Physiological and Proteomic Changes in the Apoplast Accompany Leaf Senescence in Arabidopsis

The apoplast, i.e. the cellular compartment external to the plasma membrane, undergoes important changes during senescence. Apoplastic fluid volume increases quite significantly in senescing leaves, thereby diluting its contents. Its pH elevates by about 0.8 units, similar to the apoplast alkalization in response to abiotic stresses. The levels of 159 proteins decrease, whereas 24 proteins increase in relative abundance in the apoplast of senescing leaves. Around half of the apoplastic proteins of non-senescent leaves contain a N-terminal signal peptide for secretion, while all the identified senescence-associated apoplastic proteins contain the signal peptide. Several of the apoplastic proteins that accumulate during senescence also accumulate in stress responses, suggesting that the apoplast may constitute a compartment where developmental and stress-related programs overlap. Other senescence-related apoplastic proteins are involved in cell wall modifications, proteolysis, carbohydrate, ROS and amino acid metabolism, signaling, lipid transport, etc. The most abundant senescence-associated apoplastic proteins, PR2 and PR5 (e.g. pathogenesis related proteins PR2 and PR5) are related to leaf aging rather than to the chloroplast degradation program, as their levels increase only in leaves undergoing developmental senescence, but not in dark-induced senescent leaves. Changes in the apoplastic space may be relevant for signaling and molecular trafficking underlying senescence.


INTRODUCTION
At the leaf level, senescence represents a degenerative process, but, at the same time it constitutes an efficient nutrient recycling mechanism for the plant. Most of the nitrogen released from senescing leaves comes from Rubisco and other plastid proteins degradation; therefore a major focus of senescence research has been to elucidate the mechanism for chloroplast dismantling (Noodén et al., 1997;Guiamet et al., 2002;Otegui, 2018). Other cellular changes occurring in senescing leaves are less characterized, and even less is known about the adjustments taking place outside the plasma membrane as leaf senescence progresses.
The apoplast is the space external to the plasma membrane that includes cell walls, middle lamella, intercellular spaces, and the fluid that moves freely within the walls and intercellular spaces, known as extracellular fluid or apoplastic fluid (AF). The AF represents around 4%to11%ofthetotalleaffreshweight (SattelmacherandHorst,2007). Its composition consists of molecules related to metabolism and signaling, such as amino acids, polysaccharides, secondary metabolites, ions, and secreted mRNA and proteins with several functions, particularly enzymes related to defense and proteolysis (Sattelmacher and Horst, 2007;Martínez-González et al., 2018). The apoplast plays a critical role in plant development, as it provides the environment for cell expansion and cell wall maintenance, ion and molecule trafficking, source-to-sink translocation of nutrients, intercellular and systemic signaling, stress perception, and response (Hoson, 1998;O'Leary et al., 2016;Rodríguez-Celma et al., 2016). Stress-elicitedchangesintheapoplasthavebeenextensivelyexamined, and these include modulation of the pH and fine tuning of protein abundance and composition (Felle, 2001;Dani et al., 2005;Floerl et al., 2012;Geilfus, 2017). Apoplastic pH (pH apo ) transiently increased under drought and salinity becoming more alkaline as the stresses intensify Zörb et al., 2015;Geilfus, 2017). Also, pathogen and peptide signals, such as systemin and RALF (rapid alkanilization factor), trigger apoplast alkalization (Felle et al., 2004;Higgins et al., 2007;Masachis et al., 2016). The pH apo varies along ontogeny (Mattssonand Schjoerring, 2003;Wu etal., 2016), and along with pH apo, the extracellular space volume might adjust as the leaf ages (HustedandSchjoerring,1995;Lohausetal.,2001;Nouchietal.,2012). Apoplast conditions influence chloroplast metabolism and leaf physiology (Karpinska et al., 2018) and therefore might influence senescence-related processes, for instance the regulation of intra to extracellular exchange of molecules and long distance remobilization of nutrients. Transcription analysis of Arabidopsis plasma membrane transporters suggest that molecule trafficking across the plasma membrane might increase during leaf senescence (Van Der Graaff et al., 2006).
Proteomic approaches identified chitinases, pathogenesis related proteins (PR), and other defense related enzymes, as the mayor leaf apoplastic proteins (Boudart et al., 2005;Rutter and Innes, 2017;Soares et al., 2017). The extracellular accumulation of these enzymes along with the transient alkalinization of pH apo as signature of different biotic and abiotic stress responses suggest either a cross-talk between stress pathways or a common apoplastic signal-transducing element or node (Geilfus, 2017).
Many of the stress-related extracellular enzymes are constitutive members in the AF that activate and/or accumulate upon specific signals, some of them are also up-regulated during senescence (Grudkowska and Zagdanska, 2004;Goulet et al., 2010). Other stress-related enzymes relocate inside/outside the cell in response to external stimulus. Caspase-like serine proteases from Avena sativa relocate from the cytosol to the apoplast upon programmed cell death (PCD) induction (Coffeen and Wolpert, 2004), whereas some apoplastic subtilisin proteases re-enter cells committed to PCD (Trusova et al., 2019).
Similar intra-extracellular pathways might be involved in the regulation and/or execution of leaf senescence.
Compared to stress-related broad analysis of AF proteomes (Kosová et al., 2011;Gupta et al., 2015), there is not enough information on the AF proteome dynamics in senescing leaves, however different studies evidence relevant roles for apoplastic proteins during this leaf stage. By regulating long-distance movement of sucrose, the extracellular invertase (cwINV) and its inhibitor (INVINH) probably play a crucial role in the regulation of senescence by controlling source-sink relations (Lara et al., 2004;Jin et al., 2009). The apoplastic subtilisin protease SASP is highly up-regulated during senescence, and whereas at-sasp plants do not differ from wild type plants at vegetative stage, they produce more branched inflorescences, siliques, and seeds . The Arabidopsis extracellular metalloprotease At2-MMP is up-regulated as the plant ages, and at2-mmp-1 plants show accelerated chlorophyll (Chl) degradation and delayed flowering (Delorme et al., 2000;Golldack et al., 2002). Other apoplastic proteases from different mechanistic classes (cysteine-, metallo-, and serine-proteases) are up-regulated during leaf senescence (Martínez and Guiamet, 2014).
This study aimed to shed light on the dynamics of the extracellular space during leaf senescence by analyzing physiological parameters of the apoplast space and the AF, including a large-scale quantitative proteomic approach to compare the AF proteomes of senescent and non-senescent leaves.

