STAYGREEN-mediated chlorophyll a catabolism is critical for photosystem stability upon heat stress in ryegrass

Chlorophyll (Chl) loss is one of the most visible symptoms of heat-induced leaf senescence, especially for cool-season grass species. Suppression of the Chl a Me-dechelatase gene, SGR (also named as nye1), blocked the degradation of Chl a and resulted in the ‘stay-green’ trait during leaf senescence. However, effect of Chl a catabolism on plant tolerance to long-term moderate heat stress (35-40°C) remains unclear. In this study, we suppressed the expression of Chl a catabolic gene, LpSGR, in both constitutive and inducible manners in perennial ryegrass. Constitutive suppression of LpSGR aggravated heat stress-induced chloroplast structure and photosystem damages, disrupted energy utilization/dissipation during photosynthesis, activated ROS generation with weakened ROS-scavenging enzyme activities. Transcriptome comparison among wildtype (WT) and transgenic RNAi plants under either the optimum or high temperature conditions also emphasized the effect of Chl a catabolism on expression of genes encoding photosynthesis system, ROS-generation and scavenging system, and heat shock transcription factors. Furthermore, making use of a modified ethanol-inducible system, we generated stable transgenic perennial ryegrass to suppress LpSGR in an inducible manner. Without ethanol induction, these transgenic lines exhibited the same growth and heat tolerance traits to WT, while under the induction of ethanol spray, the transgenic lines also showed compromised heat tolerance. Taken together, our data suggest that Chl a catabolism is critical for energy dissipation and electron transfer in photosynthesis, ROS-balancing and chloroplast membrane system stability upon long-term moderate heat stress.

When exposed to prolonged heat stress for 21 days, the growth of SGRi lines were more 115 severely affected than WT plants (Fig. 1A). Under the optimum growth condition, there was no 116 significant difference between SGRi and WT plants for their Chl b content, net photosynthesis 117 rate (Pn), electrolyte leakage (EL), and leaf relative water content (RWC) (Fig 1B-G). When 118 under heat stress conditions, SGRi lines had significantly lower Chl a and b contents, Pn, and 119 RWC, but higher Chl a/b ratio and EL than WT plants (Fig 1B-G).

Knockdown of LpSGR altered chloroplast ultrastructure and LHC protein 133
The chloroplast ultrastructure of SGRi lines was different from WT plants when grown under 134 non-stress or heat stress conditions (Fig. 2). Under the optimum growth temperature, SGRi lines 135 had higher number of grana per chloroplast, smaller grana size, and more tightly stacked grana 136 thylakoids than those of WT ( Fig. 2A-C). Under heat stress, the number and size of 137 plastoglobule (PG) was significantly increased by heat stress treatment in both SGRi lines and 138 WT plants, while the chloroplast and thylakoid membranes were more severely degraded, the 139 grana stacks were thicker, and intergrana thylakoids were fewer in SGRi lines than those of WT 140 plants ( Fig. 2D-F). 141

