Senescence-inducible cell wall and intracellular purple acid phosphatases: implications for phosphorus remobilization in Hakea prostrata (Proteaceae) and Arabidopsis thaliana (Brassicaceae)

Summary Targeting of senescence-inducible acid phosphatases and RNases to the cell wall and vacuolar compartments appears to make a crucial contribution to efficient P remobilization networks of senescing tissues of Hakea prostrata and Arabidopsis.


Introduction
Senescence is a tightly controlled developmental process involving the induction of senescence-associated genes that play key roles in protein, lipid, and nucleic acid breakdown and nutrient recycling (Thomas, 2013). Studies of the critical process of senescence-induced macromolecule catabolism together with remobilization of released nutrients to developing seeds and growing tissues have focused primarily on nitrogen rather than phosphorus (P) (Buchanan-Wollaston et al., 2003;Lim et al., 2007;Thomas, 2013). P-deficient plants remobilize P from their older leaves and roots to younger tissues, but the metabolic networks mediating P remobilization during senescence are poorly understood (Buchanan-Wollaston et al., 2003;Robinson et al., 2012a). However, maximizing the effectiveness of P remobilization from senescing tissues will probably make an important contribution to development of crops with enhanced P use efficiency (PUE) since most modern crop varieties exhibit relatively poor P remobilization efficiency during leaf senescence (Veneklaas et al., 2012). Even a small improvement in this capacity would probably lead to an important reduction in the use of non-renewable P-containing fertilizers in agriculture, thus reducing P i run-off and pollution of aquatic ecosystems (Veneklaas et al., 2012). Cellular P occurs in various pools, including nucleic acids, phospholipids, phosphorylated metabolites, and P i . Ribosomal RNA (rRNA) is the largest organic-P (P o ) component of mature leaves, accounting for up to 60% of their total P o content (Raven, 2012;Veneklaas et al., 2012). The DNA content remains relatively constant as leaves senesce, whereas RNA levels steadily decrease owing to the up-regulation of specific RNase isozymes (Lers et al., 1998;Buchanan-Wollaston et al., 2003). Genomic, proteomic, and biochemical approaches using the model plant Arabidopsis thaliana have made an important contribution to the identification and characterization of genes involved in plant senescence and PUE. For example, the S-like RNase AtRNS2 and the purple acid phosphatase (PAP) AtPAP26 are both induced during Arabidopsis senescence or P deprivation (Bariola et al., 1999;Robinson et al., 2012a, b). AtRNS2 has been localized to the cell wall, endoplasmic reticulum, and cell vacuole, where it mediates degradation of rRNA (Bariola et al., 1999;Borderies et al., 2003;Hillwig et al., 2011).
Acid phosphatases (APases, EC 3.1.3.2) catalyse the release of P i from a broad and overlapping spectrum of phosphomonoesters with an acidic pH optimum. PAPs are an important class of plant APases that function in P i production and recycling (Tran et al., 2010a). AtPAP26, one of 29 PAP isozymes encoded by the Arabidopsis genome, is targeted to the cell vacuole where it has a crucial P-scavenging function during leaf senescence or nutritional P deprivation (Hurley et al., 2010;Robinson et al., 2012a, b;Wang et al., 2014). Kinetic studies with purified AtPAP26 demonstrated that this PAP is well suited for P scavenging as it effectively cleaves P i from a broad range of substrates with a high catalytic efficiency (Veljanovski et al., 2006;Tran et al., 2010b). During P deficiency, AtPAP26 is also up-regulated and secreted into the cell wall, where it was hypothesized to scavenge apoplastic P i by hydrolysing P-esters leaked from the cytoplasm (Robinson et al., 2012b). However, the possible involvement of cell wall-localized acid hydrolases such as AtPAP26 or AtRNS2 in nutrient remobilization during senescence is currently unknown. The cell wall is a dynamic structure intimately involved in plant growth and development, abiotic stress response, and interactions with pathogens and symbionts. The complexity and importance of the cell wall are reflected by the large number of genes suspected to play a role in its biogenesis, assembly, and modifications (Albenne et al., 2013). Although cell wall proteins represent only ~10% of the cell wall mass, they comprise hundreds of different proteins that function as cell wall-modifying enzymes, structural proteins, or defence proteins synthesized in response to biotic or abiotic stress (Albenne et al., 2013).