Plant Material and Growth Conditions
Arabidopsis thaliana Col-0, wild type, and the transgenic line apo-pHusion (Gjetting et al., 2012) were used. Apo-pHusion plants express the chimeric apo-mRFP1-EGFP protein targeted to the apoplast, where it functions as a pH sensor (Gjetting et al., 2012). The plants were cultivated in 550 mL pots filled with soil and vermiculite (2:1 v/v). Nitrofoska ® was applied (30 mL, 1 g/L per pot) every 30 days. The plants were grown in growth chambers, at 24°C and 120 mmol m −2 s −1 photosynthetic photon flux density under a 10 h light/14 h dark photoperiod.
Each rosette was separated in groups of leaves based on the phyllotaxis and leaf size. Vegetative rosettes were separated in two groups of leaves, the youngest named S1 (Stage 1), whereas rosettes at reproductive stages were separated in three, four, or five groups of leaves, according to the rosette age and leaf number (stages S2, S3, and S4). Around six to eight consecutive leaves representing one stage (S1, 2, 3, or 4) were sampled per plant. Each plant was harvested only once and for one particular Stage only.

Physiological Parameters
Leaf chlorophyll content was measured non-destructively with the SPAD 502 Portable Chlorophyll Meter (Konica-Minolta ® ).
Maximum quantum yield of photosystem II (Fv/Fm) measurements were taken with a modulated pulse fluorimeter FMS 2 (Hansatech ® ). The leaves were acclimated for 30 min in the dark before the measurements.
Leaf water content, LWC, was determined as: Relative water content, RWC, was determined as: where FW = fresh weight, DW = dry weight, SW = water saturated weight. Water potential (Y w ) of 5 mm diameter leaf discs was measured with a dew point psychrometer and C-52 chambers (Wescor ® ). Sodium chloride solutions of different concentrations were absorbed in filter paper discs and used for calibration.
Solute potential (Ys): measurements were carried out in a vapor pressure micro osmometer Vapro 5520 (Wescor ® ), using 10 mL per sample. To measure leaf Ys, leaves without the midrib were frozen in liquid nitrogen and placed in a 1 mL syringe containing glass wool at the bottom. Pressure was applied pushing the plunger and the resulting leaf macerates were analyzed. The Ys of the AF was estimated measuring the Ys of the apoplastic wash fluids (AWFs), and correcting the obtained values by their corresponding dilution factors (F dil ) (See below).

Leaf AWF Extraction
Leaf infiltration was performed with cold deionized water using a 60 mL syringe as described in O'Leary et al. (2014). Infiltrated leaves were stuck, with petioles facing up into a 50-mL centrifuge tube and centrifuged at 600 g for 40 min at 4°C. For protein analysis, 5 mL of 300 mM phenylmethylsulfonyl fluoride (PMSF) was added at the bottom of the tube prior to centrifugation.
Estimation of the Apoplast Volume Occupied by Fluid (V AF ) and Air (V air ) and of AWF Dilution Factor Estimations of V air and V AF were carried out by the Indigo Carmine (indigo-5.5′disulfonic acid, disodium salt, IC) method (Husted and Schjoerring, 1995) with modifications. Fully hydrated leaves were excised from Arabidopsis plants, weighed, and vacuum infiltrated, as described above, but with 50 mM IC, in 10 mM sodium phosphate buffer pH 6.2. Once infiltrated, leaf surfaces were blotted, immediately re-weighed, and centrifuged. The difference in the leaf weight before and after infiltration (corrected by the density of IC solution) was used for calculation of the apoplastic air volume (V air ) (Lohaus et al., 2001).
The IC infiltration solution and AWFs were measured spectrophotometrically at 610 nm to calculate the IC concentration. The volume of AF (V AF ) was calculated using the following equation: The dilution factor (F dil ) of AWF was calculated based on V AF and V air : The physiological concentration of solutes and metabolites in the AF was calculated by multiplying the solute or metabolite concentration in the AWF by its dilution factor, F dil .
In Vivo Estimation of Leaf Apoplast pH by Confocal Fluorescence Microscopy

Ex Vivo Calibration of pHusion
Aliquots of AWF extracted from apo-pHusion plants were incubated with 300 mM Tris-HCl or 300 mM MES buffers adjusted with HCl or KOH to different pH values, placed in microscope slides with cavities, and detected by confocal microscopy. Confocal data acquisition was performed with a Leica TCS SP5 II CLSM with the following setting: Ex 488/Em 524-550 nm for EGFP, and Ex 543/Em 566-634 nm for mRFP1. For each pH value, pixel intensity of the fluorescence ratio EGFP/mRFP1 in a region of interest (ROI) was calculated using the software ImageJ. A non-linear fitting of average pixel intensities was calculated using ImageJ software, and the software GraphPad Prism was used to generate a sigmoid curve.

In Vivo Observation of Apoplastic Apo-pHusion
Images of apo-pHusion S2 and S3 leaves were acquired by confocal microscopy, with the setting described above. Ratio images EGFP/mRFP1 were generated using the software ImageJ and the average pixel intensities EGFP/mRFP1 for each ROI were calculated. PH values were estimated by extrapolation on the calibration curve. LAS AF software version 2.2.1 was used for image acquisition. For visual presentation of ratio images, a pseudocolor look-up table was designed using the software ImageJ.

Isolation of Cell Wall Polysaccharides and Cell Wall Quantitation
Cell wall polysaccharides were obtained from alcohol insoluble residues (AIR) according to Nardi et al. (2015). To estimate the amount of cell wall, the starch was subtracted from each AIR with Dimethyl sulfoxide (DMSO) according to Carpita and Kanabus (1987). After removing the starch, the resulting AIR was incubated with 0.01 M HCl in a boiling water bath for 2 h and filtered for polyuronide extraction. The obtained solution was used for pectin quantification (see below).

