Ac-YVAD-cmk, diisopropylfluorophosphate (DFP), E-64 and epoxomicin were provided by SIGMA-ALDRICH^®. The probes MV201, DCG04, JOPD1, FY01, MVB072, FP-Rh and FP-biotin (Patricelli et al., 2001; Kolodziejek et al., 2011; Richau et al., 2012; Lu et al., 2015) were available in the laboratory. Details are given in Supplementary Table S1 for correspondence between the classes of proteases, their specific inhibitors and activity-based probes.

Plants of B. napus L. (genotypes Ténor and Samouraï) were cultivated in a greenhouse at the vegetative stage according to previous experiments detailed by Poret et al. (2016). Seedlings were cultivated with 25% Hoagland nutrient solution [1.25 mM Ca(NO₃)₂.4H₂O, 1.25 mM KNO₃, 0.5 mM MgSO_4, 0.25 mM KH₂PO₄, 0.2 mM EDTA.2NaFe.3H₂O, 14 μM H₃BO₃, 5 μM MnSO₄, 3 μM ZnSO₄, 0.7 μM (NH₄)₆Mo₇O₂₄, 0.7 μM CuSO₄, 0.1 μM CoCl₂] for 6 weeks. Throughout their growth the plants were subjected to a 16 h photoperiod with a mean temperature of 20°C (day)/15°C (night) and received 400 μmoles photon. s^-1 m^-2 of photosynthetically active radiation at the canopy. During this experiment, the N remobilization in source leaves was determined precisely using a pulse-chase ¹⁵N labeling method. Thus, during the first 6 weeks of culture, plants received nitrate labeled with ¹⁵N at 2 atom% excess in order to obtain homogenous labeling throughout the plants. After 6 weeks of growth, corresponding to the beginning of treatments [Day 0 (D0)], the ¹⁵N-nitrate labeling was stopped. Plants were separated into two groups and supplied with 25% Hoagland solution containing two different nitrate concentrations: high (HN: 3.75 mM of CaNO₃) or low nitrate levels (LN: 0.375 mM CaNO₃ with compensation for Ca and K elements by adding 1.25 mM CaCl₂,2H₂O and 0.875 mM KCl). Leaves were numbered in order of their date of emergence with leaf rank no. 1 (L1) as the first emerged leaf. At D0, i.e., after 6 weeks of culture, leaf rank no. 12 (L12), a mature leaf becoming senescent during the experiment was chosen on the basis of its leaf area determined with a LI-COR 300 area meter (LI-COR, Lincoln, NE, United States) and chlorophyll content measured with a SPAD meter (Soil Plant Analysis Development; Minolta, SPAD-502 model). This leaf rank (L12) has a mean leaf area value of 57.88 cm² ± 1.96% variation and 33.31 cm² ± 6.01% variation for Ténor and Samouraï, respectively; mean SPAD value of 54.16 ± 4.55% variation and 55.33 ± 6.65% variation for Ténor and Samouraï, respectively. During the harvests after 0, 16, and 23 days of treatment (D0, D16, and D23), the chlorophyll and anthocyanin levels of L12 were measured by SPAD and an optical sensor system (Multiplex^®, Orsay, France; D’Hooghe et al., 2013), respectively. The leaf blade (laminae) of L12 was separated from the petiole and the midrib and was directly frozen at -80°C before further biochemical, proteomic and molecular analyses.

The N and ¹⁵N contents were quantified at D0, D16, and D23 on L12 of Ténor and Samouraï plants subjected to HN or LN supply. The determination was performed with an elemental analyser (EA3000, EuroVector, Milan, Italy) linked to a continuous flow isotope ratio mass spectrometer (IRMS, IsoPrime GV Instruments, Manchester, United Kingdom).

Soluble proteins were extracted from 200 mg of L12 fresh tissue with 1 mL of water. After centrifugation (5 min, 13,000 g, 4°C), concentrations were quantified by protein-dye staining (Bradford, 1976) in equivalent bovine serum albumin (BSA).