The objectives of the current study were therefore to: (i) investigate the impact of senescence on APase and RNase activities in the cell wall and corresponding intracellular fractions of leaves and proteoid roots of harsh hakea; and (ii) employ wild-type Arabidopsis and an atpap26 loss-of-function mutant to assess the possible involvement of cell wall-targeted AtPAP26 isoforms during senescence-induced P recycling. Parallel up-regulation and dual targeting of PAPs and RNases to the cell wall and intracellular compartments was demonstrated in senescing tissues of both species. This is hypothesized to contribute to efficient P remobilization during senescence by effectively recycling P i from endogenous P o compounds.

Plant material and growth conditions
Non-senescent, senescing, and fully senesced leaves were harvested from mature harsh hakea (Hakea prostrata R.Br.) plants growing in a natural bushland at Shenton Park Field Station (31°S57'00.27'', 115°E47'52.14'', University of Western Australia). Proteoid roots (Fig. 1B) were harvested from hydroponically cultivated harsh hakea plants propagated in a glasshouse at the University of Western Australia as described in Shane et al. (2013). The non-senescing and senescing proteoid roots correspond to stages III and IV (Shane et al., 2013), while fully senesced proteoid roots were ≥25 d old with no detectible respiration. Harvested tissues were immediately frozen in liquid N 2 . Samples were transported on dry ice to Queen's University (Kingston, Canada) and stored at -80 °C.
Wild-type and homozygous atpap26 T-DNA insertion mutant seeds (Hurley et al., 2010) of Arabidopsis thaliana (Col-0 ecotype) were sown in a standard soil mixture (Sunshine Aggregate Plus Mix 1; SunGro, Vancouver, BC, Canada) and stratified for 3 d at 4 °C before being moved to growth chambers (Model MTR30; Conviron, Winnipeg, MB, Canada), where plants were cultivated at 23 °C (16:8 photoperiod at 100 μmol m -2 s -1 photosynthetically active radiation). Plants were fertilized biweekly by subirrigation with 0.25× Hoagland's solution. Senescence in mature fully expanded rosette leaves of 28-day-old plants was induced according to Robinson et al. (2012a). Fully expanded leaves (up to two per plant) of synchronously growing plants were covered with aluminium foil. After 6-9 d, leaves were determined to be senescing when they appeared 90-100% yellow. Brown/shrivelled leaves were classified as being fully senesced. Harvested leaves were immediately frozen in liquid N 2 and stored at -80 °C.

Preparation of cell wall extracts optimized for Arabidopsis
Leaves were powdered under liquid N 2 and homogenized (1:15, w/v) using a mortar and pestle in ice-cold homogenizing buffer [25 mM HEPES-KOH (pH 7.4) containing 10 mM MgCl 2 , 1 mM EDTA, 1 mM dithiothreitol, 1% (v/v) Triton X-100, and 1% (w/v) PVPP]. Extracts were clarified by centrifugation for 20 min at 14 000 g and 4 °C, and supernatants were collected as the intracellular extracts. Pellets were washed three times by resuspending in homogenizing buffer and centrifuging as above. Extraction of cell wall proteins from the insoluble fractions was performed using a Polytron and 0.2 M CaCl 2 in 5 mM acetate-NaOH (pH 4.6), followed by incubation at 24 °C for 15 min. The mixtures were centrifuged as above, and supernatants were collected as the cell wall extract (Robinson et al., 2012b). Pellets were re-extracted twice with the same buffer and extracts were pooled. Intracellular and cell wall extracts were filtered through Miracloth before concentration as above to at least 2 mg ml -1 . Cell wall extracts were dialysed overnight against 500 ml of 40 mM HEPES-KOH (pH 7.4) containing 10 mM MgCl 2 , 1 mM EDTA, 1 mM dithiothreitol, and 1% (v/v) Triton X-100.