Hydrolysis and Quantification of Pectins
100 mL of the polyuronide filtrate was mixed with 600 mL of 0.0125 M sodium tetraborate in saturated H 2 SO 4 , incubated in a water bath for 20 min. and cooled down. Uronic acid concentrations were estimated by the m-hydroxydiphenyl method (Blumenkrantz and Asboe-Hansen, 1973) using galacturonic acid as standard.

Starch Quantification
Around 0.1 mL of DMSO extract was mixed with 1 mL of 0.05% w/v anthrone in 66% v/v H 2 SO 4 and incubated at 100°C for 10 min and cooled down. The starch content was measured spectrophotometrically at 620 nm using glucose as standard.

Transmission Electron Microscopy
Around 2 mm 2 pieces were sectioned from the middle region of the leaf and fixed by immersion in 2% glutaraldehyde, post fixed in 1% osmium tetroxide and dehydrated through a graded series of ethanol and embedded in epoxy resin. Ultrathin sections were cut with a Super Nova Reichert-Jung Ultra-microtome (Wien Austria), and examined with a JEM 1200 EX II transmission electron microscope (JEOL Ltd., Tokio, Japan). Cell wall analysis was performed with Image J software on at least three different leaves of each stage.

Protein Analysis
Leaves were homogenized in 50 mM Tris-HCl pH 8, 20 mM ethylenediaminetetraacetic acid (EDTA), 10 mM PMSF and 1% insoluble polyvinylpyrrolidone, and cleared at 15,000 g, 4°C for 15 min. Laemmli's Sample Buffer, LSB, was added before electrophoresis. Aliquots of AWF were lyophilized in a vacuum freeze-dryer (Scientz ® -10 N) and the pellets were re-suspended in LSB.
Densitometric quantification of protein bands in CBB-stained gels was performed by using the ImageJ 1.6 software (National Institutes of Health, USA).

Peptide and Protein Quantification
The Pierce ® BCA Protein Assay Kit (Thermo Scientific) using bovine serum albumin as standard was used with AWFs and leaf extracts Western Blot and Immunodetection SDS-PAGE gels were transferred to nitrocellulose membranes using a Bio Rad Miniprotean System. The membrane was blocked with 10% milk in PBST buffer, incubated with a polyclonal antibody against Rubisco Large Subunit, and developed with a horseradish peroxidase-conjugated secondary antibody.

Protein Identification and Quantification by LC-MS/MS and Bioinformatics
For the identification of proteins in SDS-PAGE gels the bands were excised from G-250 stained gels, washed, dried, reduced, alkylated, and trypsin digested. Peptides were extracted with three steps of 50% acetonitrile/0.5% trifluoroacetic acid, and desalted with a Zip-Tip C18. Samples were analyzed by LC-MSMS using a Q Exactive™ nano HPLC-ESI-Orbitrap mass spectrometer (Thermo Fisher Scientific). Protein identification was performed with Proteome Discoverer 1.4 (Thermo Fisher Scientific) connected to a Mascot search engine server (Matrix Science, London, UK). Only proteins identified with at least two unique peptides with 0.99 confidence (FDR = 0.01) were considered.
For large-scale protein analysis by shotgun proteomics, lyophilized AWF aliquots were resuspended in 8 M urea; aliquots containing 20 µg of protein were reduced, alkylated, trypsin digested, and desalted. LC-MSMS analysis was performed as described previously. A label-free-quantification method was performed using Proteome Discoverer 1.4. Four biological replicas for each leaf stage were analyzed. Identified proteins were submitted to a Student's t-test (P ≤ 0.05) with Perseus 1.6.1.3 (Max Planck Institute of Biochemistry). Only proteins exhibiting at least a ±100% fold change in their amount were considered.

Statistical Analysis
Data from parameters (Chl, protein content) were tested either with one-way ANOVA and Tukey tests (P ≤ 0.05), with two-way ANOVA and Holm-Sidak test (p ≤ 0.05), or with Dunnett's multiple comparisons test. For paired analysis of S2 and S3 leaf stages, means were compared using the Student's t-test (p ≤ 0.05). Statistical analysis was plotted using the GraphPad Prism 6.01 software (GraphPad Software, San Diego, CA, USA). The number of independent replicates, i.e, biological samples (n), and the corresponding technical replicates are indicated in each table and figure caption.
Cell wall thickness was analyzed by one-way ANOVA, and univariate tests of significance, Sigma-restricted parameterization Effective hypothesis decomposition (P ≤ 0.0001).

Leaf Developmental Stages and Apoplastic Fluid Isolation
In order to characterize changes in the apoplast accompanying senescence, leaf developmental stages were defined and apoplast fluids (AFs) were isolated. Four leaf stages were characterized: non-senescent leaves from plants at vegetative (S1) and early reproductive stages (S2), and senescing leaves from plants at mid (S3) and late reproductive stages (S4) ( Figures 1A, B). Chl and protein content per leaf area and Photosystem II maximum photochemical efficiency (Fv/Fm) were determined. S1 and S2 leaves showed no difference in Rubisco and total leaf protein content ( Figure 1C), PSII integrity, as assessed by Fv/Fm, ( Figure  1E), and S2 Chl concentration was slightly higher than in S1 leaves ( Figure 1D). S3 and S4 leaves showed clear signs of senescence, i.e. Chl and protein levels and Fv/Fm decreased in S3 and S4 compared to S1 and S2. Particularly, the amount of leaf protein dropped down to 40 and 20% in S3 and S4 respectively, compared to non-senescing S1 and S2 leaves. AF isolation was performed following the vacuum infiltration-centrifugation method (Lohaus et al., 2001;O'Leary et al., 2014) and obtained AF fractions (from here on AWFs) were tested for potential contamination with intracellular compounds. AWFs and whole leaf extracts were assayed for the cytosol marker Glucose 6P Dehydrogenase (G6PDH), which was expressed as G6PDH activity in AWF as a percentage of the whole leaf extract activity ( Figures 1F, G). AWF from S1 to S3 leaves were devoid of detectable G6PDH activity, whereas AWF S4 showed high activity of the cytosolic enzyme, indicating the presence of intracellular compounds in the AWF, probably due to age-related loss of cell membrane integrity ( Figure 1G). Based on whole-leaf and AWF analysis ( Figures  1A-G), S3 leaves show clear signs of senescence while still retaining cell integrity, whereas S4 leaves would represent the culmination of the remobilization process followed by membrane permeabilization and cell death. S1 and S2 showed no difference in senescence-related parameters; therefore S2 was selected as non senescent leaf stage to compare its apoplastic space and AF with that of senescing S3 leaves. Both S2 and S3 leaves belong to plants at reproductive stage and that would avoid or minimize any variation in the apoplast related to different phases of plant development.

