A positive feedback regulatory loop, SA-AtNAP-SAG202/SARD1-ICS1-SA, in SA biosynthesis involved in leaf senescence but not defense response

Abstract

Salicylic acid (SA) is an important plant hormone that regulates defense responses and leaf senescence. It is imperative to understand upstream factors that regulate genes of SA biosynthesis. SAG202/SARD1 is a key regulator for isochorismate synthase 1 (ICS1) induction and SA biosynthesis in defense responses. The regulatory mechanism of SA biosynthesis during leaf senescence is not well understood. Here we show that AtNAP, a senescence-specific NAC family transcription factor, directly regulates a senescence-associated gene named SAG202 as revealed in yeast one-hybrid and in planta assays. Inducible overexpreesion of AtNAP and SAG202 lead to high levels of SA and precocious senescence in leaves. Individual knockout mutants of sag202 and ics1 have markedly reduced SA levels and display a significantly delayed leaf senescence phenotype. Furthermore, SA positively feedback regulates AtNAP and SAG202. Our research has uncovered a unique positive feedback regulatory loop, SA-AtNAP-SAG202-ICS1-SA, that operates to control SA biosynthesis associated with leaf senescence but not defense response.

Core

A unique positive feedback regulatory loop, SA-AtNAP-SAG202/SARD1-ICS1-SA is found to operate and modulate SA biosynthesis to regulate leaf senescence in Arabidopsis. Although part of the loop, SAG202/SARD1-ICS1-SA is shared by defense response, the whole loop is not responsive to pathogen attack.

Gene & accession numbers

Sequence data from this article can be found in the GenBank/EMBL databases under the following accession numbers: AT1G73805 (SAG202, SARD1), AT4G10500 (S3H), AT1G69490 (AtNAP), AT1G74710 (ICS1), AT1G18870 (ICS2), AT5G26920 (CBP60g) and AT3G18780 (Actin2, ACT2).

Introduction

Salicylic acid (SA, 2-hydroxy benzoic acid) has pivotal roles in the regulation of many aspects of plant growth and physiological processes such as defense responses, thermogenesis, seed germination, flowering and senescence (Raskin, 1992; Rivas-San Vicente and Plasencia, 2011). It is generally accepted that there are two SA biosynthesis pathways in plants: the isochorismate (IC) pathway and the phenylalanine ammonia-lyase (PAL) pathway (Métraux, 2002; Chen et al., 2009b). In Arabidopsis, the IC pathway contributes to most of the SA production induced by pathogens and UV light (Garcion et al., 2008; Dempsey et al., 2011). Although two genes, namely isochorismate synthase 1 (ICS1) and ICS2, are involved in isochorismate synthesis in the IC pathway, ICS1 accounts for approximately 90% of the total amount of isochorismate produced in response to pathogens or UV light (Surplus et al., 1998; Wildermuth et al., 2001; Garcion et al., 2008). It is known that SA levels increase with progression of leaf senescence (Morris et al., 2000; Zhang et al., 2013; Zhang et al., 2017b); however, whether the IC pathway operates and functions during leaf senescence is not well known.

The regulation of SA biosynthesis and the SA signaling in local and systemic acquired resistance (LAR and SAR) responses against pathogens have been intensively investigated (Shirasu et al., 1997; Dangl, 1998; Shah, 2003). Ethylene insensitive 3 (EIN3) and EIN3-like 1 (EIL1) suppress ICS1 to negatively regulate SA biosynthesis (Chen et al., 2009a), and two closely related transcription factors, calmodulin binding protein 60 g (CBP60g) and systemic acquired resistance deficient 1 (SARD1), bind to the core sequence 5’GAAATTTTGG3′ in the promoter of ICS1 to positively modulate SAR-related SA production (Zhang et al., 2010a). Gene expression profiling revealed that 5’GAAATT3’ motifs were significantly over-represented in the promoters of SARD1 and CBP60g putative target genes (Truman and Glazebrook, 2012). SARD1 and CBP60g are functionally partially redundant (Zhang et al., 2010b; Wang et al., 2011). The upstream factor(s) that regulates the expression of SARD1 and CBP60g have yet to be identified and the regulatory mechanism of SA biosynthesis during leaf senescence remains unknown.