Enzyme and protein assays
All enzyme assays were linear with respect to time and concentration of enzyme assayed. APase activity was determined by coupling the hydrolysis of phosphoenolpyruvate (PEP) to pyruvate to the lactate dehydrogenase reaction at 25 °C and continuously monitoring NADH oxidation at 340 nm using a Molecular Devices Spectramax Plus microplate spectrophotometer (Sunnyvale, CA, USA). The optimized assay mix contained 50 mM sodium acetate (pH 5.0), 2.5 mM PEP, 5 mM MgCl 2 , 0.2 mM NADH, and 3 U ml -1 of desalted rabbit muscle lactate dehydrogenase in a final volume of 0.2 ml. Assays were corrected for any background NADH oxidation by omitting PEP from the reaction mixture. Substrate selectivity studies were Leaves and proteoid roots of harsh hakea have very different lifespans but similar P remobilization efficiencies. Non-senescing (NS), senescing (S), and fully senesced (FS) leaves and proteoid roots were harvested as described in the Material and methods. (A) Senescing leaves were 4-5 years old and 50-75% yellow compared with 1-2 year old non-senescing, mature green leaves. Scale bar=0.75 cm. (B) Senescing proteoid roots were 18-22 d old and greyish-brown compared with mature 11-15-day-old non-senescing mature, bright white, proteoid roots. Scale bar=1 cm. (C) P i and P o were determined for non-senescing and senescing tissues, whereas total P was determined for fully senesced tissues. Values above the bars indicate the percentage of P that was remobilized. The asterisks denote statistically significant differences (P<0.05). All values represent the mean (±SE) of triplicate determinations of n=4 biological replicates. performed by quantifying the P i released by the APase reaction (Tran et al., 2010b). Controls were run for background amounts of P i present at each substrate concentration tested. To calculate activities, a standard curve over the range 1-133 nmol P i was constructed for each set of assays. One unit of APase activity was defined as the amount of enzyme resulting in the hydrolysis of 1 μmol of substrate min -1 at 25 °C (Veljanovski et al., 2006). RNase activity was determined using a modification of the procedure of Fennoy et al. (1997) and the following assay mixture: 50 mM imidazole-HCl (pH 7.0) containing 0.4% (w/v) Torula yeast RNA (Sigma-Aldrich, Canada) in a final volume of 1 ml, and transferred to a water bath at 35 °C. Aliquots (200 μl) were removed every 15 min for 1 h and immediately added to 40 μl of a stop solution containing 25% (v/v) perchloric acid and 0.75% (w/v) uranyl acetate. Samples were incubated on ice for 5 min and centrifuged at 17 500 g for 5 min at 4 °C. Supernatants were diluted 40-fold with water and the A 260 determined. One unit of RNase activity was defined as the amount of enzyme resulting in an increased A 260 of 1.0 absorbance units owing to the release of acidsoluble nucleotides. Protein concentrations were determined using a Coomassie Blue G-250 dye binding method (Veljanovski et al., 2006) with bovine γ-globulin as the standard.

Partial APase purification from senescing harsh hakea leaves
Both chromatography steps were carried out at room temperature (24 °C) using an ÄKTA FPLC system (GE Healthcare Bio-Sciences Inc., Baie-D'Urfé, QC, Canada). Quick-frozen senescing leaves (40 g) were ground under liquid N 2 using a mortar and pestle, homogenized in ice-cold buffer A (1:4; w/v) using a Polytron, and centrifuged at 4 °C and 18 000 g for 20 min. The supernatant was brought to 25% (w/v) polyethylene glycol 8000, stirred for 1 h at 4 °C, and centrifuged. The resulting pellet was resuspended in 15 ml of buffer B, brought to 35% (saturation) (NH 4 ) 2 SO 4 , stirred for 30 min at 4 °C, and centrifuged. The pellet was resuspended again in buffer B, brought to 35% (saturation) (NH 4 ) 2 SO 4 , and loaded at 2 ml min -1 on to a column (1 × 1.0 cm) of butyl-Sepharose 4 Fast Flow (GE Healthcare Bio-Sciences Inc.) pre-equilibrated in buffer B. The column was washed until the A 280 decreased to baseline, and APase activity was eluted with a linear gradient (50 ml) of a simultaneously decreasing concentration of buffer B (100-0%) and increasing concentration of buffer C (0-100%). Pooled peak activity fractions were exchanged into buffer D and concentrated to 3 ml using AMICON Ultra-15 ultrafiltration devices (10 kDa cut-off), and applied at 0.5 ml min -1 onto an HR 5/5 Mono-S column (GE Healthcare Bio-Sciences Inc.) pre-equilibrated with buffer D. The column was washed until the A 280 decreased to baseline and APase (designated as HpPAP1) was eluted with a linear gradient (50 ml) of 0-0.5 M KCl in buffer D. Peak activity fractions were pooled and concentrated as above to 0.4 ml. APase activity in the unbound Mono-S flow-through fractions was exchanged to buffer E, concentrated as above to 1.5 ml, and applied at 0.5 ml min -1 onto a Mono-Q GL 5/50 column (GE Healthcare Bio-Sciences Inc.) pre-equilibrated with buffer E. The column was washed until the A 280 decreased to baseline and APase (designated as HpPAP2) was eluted with a linear gradient (50 ml) of 0-1 M KCl in buffer E. Peak activity fractions were pooled and concentrated as above to 0.4 ml. Aliquots (25 μl) of partially purified HpPAP1 and HpPAP2 were frozen in liquid N 2 , and stored at -80 °C.