148
Consistent to the chloroplast ultrastructure observation, immunoblotting analysis also 149 showed that heat stress significantly decreased the abundance levels of photosynthesis proteins 150 which are involved in light harvesting, electron transport, and carbon dioxide assimilation, such 151 as Lhca3, Lhcb1/2/3/5, PsaA, PsbA (D1), PsbD (D2), and RbcL, and such protein abundance 152 decreases were more severe in SGRi than in WT plants (Fig. 3A). The total protein content of 153 leaves of SGRi lines were also significantly lower in leaves of SGRi lines than those in WT 154 plants under heat stress condition (Fig. 3B).  (Fig. 4). 172 Yet, under heat stress, SGRi lines showed significantly lower Fv/Fm, Y(Ⅱ), qP, and ETR, but 173 significantly increased Y(NPQ) than WT plants. In particular, SGRi lines showed significantly 174 higher ETR and Y(Ⅱ), but lower Y(NPQ) under the optimum growth temperature, but the 175 contrary was true for SGRi lines under heat stress, suggesting that knockdown of LpSGR 176 affected photosynthetic electron transport rate and quantum yield of PSⅡ in a temperature-dependent manner. 178  (abbreviated as CS1 for control-SGRi-1, and CW for control-WT), or under heat stress (HS1 221 for heat-SGRi-1, and HW for heat-WT) were compared pair-wisely ( Fig. S2-S4).Venn diagram 222 analysis showing that 972 differentially expressed genes (DEGs) were found in all four pair-223 wise comparisons (Fig. S3). KEGG pathway enrichment analysis showed that the enriched 224 pathways in 'CW vs. HW' were significantly different from those in 'CS1 vs. HS1' (Fig. S4), 225 indicating effects of heat stress and LpSGR knockdown were interacted at the transcriptome 226 level. DEGs in CS1 vs. CW were mainly enriched in pathways, such as "diterpenoid 227 biosynthesis", "plant-pathogen interaction", "indole alkaloid biosynthesis", "homologous 228 recombination", and "zeatin biosynthesis". It was notable that such similar KEGG pathways 229 were also identified in HS1 vs. HW (Fig. S4).  LpSGR as well (table S3). Under non-stress control condition, eight out of nine unigenes coding 259 except one unigen coding HSFB1, was upregulated by knockdown of LpSGR (table S3). 261 qRT-PCR analysis was performed to validate the reliability of the RNA-seq data. As shown 262 in figure S5, the relative expression levels of 12 DEGs measured by qRT-PCR were consistent 263 with RPKM values of transcriptomic data, supporting the validity of the transcriptome data. 264 Compromised heat tolerance of SGRi was due to the suppression of LpSGR but no other 265

long-term side effect on vegetative growth 266
Since constitutive suppression of LpSGR not only led to the cosmetic stay-green trait, reduced 267 tiller number and therefor biomass production, but also compromised heat tolerance. It is 268 arguable whether the compromised heat tolerance was merely a side effect due to the inhibited 269 vegetative growth (tillering) but directly due to the suppression of LpSGR. Therefore, we SGRi lines showed the typical cosmetic stay-green trait, while Eth-SGRi lines showed the same 283 senescence rate to WT if sprayed with water, but turned to be cosmetic stay-green if sprayed 284 with 2% ethanol (Fig. 7B&C). We also proved that this ethanol-inducible system was effective 285 to prevent post-harvest leaf yellowing upon ethanol treatment before harvest in the Eth-SGRi 286 lines (Fig. 7D). Together, these results proved the stringency and efficacy of the ethanol-287 inducible system in perennial ryegrass. 288

297
Since there was no other long-term effect on vegetative growth in Eth-SGRi lines, we 298 further compared their heat tolerance to WT and SGRi plants. As shown in figure 8, under the 299 optimum growth temperature, there was no significant difference among these plants in terms 300 of Chl content, Fv/Fm, Pn, and EL of leaves treated with water or 2% ethanol. Under heat stress, 301 when all plants were sprayed with water, Eth-SGRi lines and WT had similar values of the 302 above-mentioned physiological parameters which were all significantly higher than those of 303 SGRi lines; when sprayed with ethanol, Eth-SGRi lines also showed compromised heat 304 tolerance with significantly lowers values of these physiological parameters than WT (Fig. 8). 305 Together, these results showed that the compromised heat tolerance trait of SGRi and Eth-SGRi 306 lines was solely dependent on the suppression of LpSGR but no other long-term effect on 307 vegetative growth.