Leaf senescence is a genetically programmed cell suicide process that is accompanied by mobilization of nutrients released during cell attrition to active growing regions, seeds or trunks (Gan and Amasino, 1997; Guo et al., 2021). The regulation of senescence is rather complex, and it involves activation of thousands of senescence-associated genes (SAGs) and/or inactivation of many senescence-down-regulated genes (Guo et al., 2004; Guo and Gan, 2012). TFs have been shown to have critical roles in regulating SAG expression and leaf senescence. For example, AtNAP, a NAC family TF gene, acts as a master regulator of leaf senescence because atnap null mutants display a 10-day delay in leaf senescence whereas its inducible expression in young leaves readily causes precocious senescence (Guo and Gan, 2006). The role of NAP orthologues in leaf senescence has been demonstrated in wheat (Uauy et al., 2006), maize (Zhang et al., 2012b), rice (Liang et al., 2014), cotton (Fan et al., 2015), peach (Li et al., 2016), and cabbage (Li et al., 2021). NAP also has a major role in senescence of rose petals (Zou et al., 2021) and Arabidopsis fruits (Kou et al., 2012). The direct target genes of AtNAP are of significant interest for understanding the molecular circuitry of leaf senescence regulation. It is known that AtNAP TF directly binds to the promoter of SAG113 to activate the expression of a senescence-specific and Golgi-localized protein phosphatase 2C gene to promote senescence (Zhang and Gan, 2012; Zhang et al., 2012a). The TF also physically binds to the promoter of cytokinin oxidase genes in Arabidopsis, rice (Hu et al., 2021) and rose (Zou et al., 2021) to degrade the senescence-retardant cytokinins, which facilitates the senescence process.

Here we report that a senescence up-regulated gene named SAG202 (At1G73805) is a direct target gene of AtNAP; sequence analysis reveals that SAG202 is identical to SARD1. AtNAP physically binds to the promoter region of SAG202, but does not bind to CBP60g, and SAG202 binds to the promoter region of ICS1 (but not ICS2), as revealed by yeast one-hybrid experiments. Knockouts of SAG202 and ICS1 have lower levels of SA and display a significant delay in leaf senescence whereas inducible overexpression of SAG202 leads to high levels of SA and premature leaf senescence. Quantitative PCR analyses further reveal that elevated SA levels can feedback up-regulate AtNAP and SAG202. These findings suggest that there is a unique feedback regulatory loop consisting of SA-AtNAP-SAG202-ICS1-SA that modulates the SA biosynthesis to control leaf senescence in Arabidopsis.

Results

SAG202 is up-regulated during leaf senescence

SAG202 (At1G73805) was initially identified during our analysis of the Arabidopsis leaf senescence transcriptome (Guo et al., 2004) and was later reported as SARD1 (Zhang et al., 2010b; Wang et al., 2011) (hereafter SAG202 is used). The transcript levels of SAG202 were examined in leaves at different senescence stages using qPCR (Fig. 1a). To further investigate the expression pattern of SAG202, the GUS reporter gene was fused to the 3’ end of a 2.2-kb region of the SAG202 promoter (P_SAG202). The GUS staining patterns of the rosette leaves from P_SAG202-GUS transgenic Arabidopsis showed that SAG202 was expressed quite specifically in senescing leaves (Fig. 1b).

Fig. 1

Phenotypic and molecular analyses of SAG202 in Arabidopsis. (a) qPCR analysis of the transcript levels of SAG202/SARD1 in WT leaves at different developmental stages. NS, fully expanded non-senescing stage; ES, early senescence stage (< 25% yellowing); MS, mid-senescence stage (~ 50% yellowing); LS, late senescence stage (> 75% yellowing). Relative expression levels were calculated and normalized with respect to Actin 2 (ACT2) transcripts. Error bars indicate SD of three biological repeats. (b) GUS staining of the fifth leaves from P_SAG202-GUS transgenic plants at different senescing stages. (c) Diagram of the T-DNA insertion locations of two sag202 mutants. (d) The expression of SAG202 is knocked out in the mutants shown in C as revealed by RT-PCR. (e) Age-matched 35 DAG WT and sag202 null mutants. DAG, days after germination. (f) Precocious leaf senescence in SAG202-inducible expression line (SAG202ⁱⁿ or 8004/7001) (photo was taken 4 days after DEX induction). (g) Leaves detached from the age-matched 35 DAG plants in e (counted from bottom with the oldest leaf as 1 and the youngest leaf as 12). (h) The chlorophyll contents of the fifth and sixth leaves from control (containing pTA7001 only) and SAG202ⁱⁿ lines. (i, j) Chlorophyll contents and F_v/F_m of the sixth to tenth rosette leaves of the age-matched 35 DAG plants from WT and the sag202 mutants. Error bars indicate SD of three biological repeats. *P < 0.05 using Student’s t-test

Leaf senescence is significantly delayed in sag202 knockout mutants and precociously accelerated in SAG202 inducible overexpression lines

Two T-DNA lines, namely sag202–1 (SALK_052422) and sag202–2 (SALK_138476C) (Fig. 1c) in which SAG202 was knocked out (Fig. 1d), were used to investigate the role of SAG202 in leaf senescence. Compared with the wild type (WT), both knockout lines displayed a significant delay in leaf senescence phenotypically (Fig. 1e, g, and Supplementary Fig. S1) and physiologically (Fig. 1i, j). Because both knockout lines had the same phenotype, only sag202–1 was used in the following experiments and referred to as sag202 for simplicity.