Protein electrophoresis, immunoblotting, and in-gel acid phosphatase activity staining Non-denaturing PAGE and SDS-PAGE, as well as immunoblotting and chromogenic detection of antigenic polypeptides using an alkaline phosphatase-tagged secondary antibody, were conducted as previously described (Tran et al., 2010b). APase activity was visualized following non-denaturing PAGE by in-gel activity staining. Following electrophoresis, the gel was incubated for 20 min at 25 °C in 40 mM TRIs-HCl (pH 9.0), 2 mM EDTA, and 1% (w/v) casein, and then for 20 min in 100 mM Na-acetate (pH 5.3). APase activity staining bands were revealed by placing gels in 100 mM Na-acetate (pH 5.3) containing 10 mM MgCl 2 , 0.02% (w/v) Fast Garnet GBC salt, and 0.02% (w/v) β-naphthyl-P. All PAGE and immunoblot results were replicated a minimum of three times, with representative results shown in the various figures.
Leaf free P i and total P Free P i in leaves was extracted and quantified as previously described (Delhaize and Randall, 1995). For analysis of total P, leaves were digested in concentrated HNO 3 :HClO 4 (3:1) at 175 °C and the P concentration was determined using the Malachite green method (Motomizu et al., 1983). Organic phosphorus (P o ) was estimated by subtracting P i from total P. Leaf P resorption efficiency was calculated according to % efficiency=[(total [P] mature leaf-total [P] senesced leaf)/total [P] mature leaf]×100).

Statistics
All values are presented as means ±SE. Data were analysed using the Student's t-test, and deemed significant if P<0.05.

Influence of senescence on P remobilization in harsh hakea leaves and proteoid roots
The lifespan of harsh hakea leaves was estimated to be 4-5 years, in contrast to 20-25 d for the proteoid roots (Fig. 1A, B). These results agree with previous studies for a range of Proteaceae from south-western Australia (Witkowski et al., 2001;Denton et al., 2007;Shane et al., 2013), and also highlight a marked contrast in the time interval from senescence initiation to death between leaves (2-3 months) and proteoid roots (5-7 d) (Fig. 1A, B). In a Mediterranean climate and severely P-impoverished environment such as south-western Australia, many native plants have long-lived leaves, which improves their overall PUE. For their proteoid roots, however, living longer than several weeks would elicit a carbon drain without P gains (Shane and Lambers, 2005). As expected, cells with thick walls were prominent in cross-sections of tough leaves of harsh hakea compared with proteoid rootlets (Supplementary Fig. S1A, B available at JXB online). Dry matter content of leaves [0.45 g dry weight (DW) g -1 fresh weight (FW)] was ~10-fold greater than that of proteoid roots (0.05 g DW g -1 FW) ( Supplementary Fig. S1C at JXB online), and is consistent with earlier studies (Shane and Lambers, 2005;Sulpice et al., 2014).
