Arabidopsis thaliana ecotype Columbia-0 was used throughout the study apart from the mutant assay where Arabidopsis ecotype Wassilewskija-4 was used for both the wild-type and the T-DNA insertion line of ACS7. The sources of the two materials in the latter case were described in Dong et al. (2011). The T-DNA insertion line of ACS7 was further confirmed to have a single insertion site by southern blot analysis (Supplementary Figure S1) in this study following standard protocols (Sambrook and Russell, 2001). The generation of transgenic Arabidopsis GVG:ACS7^Δ1-14-Flag, GVG:ACS7-Flag, ACS7:GUS and ACS7:N^7(1-54)-GUS were described in Xiong et al. (2014). In order to generate 35S:ACS7-eGFP and 35S:ACS7^Δ1-14-eGFP, ACS7 and ACS7^Δ1-14 were inserted into the binary vector pCABMBIA1301 and fused with a GFP gene to create the recombinant transcription units, ACS7-eGFP and ACS7^Δ1-14-eGFP, respectively. Similarly, in order to generate 35S:N^7(1-14)-SSPP-HA, a fusion construct was obtained via PCR by initially using the 35S-N7-SSPP-F2 and 35S-N7-SSPP-R2 primer pair and the SSPP gene (Xiao et al., 2015) as template. The diluted PCR product was then used as a template while 35S-N7-SSPP-F1 and 35S-N7-SSPP-R1 were used as the second primer pair to accomplish construction. Constructions of this fusion gene have been patented by our group in the State Intellectual Property Office of the People’s Republic of China with number 201510050322.1 and we have also applied for the international patent with application number PCT/CN2016/070504. The fusion gene was subsequently transferred into the pBI121 expression vector. The construction of SSPP-HA was generated on the basis of N^7(1-14)-SSPP-HA by removing the N^7(1-14) part through double digestion. All the primer sequences used in this study are listed in Supplementary Table S1. All recombinant constructs were transferred into Arabidopsis by Agrobacteria-mediated floral dip method (Clough and Bent, 1998). Recombinant constructs in transgenic plants were confirmed by PCR genotyping and homozygous plants were used for experiments.

Seeds were surface sterilized in 10% (v/v) sodium hypochlorite (Tianjin Chemicals, 559) for 2 min, washed at least 10 times with sterilized water and geminated on one-half-strength Murashige Skoog (MS) medium (Duchefa Biochemie, M0222) containing 0.8% (w/v) agar (Solarbio, A8190), pH5.7, 1% (w/v) Suc (Jiangtian Chemicals, 11411), supplemented with or without antibiotics, stratified at 4°C for 2 days in the dark, and then grown in plant growth chamber at 22/19°C with cycles of 16 h light and 8 h darkness under 100 to 150 μmol m^-2 s^-1 light intensity. The 10-day-old seedlings were then transferred to soil and grown under the same conditions for further experiments. Where dexamethasone (DEX, Sigma, D1756) spray was applied, the compound was dissolved in ethanol and the final concentration was at 30 μM.

Ethylene emission of the 41-day-old wild-type Arabidopsis was determined in 10 different leaves of one rosette by gas chromatography (Agilent 7890A) as described in Li and Mattoo (1994). Approximately two leaves of each were separately collected depending on their sizes and vial spacing. Detached leaves were subsequently incubated in vials under light before analysis. The chlorophyll content in mesophyll cells was spectrophotometrically measured as described in Arnon (1949). At least six independent biological replicates were carried out.

GUS (β-glucuronidase) histochemical staining of transgenic Arabidopsis was carried out as described previously in Wang et al. (2005). The images of blue-colored whole plants were recorded with a scanner (Epson V30). Three to six biological replicates for each transgenic line were performed in this study. In total five and seven independent transgenic lines were generated for ACS7:GUS and ACS7:N^7(1-54)-GUS, respectively, and data shown were the typical results of lines with approximately same levels of GUS gene expression.

RNA extraction, cDNA synthesis and RT-PCR analysis were performed as described previously in Liu et al. (2010). The real-time RT-PCR was done using SYBR Green Perfect Mix (Tiangen, FP205-02) on iQ5 (Bio-Rad) machine following the manufacturer’s instructions. All reactions were performed under the following conditions: 95°C for 2 min, 40 cycles of 95°C for 10 s, and 56°C or 58°C in case of SSPP amplification for 30 s. TIP41-like gene was used as an internal control. The calculation of relative gene expression levels was as described in Xiong et al. (2014). At least three independent replicates were performed to give typical results shown here. All primers used in RT-PCR analysis were listed in Supplementary Table S1.