The role of SAG202 in leaf senescence was also investigated in dexamethasone (DEX) inducible gain-of-function lines harboring both pTA7001 and pGL8004 constructs. pTA7001 (control) provides constitutive expression of a recombinant transcription factor (TF), GAL4^BD-VP16^AD-GR, in transgenic plants (Guo and Gan, 2006). In pGL8004 construct, SAG202 is driven by a promoter containing six tandem copies of the GAL4 upstream activation sequence. When DEX (a synthetic glucocorticoid) binds to GR and causes conformational changes, VP16 is able to activate transcription of SAG202 in plants harboring both pTA7001 and pGL8004 (SAG202ⁱⁿ). As shown in Fig. 1f and h, treatment of 20-day-old non-senescing plants with 30 μM DEX caused precocious leaf senescence in SAG202ⁱⁿ lines but not in the control lines. qPCR analyses showed that SAG202 was strongly induced in SAG202ⁱⁿ lines but not in the control lines (Fig. 2f).

Fig. 2

qPCR analyses of gene expression upon chemical induction of AtNAP or SAG202. (a-e) The transcript levels of AtNAP, SAG202/SARD1, CBP60g, ICS1 and ICS2 in the AtNAP inducible (AtNAPⁱⁿ) and control lines at 0 h, 3 h, 6 h and 24 h after DEX treatment. (f-h) The transcript levels of SAG202, ICS1 and ICS2 in the SAG202ⁱⁿ and control lines at 0 h, 3 h, 6 h and 24 h after DEX treatment. Relative expression levels were calculated and normalized with respect to Actin 2 (ACT2) transcripts. Error bars indicate SD of three biological repeats. *P < 0.05 using Student’s t-test

SAG202 (but not CBP60g) and ICS1 (but not ICS2) are co-induced with AtNAP

AtNAP is a NAC family TF that is up-regulated during senescence, and its DEX-inducible expression lines (AtNAPⁱⁿ) are readily available (Guo and Gan, 2006). Upon DEX treatment, the expression of AtNAP was significantly induced in AtNAPⁱⁿ lines but not in control plants (Fig. 2a). qPCR analyses revealed that SAG202 and ICS1 were also induced (Fig. 2b, d), but CBP60g, a gene closely related to SAG202, and ICS2 were not induced (Fig. 2c, e). ICS1 expression was co-induced with the induction of SAG202 in SAG202ⁱⁿ lines while the expression of ICS2 was not induced (Fig. 2g, h).

AtNAP TF physically binds to the promoter region of SAG202 (but not CBP60g) in yeast cells and in planta

The above co-induction of SAG202 with AtNAP raised the possibility of SAG202 being a direct target gene of AtNAP. To test this, we performed yeast one-hybrid experiments in which a series of truncated promoter fragments of SAG202 (Fig. 3a) were cloned in front of a LacZ reporter gene as promoter baits to form various reporter constructs; the AtNAP coding sequence was fused with the yeast GAL4 activation domain (GAD) to form the effector GAD-AtNAP construct (Zhang and Gan, 2012). The AtNAP TF was able to physically bind to a specific region of SAG202 promoter containing 5’CACGcgAaT3’ that is very similar to the 9-bp AtNAP core binding sequence, 5’CACGtaAgT3’ (nucleotides in lower case are variable), in the promoter of SAG113 (Zhang and Gan, 2012) (Fig. 3a). In contrast, AtNAP TF did not bind to the promoter of CBP60g (Fig. 3a). The binding of AtNAP TF to the motif 5’CACGcgAaT3’ on the SAG202 promoter (P_SAG202) in yeast cells was confirmed by transversion and deletion mutants of the sequence: when mutated, AtNAP TF could no longer bind to the promoter (Fig. 3b).

Fig. 3

Binding of AtNAP to the SAG202 promoter truncations and mutants in yeast and in planta. (a) Binding of AtNAP to the SAG202 promoter truncations revealed by yeast one-hybrid assay. The LacZ reporter gene driven by various SAG202 promoter truncations was used to test the binding ability of the GAD-AtNAP fusion protein. Red dash lines indicate promoter sequence that is highly conserved to the 9-bp (red letters) AtNAP binding site of the SAG113 promoter (Zhang and Gan, 2012). The immediate upstream bp of the translation start site was numbered as − 1. The CBP60g promoter (1727 bp in length) was also tested. (b) Failure of AtNAP binding to the SAG202 promoter that contains either transversion mutation or deletion of the motif sequence in yeast cells. (c) GUS staining of senescent leaves in transgenic Arabidopsis plants (WT or annap mutant background) harboring P_SAG202W-GUS (_W for WT), P_SAG202T-GUS (_T for transversion; the promoter contains the transversion mutation shown in (b)) or P_SAG202D-GUS (_D for deletion; the promoter contains the deletion mutation shown in (b)). (d) GUS enzymatic activities (expressed as nmol methylumbelliferone produced min ^− 1 mg ^− 1 protein) of senescent leaves of transgenic plants shown in (c). Data are mean values ± SD of five samples. Significant (P > 0.05) differences between means are indicated by different letters. ANOVA analysis with LSD test was used