To assess the influence of senescence on P remobilization from harsh hakea leaves and proteoid roots, total P was fractionated into P i and P o . The P remobilization efficiency of leaves and proteoid roots was ≥85% (Fig. 1C), corroborating previous studies (Shane et al., 2004;de Campos et al., 2013). For non-senescing tissues, P o comprised the major fraction (~80%) of total [P] (Fig. 1C). Interestingly, for the senescing tissues, the fraction of [P o ]/[total P] was ~0.4 (Fig. 1C), demonstrating that senescing leaves and proteoid roots had been collected at the approximate mid-point of P remobilization (Fig. 1C) when the breakdown of P o substrates exceeded translocation of P i out of the senescing tissues. At this stage of senescence, ~75% of the original P o had disappeared and free P i had doubled (Fig. 1C). The findings suggest that cell wall thickness, tissue longevity, and rate of senescence exert little influence on P remobilization efficiency between the functionally distinct, short-lived proteoid roots and relatively long-lived (harsh hakea) or short-lived Arabidopsis foliage (Fig. 1C).

Influence of senescence on intracellular and cell wall-targeted acid phosphatases and RNases of harsh hakea
The cell wall extracts were free of contaminating intracellular proteins as indicated by the absence of cytoplasmic marker proteins (aldolase and PEP carboxylase) on immunoblots of cell wall fractions ( Supplementary Fig. S2A at JXB online). Comparison of the intracellular and cell wall fractions on protein-stained SDS gels indicated clear differences in their respective proteomes ( Supplementary Fig. S2B). The 200-300% up-regulation of RNase and APase specific activities in intracellular and cell wall fractions of senescing leaves and proteoid roots ( Fig. 2A, B) was paralleled by significant P remobilization during senescence (Fig. 1C). The specific activities of intracellular and cell wall RNase in harsh hakea were at least 400% greater in proteoid roots than in leaves ( Fig. 2A). It is unknown whether this was due to the potential additional role of proteoid roots in extracellular P i scavenging from the soil's RNA pool and/or to higher relative amounts of RNA in proteoid root cells. Relatively low RNA levels reported for mature non-senescing harsh hakea leaves are considered an adaptive mechanism that increases photosynthetic PUE (Sulpice et al., 2014). RNA levels in proteoid roots have not been determined, but might be expected to be higher, given their much more rapid rate of development and turnover, greater protein concentration, and potentially greater metabolic activity (Shane et al., 2004(Shane et al., , 2013. In contrast, cell wall APase activity in either non-senescing or senescing leaves was significantly greater than that of proteoid roots (Fig. 2B). It is possible that the difference might be made up by APases excreted in to the rhizosphere by the proteoid roots, and additional studies will be required to assess the cell wall proteomes of leaves versus proteoid roots.
The present results suggest a key P remobilization role for intracellular and cell wall-targeted hydrolases during leaf and proteoid root senescence in harsh hakea. This is hypothesized to contribute to the overall high PUE of harsh hakea, and may apply to similar taxa well adapted to P-impoverished natural habitats (Lambers et al., 2011;Shane et al., 2013). This senescence-induced alteration in P metabolism is potentially very important for proteoid roots which represent significant P sinks but have relatively short lifespans (Fig. 1) (Shane et al., 2004).

Senescence-inducible acid phosphatases of harsh hakea are purple acid phosphatases
Enhanced APase activity during senescence of harsh hakea leaves and proteoid roots was correlated with the up-regulation of a pair of APase activity-staining bands following non-denaturing PAGE of clarified extracts; a 120 kDa band was abundant in both intracellular and cell wall proteomes, whereas a 60 kDa band was only detected in the intracellular proteomes (Fig. 2C). Absence of the 60 kDa activity-staining band in the cell wall extracts provides further evidence that  (Tran et al., 2010b) were resolved by non-denaturing PAGE and subjected to in-gel APase activity staining. they were uncontaminated by intracellular proteins. Although extracts from senescing tissues exhibited greater APase activity staining, the major activity-staining bands were similar in extracts from senescing and non-senescing tissue (Fig. 2C). The higher molecular weight APase activity-staining bands co-migrated with AtPAP12 purified from the secretome (cell culture filtrates) of P-deprived Arabidopsis suspension cells. Native AtPAP12 exists as a 120 kDa homodimer composed of identical 60 kDa subunits covalently linked by a disulphide bond (Tran et al., 2010b).