The labeling of active proteases was carried out according to Poret et al. (2017). Protein extracts (20 μL; see Supplementary Table S2 for concentrations of soluble proteins) were incubated in a mix containing 0.5 μM of probe (MV201, FY01, or JOPD1), 50 mM of sodium acetate buffer (NaAc, pH 5.5) and 2 mM DTT (Mix final volume: 200 μL). In parallel, 20 μL of protein extract were incubated in a mix of 50 mM Tris-base buffer (pH 7.5) containing 0.5 μM of MVB072 or 0.25 μM of FP-Rh (200 μL final volume). Because probes have not the same affinity for the active site of the targeted hydrolases, the two different mixtures were incubated for 4 h (MV201, FY01, or JOPD1) or 1 h (MVB072 or FP-Rh) under gentle agitation in the dark for better results. As control, equal volumes of soluble protein extracts of L12 (D0, D16, and D23) treated under HN or LN conditions were combined and 20 μL of each mixture were treated as described above. An equal volume of DMSO was added to the No-Probe-Control (NPC) and competition experiments were carried out by performing a pre-treatment for 30 min with 50 μM of E-64 (competition with MV201 and FY01), ac-YVAD-cmk (competition with JOPD1), epoxomicin (competition with MVB072) or DFP (competition with FP-Rh) before adding probes (see Supplementary Table S1 for correspondence between the classes of proteases, their specific inhibitors and activity-based probes). Reactions were stopped by adding 1 mL of ice-cold acetone to precipitate the proteins. After centrifugation (15 min, 16,000 g, 4°C), the pellet was dissolved in 2X SDS-PAGE gel-loading buffer (140 mM sodium dodecyl sulfate, 200 mM Tris, 20% glycerol, 5% β-mercaptoethanol, 0.3 mM Bromophenol Blue), heated at 90°C for 10 min and separated on 12% SDS-PAGE gels. The fluorescence of labeled proteins was visualized by scanning the gels using a Typhoon 9400 scanner (GE Healthcare Life Science, excitation wavelength: 532 nm; emission wavelength: 580 nm). Signals were quantified by ImageJ software. To control the protein quantity after electrophoresis, gels were stained with Coomassie Brilliant Blue stain [0.5 g CBB G250, 10% acetic acid, 45% methanol in ultra-pure water (v/v)], destained [10% acetic acid, 40% methanol in ultra-pure water (v/v)] and scanned. Finally, specific protease activity was expressed as fluorescence intensity.mg^-1 protein (see Supplementary Table S2 for concentrations of soluble proteins).

To identify active proteases previously detected on gels as described above, protein extracts were labeled with biotin-tagged probes and a pull-down of biotinylated proteins was performed with a modified protocol from Poret et al. (2016). Briefly, 900 μg of protein was labeled with 10 μM of DCG04 or FP-biotin in labeling buffer [50 mM sodium acetate buffer (NaAc, pH 5.5), 2 mM DTT for DCG04 or 50 mM Tris-buffer, pH 7.5 for FP-biotin]. Samples were incubated under gentle agitation at room temperature for 4 h (for DCG04) or 1 h (for FP-biotin). An additional aliquot was treated as described above but without probes as the control. As described by Chandrasekar et al. (2014), the reaction was stopped and the biotin-proteins were purified using streptavidin beads. Finally, after separation on 12% SDS-PAGE gels, the eluted proteins were stained using the silver staining procedure described by Blum et al. (1987).

Bands of interest were manually excised and were reduced at 50°C for 1 h with 10 mM dithiothreitol (DTT, GE Healthcare) and alkylated for 1 h in the dark with 55 mM iodoacetamide (IAA, Sigma). The gel fragments were washed several times with water and ammonium carbonate, dehydrated with 0.1% acetonitrile and dried. Trypsin digestion was performed overnight with a dedicated automated system (MultiPROBE II, PerkinElmer). The gel fragments were subsequently incubated twice for 15 min in acetonitrile solution to extract peptides from the gel pieces. Peptide extracts were then dried and dissolved in a buffer containing 3% acetonitrile and 0.1% formic acid for chromatographic elution. Peptides were enriched, separated and analyzed using a 6520 Accurate-Mass Q-TOF LC/MS equipped with an HPLC-chip cube interface (Agilent Technologies, Massy, France). The fragmentation data were interpreted using the Mass Hunter software (version B.03.01, Agilent Technologies). For protein identification, MS/MS peak lists were extracted, converted into mzdata.xml format files and compared with the protein database (NCBInr-Viridiplantae) using the MASCOT Daemon search engine (version 2.1.3; Matrix Science, London, United Kingdom). The searches were performed with no fixed modification and with variable modifications for oxidation of methionine, and with a maximum of two missed cleavage sites. MS/MS spectra were searched with a mass tolerance of 20 ppm for precursor ions and 0.6 Da for MS/MS fragments. Only peptides matching an individual ion score > 48 were considered. Proteins with two or more unique peptides matching the protein sequence were considered as a positive identification. The assigned protein of best match is provided alongside the UniProt or NCBI/GenBank accession number. Score, queries matched, peptide matches, different peptide matches, experimental mass and theoretical mass are also presented. In addition, other proteins identified in B. napus with the same peptides by MASCOT are presented. Protein sequences were matched against sequences of Brassicacea proteins using the NCBI BLAST Protein Database (algorithm blastp) and the best BLAST results are presented with the name of the protein, the organism, the UniProt or NCBI/GenBank accession number and the percentage of sequence identity. Finally, PLCPs were classified according to the classification of Richau et al. (2012) while SPs were classified according to the MEROPS database.