The physical interaction of AtNAP TF to P_SAG202 was further examined in planta. The P_SAG202W (wild-type promoter) and its variants with either transversion (P_SAG202T) or deletion mutation (P_SAG202D) shown in Fig. 3b were fused with the GUS reporter gene and transformed into WT or atnap null background. The GUS activities in senescent leaves (WT-S or atnap-S) were analyzed by histochemical staining and enzymatic assay, which revealed that the GUS activities with either of the mutated promoter motif in senescent leaves of WT were reduced to less than 30% of the P_SAGW-GUS plants and that the GUS activities in the atnap null background were less than 45% of the activities in WT background (Fig. 3c,d). The data further supported the AtNAP TF physically bound to the motif 5’CACGcgAaT3’ of the SAG202 promoter to direct the gene expression in planta.

SAG202 TF physically binds to the promoter of ICS1 (but not ICS2) in yeast cells and in planta

To investigate if the SAG202 TF physically interacts with the ICS1 promoter, various ICS1 promoter fragments with truncations were generated and yeast one-hybrid system was utilized to show that the SAG202 TF was indeed able to bind to a 104 bp region (− 1178 ~ − 1281) of the promoter of ICS1. The 1958-bp promoter of ICS2 (− 2 ~ − 1959) was also used in the yeast one-hybrid experiment, which revealed there was no physical interaction between the SAG202 TF and the ICS2 promoter (Fig. 4a). The 104 bp region contained a 6 bp motif 5’GAAATT3’ that was believed to be the binding site of SAG202. To test this, ICS1 promoters with either transversion or deletion mutation were used in the yeast one-hybrid experiments, and the mutations abolished the binding of SAG202 to the ICS1 promoter variants in the yeast cells (Fig. 4b). Further analyses involving the use of the ICS1 promoters with or without the transversion or deletion mutation to direct the GUS expression in WT and sag202 null background revealed that SAG202 also physically interacted with the 5’GAAATT3’ cis element of the ICS1 promoter in planta (Fig. 4c,d).

Fig. 4

Binding of SAG202 to the ICS1 promoter truncations and mutants in yeast and in planta. (a) Binding of SAG202 to the ICS1 promoter truncations. The LacZ reporter gene driven by various ICS1 promoter truncations was used to test binding ability of the GAD-SAG202 fusion protein. Red dash lines indicate the 104 bp promoter region containing the SAG202 binding sequence. The ICS2 promoter (1626 bp in length) was also tested. (b) Failure of SAG202 binding to the ICS1 promoter that contains either transversion mutation or deletion of the motif sequence GAAATT (red letters) within the 104 bp region in yeast cells. (c) GUS staining of senescent leaves in transgenic Arabidopsis plants (WT or sag202 mutant background) harboring P_ICS1W-GUS (_W for WT), P_ICS1T-GUS (_T for transversion; the promoter contains the transversion mutation shown in b) or P_ICS1D-GUS (_D for deletion; the promoter contains the deletion mutation shown in b). (d) GUS enzymatic activities (expressed as nmol methylumbelliferone produced min ^− 1 mg ^− 1 protein) of senescent leaves of transgenic plants shown in c. Data are mean values ± SD of five samples. Significant (P > 0.05) differences between means are indicated by different letters. ANOVA analysis with LSD test was used

Both AtNAP and SAG202 are positively regulated by SA

The above data revealed a regulatory chain consisting of AtNAP-SAG202-ICS1 operating to produce SA during leaf senescence. If so, knocking out of an upstream gene should effect the expression of its downstream gene(s). We thus performed qPCR to analyze the expression levels of these genes in WT, atnap, sag202 and ics1 null mutants at different senescence stages (Fig. 5a-c). As expected, the transcript levels of both SAG202 and ICS1 were significantly reduced in the absence of AtNAP (Fig. 5b, c); similarly, the ICS1 expression levels were remarkably lowered in the sag202 mutants (Fig. 5c). The expression levels of CBP60g and ICS2, two genes outside of the regulatory chain, were not altered in any of the mutant backgrounds (Supplementary Fig. S2).

Fig. 5

qPCR analyses of transcript levels of AtNAP, SAG202, and ICS1 in different mutants during senescence or after SA treatment. (a-c) Transcript levels of AtNAP, SAG202 and ICS1 in leaves of WT, atnap, sag202 and ics1 null mutants at different senescence stages (1, NS; 2, ES; 3, MS; 4, LS as described in legend to Fig. 1a). (d-f) Transcript levels of AtNAP, SAG202 and ICS1 in NS leaves of WT, atnap and sag202 null mutants at time points after treatment with 5 mM SA. Relative expression levels were calculated and normalized with respect to Actin 2 (ACT2) transcripts. Error bars indicate SD of three biological repeats. Significant (P < 0.05) differences between means are indicated by different letters using Tukey’s HSD test

Interestingly, the expression levels of AtNAP in either sag202 or ics1 null mutants were reduced (Fig. 5a), and the transcript levels of SAG202 in leaves at the mid-senescence stage (MS) in the ics1 background were also decreased (Fig. 5b). These data suggested the possibility that the end product SA of the regulatory chain might feedback regulate those genes. To test this hypothesis, we analyzed the expression levels of AtNAP, SAG202, ICS1 in WT, atnap and sag202 mutants upon SA treatments. AtNAP (Fig. 5d) and SAG202 (Fig. 5e) were significantly induced by SA whereas the induction of ICS1 in the sag202 null mutants was not as significant (Fig. 5f), suggesting that AtNAP and SAG202 were positively feedback regulated by SA.