Intracellular APase activity was resolved as two distinct peaks (HpPAP1 and HpPAP2) during fast protein liquid chromatography (FPLC) of a butyl-Sepharose-enriched senescing leaf extract on a Mono-S cation-exchange column. HpPAP2 failed to bind to a Mono-S column, whereas HpPAP1 was bound and eluted following application of a linear KCl gradient. HpPAP1 and HpPAP2 were purified to final specific APases activities of 0.40 U mg -1 and 3.4 U mg -1 , respectively (Supplementary Table S1 at JXB online). These values are far lower than specific activities of 110 U mg -1 and 1710 U mg -1 reported for AtPAP12 and AtPAP26 isozymes purified to homogeneity from culture media of P-deprived Arabidopsis suspension cells (Tran et al., 2010b). SDS-PAGE followed by protein staining confirmed that the final HpPAP1 and HpPAP2 preparations were relatively impure (results not shown). However, non-denaturing PAGE of partially purified HpPAP1 and HpPAP2 yielded single 120 kDa and 60 kDa APase activity-staining bands, respectively, that cross-reacted with anti-(recombinant AtPAP12) immune serum (Fig. 3A). Anti-AtPAP12 immune serum is monospecific to high molecular weight members of the plant PAP family, but not to individual PAP isozymes (Tran et al., 2010b). To determine subunit molecular masses of HpPAP1 and HpPAP2, samples were resolved by SDS-PAGE and subjected to immunoblotting with the anti-AtPAP12 immune serum (Fig. 3B). In each case an ~60 kDa immunoreactive polypeptide was observed, similar to the subunit sizes of high molecular weight plant PAPs (Tran et al., 2010a, b). The combined results indicate that native HpPAP1 exists as a 120 kDa homodimer composed of 60 kDa subunits, whereas native HpPAP2 is a 60 kDa monomer (Fig. 3B).
HpPAP1, but not HpPAP2, was activated (~70%) by Mg 2+ (Supplementary Table S2 at JXB online). There was no effect on the activity of either HpPAP when the reaction mixture (lacking Mg 2+ ) contained 5 mM EDTA. The most potent HpPAP inhibitors were molybdate, vanadate, and fluoride, with some variation in inhibition between the two isoforms (Supplementary Table S1). Similar findings have been reported for other plant PAPs (Bozzo et al., 2002;Veljanovski et al., 2006;Tran et al., 2010a, b). In particular, the APase activity of HpPAP1 and HpPAP2 was insensitive to 10 mM l-tartrate, a diagnostic kinetic feature of all PAPs (Tran et al., 2010a). The relative insensitivity of HpPAP1 and HpPAP2 to feedback inhibition by P i (Supplementary Table S1) was surprising, given that PAPs from P-deprived Arabidopsis (Veljanovski et al., 2006) and tomato (Bozzo et al., 2002) are subject to potent product inhibition by P i . However, such feedback inhibition could be counterproductive during senescence when P-ester hydrolysis needs to continue in the presence of elevated P i levels (Fig. 1C).
The overlapping, but non-identical, substrate selectivities (Table 1), and pH activity profiles of HpPAP1 and HpPAP2 (optimal activities occurring in the range of pH 4.4-5.2, Supplementary Fig. S3 at JXB online) support the hypothesis that senescence-inducible PAPs of harsh hakea function to scavenge P i efficiently from a wide range of phosphomonoesters over a broad pH range. Although neither PAP showed activity with phytic acid, they both readily hydrolysed glycerol-3-P, ATP, ADP, and P-tyrosine as substrates (Table 1). However, some clear differences in their substrate selectivity were noted; for example, HpPAP2 exhibited much greater activity with phenyl-P and PP i relative to HpPAP1.

