Photosynthetic analysis of mid-vein and leaf lamina in high-yield hybrid rice in elds during senescence

Previous studies on rice (Oryza sativa L.) have shown that different components of the photosynthetic apparatus are not uniformly synthesized or degraded during senescence. However, most of these senescence-related studies focused on leaf lamina, while few have addressed functional aspects on chloroplasts or leaf physiology. Here, we investigated the photosynthetic properties of the mid-vein and leaf lamina in a super high-yield hybrid rice (LYP9) during senescence. We found that assimilation and transpiration decreased more slowly in the mid-vein than in the lamina during senescence, suggesting more sustained photosynthesis in the mid-vein, as well as stronger heat dissipation. Two-dimensional gel electrophoresis revealed that the mid-vein had a higher abundance of proteins involved with energy and lower levels of disease or defense-related proteins, suggesting that photosynthesis and energy metabolism were less affected by senescence in the mid-vein than in leaf lamina. In late senescence stage, the excess energy dissipation in the mid-vein through the xanthophyll cycle had a higher active photosynthetic capacity than in the leaf lamina, and we inferred that the mid-vein and leaf lamina of LYP9 rice aged heterogeneously. Taken together, these results provide new insights into the underlying mechanisms of senescence and associated physiology of the rice mid-vein. stem, root, ower and fruit (Aschan and Pfanz 2003; Dima et al. 2006; Kalachanis and Manetas 2010; Pfanz et al. 2002; Shen et al. 2016). In the mid-vein of tobacco (Nicotiana tabacum), celery (Apium graveolens) and A. thaliana (Brown et al. 2010; Hibberd and Quick 2002), the C 4 photosynthesis pathway involves C 4 acid decarboxylases, whose activity is required for sugar and amino acid metabolism. These studies have revealed aspects of C 3 plants that are potentially involved in the preconditioning of C 4 pathway evolution. Enzymatic activities that are crucial for C 4 photosynthesis have also been identied recently in the mid-vein of rice (Shen et al. 2016; Gao and Shen, 2018), suggesting a potential clue for transgenically providing rice with C 4 photosynthetic pathways. Hence, photosynthesis in the mid-vein could be an important factor for controlling grain yield. C 4 -like photosynthesis pathways uncovered in C 3 plants were based on the characterization of anatomical structures and C 4 related enzymes (Hibberd and Quick, 2002; Hibberd and Covshoff, 2010; Aubry et al., 2011), however, the associated specic photosynthetic machinery of proteome is rarely reported. Here, we selected the high-yield rice cultivar, Liangyoupei9 (LYP9), which has particularly large mid-veins to study the regulation and coordination of senescence in the mid-vein. We presented a comparative analysis of changes in photosynthetic performance of the leaf lamina and mid-vein during senescence. The assimilation and transpiration rates showed slower decreases in the mid-vein compared to the leaf lamina, suggesting that photosynthesis is likely to occur in the mid-vein. By measuring photosynthetic parameters, photosynthetic pigments and protein levels, we were able to determine whether the rice mid-vein has signicant photosynthesis properties during leaf senescence. Heat dissipation and xanthophyll cycle related parameters were also determined, and we found that the heat dissipation performance was stronger in the mid-vein than in the leaf lamina. In addition, compared to the leaf lamina, various energy-related proteins were more abundant in the mid-vein than in the lamina, such as ATP synthase-related enzymes and TCA cycle-related enzymes. Meanwhile, some disease/defense-related proteins were less abundant in the mid-vein, suggesting less cellular damage in this tissue than in the leaf lamina. Taken together, these results suggest that the photosynthetic pathway and energy metabolism were less affected by senescence in the mid-vein than in the lamina. We conclude that the mid-vein and leaf lamina of rice LYP9 age differently. The mid-vein may play an important role during leaf senescence, and our data might provide new insights into the underlying mechanisms of senescence and the physiology of the rice mid-vein.