Seedling or leaf samples were ground in liquid nitrogen and total soluble protein was extracted as described in Wang et al. (2005). The concentration of protein extract was measured using the Bradford reagent (USB, DH078) and total protein was separated by 12% SDS-PAGE gel and detected by antibodies after being transferred to a polyvinylidene fluoride (PVDF) membrane (GE Healthcare, 10600023). The antibodies used in this study include rabbit anti-GFP (Abcam, ab290), mouse anti-Flag (Sigma, F3165), rabbit anti-HA (Abcam, ab9110), mouse anti-actin (Abmart, 26F7), mouse anti-Tubulin (Abmart, 10B1), goat anti-rabbit IgG H&L (HRP) (Abcam, ab6721) and rabbit anti-mouse IgG H&L (HRP) (Abcam, ab6728). Protein bands were visualized using an ECL western blotting reagent pack (GE Healthcare, RPN2232) flowing the manufacturer’s instructions. Images were recorded by a chemiluminescence imaging system (CELVIN’s, Biostep). Quantification of the signals was performed in Image J software as described in Xiong et al. (2014).

To evaluate the spatiotemporal distribution of ethylene biosynthesis and the correlations between ethylene production and leaf senescence, we measured changes in ethylene emissions and chlorophyll contents in the developing leaves of 41-day-old Arabidopsis (Figure 1). The results showed that the content of chlorophyll in mesophyll cells was gradually decreased during leaf senescence and the most significant decline was between leaves ten and nine. At the same time, however, the level of ethylene production was gradually increased from the mature 10th leaf, reaching a peak in the seventh. In the older leaves, from the sixth to the fourth, both the chlorophyll content and ethylene emission were continuously dropped (Figure 1). Indeed, both chlorophyll content and ethylene production were hardly measurable in the much more senescent leaves, including the third, second and first (data not shown).

The quantitative RT-PCR assay was performed to determine the involvement of ACS7 in Arabidopsis leaf senescence. Results showed that during natural senescence, the transcription levels of ACS7 remained relatively low in young leaves but increased gradually as they developed to late senescence stages (Figure 2A). It was also shown that 4-day dark treatment resulted in a significant decrease in the chlorophyll content in the wild-type Arabidopsis plants at stage 5.10 (Boyes et al., 2001). In the meantime, however, an obvious increase in the level of ACS7 transcript was revealed during the dark-induced senescence process (Figure 2B).

As we have previously reported, the N-terminal 1–14 residues of ACS7 negatively regulate the protein stability. Consistently, the truncated form of ACS7 lacking the 14 residues is more stable than the full-length form when transgenically expressed in the light-grown Arabidopsis seedlings (Xiong et al., 2014). We therefore used previously generated transgenic Arabidopsis plants expressing Flag-tagged full-length ACS7 as well as truncated ACS7 with 1–14 residues deleted under the control of a DEX-inducible promoter, namely GVG:ACS7-Flag and GVG:ACS7^Δ1-14-Flag (Xiong et al., 2014) in this study to examine the detailed effects of ACS7 on leaf senescence. Quantitative RT-PCR revealed comparable transcript levels of ACS7 in these two transgenic materials (Supplementary Figure S2). It was shown that the GVG:ACS7-Flag transgenic Arabidopsis exhibited normal growth and development compared to the GVG:GUS control plants after 4 days following DEX spray. In contrast, the GVG:ACS7^Δ1-14-Flag transgenic plants displayed prominent leaf yellowing and had significantly smaller rosettes (Figure 3A), as initially noted by Xiong et al. (2014). We further sampled the fifth and sixth leaves of GVG:GUS, GVG:ACS7-Flag and GVG:ACS7^Δ1-14-Flag after DEX treatment and performed quantitative RT-PCR to determine the transcription levels of several known senescence-associated marker genes including two transcriptional factors NAP1 (Guo and Gan, 2006) and WRKY6 (Robatzek and Somssich, 2002), as well as one PP2C family protein phosphatase gene SAG113 (Zhang et al., 2012). No significant increases in the transcription levels of these marker genes were found in the DEX-treated GVG:GUS and GVG:ACS7-Flag plants while the expressions of these senescence-associated markers in the DEX-treated Arabidopsis expressing ACS7^Δ1-14-Flag were greatly up-regulated (Figure 3B). Moreover, western blot assay using anti-Flag antibody revealed a much higher level of ACS7 protein accumulation in the fifth leaves of DEX-treated GVG:ACS7^Δ1-14-Flag plants than in their GVG:ACS7-Flag counterparts (Figure 3C, for original images of the blots please see Supplementary Figure S6). These results confirmed that the induced high-level accumulation of ACS7 protein led to precocious leaf senescence in Arabidopsis.