Free SA levels were reduced in atnap and sag202 mutants and elevated in AtNAP ⁱⁿ and SAG202 ⁱⁿ lines

The free SA levels in fully expanded non-senescing leaves (NS) and senescing leaves (S) of WT, atnap, sag202 and ics1 mutants were quantitatively analyzed using LC-MS/MS. The SA levels in the senescing leaves were significantly reduced in these null mutants but remained unchanged in the non-senescing leaves of any of the plants (Fig. 6a).

Fig. 6

LC-MS/MS analyses of free SA levels in WT, atnap, sag202, ics1 null mutants, AtNAPⁱⁿ and SAG202ⁱⁿ lines. (a) Free SA levels in NS and S (~ 50% yellowing) leaves of WT, atnap, sag202 and ics1 mutants, respectively. (b) Free SA levels in young leaves of AtNAP inducible lines (AtNAPⁱⁿ) and SAG202 inducible lines (SAG202ⁱⁿ) at 0, 1, 2, 4, and 7 days after DEX induction. Error bars indicate SD of three biological repeats. Significant (P < 0.05) differences between means are indicated by different letters using Tukey’s HSD test

The free SA levels in leaves with induced expression of AtNAP (AtNAPⁱⁿ) or SAG202 (SAG202ⁱⁿ) were also quantitated. As shown in Fig. 6b, the SA levels were significantly increased readily one day after the DEX induction.

The SAG202-ICS1-SA regulatory chain is shared between leaf senescence and defense response

Our studies showed that the positive feedback regulatory loop consisting of SA-AtNAP-SAG202-ICS1-SA operates during leaf senescence, and the SAG202-ICS1 node has been clearly shown to function in plant defense response (Wildermuth et al., 2001; Zhang et al., 2010b; Wang et al., 2011). To investigate whether AtNAP also has any roles in disease resistance, we inoculated mature non-senescing leaves of atnap, sag202, ics1 mutants and WT with Pseudomonas syringae pv. Tomato DC3000 and found that the pathogen resistance in the atnap mutant was not changed compared with that in WT while the sag202 and ics1 mutants became more susceptible to the pathogen infection (Fig. 7). These data strongly suggest that the SAG202-ICS1-SA regulatory chain is shared by leaf senescence and defense response and that the up-stream component AtNAP appears to be leaf senescence specific (Fig. 8).

Fig. 7

Growth of P. syringae pv. Tomato DC3000 in leaves of WT, atnap, sag202, ics1, s3h null mutants and S3HOE1 transgenic plants. The number of colony-forming units (cfu) per square centimeter of leaf area was determined 0, 1, and 2 days after inoculation. Error bars indicate SD of three biological repeats. *, ** and *** indicate P < 0.05, P < 0.01 and P < 0.001, respectively. Student’s t-test was used. The s3h null mutant (resistant to the infection) and S3HOE1 transgenic plants (susceptible to the infection) (Zhang et al., 2013) were included for comparison purpose

Fig. 8

A working model of SA-AtNAP-SAG202-ICS1-SA positive feedback loop in leaf senescence and its convergence/divergence in defense response in Arabidopsis. At the onset of and during leaf senescence, AtNAP TF physically binds to the promoter of SAG202 to direct the target gene expression. Subsequently the SAG202 TF activates its direct target gene ICS1 that is involved in the SA biosynthesis. The produced SA in turn feedback upregulates both AtNAP and SAG202. When the SA levels increase to a threshold, S3H (encoding an SA 3-hydroxylase) and S5H is induced to prevent overaccumulation of SA (Zhang et al., 2013; Zhang et al., 2017b). Too high levels of SA will cause hypersensitive response (HR)-like fast cell death. Leaf senescence is a slow programmed cell death process to allow nutrients released from degradation of proteins and other macromolecules to be recycled to active growing region or storage organs. Insert: A diagram showing convergence and divergence between leaf senescence and defense response with regard to the newly uncovered SA-AtNAP-SAG202-ICS1-SA regulatory loop. SA-AtNAP-SAG202-ICS1-SA feedback loop is activated to modulate endogenous SA levels in leaf senescence but not in defense responses. In defense response part of the loop, SAG202-ICS1-SA is activated