Water Relations of Non Senescing and Senescing Leaves. Relative Water Content and Water Potential in S2 and S3 Leaf Stages
Leaf water status directly affects the apoplast water content and therefore its metabolite concentration (Tyree, 1976;Andersen et al., 1991;Wardlaw, 2005). Water relations vary throughout Arabidopsis leaf development, particularly right after the leaf reaches its maximum extension and starts yellowing (Breeze et al., 2011). Relative water content (RWC), moisture, and water Values in the graphs show the average ± standard error, SEM, of independent measurements. Data were normally distributed. SPAD readings were taken in 10 to 15 different leaves for each stage, two readings per leaf blade. Fv/Fm values are the average of 16 leaves (S1), 30 leaves (S2 and S3), and six different S4 leaves, one reading per leaf blade. G6PDH activity (D and E) represents the average of biological samples (n), where each FA and Ext biological sample was made up of at least 4 leaves. S1: n = 6 FA and Ext n = 6, S2: n = 4 FA and n = 3 Ext, S3: n = 9 FA and n = 5 Ext, S4: n = 3 FA and 3 Ext. Significant differences were calculated at P ≤ 0.05 between the conditions using the Tukey test; letters indicate significant differences between stages. Asterisks indicate significant differences between AF and Ext; ****, P < 0.0001, calculated using a two-way ANOVA test. potential (Y w ) were determined in S2 and S3 leaves. RWC remained constant between S2 and S3 stages (S2 RWC = 0.949 ± 0.002 and S3 RWC = 0.955 ± 0.004) ( Figure 2A). However, percentage leaf water content decreased and leaf fresh weight per area (LFWA) increased in S3 compared to S2 leaves ( Figures 2B, C). A higher LFWA might relate to starch accumulation during senescence (Oda-Yamamizo et al., 2016). The 3-fold increase in starch content detected from S2 to S3 leaves ( Figure 2D) might explain the higher LFWA in S3 leaves.
Leaf water potential dropped from S2 to S3 (Y w = −1.00 ± 0.21 to −1.26 ± 0.25 MPa, respectively), whereas the solute potential remained stable between stages (Ys S2 = −0.94 ± 0.03 and Ys S3 = −1.01 ± 0.08) ( Figure 2E, left and middle parts of the graph). Intriguingly, experimental Ys values were the same or higher than Y w values. Since extracts from frozenthawed squeezed leaves are composed of cellular, AF and xylem contents, it would be plausible that AF and xylem fluids diluted the cellular sap, leading to higher Ys than expected based on leaf disk Y w values. Yet, a methodological artifact cannot be dismissed since Y w was measured in fresh leaf disks with a dew point psychrometer, whereas Ys was measured in frozensqueezed leaves using a vapor pressure micro-osmometer.
S2 and S3 AF solute potential (Y S AF) was estimated measuring AWFs with the vapor pressure microosmometer, and correcting the obtained AWF Y S values by the corresponding dilution factor (see Materials and Methods, Leaf AWF Extraction). Estimated apoplastic Y S was significantly higher than leaf Ys, and differs between leaf stages; Y S AF values were higher in S3 with respect to S2, thus evidencing a more diluted AF in senescing leaves ( Figure 2E, right side of the graph).