In order to further confirm the function of ACS7 in leaf senescence, we examined the phenotype of a loss-of-function mutant of ACS7, acs7-1 (Tsuchisaka et al., 2009; Dong et al., 2011). As was shown in Figures 4A,B, the acs7-1 mutant exhibited delayed leaf senescence compared to its wild-type control (Wassilewskija-4). In line with the phenotypic observations, the chlorophyll content in the acs7-1 mesophyll cells was significantly higher than in the wild-type control when measured at 36 days after emergence (DAE) (Figure 4C). The acs7-1 mutants exhibited larger rosettes, faster growth, and reached greater heights (Supplementary Figure S3). We also observed an early flowering phenotype within these mutant plants, consistent with a previous report (Tsuchisaka et al., 2009), as well as a shorter blooming period in comparison to their wild-type control (Supplementary Figure S3).

To further explore the underlying mechanisms of ACS7-regulated leaf senescence, we generated transgenic Arabidopsis expressing the GUS gene fused to the N-terminal 1–54 residues of ACS7, namely N^7(1-54)-GUS, or a GUS gene alone driven by ACS7 native promoter. Two lines that exhibited no significant differences in levels of transcriptional expression of the GUS genes were selected for further study (Supplementary Figure S4) and photographed at six different developmental stages including stage 0.5, stage 0.7, stage 1.02, stage 1.04, stage 5.10, and stage 6.00 as previously described by Boyes et al. (2001) (Figure 5, upper row). In view of the fact that N-terminal 1–14 residues play an important role in the post-translational regulation of ACS7, the activities of N^7(1-54)-GUS are thus indicative of changes in the accumulation levels of internal ACS7 protein during leaf development (Xiong et al., 2014). The histochemical GUS staining assay revealed that ACS7 promoter activity and the accumulation of ACS7 protein were both at very high levels in the late stages of seed germination. With the growth of seedlings, the promoter activity of ACS7 was maintained at a very high level while its protein accumulation varied from stage to stage (Figure 5). For instance, the blue coloring of ACS7:N^7(1-54)-GUS was significantly decreased at vegetative stages such as stages 1.02 and 1.04 but was gradually restored at stages 5.10 and 6.00 along with the initiation of leaf senescence. A closer look revealed that the blue colors in ACS7:N^7(1-54)-GUS at the last two stages were generally observed in mature and senescent leaves while almost no traces were seen in young leaves (Figure 5, middle row). All results presented here suggested that the N-terminal mediated ACS7 degradation was restricted during the process of seed germination, and that the restriction was released at vegetative stages but was tighten again with the progressing of leaf senescence.

In order to confirm the role of senescence signals in the regulation of ACS7 stability, we generated transgenic Arabidopsis expressing GFP-tagged full-length ACS7 (ACS7-eGFP) or truncated ACS7 lacking N-terminal 1–14 residues (ACS7^Δ1-14-eGFP) under the control of 35S promoter.

To begin with, we evaluated expression levels of the ACS7 transgenes using quantitative RT-PCR. The results showed that the transcript levels of ACS7 in line 7 of 35S:ACS7-eGFP and in line 16 of 35S:ACS7^Δ1-14-eGFP were about 500 and 400 times higher, respectively, than that of the internal control TIP41-like. In line 3 of 35S:ACS7-eGFP and line 12 of 35S:ACS7^Δ1-14-eGFP, the expressions of the transgenes were at approximately the same but relatively much lower levels (Figure 6A).