LpSGR knockdown transgenic lines had no stay-green phenotype under heat stress 327
The stay-green trait is defined by the retention of Chl during senescence (Hörtensteiner, 2013). 328 Stay-green mutants are further divided into two types, that is functional stay-green and cosmetic 329 stay-green mutants. The functional stay-green mutants show delayed initiation of senescence 330 or slower progression of senescence than their reference plants, while the cosmetic stay-green 331 mutants show normal progression of senescence but remains greenness during leaf senescence. 332 Null mutants in sgr were responsible for Mendel's green pea (Armstead et al., 2007), and were In addition to the side effect on nitrogen remobilization in the sgr mutant, double mutation 344 of SGR1 and SGR2 caused a severe phototoxic injury to maturing seeds in Arabidopsis thaliana 345 and thus reduced seed longevity (Li et al., 2017). In this study, we found that SGRi lines not 346 only lost the stay-green phenotype under heat stress but were more heat susceptible than WT. 347 The rationale behind this observation was likely due to disrupted electron chain transport, 348 quantum yield with energy dissipation, ROS yield in chloroplast and ROS-scavenging systems 349 when under heat stress discussed as follows. exacerbated the degree of non-regulated energy dissipation from the PS system (Fig. 4). In 408 photosynthetic organism, energy absorbed by PSⅡ is normally consumed for driving 409 photosynthetic electron transport for ATP and NADPH synthesis. When the excitation energy 410 transferred to PSⅡ exceeds that can be utilized, the excessive energy can be dissipated through 411 non-photochemical quenching (NPQ) or transferred to O2 to produce ROS (Wilson et al., 2006). The production of O2can be catalyzed by respiratory burst oxidase homologue (Rboh, a 420 subunit of NADPH oxidase) proteins, and O2can be convert to H2O2 dismutated by SOD (Liu 421  Based on these results, we proposed that blocking SGR leads to more efficient quantum 435 yield and ETR, and smaller but more stacked grana/thylakoid at optimum temperature, but heat 436 stress induced excess energy dissipation in both regulated and non-regulated manners. The 437 excessive non-regulated energy dissipation further resulted in burst and accumulation of ROS, 438 causing severe degradation of LHC proteins and the thylakoid system in SGRi plants. with ethanol, these lines showed the same stay-green trait to SGRi lines that is highly desirable 460 for forage purpose. Thus, these results not only proved the stringency and efficacy of the 461 ethanol-inducible system in perennial ryegrass, but also holds a promise for its application in forage grass breeding in the future. 463

Conclusion 464
In summary, Chl a catabolism is essential for plant heat tolerance that blocking this 465 catabolic step by suppressing SGR caused severe heat-induced phototoxic injury with disrupted 466 photosystem and activated ROS generation in perennial ryegrass. As illustrated in figure 9, 467 under optimum temperature, suppressing SGR leads to the typical stay-green phenotype with 468 Chl retention in senescent leaves; while under heat stress, suppressing SGR leads to 469 compromised heat tolerance with more rapid leaf senescence than WT plants. This 470 compromised heat tolerance in SGRi plants can be remedied by using ethanol inducible system 471 to avoid side effect of SGR suppression but to keep the stay-green trait in forage production. 472

Construction of Gateway-compatible and ethanol inducible RNAi vector 480
The ethanol inducible gene expression system consists of the AlcR transcription factor and its target promoter, alcA (Felenbok, 1991). AlcR binds to specific site in the alcA promoter and, in 482 the presence of ethanol, initiates downstream gene's transcription. The modified alcA_V5 483 promoter with higher specificity and activity (Kinkema et al., 2014)  structure. In short, the entry vector pGMKannibal-2×LpSGR and the destination vector of 492 pCAMBIA1300-UbiP-alcR-alcA_V5-ccdB was recombined through LR reaction (Invitrogen,493 Carlsbad, CA, USA) to generate pCAMBIA1300-UbiP-alcR-alcA_V5-2×SGRi (Fig. S1). The 494 generated vector was verified by vector size and enzyme digestion for the presence of 2×SGRi 495 fragment and then electro-transformed into Agrobacterium tumefaciens strain 'AGL1' cells.

Plant materials, growth conditions and treatment 505
The non-transgenic (wildtype, WT) and transgenic lines (SGRi lines and Eth-SGRi lines) of the 506 same variety 'Buena vista' were compared in this study. All plants were grown in a mixture of 507 peat, vermiculite, and perlite (3:3:1 v/v) in plastic pots at photoperiod of 14/10 hr, 25/20℃, 508 day/night except for heat stress, relative humidity of 70%, and light intensity of 750 μmol 509 photos m -2 s -1 . Plants were watered and fertilized weekly with half-strength Hoagland's nutrient 510 solution (Hoagland & Arnon, 1950).
For testing whether knockdown of LpSGR may affect plant phenotypic traits, the single 512 tillers of SGRi, Eth-SGRi and WT and were split from the mother plants and grown under the 513 same condition. After one month of growth, the plant height, width and length of mature leaf 514 (3 rd leaf from the top), tiller number, and shoot fresh weight of each plant were measured. 515 For heat stress treatment, ryegrass plants were vegetative propagated and maintained at a 516 height of 12 cm by weekly mowing. After two-months growth in the growth chamber, plants 517 were either grown under the optimum temperature at 25/20℃ (day/night) or exposed to high 518 temperature at 38/35℃ (day/night) in growth chambers for 21 days. The photoperiod, relative 519 humidity, and light intensity were set at the same values as described above. As for the ethanol 520 treatment, ryegrass plants were treated with water (control) or 2% ethanol one week prior to 521 and once a week during the heat stress treatment. 522 For dark-induced leaf senescence, the excised mature leaves (3 rd leaf from the top) were 523 wrapped in paper towels moistened with 3 mM 2-(N-morpholino) ethanesulfonic (MES) buffer 524 (pH 5.8), with or without 2% (v/v) ethanol, in the dark at 25℃ for 8 days (Zhang et al., 2016b). 525 For the forage post-harvest storage test, the Eth-SGRi, SGRi, and WT plants were pretreated 526 with 2% ethanol or water control for one week prior to harvest, then the harvested shoot were 527 dried for 21 days under natural condition. 528