Introduction
It is generally accepted that carbon isotopic composition of plant material is correlated with C 3 or C 4 pathways of carbon xation in photosynthesis (Sage, 2004). The C 4 plants are anatomically different from C 3 plants and are more e ciently at concentrating carbon dioxide (CO 2 ) around a particular enzyme named RuBisCO, which is crucial for photosynthesis. Comparing with C 4 plants, C 3 plants need more CO 2 because of their high light respiration rate and low photosynthesis rate. Although genes required for C 4 photosynthesis are also existed in C 3 plants, few of them exert related functions in C 3 plants. Many C 3 plants have several genes needed for C 4 photosynthesis, but do not use them in the same way as C 4 plants.
That is why transforming key genes from C 4 plants into C 3 plants is intriguing for improving C 3 plants photosynthesis and water or nitrogen (N) use e ciency.
The staple food crop rice (Oryza sativa L.) is a typical C 3 plant. During rice senescence, rice leaves turn yellow and lose their chlorophyll (Ginsburg et al. 1994). In addition, their chloroplasts undergo ultrastructural changes, resulting in reduced photochemical activity in the leaves and limited photosynthesis e ciency (Harding et al. 1990). Damages to oxygen evolving complex contained photosystem II (PS II) have been reported to occur in many plant species during leaf senescence (Biswal et al. 2012; Deoa and Biswalb 2001;Kusaba et al. 2007; Lu et al. 2001;Clermont., 2004). For example, lower thermoluminescence values (indicative of damage to PSII) were recorded in Arabidopsis thaliana during senescence (Wang et al.2016). Speci cally, activities of the whole electron transport chain and reaction center decline acutely at the onset of senescence (Biswal and Prasanna 1978;Prakash et al. 1998). This is coincident with a loss in redox homeostasis in the electron transport chain between PSI and PSII, caused by an increase in the quantity of reduced quinones, representing an energy imbalance. This premise is supported by a decline in the actual quantum yield of PSII in the light adapted state and maximum quantum yield of primary photochemistry in the dark-adapted state of chlorophyll uorescence.
In addition, leaf senescence is also accompanied by a decline in oxygen evolution, stomatal conductance, CO 2 xation and an up-regulation of certain enzymes (Mohapatra et al. 2010). Since leaf senescence has a tremendous negative effect on rice, delaying leaf senescence could be a possible practice to elevate the global yields (Grover 1993;Quirino et al. 2000). Humbeck et al. (1996) demonstrated that different components of the photosynthetic apparatus were not synthesized or degraded uniformly during senescence. However, senescence-related studies have generally focused on the leaf lamina, and very few focused on chloroplast function, which can also be found in other free heterotrophic plant parts, such as the mid-vein, stem, root, ower and fruit (Aschan and Pfanz 2003;Dima et al. 2006; Kalachanis and Manetas 2010;Pfanz et al. 2002;Shen et al. 2016). In the mid-vein of tobacco (Nicotiana tabacum), celery (Apium graveolens) and A. thaliana (Brown et al. 2010;Hibberd and Quick 2002), the C 4 photosynthesis pathway involves C 4 acid decarboxylases, whose activity is required for sugar and amino acid metabolism. These studies have revealed aspects of C 3 plants that are potentially involved in the preconditioning of C 4 pathway evolution. Enzymatic activities that are crucial for C 4 photosynthesis have also been identi ed recently in the mid-vein of rice (Shen et al. 2016; Gao and Shen, 2018), suggesting a potential clue for transgenically providing rice with C 4 photosynthetic pathways. Hence, photosynthesis in the mid-vein could be an important factor for controlling grain yield. C 4 -like photosynthesis pathways uncovered in C 3 plants were based on the characterization of anatomical structures and C 4 related enzymes (Hibberd and Quick, 2002;Hibberd and Covshoff, 2010;Aubry et al., 2011), however, the associated speci c photosynthetic machinery of proteome is rarely reported.