Influence of senescence on cell wall-targeted acid phosphatase (AtPAP26) and RNase of Arabidopsis
When Arabidopsis is subjected to P deficiency, AtPAP26 is up-regulated and dual targeted to the cell vacuole and cell wall to recycle and scavenge P i from intra-and extracellular P o sources, respectively (Veljanovski et al., 2006;Hurely et al., 2010;Robinson et al., 2012a;Wang et al., 2014). Although  (Veljanovski et al., 2006;Tran et al., 2010b) were resolved by non-denaturing PAGE and subjected to in-gel APase activity staining and/or immunoblotting with anti-AtPAP12 immune serum as indicated. (B) HpPAP1 and HpPAP2 (0.5 μg per lane) and purified AtPAP12 (0.1 μg) were subjected to SDS-PAGE and immunoblotting with anti-AtPAP12.
AtPAP26 also plays a pivotal P remobilization function during dark-induced leaf senescence of P-sufficient Arabidopsis, it was not determined if cell wall-targeted AtPAP26 is also up-regulated during senescence (Robinson et al., 2012b). In the current investigation, cell wall RNase activity in senescing leaves of the wild type (Col-0) or an atpap26 loss-of-function mutant was increased by ~300% relative to non-senescing controls (Fig. 4A). Based upon the findings of Borderies et al. (2003), along with pronounced induction of AtRNS2 in senescing Arabidopsis leaves (Robinson et al., 2012b), it is hypothesized that the cell wall RNase activity measured in the current study was due to AtRNS2.
Senescing Col-0 leaves also exhibited a significant (~200%) increase in cell wall-localized APase activity (Fig. 4B), whereas cell wall APase activity of the atpap26 mutant was very low in non-senescing controls and remained unchanged during senescence (Fig. 4B). When cell wall proteins were resolved by non-denaturing PAGE and subjected to in-gel APase activity staining, a 100 kDa band was observed that showed a pronounced increase in senescing Col-0 leaves, and that co-migrated with the 100 kDa native AtPAP26 homodimer previously purified from P-deprived Arabidopsis suspension cells (Fig. 4C). The 100 kDa APase (AtPAP26) activity staining band was absent in cell wall extracts from atpap26 leaves (Fig. 4C). Immunoblotting confirmed the accumulation of immunoreactive 55 kDa AtPAP26 polypeptides in cell wall extracts of senescing Col-0 leaves, whereas AtPAP26 polypeptides were absent on immunoblots of atpap26 cell wall extracts (Fig. 4D). The Arabidopsis cell wall extracts used in the previous and present studies APase activity was determined with 5 mM of each compound using the spectrophotometric P i assay described in the Materials and methods. Activity is expressed relative to the rate of P i hydrolysis from 5 mM PEP set at 100%. All values represent the means of n=4 separate determinations and are reproducible to within 10% of the mean value.

Substrate
Relative activity  Fig. 4. Influence of leaf senescence on RNase and APase activities, and relative levels of AtPAP26 polypeptides in cell wall extracts of wild-type and atpap26 mutant plants. Arabidopsis leaves were considered to be senesced when they appeared as 90-100% yellow which required 6 d and 9 d of dark treatment for wild-type and atpap26 leaves, respectively (Robinson et al., 2012a). were free of contamination by a cytoplasmic marker protein (PEP carboxylase), and protein-stained SDS gels indicated clear differences between the leaf cell wall and corresponding intracellular proteomes ( Supplementary Fig. S1D at JXB online) (Robinson et al., 2012a).