The phenotypic analysis of etiolated seedlings showed that all of the four lines mentioned above exhibited prominent triple responses, implying that expression of ACS7^Δ1-14-eGFP and ACS7-eGFP both led to over-production of ethylene at this developmental stage (Figure 6B, upper panel). Immunoblot analysis also revealed approximately equally high levels of ACS7 accumulation in the transgenic etiolated seedlings of line 3 of 35S:ACS7-eGFP and line 12 of 35S:ACS7^Δ1-14-eGFP, however, the light-grown line 3 seedlings of 35S:ACS7-eGFP exhibited a much lower accumulation level of ACS7 protein in comparison to line 12 of 35S:ACS7^Δ1-14-eGFP (Figure 6B, for original images of the blots please see Supplementary Figure S6), even though the transcripts of ACS7 in both lines were at similar levels (Figure 6A). In addition, line 7 of 35S:ACS7-eGFP, which had extremely high levels of ACS7 transcript and protein accumulation in etiolated seedlings (Figure 6B, upper panel), also displayed a remarkable decrease in the accumulation level of ACS7 protein in light-grown seedlings compared to the line 16 of 35S:ACS7^Δ1-14-eGFP (Figure 6B, lower panel, for original images of the blots please see Supplementary Figure S6). These results implied that the accumulation of full-length ACS7 was inhibited and further supported our previous findings that the truncated ACS7 protein lacking the N-terminal 1–14 residues was more stable than the full-length one in the vigorously growing seedlings under normal condition (Xiong et al., 2014).

To further evaluate the phenotypic differences in leaf senescence, line 12 of 35S:ACS7^Δ1-14-eGFP and line 3 of 35S:ACS7-eGFP that had approximately same levels of ACS7 transcript as well as wild type Arabidopsis were grown under light-grown conditions and photographed at three different stages including 17, 29, and 40 DAE (Figure 6C). The results revealed no significant differences between 35S:ACS7-eGFP and the wild-type control until 40 DAE when the transgenic plants exhibited slightly earlier senescence and smaller rosette, indicating a restoration of ACS7 protein level in late stages. In contrast, transgenic plants overexpressing the truncated ACS7 with the N-terminal 14 residues deleted exhibited a number of significant ethylene response phenotypes including smaller rosette since 17 DAE and much earlier senescence at 40 DAE. In line with this, the accumulation of ACS7 protein in the fifth and sixth leaves of line 3 of 35S:ACS7-eGFP was at very low level when the plant was at vegetative stages such as 9 and 17 DAE, however, it was increased when the plant approached its first-flower opening stage at 29 DAE and reached a much higher level at 40 DAE as plant aging. By contrast, the ACS7 proteins in the fifth and sixth leaves of line 12 of 35S:ACS7^Δ1-14-eGFP were maintained at a high level throughout all these time points and did not show any significant response to senescence signaling (Figures 6B,D). All these results demonstrated again that the N-terminus mediated ACS7 degradation was negatively regulated by senescence signals.

In order to further confirm the role of senescence signaling in the regulation of N-terminus mediated ACS7 protein degradation, we generated transgenic Arabidopsis plants over-expressing HA-tagged SSPP, a negative regulator of Arabidopsis leaf senescence, or HA-tagged SSPP fused with the N-terminal 1–14 residues of ACS7 at its N-terminus under the control of 35S promoters [namely 35S:SSPP-HA and 35S:N^7(1-14)-SSPP-HA, respectively]. Two transgenic lines that had approximately the same transcript levels of SSPP were used in this study (Supplementary Figure S5) and photographed at 10, 21, 48, and 61 DAE, respectively. In line with our previous findings from the 35S:SSPP plants (Xiao et al., 2015), the 35S:SSPP-HA transgenic lines displayed a remarkable decrease in rosette size. However, all the 35S:N^7(1-14)-SSPP-HA transgenic plants developed normally before the onset of leaf senescence, regardless of the similar transcript levels of SSPP (Figure 7A). And more interestingly, after the initiation of leaf senescence, the 35S:N^7(1-14)-SSPP-HA transgenic plants also exhibited significantly delayed leaf senescence in comparison to the wild-type control at 48 DAE and much later stages such as 61 DAE (Figure 7A). We further carried out western blot assay to evaluate SSPP accumulations in both 35S:N^7(1-14)-SSPP-HA and 35S:SSPP-HA transgenic plants at different developmental stages. As expected, while the accumulation levels of SSPP in the 35S:SSPP-HA plants were constantly maintained at a relatively high and stable level from 10 to 48 DAE stages, the SSPP accumulation in the 35S:N^7(1-14)-SSPP-HA seedlings was very low at 10 DAE, but it was increased at 21 DAE and reached a higher level at 48 DAE as plants aging (Figure 7B). These observations suggested that the N-terminus-mediated degradation of SSPP was also suppressed by senescence signaling. The senescence-delayed effect of N^7(1-14)-SSPP fusion gene was further confirmed by the quantitative analysis of several senescence-associated marker genes. As shown in Figure 8, the expressions of SAG12, WRKY6, NAP1, and NAC1 were significantly down-regulated in both 35S:N^7(1-14)-SSPP-HA and 35S:SSPP-HA transgenic plants in comparison to WT at 32 DAE.