Here, we selected the high-yield rice cultivar, Liangyoupei9 (LYP9), which has particularly large mid-veins to study the regulation and coordination of senescence in the mid-vein. We presented a comparative analysis of changes in photosynthetic performance of the leaf lamina and mid-vein during senescence. The assimilation and transpiration rates showed slower decreases in the mid-vein compared to the leaf lamina, suggesting that photosynthesis is likely to occur in the mid-vein. By measuring photosynthetic parameters, photosynthetic pigments and protein levels, we were able to determine whether the rice mid-vein has signi cant photosynthesis properties during leaf senescence. Heat dissipation and xanthophyll cycle related parameters were also determined, and we found that the heat dissipation performance was stronger in the mid-vein than in the leaf lamina.
Protein analysis by two-dimensional gel electrophoresis uncovered a higher abundance of energy-related proteins and lower abundance of disease/defense-related proteins in the mid-vein, which in turn meant that the photosynthetic pathway and energy metabolism of the mid-vein were less affected by senescence than in the leaf lamina. Major photosynthetic activity was observed and processes determined in the mid-vein during senescence, providing new insights into the underlying mechanisms of senescence and physiology in the rice mid-vein.

Plant materials and growth conditions
The LYP9 rice cultivar was cultivated in experimental elds of Nanjing Normal University. Regular management was performed according to Yu et al. (2012). Sampling was performed in the mornings (09:30 − 10:30 a.m.) on clear days at approximately 7-day intervals from September 11 (premature senescence) to October 10 (near grain harvesting time). A leaf was carefully detached from the petiole with ne forceps and for in vitro experiments, the mid-vein was removed from the leaf, and where necessary any contaminating leaf tissue was stripped removed, following procedures described by Brown et al. (2010). Several plant samples were pooled to obtain su cient material, frozen in liquid nitrogen and stored at − 80 °C.

Detection of photosynthetic parameters
Measurements of the assimilation rate (A), transpiration rate (E), internal CO 2 concentration (Ci) and water use e ciency (WUE) of the leaf lamina and mid-vein were carried out in the eld using a portable photosynthesis system (CIRAS-3, PP-Systems Hitchin, UK). The conditions were: ambient CO 2 concentration was 390±10 mmol mol -1 , PAR intensity was at 1, 200 ± 50 μmol m -2 s -1 , ow rate was 300 ml min -1 , leaf temperature was 25±1°C, and relative air humidity was 65-70% (Zhang et al. 2006). In order to accurately measure the photosynthetic rate of the leaf lamina and mid-vein, a 3 mm × 30 mm rectangular area was used to cover other tissues and tted between clips of the CIRAS-3 and the mid-vein or leaf lamina during measurements (Pavlovic et al. 2009). The mid-vein was enclosed in a leaf cuvette and measurements were started in the morning between 09:00 and 10:00 a.m.with 10 repeats for each leaf analyzed, between the period from September 11 (premature senescence) to October 10 (near grain harvesting time).

Measurement of chlorophyll a uorescence
Chlorophyll uorescence parameters in the leaf lamina and mid-vein were estimated simultaneously with a portable uorometer (Handy PEA, Hansatech, UK) as previously described (Strasser et al. 1995). Samples of leaf lamina and mid-vein still attached to the plants were collected from the midsection of the same leaf during the period from September 11 (premature senescence) to October 10 (near grain harvesting time). To make sure that photon exchange between the instrument and the mid-vein or leaf lamina did not interfere with each other when measuring light intensity, a 3 mm × 15 mm rectangular area and a non-uorescing piece of black tape, were used as in Manetas (2004) and Panda et al. (2013). Prior to each measurement, leaf clips for dark adaptation were placed on the leaves for 30 min and the leaves were then illuminated with continuous red LED light (peak at 650 nm) at an excitation irradiance of 3,000 μmol m -2 s -1 with a duration of 800 ms. We repeated and recorded these measurements 10 times for each leaf, and the data analysis was performed using the professional PEA Plus and Biolyzer HP3 software (Hansatech, UK). Parameters are described in Table 1.