For immunodetection of the CPs SAG12 and RD21, soluble proteins (30 μg) were first denatured with Laemmli 2X buffer (Laemmli, 1970) containing 5% β-mercaptoethanol (v/v) and separated on a 4–15% gradient in SDS-PAGE precast Stain-free gels (Mini-PROTEAN^® TGXTM Stain Free, Bio-Rad, Marne-la-Coquette, France). The gels were scanned under UV light with a Gel Doc^TM EZ scanner (Bio-Rad^®, Marne-la-Coquette, France) for detection of proteins and the Western blot to the polyvinylidene difluoride (PVDF) membrane was performed as described by Desclos et al. (2008). The immunodetection of SAG12 was made using specific polyclonal antibody from rabbit provided by Agrisera^® [1/2000, diluted in Tris buffer saline – Tween 20 (TBST; Tris 10 mM, NaCl 150 mM, pH8, Tween 20 0.15% (v/v)) containing 5% milk (w/v) to avoid non-specific hybridization]. For immunodetection of RD21, specific polyclonal antibody from rabbit kindly provided by Dr. Hara-Nishimura (Yamada et al., 2001) was used (1/1000, diluted in TBST containing 5% milk). The primary antibody was detected by a secondary antibody from goat coupled with horse-radish peroxidase [(1/10000 and 1/2000 for immunodetection of SAG12 and RD21, respectively) diluted in TBST, Bio-Rad^®] and detected by enhanced chemiluminescence (ECL kit, Bio-Rad^®) using a ProXPRESS 2D proteomic Imaging System (PerkinElmer, Courtaboeuf, France).

Phytohormones were extracted by grinding 200 mg of frozen L12 tissue with 1 mL of extraction buffer (methanol 80%, formic acid 1% in ultra-pure water). After 5 min in an ultrasonic bath at 4°C, samples were agitated for 30 min at room temperature and centrifuged at 12,000 g (10 min at 4°C). The resulting supernatant containing phytohormones was collected and a second extraction was performed from the resulting pellet. Both supernatants were mixed and evaporated using a SpeedVac. Finally, the pellet was resuspended in 100 μL of methanol containing 1% of formic acid buffer and then analyzed by UPLC-ESI (+/-) –TQD as described in Pan et al. (2010).

Briefly, the extract obtained is analyzed by ultra-high performance liquid chromatography (UPLC) (Acquity of Waters, Guyancourt, France) on a reverse phase column coupled with a double detection in UV-Visible and by mass spectrometry triple quadrupole (Acquity of Waters, Guyancourt, France), which allows a specificity and sensitivity compatible with the quantification of the phytohormones.

The normality of the data was studied by using the Ryan-Joiner test at 95%. Analysis of variance (ANOVA) and the Newman–Keuls test were performed by using Microsoft^® Excel 2010/XLStat^® 2014 to compare the means. The non-parametric test of Kruskal–Wallis was carried out if the normality law of the data was not respected. Statistical significance was postulated at P < 0.05. Three biological repetitions were analyzed (n = 3) for all measurements and all the data are presented as the mean ± standard deviation (SD).

The two B. napus L. genotypes, Ténor and Samouraï, were subjected to ample (HN: 3.75 mM NO₃^-) or low nitrogen supply (LN: 0.375 mM NO₃^-) for 23 days after a pulse-chase labeling treatment with ¹⁵N-NO₃^-. Chlorophylls and anthocyanin contents as well as the soluble proteins and the total N and ¹⁵N contents were followed in a mature leaf (L12, assumed to commence aging and develop as a source tissue during the time course of the experiment) to study senescence progression during the 23 days of the HN or LN treatments (see section “Materials and Methods”).