Discussion

Because of the significant role of SA in plant defense, much research has been performed to decipher its biosynthesis and signaling in plant (Shirasu et al., 1997; Dangl, 1998; Fu et al., 2012). There are two pathways leading to the production of SA in plants: one from phenylalanine and the other from chorismate via isochorismate (IC) (Dempsey et al., 2011). In Arabidopsis, the IC pathway contributes predominantly to SA accumulation during defense responses and isochorismate synthase 1 (ICS1) has the major role in this accumulation (Wildermuth et al., 2001). The IC pathway appeared to be predominant during leaf senescence in Arabidopsis because the SA levels in senescent leaves of ics1 were less than 5% of WT (Fig. 6a). Further studies showed that SARD1 and CBP60g bind to the promoter of ICS1 to regulate this gene’s expression (Zhang et al., 2010b; Wang et al., 2011). Which TFs regulate SARD1 (and CBP60g) is unknown. Our research addressed this question by identifying AtNAP (a NAC family TF) as a direct upstream regulator of SARD1(Fig. 8); this was supported by at least three lines of evidence: (i) the yeast one-hybrid experiments showed that AtNAP could physically bind to a promoter region of SAG202 (identical with SARD1) that contains a highly conserved sequence to which AtNAP binds (Fig. 3a, b), (ii) AtNAP binds to the 9 bp motif of the SAG202 promoter in planta (Fig. 3c, d), and (iii) SAG202 was co-induced when AtNAP was chemically induced (Fig. 2b). Interestingly, CBP60g, the close homolog of SAG202, is unlikely to be directly regulated by AtNAP because AtNAP could not bind to the promoter region of CBP60g (Fig. 3a) and that CBP60g was not co-induced with AtNAP (Supplementary Fig. S2). In addition to uncovering the AtNAP-SAG202 chain, we also provided new lines of evidence that SAG202 physically binds to the promoter of ICS1 (but not ICS2) as shown by our yeast one-hybrid experiment results (Fig. 4a, b), by in planta analyses (Fig. 4c, d), and by induction of the expression of ICS1 (Fig. 2d, g) but not ICS2 (Fig. 2e, h) through chemical activation of AtNAP or SAG202. SAG202 TF/SARD1 has been previously shown to directly regulate ICS1 using chromatin immunoprecipitation (ChIP), electrophoretic mobility shift assay (EMSA) and promoter sequence analysis (Zhang et al., 2010b; Truman and Glazebrook, 2012). These data reveal a unique regulatory chain consisting of AtNAP-SAG202-ICS1 (Fig. 8), which significantly advanced our understanding of molecular regulatory mechanism of the SA biosynthesis.

It is known that the SA levels are higher in senescing leaves than in non-senescing leaves and SA has an important role in controlling leaf senescence in Arabidopsis (Morris et al., 2000; Zhang et al., 2013; Zhang et al., 2017b; Yu et al., 2021). SARD1-ICS1 was shown to contribute to the SA production during defense responses (Wildermuth et al., 2001; Zhang et al., 2010b; Wang et al., 2011), but whether it, together with its upstream component AtNAP, also operates and functions during leaf senescence was not known. This research provided several lines of evidence that the regulatory chain operates in leaf senescence. The first line of evidence comes from the qPCR analysis of transcript levels of individual genes in the chain. As shown in Fig. 2a-d, the expression levels of AtNAP, SAG202 and ICS1 were all up-regulated upon chemical induction of AtNAP. The second line of evidence is from the quantification of SA levels in senescing leaves of respective null mutants. When individual genes of the regulatory chain were knocked out, the SA levels in senescing leaves were all significantly reduced (Fig. 6a). It should be noted that there is still ~ 30% SA in the senescent leaves of sag202 and ~ 75% in atnap (Fig. 6a), which could be due to ICS2 that is upregulated during senescence (Fig. S2) and/or ICS1 that is activated by such retrograde signaling protein WHIRLY1 (Lin et al., 2020) and senescence-associated TFs as WRKY75, WRKY51, WRKY28, WRKY55 and WRKY46 that have been shown to directly bind to the promoter of ICS1 (Guo et al., 2017; Zhang et al., 2017a; Tian et al., 2020; Wang et al., 2020).

In the absence of either SAG202 or ICS1, the transcript levels of AtNAP were significantly reduced in senescing leaves, in particular late-senescence (LS) leaves (Fig. 5a). Similarly, the expression levels of SAG202 in the ics1 null background were also decreased (Fig. 5b). These data suggested that SA, the end product of the regulatory chain, might positively feedback regulate AtNAP and SAG202 as shown in Fig. 8. This feedback regulation is supported by the fact that exogenous SA markedly elevated the AtNAP transcript levels in the sag202 mutants (and WT) (Fig. 5d). In the absence of AtNAP, external SA was able to highly induce the SAG202 expression (Fig. 5e), suggesting that SA may have a positive feedback regulation on SAG202 beyond AtNAP. In contrast, in the absence of SAG202, the ICS1 expression levels were not significantly altered by the external SA (Fig. 5f), indicating that ICS1 is unlikely to be positively feedback regulated by SA.