The extraction and analysis of xanthophyll by high performance liquid chromatography (HPLC) was performed as previously described (Wright et al. 2011). Under dimmed room lighting, a sample was extracted separately with 85% acetone and 100% acetone, before centrifugation for 4 min at 12,000 g at 4°C. The supernatant was further cleared by passing through a 0.22 μm nylon lter of organic phase (Nylon 66, Jinteng, China). The xanthophyll was eluted using 100% of solution A (acetonitrile: methanol, 85:15 v/v) for the rst 14.5 min followed by a 2 min linear gradient with 100% solution B (methanol: ethyl acetate, 68:32 v/v) for an additional 28 min. The xanthophylls were then separated using a non-endcapped Zorbax Elipse (250 mm × 4.6 mm ID, XPB-C18, 5 μm) analytical column (Agilent 1100, USA). Pigments were detected by measuring absorbance at 445 nm.

Two-dimensional electrophoresis (2-DE) and image analysis
Rice leaves grown for 120 days were collected as material for 2-DE. Protein samples were isolated separately from 1 g of the leaf lamina and mid-vein using a trichloroacetic acid (TCA)-acetone/phenol extraction method (Wang et al. 2006). Samples were extracted with 10 % (w/v) trichloromethane in acetone and centrifuged at 16,000 g for 5 min at 4ºC. The pellet was washed with 0.1M ammonium acetate and 80% acetone, incubated at -20ºC for 1 h, and centrifuged at 16,000 g for 20 min at 4ºC. The pellet was air dried at room temperature. Approximately 0.1 g pellet was added to 0.6 mL of a Tris-saturated phenol solution (pH 8.0) and 0.6 mL of sodium dodecyl sulfate (SDS) buffer (30% sucrose, 2% SDS, 0.1M Tris-HCl, pH 8.0, 5% mercaptoethanol), incubated for 5 min, and centrifuged at 16,000 g for 20 min at 4ºC. The phenol phase was transferred to a new tube containing four to ve times the volume of 0.1M ammonium acetate in 80% (v/v) methanol (100 mL), and the samples were incubated overnight at -20ºC. After centrifugation, the pellet was air dried at room temperature and resuspended in an immobilized aqueous solution (8M Urea, 20% CHAPS, 0.5% IPG buffer, 0.002% bromophenol blue, 10 mL). Three independent samples were extracted as biological replicates.
2-DE and image analysis were performed as described by Carpentier et al. (2005). The rst dimension isoelectric focusing (IEF) was performed with IPG strips (Bio-Rad, USA, pH 4-7, 24 cm) with an Ettan IPGphor 3 system (GE Healthcare, USA). The second-dimension electrophoresis was performed on 24 × 9 × 24 cm SDS-PAGE gels (12.5% acrylamide) without a stacking gel using an Ettan DALT six Large Vertical System (GE Healthcare, USA). A total of six 2-DE gels were loaded with equal amounts of protein (1.25 mg) dissolved in the aqueous solution to a nal volume of 425 mL. The gels were stained by 0.1% (w/v) Coomassie brilliant blue R-250 for 2-3h, then the gel images ( Figure S1) were analyzed using the method described by Carpentier et al. (2005), using the Image-Master 2-D Elite software version 4.01 (Amersham Biosciences). Protein spots showed differences in size that were observed in all the replicates were selected as targets.

In-gel digestion and MALDI-TOF/TOF MS analysis
In-gel tryptic digestion of proteins in the selected spots was performed as in Guha et al. (2013). Samples were analyzed using matrix-assisted laser desorption/ionization (MALDI) time-of-ight (TOF) mass spectrometry (MS) with a proteomics analyzer (4800 Plus, Applied Biosystems, USA), and were internally calibrated using tryptic peptides from auto-digestion. Database searching and PMF (peptide mass ngerprinting) was performed using the in-house Mascot server (http://www.matrix science.com) for matching against to the National Center for Biotechnology non-redundant (NCBI nr) database.
Protein functions were assigned using the protein functional database UniProt (http://www.uniprot.org) and Inter-Pro (http://www.ebi.ac.uk/interpro/) (Apweiler et al. 2002). Proteins identi ed were then categorized according to their assigned biological functions as described by Bevan et al. (1998). The subcellular locations of the unique proteins identi ed in this study were predicted using WolfPsort (http://wolfpsort.org) (Wu et al. 2013).