Leaf Senescence-Related Changes in the Apoplast Volume, Apoplastic Air-Fluid Ratio, and Cell Wall Properties
To explore potential changes in the extracellular space volume along leaf senescence the apoplastic air (V air ) and AF (V AF ) volumes were determined in S2 and S3 leaves ( Table 1).
Volume determinations were performed with the Indigo carmine, IC, technique (Husted and Schjoerring, 1995;Lohaus et al., 2001) in multiple independent assays. Determined V air and V AF values were similar to those reported for other species, in the range between 40 and 200 µL per gr of leaf FW (Dietz, 1997;Lohaus et al., 2001;Nouchi et al., 2012). The V air remained stable between S2 and S3 stages, whereas V air was shown to increase with age in Brassica napus and Vicia faba (Husted and Schjoerring, 1995;Lohaus et al., 2001). When expressed on a leaf fresh weight, LFW, basis there was a slight decrease of 4.6% V air in S3 compared to S2 (303.50 ± 4.17 mL g −1 LFW in S2 and 289.60 ± 2.05 mL g −1 LFW in S3) whereas when the V air was expressed per unit leaf area there was no significant difference between S2 and S3. On the other hand, V AF increased significantly as leaves aged. S3 leaves contained 157% more AF than S2 leaves (21.40 ± 1.57 mL g −1 LFW in S2 and 55.01 ± 1.95 mL g −1 LFW in S3). When V AF was expressed per unit leaf area S3 leaves showed a 173% increase compared to S2 (0.49 ± 0.04 mL cm −2 and 1.34 ± 0.04 mL cm −2 in S2 and S3, respectively). The higher fresh weight per unit area in S3 with respect to S2 leaves ( Figure 2B) might explain the differences observed when apoplast volumes (V AF or V air ) are expressed on a leaf area or on a LFW basis. The increased V AF in S3 leaves is (E) Leaf water potential (Y w ) measured on leaf disks, solute potential of freeze-thaw squeezed leaves (Y s ) and AF solute potential (Y S AF ). Data represents the average of biological samples (n). (A) RWC: n = 6 (S2) and n = 6 (S3), (B) LFWA: n = 11 (S2) and n = 13 (S3), (C) LWC: n = 6 (S2) and n = 6 (S3), (D) starch: n = 5 (S2) and n = 6 (S3), (E) Y w : n = 13 (S2) and n = 13 (S3), Y s : n = 4 (S2) and n = 6 (S3), Y S AF : n = 4 (S2) and n = 4 (S3). Values were normally distributed and significant differences were calculated between S2 and S3 at P ≤ 0.05 using the Student's t test: *, P < 0.05; **, P < 0.01; ***, P < 0.001. Values in the graphs show the average ± SEM. consistent with their higher Y S AF ( Figure 2G). An age-related increase in V AF was also observed in leaves of B. napus and V. faba (Husted and Schjoerring, 1995;Lohaus et al., 2001), whereas V AF was reported to decrease during rice leaf senescence (Nouchi et al., 2012). An increase in V AF has also been observed in plants under different types of stress (Rascio et al., 1992;Tetlow and Farrar, 1993;Lohaus et al., 2001;Hamouda et al., 2016).
It has been proposed that cell wall mass and structure, particularly the type and content of pectin polymers that affect the water retention potential of the cell walls (Levesque-Tremblay et al., 2015) influence the V AF . (Boyer, 1967;Tyree, 1976). We surmised that the steep increase in V AF in S3 leaves would be due to differences in the cell wall properties between leaf stages. To test this hypothesis, the amount of cell wall and pectins, quantified as AIR without starch, and uronic acid concentration, respectively, as well as cell wall thickness were compared between S2 and S3 leaves. Cell wall and pectin content did not differ significantly between S2 and S3, even though there was a slight increase in the pectin content in S3 leaves ( Table 2). It has been demonstrated that cellwall components continue to be deposited into the wall even after the cell no longer expands leading to an increase in cell wall thickness (Wei and Lintilhac, 2007), however the cell wall content is similar in S2 and S3. Interestingly, senescing leaves showed slightly thicker spongy mesophyll cell walls compared to those of non-senescing leaves (247.9 ± 3.0 and 278.8 ± 2.9 nm in S2 and S3, respectively) ( Figure 3). Water retained inside the wall might lead to increased thickness (Forouzesh et al., 2013). Also, modifications such as methylesterifications and ion cross-links directly affect the water binding capacity of pectins, and these changes occur in response to abiotic stress and along leaf development (Levesque-Tremblay et al., 2015). Forouzesh et al. (2013) showed that cell wall thickness increases with plant age, and the oldest leaves have thinner walls compared to the newer ones. The results presented here can be reconciled with those of Forouzesh et al. (2013) since 1.34 ± 0.04 **** 55.01 ± 1.95 **** 7.13 ± 0.11 289.60 ± 2.05 *** 6.54 ± 0.22 **** Apoplastic volumes were expressed in mL per gram of leaf fresh weight (LFW) and in mL per unit leaf area. Obtained V air and V AF values were used to estimate AWF dilution factor (F dil , see equation in Materials and Methods). Data were normally distributed and significant differences were calculated at P ≤ 0.05 between leaf stages using the Student's t test.
Each value represents the mean of 35 to 55 biological samples, n, where each sample was made up of at least four leaves. Asterisks indicate significant differences: ***, P < 0.001; ****, P < 0.0001.  these authors examined cell wall thickness in young and older leaves in the plant, whereas the present study compares nonsenescing and senescing leaves that represent the same cohort at two different times of plant development.

Apoplastic Alkalinization Accompanies Developmental Leaf Senescence
Transient oscillations and long term changes in apoplastic pH (pH Apo ) trigger and/or modulate a myriad of signaling and metabolic pathways (Felle, 2001). Possible fluctuations in the pH Apo between S2 and S3 leaves were evaluated by in vivo imaging using the genetically encoded pH sensor Apo-pHusion. Apo-pHusion plants stably express the mRFP1-EGFP tandem targeted to the apoplast (Gjetting et al., 2012). Confocal images were acquired at the center of the leaf blade, from epidermal cells, where the fluorescence signal is strong enough to be quantified ( Figure 4). The pH for each region of interest (ROI) was estimated based on the ratio between EGFP and mRFP1 fluorescence intensities, according to a calibration curve ( Figures 4C, D). Estimated pH Apo shifted around 0.8 units between S2 and S3, from pH 5.43 ± 0.03 in S2 leaves to 6.26 ± 0.05 in S3 leaves ( Figures 4A, B). These pH Apo values are in the range of pH 5 to 6 expected for leaf pH Apo (Felle, 2001), although absolute pH values may be difficult to estimate as they might be influenced by different factors, i.e. calibration. A pH Apo shift of around 1 unit has been reported associated to electrical waves related to long distance signaling triggered by stress (Felle and Zimmermann, 2007). Different types of stress lead to a rise in pH Apo and in extracellular ABA level, a promoter of leaf senescence (Zhao et al., 2016;Geilfus, 2017). It is controversial whether apoplast alkalinization acts as an upstream signal triggering stress responses, or it is a consequence or secondary effect of the stress response (Geilfus, 2017), this controversy could be extended to the senescence-related scenario. Changes in Apo pH affect the activity of extracellular enzymes and transporters, modulating different metabolic pathways (Hothorn et al., 2010). The NH 3 released to the extracellular space during developmental and dark-induced senescence converts into NH 4 + , that alkalinizes the apoplast (Mattsson and Schjoerring, 2003). NH 4 + accumulation and pH Apo alkalinization rates correlate with the progress of senescence in tobacco (Wu et al., 2016). A rise in pH Apo leads to the ionization of wall molecules, i.e. pectin carboxyl groups, therefore more molecules of water are able to interact with these moieties, which could lead to an increase of cell wall thickness and V AF.