Previous studies suggested an important role of SA in leaf senescence (Morris et al., 2000; Gan, 2010; Zhang et al., 2013; Zhang et al., 2017b). Examples include the observations that there are higher levels of SA in senescing leaves compared with those in non-senescing leaves, leaf senescence is delayed in NahG or S3HOE plants in which a SA-degrading enzyme of bacterial or Arabidopsis origin is overexpressed, and the leaf senescence is accelerated in the s3h null plants in which SA are over-accumulated (Zhang et al., 2013). In this research, we found that when any of the genes in the regulatory loop are knocked out, the endogenous SA levels are significantly reduced (Fig. 5a) and the leaf longevity is substantially extended (Fig. 1a-b, Supplementary Fig. S1). Conversely, when AtNAP and SAG202 were individually chemically induced, the endogenous SA levels were enhanced (Fig. 6b) and the plants displayed precocious leaf senescence (Fig. 1f, h). These data reinforce SA’s role in promoting leaf senescence.

The shared regulatory chain of SAG202/SARD1-ICS1-SA is initiated by different cues and regulated to different extent of SA accumulation (insert in Fig. 8). The upstream regulator of SAG202 during disease resistance is not known yet, while in leaf senescence, SAG202 is regulated by AtNAP TF. CBP60g, the closely related family member to SAG202, was not directly regulated by AtNAP TF; however, CBP60g might also have a role in leaf senescence because its expression profile showed a senescence-associated elevation (Supplementary Fig. S2). In defense response, both CBP60g and SAG202 are involved in the induction of SA; after pathogen infection, SA level was elevated to a very high level in local leaves and leaded to a suicide cell death (Raskin, 1992; Zhang et al., 2010b). However, in age-dependent leaf senescence, SA level in senescing leaves was up-regulated to about 4 times higher than that in non-senescing leaves (Morris et al., 2000; Zhang et al., 2013; Zhang et al., 2017b) and resulted in a gentle and slow programmed cell death (PCD) necessary for remobilization of nutrients released during senescence to active growing tissues or storage organs such as seeds and trunk (Gan and Amasino, 1997).

Another difference is the significance of ICS1 in SA production between disease resistance and leaf senescence. SA level was almost undetectable in senescing leaves of ics1 mutants, which was even lower than its level in non-senescing leaves (Fig. 6a). This can be interpreted that ICS1 contributes to almost all SA production in senescing Arabidopsis leaves. In plant defense responses, however, there were still SA production in ics1 mutants, suggesting that there are other genes such as ICS2 (Garcion et al., 2008) or other SA biosynthesis pathways such as PAL pathway (Dempsey et al., 2011) functions in the SA biosynthesis. In addition to the regulation of SA anabolism, the level of SA is regulated by SA catabolism during leaf senescence. Studies showed that SA 3-hydroxylase (S3H) and SA 5-hydroxylase (S5H) are induced by SA and converts SA to its inactive forms 2,3-DHBA and 2,5-DHBA, respectively, which constitutes the negative feedback regulation of SA in leaf senescence to prevents SA over accumulation (Zhang et al., 2013; Zhang et al., 2017b).

Materials and methods

Plant materials and growth conditions

Arabidopsis thaliana ecotype Columbia was used in this study. The atnap knockout mutants, the AtNAP-inducible expression lines (Guo and Gan, 2006), and two T-DNA insertion lines (SALK_052422 and SALK_128476C, from ABRC) were all in the Columbia background. Per http://signal.salk.edu/tdnaprimers.2.html, a PCR-based method was used to identify homozygous T-DNA insertion mutants. The T-DNA left border primer G2325 (LBb1.3) and the gene-specific primers, G3832 and G3833 for sag202–1 (SALK_052422) and G3809 and G3810 for sag202–2 (SALK_128476C), were used. Plants homozygous for the T-DNA insertion were used in this study. All primers used in this research are listed in Supplemental Table S1.

Seed sterilization and growth were as previously described (Guo and Gan, 2006). The mutants, transgenic plants, and WT were grown side by side.

Plasmid construction

For the P_SAG202-GUS construct (pGL8002), a 2201-bp promoter fragment of SAG202 (At1G73805) was amplified from Arabidopsis genomic DNA by PCR with primers G3830 and G3831, cloned into pGEM-T easy vector (Promega, Madison, USA), sequenced, digested with Pst I and Nco I and inserted into pBI211 to form pGL8002.

To generate DEX-inducible SAG202 overexpression construct (pGL8004), the 1357-bp full length CDS of SAG202 was amplified from Arabidopsis cDNA by PCR with primers G3828 and G3829, ligated to pGEM-T easy vector, sequenced, digested with Hind III (Klenow fill-in) and Pst I, and cloned into the inducible binary vector pGL1152 (Guo and Gan, 2006) that was digested with Spe I (Klenow fill-in) and Pst I to form pGL8004.