Statistical analysis
Values are presented as mean ± standard deviation from at least three individual experiments. Data were assessed by independent samples t test analysis of variance using GraphPad Prism 6 and SPSS 17.0 (SPSS Inc., Chicago, IL, USA). Differences between mid-vein and leaf lamina samples were considered signi cant at P<0.05.

Variations in photosynthetic parameters
As shown in Fig. 1, photosynthesis was detected in the mid-vein as well as in the leaf lamina. During senescence, a large decrease trend in the rate of assimilation (A) was observed in the whole blade. The assimilation rate in the lamina decreased by 63% (P < 0.05) on day 35 compared with the day 7, while in the mid-vein the decrease was smaller (52%) (Fig. 1A, P < 0.05). The reduced transpiration rate (E) in the leaf lamina and mid-vein was approximately 13% throughout senescence (Fig. 1B, P < 0.05). The gradual decrease of assimilation and transportation in the mid-vein maybe was due to the stomatal closure induced by senescence. In contrast, as the value of g s slowly dropped, stomatal conductance in the leaf lamina sharply declined (Fig. 1C). This suggests that mid-vein senescence was slower than leaf lamina, which might be due to the decreasing assimilation values.
Overall, the internal CO 2 concentration (Ci) in the mid-vein was higher, and consistently increased until day 28, when it began to decrease, whereas Ci in the leaf lamina started to decrease gradually on day 21 (Fig. 1D). Water use e ciency (WUE) in the mid-vein followed the same trend as the Ci, but slowly declined in the lamina during senescence (Fig. 1E). The variation in vapor pressure de cit (VPD) was opposite that of the transpiration rate ( Fig. 1F).

Changes in chlorophyll uorescence
As shown in the radar plot graphs of the photosynthesis parameters, when averaged, the overall days of measurement were equivalent to each speci c sampling date (Figure 2). In the mid-vein, both the maximum quantum yield for primary photochemistry (TR 0 /ABS) and the potential activity of PSII (F V /F 0 ) showed a substantial decrease after day 14, while the electron transport ux (further than Q A ) per reaction center (RC) at t = 0 (ET 0 /RC) began to decline on day 7. Conversely, the absorption ux per RC (re ecting an average antenna size) (ABS/RC) and the trapped energy ux (leading to Q A reduction) per RC at t = 0 (TR 0 /RC) in the mid-vein showed an incremental variation. In addition, 1/V i , RE 0 /RC, PI total and PI abs were reduced in the mid-vein during senescence, while RE 0 /ET 0 and RE 0 /CS 0 rose from day 7 and day 28, respectively. Thus, the terminal electron acceptors at the PS I electron acceptor side (RE) driven by PSI were also inhibited. DI 0 /ABS, DI 0 /RC, DI 0 /CS 0 and DI 0 /CS m values in the mid-vein all started to increase on day 7, suggesting increased energy dissipation.

Xanthophyll cycle pigment
The photosynthetic pigment pro le in the mid-vein and leaf lamina is shown in Figure 3. Compared to the leaf blade, total chlorophyll (Chl) and carotenoid (Car) levels were approximately 1.23 and 1.71 higher, respectively, in the mid-vein ( Figures 3A and 3B, P<0.05). The Car/Chl ratio continuously increased in the mid-vein, whereas the Car/Chl ratio began to decrease on day 28 in the leaf lamina ( Figure 3C). Hence, green mid-veins were characterized by a higher Car/Chl ratio, mainly caused by the increased pools of VAZ cycle (VAZ=V + A + Z, V: violaxantin, A: antheraxanthin, Z: zeaxanthin) components that whose concentrations were determined based on the total chlorophyll and total carotenoids basis. VAZ in the leaf lamina decreased sharply and more rapidly than in the mid-veins ( Figure 3D). The average VAZ/Car percentage was 33%, yet it was signi cantly higher in the mid-vein ( Figure 3E, P<0.05) and the enhanced xanthophyll cycle pool size was accompanied by higher De-epoxidation state (DEPS, DEPS =Z + 0.5A/VAZ) values ( Figure 3F), indicating that the cycle was more dynamic in green mid-veins than in the equally exposed leaf lamina.