Protein Profile of Apoplastic Fluids. Pathogenesis Related Proteins, PR, Accumulate in the Apoplast of Senescing Leaves
AWFs along with the corresponding whole leaf extracts from S2 and S3 leaves were examined in SDS-PAGE gels. In the range of 14 to 100 kDa, the AF protein patterns from S2 and S3 leaves (AFS2 and AFS3) displayed clear quantitative and qualitative differences ( Figure 5). The AFS2 profile shows less protein per unit leaf area than the AFS3, and it is composed of several uniformly distributed protein bands. Few bands from 45 to 97 kDa are distinguished in AFS2 but not in AFS3 ( Figure 5A, small arrowheads). The AFS3 protein profile is characterized by two conspicuous bands, one of apparent molecular mass of 25 kDa and another of 37 kDa, that are not clearly recognizable in AFS2 or in total leaf protein profiles ( Figures 5A, B, big For each biological sample two technical replicates were taken. Significant differences were calculated at P ≤ 0.05 between leaf stages using the Student's t-test: *, P < 0.05; ** , P < 0.01. arrowheads). The densitometric analysis of SDS-PAGE gels measured a higher protein content per unit leaf area in AFS3 than in AFS2 ( Figure 5C) whereas, on the other hand, a colorimetric absorbance assay (Pierce™ BCA Protein Assay Kit), that detects all molecular weight proteins and peptides, reveals a higher protein and peptide content in AFS2 than inAFS3 either on a leaf area or LFW basis ( Figures 5D, E). The discrepancy between methods might indicate a decline in the relative abundance of low molecular weight proteins (<14 kDa) and peptides in AFS3.
To identify the proteins corresponding to the two main bands detected in AFS3 protein profiles ( Figure 5A) equal amounts of AFS2 and AFS3 proteins were loaded on an SDS-PAGE gel ( Figure 5B), and the 25 and 37 kDa bands from S2 and S3 were cut off the gel and eluted for mass spectrometry analysis.
Peptide sequences identified different proteins in each band. According to the total number of peptide sequences for each identified protein (#PSM), the most abundant proteins were Pathogenesis-related protein 5 (PR5, At1g75040) in the 25 kDa band and endo-1,3 Glucan-beta-glucosidase, acidic isoform (PR2, At3g57260) in the 37 kDa band. Other less represented proteins co-migrating with PR2 and PR5 in the SDS-PAGE were also identified (Supplemental Tables S1 and S2). Both PR2 and PR5 contain an N-terminal signal peptide (SP) for secretion, and PR5 was previously detected in the N-glycosylated sub-proteome of Arabidopsis mature stems (Wei et al., 2015). According to sequence homology, PR5 is a thaumatin-like protein (TLP) (Uknes et al., 1992). TLP have antifungal and antifreeze properties and accumulate in response to stress (Velazhahan et al., 1999). PR2 is a cell wall b-1,3-glucanase that hydrolyses the b-1,3 glucosidic bonds of b-1,3-glucans (callose) (Uknes et al., 1992). Besides being found in plant cells, b-1,3-glucans constitute the main structural components of fungal cell walls, hence the antifungal properties of PR2 (Doxey et al., 2007). PR5 and PR2 expression is associated with the hypersensitive response (HR), and systemic acquired resistance (SAR) (Geraats et al., 2007;Mishina and Zeier, 2007), and are widely considered hallmark genes for Salicylic acid (SA)-mediated defense and PCD (Uknes et al., 1992;Van Loon and Van Strien, 1999;Sävenstrand et al., 2004;Seo et al., 2008). However, PR2 and PR5, along with other PR genes are also induced in a SA partially independent manner, for instance, as part of signaling pathways mediated by the protein Di19 in Arabidopsis under drought-stress (Liu et al., 2013) and by the plasma membrane-localized, dark induced senescence associated Receptor-Like Kinase, OsSIK2, in rice .
PR2 and PR5 genes are up regulated during leaf senescence (Winter et al., 2007). PR5 expression is induced by the senescence-age related transcription factor WRKY75 in a SA mediated pathway (Guo et al., 2017). WRKY75 also represses catalase activity, a down regulator of PR2 expression (Jing et al., 2009).
To dissect whether the accumulation pattern of PR2 and PR5 in the apoplast, is associated to the senescence syndrome or agerelated, S2 leaves were induced to senesce in the dark for further AF evaluation. To this end, attached and detached S2 leaves where exposed to dark until their Chl content dropped down to S3 leaf values (Figures 6A, B). Dark-induced senescent S2 leaves attached and detached from the plant show lower AF protein content than that of control S2 leaves, and no accumulation of PR2 and PR5 ( Figure 6C), suggesting that PR2 and PR5 proteins accumulate in the apoplast as part of developmental senescence, but not when senescence is imposed by dark incubation.