Yeast one-hybrid assay-related constructs: pGL3175 (for producing GAD-AtNAP fusion protein in yeast) was constructed as described previously (Zhang and Gan, 2012). To construct pGL8040 (for producing GAD-SAG202 fusion protein in yeast), the SAG202 coding sequence was amplified from Arabidopsis cDNA by PCR with primers G4020 and G3992, ligated to pGEM-T easy vector, sequenced, digested with HindIII and XhoI, and cloned into the pJG4–5 (Lin et al., 2007) to form pGL8040. To construct P_SAG202-LacZ, P_ICS1-LacZ reporter genes, the 1122-bp SAG202 promoter (P_SAG202) region and the 1625-bp ICS1 promoter region (P_ICS1) were amplified from the Arabidopsis genomic DNA. The pairs of primers used were G3967 and G3918 for P_SAG202, and G3993 and G3994 for P_ICS1. The amplified fragment was ligated to the pGEM-T easy vector, sequenced, then released from the plasmid with EcoR I-Sal I and EcoR I-Xho I, respectively, and cloned into pLacZi-2 μ (Lin et al., 2007) that was digested with EcoRI-XhoI to form pGL8017 and pGL8036, respectively. Other LacZ reporter gene plasmids containing various truncated SAG202, ICS1, CBP60g and ICS2 promoter regions were similarly constructed using the primers listed in Supplementary Table S1.

Histochemical GUS staining, chlorophyll assay, and Fv/Fm assay

Histochemical GUS staining, chlorophyll assay, F_v/F_m assay were performed as previously described (Zhang and Gan, 2012; Hou et al., 2013).

qPCR analyses of transcripts

Total RNA extractions from Arabidopsis leaves and real-time PCR analyses were performed according to (Hou et al., 2013). cDNA was synthesized from 3 μg of total RNA (treated with RNase-free DNase; New England Biolabs, USA) at 42 °C with MV-Reverse Transcriptase (Promega, USA) (Hu et al., 2021). For qPCR, 1 μL of each diluted sample (40 folds) was used as a template in a 25-μL reaction. All qPCR reactions were performed on a Bio-Rad IQ-5 thermocycler with an annealing temperature around 55 °C. Cycle threshold values were determined by IQ-5 Bio-Rad software assuming 100% primer efficiency (Hu et al., 2021). Primers used for quantitative RT-PCR were listed in Supplementary Table S1. Three repetitions were performed for each combination of cDNA samples and primer pairs.

Plant transformation

Various constructs in binary vectors were transferred into Agrobacterium tumefaciens strain ABI1 that were subsequently used to transform Col-0 via the floral-dip method (Clough and Bent, 1998). Approximately 30 antibiotics-resistant T1 transgenic lines for each transgene were selected; phenotypic analyses were performed in T2 or advanced generations. Homozygous plants were used in all experiments.

Dexamthasone (DEX) treatments

The glucocorticoid treatments were performed as described by Guo and Gan (2006). 30 μm examethasone (DEX) was sprayed twice (once a day) to 2-week-old plants grown in pots. Photos were taken 2 days after the last spray.

SA treatment and chemical induction of gene expression

WT Col-0 plants, atnap and sag202 mutant plants (all 20 days old) were sprayed with 0.005% Silwet L-77 with or without 5 mM SA. Glucocorticoid treatments were performed as previously described (Guo and Gan, 2006). Twenty-day-old plants were sprayed with 30 μM dexamethasone (DEX, a synthetic glucocorticoid) containing 0.005% Silwet L-77. The 5th, 6th and 7th rosette leaves of each plant (counted from bottom) were collected for RNA extraction at different time points after the spray.

Yeast one-hybrid assay

Yeast one-hybrid assays were performed as previously described (Zhang and Gan, 2012). pGL3175 (the GAD-AtNAP fusion gene) was co-transformed with different LacZ reporter constructs containing different lengths of the SAG202 and ICS1 promoter fragments into the yeast strain EGY48. Similarly, pGL8040 (the GAD-SAG202 fusion gene) was co-transformed with different LacZ reporter constructs containing different lengths of the ICS1 promoter fragments into the yeast stain EGY48. The transformants were grown on proper dropout plates containing 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside (X-Gal) for the blue color development.

SA quantification

Non-senescing and mid-senescence leaves (0.1–0.3 g) of WT, atnap, sag202, ics1, and the leaves (also 0.1–0.3 g) of AtNAPⁱⁿ lines, SAG202ⁱⁿ lines and control lines at different time points after chemical induction were collected and analyzed for free SA using an LC–MS/MS (Zhang et al., 2013).

Bacterial growth assay

The bacterial strain Pto DC3000 suspension in sterile water (OD₆₀₀ = 0.002) were infiltrated into 6th and 7th leaves of 4-week-old plants using a needleless syringe. For determination of bacterial growth in inoculated leaves, the leaf samples were collected shortly (0d), 1d or 2d after inoculation. Bacterial inoculum preparation, syringe injection and bacterial pathogen enumeration were performed according to previously described (Katagiri et al., 2002).