Detection of ATP content, as well as Ca 2+ -ATPase and Mg 2+ -ATPase activities ATP content, and Ca 2+ -ATPase and Mg 2+ -ATPase activities decreased during senescence (Figure 4). The ATP content in the leaf lamina was markedly lower than in the mid-vein from day 14. In addition, Ca 2+ -ATPase and Mg 2+ -ATPase activities in the mid-vein were higher than in the leaf lamina. The decrease in Mg 2+ -ATPase activity in the mid-vein was much smaller during senescence compared to the leaf lamina, indicating a higher Mg 2+ -ATPase sensitivity in the leaf lamina during senescence.
Protein pro les in the mid-vein and leaf lamina Three amino acid metabolism-related proteins were identi ed ( Table 2), including cysteine synthase (spot 37) and glutamine synthases (spot 58 and 104). Spots 58 and 104 were signi cantly smaller while spot 37 was markedly larger in the mid-vein samples compared with lamina. The differential expression between the mid-vein and leaf lamina would result in a differential capacity for primary metabolism and subsequent plant growth. A total of 15 protein spots were associated with energy processes, representing the largest functional category of the differentially abundant proteins, and proteins involved in photosynthetic processes represented the largest category. Of these were 5 spots whose migration points indicated different pIs (isoelectric point) and/or MWs (molecular weight), but were all identi ed as ribulose-1, 5-bisphosphate carboxylase/oxygenase (RuBisCO) large subunits. This variation in spot position might be due to post-translational modi cations, such as glycosylation and phosphorylation, or protein degradation induced by senescence (Li et al. 2014). In the mid-vein compared with lamina, some of these RuBisCO large subunit proteins (spots 123, 125, 128 and 135) showed lower intensity, while another (spot 157) had a higher intensity. RuBisCO activase (spot 92) showed a lower abundance in the mid-vein compared with lamina, while RuBisCO large chain precursor (spot 38) and RuBisCO activase small isoform precursor (spot 70) had a higher intensity. These results indicated that senescence coincides with a major changes in the structure and abundance of RuBisCO in the leaf lamina, which would in turn lead to a comparatively more severe disruption in photosynthesisthan in the mid-vein.
The other two energy-related proteins were annotated as components of the citric acid (TCA) cycle, malate dehydrogenase (spot 50) and dihydrolipoyl dehydrogenase 1 (spot 150). Compared to the leaf lamina, the higher abundance of these proteins in the mid-vein suggests that mid-vein is more dominant during senescence. Five spots were identi ed as three proteins, and were grouped into the protein synthesis and storage category. Two higher intensity spots (spots 108 and 109) were identi ed as the reversible chloroplast translational elongation factor Tu, while another translationrelated protein was a putative mediator of RNA polymerase II transcription subunit 37c (spot 247). In addition, Hsp70, which belongs to a class of functionally related proteins involved in the folding and unfolding of other proteins, showed different abundance in different tissues of leaf blade.
Finally, three spots were categorized as proteins related to disease/defense: an L-ascorbate peroxidase 2 (spot 24), a thioredoxin-like protein CDSP32 (spot 34) and hypothetical protein OsI-29063 (spot 61) ( Table 2). All three proteins were less abundant in the mid-vein, suggesting that the mid-vein suffered less aging stress compared with lamina. We also identi ed some proteins that were annotated as being associated with signaling transduction but with unknown molecular functions ( Table 2).