Apoplastic Fluid Proteome Dynamics During Developmental Leaf Senescence
For a more comprehensive repertoire of senescence related AF proteins, AWFs of S2 and S3 leaves were compared by shotgun proteomics. As a result, 212 proteins were identified, 183 of them showed a significant change (at least twofold difference between AFS2 and AFS3), and were considered AF senescence-related proteins, AFSP. Around 13% (24 proteins) of AFSP increased in the apoplast fluid of senescent leaves, S3 (Table 3), whereas 87% (159 proteins) of AFSP decreased during senescence (Supplemental Table S3).
Approximately half of the identified AFSP proteins lack of N-terminal signal peptide (SP), as described in extracellular proteomes before (Agrawal et al., 2010). All the 24 proteins that increased in AFS3 contain SP, whereas only 41.5% of the proteins that decreased in AFS3 contain SP ( Figure 7A), s u g ge s ti ng t h a t u n c on ve n ti on a l p r o te in s ec r e ti on mechanisms that locate leaderless secreted proteins to the apoplast might be less active during leaf senescence. Nglycosylations are posttranslational modifications required for proper folding, transport and/or function of secreted proteins , and the extent and type of Nglycan structure varies as leaf ages (Elbers et al., 2001). Potential glycosylation sites are slightly more represented in proteins that increased in AFS3 compared to those that decline during senescence ( Figure 7B).
AFSPs were classified into thirteen functional classes ( Figure  7C), those proteins with several known functions were assigned to more than one functional group. AFS3 proteins relate to the main stress-response group, whereas AFS2 proteins relate mainly to two groups: stress-response and photosynthesis/ carbohydrate metabolism.
Interestingly, all the AFS3 identified proteins have been related to stress responses, in particular to biotic stress, with the exception of ENODL1 like protein At5g53870, that was shown to be promoted by drought and heat (Rizhsky et al., 2004); and only few AFS3 proteins have been previously related to leaf senescence. One of them is the amylase AMY1, that locates to the apoplast where it might degrade cell wall associated carbohydrates during Arabidopsis senescence (Doyle et al., 2007;Kamranfar et al., 2018), AMY1 expression is also induced by ABA, heat, and pathogen-related stresses (Doyle et al., 2007). Arabidopsis amy1 plants show early flowering (Jie et al., 2009). The ribunoclease RNS1 has also been associated to senescence, with a potential role in Pi recycling (Bariola et al., 1994;Bariola et al., 1999). RNS1 is markedly promoted by Pi deficit and RNS1 protein accumulates in the secretome of Pi starved cells (Tran and Plaxton, 2008). RNS1 expression is also associated to wounding induced local and systemic signaling (Lebrasseur et al., 2002), and correlates with the accumulation of specific tRNA fragments (Megel et al., 2019). The role of the other identified ribonuclease, PR4, is less known, its expression is associated with SAR (Mishina and Zeier, 2007), and its induced by ethylene and SA (Potter et al., 1993;Camargo-Ramírez et al., 2018).
The five AFS3 peroxidases have been related to pathogen responses. PRX34 is sharply up regulated during senescence (Winter et al., 2007). PRX34 was identified in the N-glycan secretome of Arabidopsis mature stems (Wei et al., 2015) and is also over accumulated under abiotic stress (Mano et al., 2014). Arabidopsis knock down plants for PRX34 and its homolog PRX33 (prx34/prx33) produce less apoplastic H 2 O 2 as MAMPs (Microbe-associated molecular patterns) compared to wild type plants  and are more sensitive to pathogen attack Mammarella et al., 2015). PRX34 along with PRX52 accumulate in the leaf apoplast upon Verticillium longisporum infection (Floerl et al., 2012). The expression of PRX54, is also upregulated in response to fungal infection (Zhao et al., 2007), whereas AtPRX53 is up regulated in response to nematodes, at their penetration sites, and also in response to wounding and jasmonic acid treatments (Jin et al., 2011). The lipid transfer protein LTP4 accumulates in the AF, along with PR5 and other PR genes, as part of the systemic resistance response induced by Trichoderma sp. and P. syringae in Arabidopsis (Brotman et al., 2012).Two proteases with unknown function accumulate in the AF of S3 leaves. The metallo-protease 3-MMP belongs to the At-MMP family, predicted to be extracellular and related to pathogen responses (Zhao et al., 2017), though at2-mmp plants exhibited early senescence and late flowering (Golldack et al., 2002). The cysteine protease At3g49340 is up regulated in response to butterfly egg deposition, causing localized cell death at the site of oviposition (Little et al., 2007). Extracellular cysteine proteases have been shown to mediate defense responses by processing an apoplast immune signaling peptide (Ziemann et al., 2018). At3g49340 might play a similar role in senescence related signaling pathways.
The Germin like protein GLP6, is the only of the identified AFS3 proteins whose expression is down regulated under stress, either abiotic stress, like salt treatment (He et al., 2005), and biotic stress, for example at the galls formed in Arabidopsis roots upon infection with parasitic nematodes (Fuller et al., 2007).

Proteins That Decrease in the Apoplast of Senescent Leaves
The transcript levels of most of the identified proteins that decrease in AFS3 leaves remains unchanged or is down regulated during senescence, except for the aldolase FBA6 (At2g36460), the acid phosphatase AtPAP26 (At5g34850), and the peroxidase PRX51 (At4g37530), that remain constant or are Four samples (biological replicates) for each leaf stage were analyzed. A confidence level (CL) was assigned to each protein based on the number of biological replicates in which it appears: **, present in two biological replicas; ***, present in three biological replicas and ****, present in the four biological replicas. Rows highlighted in gray: proteins detected only in AF S3. Gene expression patterns associated with senescence (column "gene expression during senescence") were examined using the eFP Browser software (Winter et al., 2007). Subcellular locations and protein functions were determined using the SUBA4 (Hooper et al., 2016) and Gene Ontology (Ashburner et al., 2000;Consortium, 2016) databases. In bold, manually-assigned locations and in normal font, locations that were inferred from electronic annotation (IEA) or predicted. Biotic (B) or abiotic (A) stress related function were assigned according to Gene Ontology, or otherwise experimental data from literature. Presence of SP was determined by the SignalP 4.1 software (Petersen et al., 2011), and presence of potential N-glycosylation sites was determined with the ScanProsite tool (De Castro et al., 2006). ID, identification; SL, subcellular localization; SP, signal peptide; N-g.s, potential N-glycosylation sites; CL, confidence level; M, membrane; PM, plasma membrane; G, Golgi; N, Nucleo; Cyt, cytoplasm; Chl, chloroplast; Mit, mitochondrion; Ap, apoplast; ER, endoplasmic reticulum; PX, peroxisome; V, vacuole; T, tonoplast; EV, extracellular vesicles (Rutter and Innes, 2017) slightly up regulated during leaf senescence (Winter et al., 2007). Based on Gene Ontology (Ashburner et al., 2000) some of the AFSP that decrease in senescing leaves are also stress related, even though no prevalence of biotic stress over abiotic was observed as in ASF3 (Supplemental Table S3). Among ASF2 proteins, the lipid transfer protein LTP6 (At1g55260) was shown to be involved in pathogen elicited responses, and its loss of function increases the susceptibility to penetration of the epidermal cell wall by powdery mildew (Fahlberg et al., 2019). Another LTP (At5g48490), identified only in AFS2, has also been related to SAR responses (Champigny et al., 2013).
The b-glucanase BG3 (At3g57240), the aspartic protease AED (At1g09750, Apoplastic, EDS1-Dependent), and the subtilase SBT 1.7 At5g67360 have also been described as part of pathogen responses, particularly associated to virus infection and to wounding (Golldack et al., 2003;Zavaliev et al., 2013;Breitenbach et al., 2014). The b galactosidase bG60 (At3g13750) was only detected in the AF of S2 leaves; its expression is up regulated by MBF1c (stress-response transcriptional coactivator multiprotein bridging factor) that enhances the tolerance to bacterial infection, heat, and osmotic stress (Suzuki et al., 2005).
Other abiotic stress inducible proteins specifically identified in AFS2 (absent in AFS3), are KIN2 (At5g15970), that responds FIGURE 7 | Classification of AF proteins. Presence of signal peptide, SP, (A) and potential glycosylation sites (B). Numbers inside and outside the bars indicate the number of proteins. (C) Functional classification of AF proteins. Proteins with several known functions were assigned to more than one functional group. Values refer to the percentage of each functional group within the group of proteins that decrease in S3 and within the group of proteins that increase in S3. SP, signal peptide; N-g.s, potential N-glycosylation sites; CW, cell wall; ROS, reactive oxygen species; modif., modification; metab, metabolism; CHO, carbohydrate. strongly to ABA, drought and salinity stresses (Kurkela and Borg-Franck, 1992;Zhang et al., 2018), and the pectin methyl esterase At2g46930, that is up regulated under cold stress (Provart et al., 2003).
Few of the identified AFS3 down regulated proteins have been linked to senescence, however, some of them might play specific roles related to senescence. For instance, the chaperone CPN-60 b1 (At1g55490) was previously detected in the apoplast (Bindschedler et al., 2008), and the lack of its activity leads to growth rate reduction and cell death in Arabidopsis (Ishikawa et al., 2003) and tobacco (Zabaleta et al., 1994). Anexin AnnAt1 (At1g35720) was detected before in the apoplast (Clark et al., 2005;Bindschedler et al., 2008;Ge et al., 2011), and at1g35720 plants show early senescence and hypersensitivity to osmotic stress . Ectopic expression of AnnBj1, the homolog of AnnAt1 in Brassica juncea, in tobacco and cotton leads to delayed senescence and increased H 2 O 2 levels (Jami et al., 2008;Divya et al., 2010).
Some identified enzymes related to ROS (Reactive oxygen species) synthesis and catabolism decrease in AFS3. The superoxide dismutases CSD1 (At1g08830) and CSD2 (At2g28190) are plastidic, but have also been detected in the apoplast (Cheng et al., 2009;Nguyen-Kim et al., 2016). The glyoxylate oxidases GOX1 (At3g14420) and GOX2 (At3g14415) are involved in ROS production. These enzymes lack a SP, but were detected in the apoplast of Arabidopsis exposed to oxidative stress (Bindschedler et al., 2008). Double mutant plants amiRgox1/ 2 display accelerated leaf senescence (Dellero et al., 2016). AtPAP26 expression is mildly up regulated during senescence (Winter et al., 2007), however its protein level decreases in AFS3. AtPAP26 localized to the apoplast and to the central vacuole is involved in vacuolar Pi recycling and extracellular Pi recruitment Tran et al., 2010). Mutant atpap26 plants show severe reduction in P recycling, seeds with low P content, and delayed senescence (Robinson et al., 2012).

