WS being much higher than that in PLDδ-KO plants after ethylene treatment for 5 days, especially for the molecular species PA 34:3, 36:3, and 36:6 (Tables 1 and 3; Figure 3). PA is a non-bilayer lipid and a potent promoter of the formation of the hexagonal phase and destabilization of the plasma membrane. For further assessment of the cell membrane stabilization of Arabidopsis during ethylene-promoted senescence, we calculated the PC/PE ratio in this process. This ratio in WS plants increased from 1.61 (NS) to 2.11 (leaves treated with ethylene for 3 days), and then decreased to the initial level of 1.73 (leaves treated with ethylene for 5 days). The PC/PE ratio in PLDδ-KO leaves increased constantly in the course of ethylene-promoted senescence, from 1.66 (NS) to 2.39 (leaves treated with ethylene for 5 days). In addition, the ratio of PC/PE in PLDδ-KO detached leaves was much higher than that in WS leaves after ethylene treatment for 5 days, namely, 2.39 and 1.73, respectively (Table 4). Our results indicate that the increase in the relative content of PA promoted destabilization of the plasma membrane; this may have led to the loss of membrane integrity and functions of membrane-associated proteins, thereby promoting senescence. Therefore, a reduction in the relative content of PA in PLDδ-KO leaves may have accounted for the higher ratio of PC/PE, which may have helped to maintain plasma membrane integrity and normal membrane protein function that eventually resulted in the retardation of ethylene-promoted senescence.