Discussion
Senescence is known to involve the degradation of various proteins, but the mechanisms responsible for mid-vein protein degradation remain largely unknown. Shen et al. (2016) found that in rice leaves, the mid-veins have chloroplasts exhibiting active photosynthesis during senescence. In this current study, proteome analysis revealed differential expression of photosynthesis-associated proteins in the mid-vein and leaf lamina, suggesting different photosynthetic performances in the two tissues. The abundance of NADH dehydrogenase [ubiquinone] iron-sulfur protein 1 in the mid-vein was lower than in the lamina, and Guéra and Sabater (2002) found that the total amount of the NADH dehydrogenase complex in pericarp tissue of pepper and tomato fruits is also lower in the ripening stage compared to total plastid protein. Other previous studies have suggested that the NADH dehydrogenase complex may be involved in cyclic electron transport through PSI, probably by balancing the redox state of cyclic electron transporters (Casano et al. 2000;Shikanai et al. 1998). In our study, electron transfer related parameters all showed a decreasing trend during senescence in the mid-vein, which was accompanied by a lower abundance of NADH dehydrogenase. Moreover, the physiological parameters TR 0 /ABS, F V /F 0 and ET 0 /RC related to PSII showed decreasing values in the mid-vein in late senescence, suggesting that the excitation energy during transfer between subunits of PSII in the mid-vein might be suppressed. We infer that photosynthesis in the mid-vein was perturbed during senescence.
RuBisCO large subunit proteins and RuBisCO activase were less abundant in the mid-vein than in the lamina. RuBisCO is a key enzyme in the Calvin cycle and is a high-abundance protein in plants, contributing to 50-70% of the total protein content in leaves (Feller et al. 2008). Previous studies showed that a part of the RuBisCO large subunit and RuBisCO activase are present in lower levels in A. thaliana and Trifolium repens (L.) during leaf senescence due to protein degradation (Wilson et al. 2002;Hebeler et al. 2008). Additionally, the change in RuBisCO levels might lead to differences in plasticity of the WUE (Silim et al. 2001). The lower abundance of RuBisCO in the mid-vein compared than in the leaf lamina suggested an overall down-regulation/degradation of the photosynthetic machinery in the mid-vein during senescence. This in turn potentially may lead to a signi cant decrease in the net photosynthetic rates, while WUE increased in the mid-vein during leaf senescence. As Fig. 1 shows, in the mid-vein, the assimilation rate (A) slowly declined, while the WUE increased. Furthermore, reduced water loss and higher e ciency of water use has been shown to be controlled by effective stomatal conductance (Rivelli et al. 2002) and CO 2 enrichment (Chen et al. 1997), and a decrease in g values appears to function as a major determinant of the decrease in carbon assimilation (Petrie et al. 2000;Yokota et al. 2002). Thus, the decline in g s and the competing effects of transpiration and uptake of Ci in the mid-vein suggests that the effects of senescence stress in the mid-vein were lower than in the leaf lamina, while the degree of perturbation of photosynthetic apparatus and reduced carboxylation e ciency in the mid-vein were lower than in the leaf lamina.
Other energy-related proteins involved in glycolysis, the TCA cycle, and glyoxylate showed a higher abundance in the mid-vein. Upregulation of glycolytic enzymes might lead to enhanced respiration and accelerated consumption of sugars that serve as energy reserves (Jespersen et al. 2015). GAPDH enzymes are involved in glycolysis during respiration (Plaxton 1996). Expression of GAPDH in plants leads to decreased levels of reactive oxygen species (ROS) and enhanced tolerance to heat shock-induced cell death (Baek et al. 2008). Thus, the photosynthetic ability seems to be less affected by senescence in the mid-vein. The TCA cycle is composed of many enzymes linking the oxidation product of pyruvate and malate to CO 2 with the generation of NADH for oxidation by the mitochondrial respiratory chain (Fernie et al. 2004). The higher abundance of malate dehydrogenase (MDH) and dihydrolipoyl dehydrogenase1 in the mid-vein suggest that the TCA cycle was minimally affected compared to the lamina by senescence in the mid-vein. In addition, the high abundance of phosphoglycolate phosphatase 1B has been shown to be involved in the evolution of the photorespiratory glycolate mechanism in higher plants (Fischer and Feller 1994). The higher abundance of phosphoglycolate phosphatase 1B in the mid-vein may serve to maintain normal photorespiration and ensure glutathione production during senescence in the mid-vein, thus further protecting it from damage by senescence.