CONCLUSIONS
The two most abundant proteins in the apoplast of senescing leaves, PR2 and PR5, along with all the identified apoplastic proteins that increase during senescence, have been previously linked to stress responses.
Leaf senescence-related changes in the secretome, along with variations in the AF volume and pH resemble pathogen attack responses, that in turn involve the induction of many Senescence Associated Genes, SAGs genes (Little et al., 2007;Schippers et al., 2015). PR5 markedly accumulates in the apoplast in response to a variety of biotic and abiotic stresses (Piofczyk et al., 2015;Lambertucci et al., 2019), suggesting a role in a possible universal stress response, or in a crosstalk point where different stress responses converge. It would be plausible that stress-related and senescence programs trigger common mechanisms under conditions requiring recycling. Biotic and abiotic responses are differentially prioritized in a leaf dependent manner in Arabidopsis (Berens et al., 2019). Leaves of different ages differentially control responses to stress, allowing for a balance to maintain growth or survival at the organism level (Farber and Mundt, 2017;Berens et al., 2019).
Dark induced-senescence genes are up regulated in Arabidopsis infected with tobacco rattle virus, TRV (Fernandez-Calvino et al., 2016), however PR2 and PR5 showed no accumulation in the apoplast of mature healthy leaves (S2) induced to senesce in darkness ( Figure 6). The specific accumulation of PR2 and PR5 in the apoplast during natural senescence, but not during dark induced senescence suggests that age-related signals, independently of the sinksource status of the leaf, might be priming the apoplast for stress responses in an as yet undamaged leaf. Salicylic Acid, SA, accumulates in an age-related way, and under SA basal levels defense responses are promoted and PR proteins accumulate.
Comparative proteomics of vascular sap reveals a common theme among apoplast fluid, xylem, and phloem sap proteomes collected from stressed plants, that is the accumulation of PR proteins including thaumatin-like proteins, chitinases, and glucanases (Rodríguez-Celma et al., 2016).
The extracellular space might thus represent an up-to-date dismissed scenario where developmental senescence and stress related pathways overlap and integrate.

DATA AVAILABILITY STATEMENT
This article contains previously unpublished data. The name of the repository and accession number(s) are available at https:// figshare.com/projects/FPS_manuscript_476421_Senescence_ related_changes_in_leaf_apoplast/71789.

AUTHOR CONTRIBUTIONS
MB, JG, and DM designed the experiments. MB and MM performed the experiments. DM and JG wrote the article. MB, JG, and DM discussed the data and revised the article. All authors approved the final manuscript.

FUNDING
This work was supported by grant from Agencia nacional de Promoción Científica y Tecnológica, Pres. BID. Pict 1092.

ACKNOWLEDGMENTS
ApoPhusion line was kindly provided by Dr Fuglsang (Department of Plant Biology and Biotechnology, Faculty of Life Sciences, University of Copenhagen). The Dew point Psychrometer C-52 (Wescor ® ) and the Vapor pressure micro Osmometer belong to the Instituto de Investigaciones Fisiológicas y Ecológicas vinculadas a la Agricultura (IFEVA) and Instituto Leloir, respectively. The Vacuum Freeze Dryer be lo ng s to C e ntro d e In ve s tig a ci ón de Prote ín a s Vegetales (CIProVe).