Concluding remarks
Up-regulation of intracellular and secreted APase and RNase by leaves, roots, and cell suspension cultures is widely recognized as a ubiquitous response that helps P-deprived plants recycle and scavenge their precious P resources (Abel and Glund, 1987;Tran et al., 2010b;Plaxton and Tran, 2011). Conversely, few studies have considered the contribution of these APases to P remobilization during senescence. However, optimizing P remobilization from senescing tissues using selective breeding and/or biotechnological strategies will probably make a significant improvement to the PUE of many crops, thereby triggering an important reduction in the use of polluting and non-renewable P-containing fertilizers in agriculture (Veneklaas et al., 2012). Considerable variations exist in the ability of different plants to remobilize P from senescing tissues. For example, leaves of a commercial soybean cultivar only remobilized ~50% of their total P during senescence, and similar values have been reported for several other crops (Crafts-Brandner, 1992;Veneklass et al., 2012). In contrast, the P remobilization efficiency of senescing harsh hakea (~85% in leaves and proteoid roots) (Fig. 1B) and Arabidopsis (~80% in leaves) (Himelblau and Amasino, 2001) is comparatively high. As summarized in the model presented in Fig. 5, the current study and a previous study (Robinson et al., 2012b) have correlated this remarkable P remobilization ability with the pronounced up-regulation of cell wall and intracellular APases and RNases. It is notable that leaf senescence of two phylogenetically distinct species that both display excellent P remobilization efficiencies was paralleled by the pronounced induction of intracellular and cell wall-targeted APase and RNase activities. This indicates that this phenomenon may be a universal feature of the complex suite of biochemical adaptations evolved by native plant species that typically inhabit P-deficient soils. It has also been also demonstrated that the pair of prominent APase isoforms up-regulated during senescence of harsh hakea leaves or proteoid roots are probable PAPs (Fig. 3B). Likewise, senescing Arabidopsis leaves also induce cell wall-targeted and vacuolar APase activity, which is primarily due to the action of AtPAP26 (Fig. 4) (Robinson et al., 2012b). This further implicates AtPAP26 as the predominant PAP isozyme involved in Arabidopsis PUE. Abolishing AtPAP26 expression in atpap26 knock-out plants led to dramatic (>90%) reductions in cell wall-localized and vacuolar APase activities, resulting in markedly impaired P remobilization during leaf senescence (Fig. 4) (Robinson et al., 2012b).
To the best of the authors' knowledge, the apparent involvement of cell wall-localized PAPs and RNases in remobilizing P during senescence has not been previously suggested. The induction of cell wall-targeted APase and RNase during senescence of harsh hakea and Arabidopsis suggests that these hydrolases function to remobilize extracellular P from leaked P o . Classic studies by Bieleski's group demonstrated that significant levels of cytoplasmic P o can leak during P deficiency, and failure to recapture this P would seriously diminish a plant's PUE (Bieleski and Johnson, 1972). Cell wall-targeted APases up-regulated by P-deficient plants were hypothesized to recycle P i from this leaked P o (Barrett-Lennard et al., 1993;Robinson et al., 2012a). Senescence involves the regulated catabolism of macromolecules, including conversion of membrane lipids to sugars; the progressive disintegration of the plasma membrane would also permit significant levels of P o (e.g. P-esters and RNA oligonucleotides) to escape from the cytoplasm (Fig. 5) (Lim et al., 2007). It will therefore be of considerable interest to assess the extent to which P remobilization efficiency of senescing leaves (and thus overall PUE) might be improved in crop plants overexpressing senescence-inducible, cell wall-targeted PAP and nuclease isozymes such as AtPAP26 and AtRNS2, respectively.
Many other proteins undoubtedly contribute to plant P remobilization efficiency during senescence, particularly plasma membrane P i transporters involved in P i translocation (Buchanan-Wollaston et al., 2003;Lim et al., 2007), as well as proteins that mediate the strength of P sinks, particularly developing seeds. For example, unlike their vegetative tissues, seeds of many south-western Australian Proteaceae such as harsh hakea contain relatively high P concentrations (≥10 mg g -1 dry matter) (Denton et al., 2007;Groom and Lamont, 2010). The current study augments several recent discoveries concerning fascinating metabolic adaptations of harsh hakea for survival in a severely P-impoverished habitat Shane et al., 2013;Sulpice et al., 2014). However, much is still to be learned from these 'champions of low P tolerance', especially if the aim is to apply some of the knowledge gained for agriculture in a world that features a rapidly escalating population coupled with diminishing availability of rock phosphate reserves.

Supplementary data
Supplementary data are available at JXB online. Figure S1. Differences in cell wall thickness and dry matter content of leaves and proteoid roots of harsh hakea. Figure S2. Immunoblot and SDS-PAGE analysis of intracellular and cell wall proteins extracted from harsh hakea or Arabidopsis leaves. Figure S3. Phosphatase activity of partially purified HpPAP1 and HpPAP2 as a function of assay pH. Table S1. Purification of APase from 20 g of senescing harsh hakea leaves. Table S2. Effects of various substances on the activity of partially purified HpPAP1 and HpPAP2 from senescing leaves of harsh hakea.