As previously shown by Poret et al. (2016) in oilseed rape (cv. Aviso), a limitation of nitrate supply accelerated leaf senescence compared to plants subjected to HN conditions in both genotypes (Figure 1). Indeed, in agreement with the fact that chlorophyll degradation corresponds to one of the first visible processes of aging (Pan et al., 2010), chlorophyll content decreased significantly after 23 days under LN conditions compared to the HN supply and particularly for Ténor (9.5 ± 3.8 SPAD units) compared to Samouraï (26.9 ± 2.3 SPAD units) (Figure 1A). Moreover, the anthocyanin content significantly increased during leaf senescence in the two genotypes and particularly after 23 days under LN supply (Figure 1B). This increase in anthocyanin content during leaf senescence was previously shown in oilseed rape (Poret et al., 2016) and corresponds to a stress symptom in oilseed rape (D’Hooghe et al., 2013). It was recently proposed that the photo-protective function of anthocyanins prevents the risk of photooxidation and enables a tightly controlled and efficient chlorophyll breakdown during leaf senescence (Ougham et al., 2005; Diaz et al., 2006; Koeslin-Findeklee et al., 2015).

As proteolysis efficiency is critical for N recycling performance, the quantity of soluble proteins during senescence was compared in Ténor and Samouraï (Figure 1C). After 23 days, the quantity of soluble proteins was significantly decreased in L12 and this was particularly acute for Ténor under LN conditions with a degradation of 85% of the initial pool of soluble proteins versus only 70% for Samouraï. Then, because the N requirements of growing organs is significantly fulfilled by the transport of N compounds that originate from the recycling of amino acids and soluble proteins in senescing leaves (Demirevska-Kepova et al., 2005; Thoenen et al., 2007), the total N and ¹⁵N contents were studied (Figure 1D,E). As expected, the amounts of total N and ¹⁵N strongly decreased during leaf senescence, which could correspond to an increase in N remobilization from the old source leaves to the young sink organs. Moreover, Ténor was characterized by a stronger decrease in the total N and ¹⁵N contents during the 23 days compared to Samouraï, with a decrease of 70% in ¹⁵N for Ténor versus only 50% for Samouraï.

In conclusion, Ténor, which is able to maintain its biomass production under N limitation compared to Samouraï (Girondé et al., 2015), is characterized by a stronger decrease in chlorophyll content, an enhanced degradation of soluble proteins under LN conditions and this leads to a stronger decrease in the ¹⁵N and total N quantities under LN conditions compared to Samouraï.

As shown previously, Ténor is able to strongly degrade soluble proteins compared to Samouraï particularly in plants supplied with LN treatment. From these results, two different hypotheses were postulated: (i) the activities of SAPs may be stronger for Ténor than Samouraï and/or (ii) Ténor may present other SAPs activities that could be absent in Samouraï. In order to validate or refute these hypotheses, the aim was to address the typology of protease activities associated with leaf senescence for Ténor and Samouraï. To reach this goal, an appropriate method of labeling active proteases was undertaken at pH 5.5 or 7.5 using activity-based probes specific for different protease classes (van der Hoorn and Kaiser, 2012). The following results allowed validation of the two hypotheses postulated above.

The senescence process is highly regulated by different endogenous factors that include the phytohormones (Jibran et al., 2013; Khan et al., 2014). Therefore, the difference in leaf-senescence protease activities between Ténor and Samouraï could be associated with differences in the regulation of senescence by phytohormones. To validate or refute this hypothesis, our goal was to study and compare phytohormone contents between Ténor and Samouraï during leaf senescence and correlate the hormonal patterns with the protease activities. To reach this goal, several phytohormone contents were measured by UPLC-ESI (+/-) -TQD in plants of Ténor and Samouraï subjected to HN or LN supply for 23 days (Figure 8).

Surprisingly, the MeJA content decreased during leaf senescence regardless of the treatment and the genotype (Figure 8A) whereas it has been demonstrated that exogenous application of MeJA up-regulates the transcript abundance of genes that are markers of developmental senescence in A. thaliana (Jung et al., 2007). Similarly, the JA content decreased during leaf senescence in B. napus L. regardless of the nitrate supply or the genotype (Figure 8B) while JA biosynthesis is upregulated in senescing leaves of A. thaliana (Seltmann et al., 2010). Otherwise, the auxin (IAA) content remained stable during leaf senescence irrespective of the nitrate treatment and the genotype (Figure 8C). The role of auxin (IAA) in senescence is not clear. Indeed, it has been demonstrated that the IAA-induced gene, SAUR36, was up-regulated during senescence while its dysfunction induced a delay to senescence in A. thaliana (Hou et al., 2013). Contrastingly, a senescence repressor role for IAA was demonstrated in A. thaliana with the down-regulation of several senescence-associated genes (SAGs), including SAG12, by an exogenous application of IAA (Noh and Amasino, 1999; Kim et al., 2011). MeJA, JA and IAA do not seem to be associated with the different behaviors observed during leaf senescence in Ténor and Samouraï because the levels were quite similar, and therefore their roles in senescence processes remain unclear in B. napus L.