A higher abundance of the atpB gene product was also detected in the mid-vein compared with the lamina, suggesting that an increased ATP supply may meet increased energy demands caused by stress, thereby alleviating cellular stress caused by senescence. Accumulation of the atpB protein may provide a signal to increase ATP synthesis in order to tolerate stress (Sobhanian et al. 2011), which could be bene cial for the rice plants during senescence. In our study, ATP content, as well as Ca 2+ -ATPase and Mg 2+ -ATPase activity in the mid-vein, were higher than in the leaf lamina during late senescence, suggesting a more progressed senescence in the leaf lamina. We also con rmed that photosynthesis in the mid-vein was higher than in the leaf lamina, and the dissipated energy indexes, DI 0 /RC, DI 0 /ABS, DI 0 /CS 0 and DI 0 /CS m were higher. We conclude that the higher stress caused by aging was converted into heat dissipated energy in the mid-vein, which may represent a self-protection strategy to avoid aging stress. Carotenoids also play a role in delaying senescence in leaves (Biswal., 1995), while xanthophylls are important for light harvesting as well as for processing excess excitation pressure through a singlet-and triplet state energy quenching mechanism. We observed a smaller values of fresh weight-based total Chl and Car, but a in the mid-vein than in the leaf lamina (Fig. 3). Demmig-Adams (1996) showed that an increased Car/Chl ratio may re ect a higher need for light capture. The higher total Car/Chl increase in the mid-vein compared with the lamina suggests either an increased need for dissipation of extra excitation energy or an increased requirement for photon capture, which were shown by the increased pools of VAZ cycle components (Choudhury and Behera 2001; Munne-Bosch and Penuelas 2003). The greater VAZ pool size in the mid-vein compared with the lamina combined with a higher DEPS revealed that the cycle is more active under senescence stress in the mid-vein than in the leaf lamina. Similar results have reported in studies of apple peels (Cheng and Ma 2004), where the enhanced pool and functionality of the xanthophyll cycle components were correlated with a higher thermal dissipation of excess excitation energy shown by the fruit (Cheng and Ma 2004). Thus, the xanthophyll cycle was more active in the mid-vein and favored energy dissipation, which effectively relieved the loss due to excessive accumulation of energy.
The proteins related to disease/defense, like L-ascorbate peroxidase 2 and thioredoxin-like protein CDS 32 were less abundant in the mid-vein compared with the lamina, which in turn lowered the oxidation levels and stress damage. Hsp70 and the elongation factor both assist in protein folding and refolding and are distributed ubiquitously in all living organisms (Kato and Sakamoto 2013). The high abundance of Hsp70 prevents aggregation of denatured proteins and helps in refolding non-native proteins (Sherman et al. 2007;Timperio et al. 2008;Onda and Kobori 2014). We conclude that protein degradation and synthesis in the mid-vein was hampered by a high abundance of Hsp70 and elongation factor. Accordingly, during late senescence the photosynthetic capacity was greater and the senescence rate was slower in the mid-vein than in the leaf lamina.

Conclutions
Based on the above results and discussions, we conclude that the cells around the mid-vein vascular bundle may have a relatively complete photosynthetic system that is important in late senescence (Fig. 7). Compared to the leaf lamina, the mid-vein may dissipate more excess energy in the form of heat through the xanthophyll cycle, which may be associated with a longer and more active photosynthetic capacity during the latter part of senescence. In addition, compared to the leaf lamina, various energy-related proteins were more abundant in the mid-vein than in the lamina, such as ATP synthase-related enzymes and TCA cycle-related enzymes. Meanwhile, some disease/defense-related proteins were less abundant in the midvein, suggesting less cellular damage in this tissue than in the leaf lamina. Taken together, these results suggest that the photosynthetic pathway and energy metabolism were less affected by senescence in the mid-vein than in the lamina. We conclude that the mid-vein and leaf lamina of rice LYP9 age differently. The mid-vein may play an important role during leaf senescence, and our data might provide new insights into the underlying mechanisms of senescence and the physiology of the rice mid-vein.       Summary of differences between the leaf lamina and mid-vein during senescence