Compared to the initial level (Day 0), the ABA content significantly increased during leaf senescence in Tenor after 23 days under nitrate limitation, while it decreased in Samouraï (Figure 8D). A promotive role of ABA has already been demonstrated during senescence in A. thaliana with the up-regulation of SAG113 (a senescence associated gene) linked with an increase in the ABA endogenous content (Zhang et al., 2012). Moreover, endogenous ABA content increases during senescence in many plant species such as Avena sativa (Lim et al., 2007) and A. thaliana (Zhang et al., 2012).

The SA content increased during leaf senescence in plants subjected to N limitation for 23 days, regardless of the genotype (Figure 8G). This is correlated with the fourfold increase in SA levels reported in senescent leaves of A. thaliana by Morris et al. (2000). Moreover, these authors have also reported that a dysfunction of SA synthesis in A. thaliana leads to a delay in senescence and a down-regulation of SAG12 expression (Morris et al., 2000).

The contents of two intermediate molecules of cytokinin synthesis, IP and isopentenyladenosine (IPR), decreased during senescence in plants regardless of the N treatment and the genotype (Figure 8E,F). This is correlated with the general decrease in cytokinin content observed during leaf senescence in N. tabacum while an increase in the endogenous cytokinin level leads to a delay in senescence processes (Yu et al., 2009). Our results are also in agreement with the fact that over-expression of the isopentenyltransferase gene, responsible for cytokinin biosynthesis, generally leads to senescence delay and stay-green phenotypes in numerous species including B. napus (Kant et al., 2015; Zeng and Zhao, 2016). Because ABA, SA and cytokinins seem to be strongly involved in the regulation of leaf senescence in B. napus L., we focused our analysis on the content of these phytohormones. The change in the ratio between phytohormones that accelerate leaf senescence, such as ABA and SA, and those that negatively regulate senescence, such as cytokinins was analyzed (Figure 8H). The ratio (SA + ABA)/(IP + IPR) increased greatly during leaf senescence in Ténor plants after 23 days of treatment, and particularly after N limitation, whereas it remained stable for Samouraï leaves regardless of the nitrate treatment. This ratio allowed the two genotypes to be distinguished during senescence as well as a clear separation between the N treatments in Ténor during senescence. This ratio was also clearly correlated with the activities of the PLCPs, the AALP and the SPs during senescence with Pearson correlation coefficients of 0.866, 0.932, and 0.905, respectively (p-value < 0.0001, data not shown). These data suggested that different hormonal balances during senescence between Ténor and Samouraï could be associated with stronger PLCP and SP activities and proteolytic performance in Ténor compared to Samouraï during senescence, particularly under N limitation. As already described, phytohormones like ABA and SA are able to up-regulate protease transcripts or protease abundance and dysfunction in SA synthesis during senescence leads to down-regulation of SAG12 gene expression in A. thaliana (Morris et al., 2000). Further, ABA has been shown to enhance RuBisCO degradation in detached leaves of Oryza sativa L. floated on a solution containing ABA compared to those on a solution without ABA (Fukayama et al., 2010). This was correlated with an increase in SDS-dependent protease activities (with an optimum pH 5.5 suggesting a vacuolar localization) that significantly increased with ABA treatment. These types of protease activities are also increased during leaf senescence in rice (Fukayama et al., 2010).

In addition to these data, a preliminary study that infiltrated exogenous SA or water in mature leaves of oilseed rape (cv. Aviso) was performed (Supplementary Figure S2). The degradation of RuBisCO was followed in vitro in the presence or absence of MG132 (an inhibitor of CPs and the proteasome) in these leaves. When leaf tissues were infiltrated with SA, the level of RuBisCO degradation was strongly reduced in the presence of MG132 (with a rate of degradation of 32% versus 60% in the absence of MG132) compared to water infiltration (rate of degradation of 46% versus 59% in the absence of MG132) (Supplementary Figure S2). This enhanced inhibition of RuBisCO degradation by MG132 after SA infiltration (inhibition of 46% versus 22% for water infiltration, Supplementary Figure S2) suggests that the contribution of CPs and the proteasome activities to proteolysis is more important after SA application and reinforces the assumption that SA is involved in the regulation of CP and proteasome activities in leaves of oilseed rape. According to these results observed during leaf senescence, we have recently demonstrated that the infiltration of ABA and SA in cotyledons of oilseed rape (Ténor) provokes the induction of senescence and several cysteine and SP activities in cotyledons (Poret et al., 2017). Nevertheless, additional experiments will be required to validate the postulate that these phytohormone patterns could regulate PLCP or SP protease activities.