Measurement of physiological parameters 529
Chl content was determined following dimethyl sulfoxide (DMSO) extraction protocol 530 described by Barnes et al. (1992). Membrane stability was evaluated by measuring electrolyte 531 leakage (EL) following a method described by Blum and Ebercon (1981). In brief, 0.1 g of 532 leaves were immersed in 30 ml distilled deionized water in a 50 ml centrifuge tube with constant 533 shaking for 24 hr at room temperature. The initial conductance reading (Ci) of the incubated 534 solution was taken using a conductivity meter (Thermo Scientific). Then the leaf samples were 535 autoclaved at 121 ℃ for 20 min and shaking for another 24 hr, and then the maximum 536 conductance (Cmax) of the solution was measured. The relative EL was calculated as Ci/Cmax 537 × 100. Net photosynthesis rate (Pn) was measured using Li-COR6400 portable photosynthesis 538 system (LI-COR Inc., Lincoln, NE, USA) according to the protocol described by Burgess and 539 Huang (2014). The leaf relative water content (RWC) was measured by previously described 540 protocol (Zhang et al., 2020). 541 The Chl fluorescence parameters were measured with Chl fluorometer PAM-2500 (Heinz 542 Walz GmbH, Effeltrich, Germany). In brief, leaves were dark adapted for 15 min prior to 543 measurement. The minimum fluorescence (Fo) was measured by applying 1 Hz light pulses. 544 The maximum fluorescence (Fm) was obtained by applying 10 Hz saturating blue pulse. The The methods for quantification of O2 − and H2O2 production, histochemical staining for 556 O2 − and H2O2, and activity analysis of antioxidant enzyme, including SOD, CAT, APX, and 557 GPX, were previously described by Zhang  Mercaptoethanol, pH 6.8). After 20 min incubated on ice, the homogenate was centrifuged at 571 12000 rpm for 10 min at 4 ℃ and the supernatant was saved for protein content and 572 immunoblotting analyses. The protein content in the supernatant was determined as described 573 by Bradford (1976). Ten μl of protein solution was subjected to SDS-PAGE (12% 574 polyacrylamide gel, made by Invitrogen) and the electrophoresed proteins were then transferred 575 onto a polyvinylidene difluoride membrane (Immobilon ® -P SQ membrane, Millipore GmbH, 576 Eschborn, Germany) for immunoblotting analysis. The target proteins were detected with 577 primary antibodies, including Lhca3, Lhcb1, Lhcb2, Lhcb3, Lhcb5, PsaA, PsbA (D1), PsbD 578 (D2), and RbcL, from Agrisera (http://www.agrisera.com/) and secondary HRP-conjugated 579 goat anti-rabbit IgG antibody form Invitrogen (Shanghai, China). Target proteins in PVDF 580 membrane were reacted with SuperSignal™ West Femto Maximum sensitivity substrate 581 (Thermo Scientific) and then were visualized using the Fusion Solo chemiluminescence system 582 (VILBER LOURMAT, France). 583

Transcriptomic analysis 584
Leaves of SGRi-1 and WT plants grown under the optimum temperature or 21 days of high 585 temperature and treatment were harvested and quickly frozen in liquid nitrogen. Leaves from 586 each pot of plants were regarded as one biological replicate, and three biological replicates were 587 performed for RNA-seq analysis in each treatment. Total RNA isolation, cDNA library 588 construction, Illumina sequencing, data processing, and differentially expressed gene (DEGs) 589 analysis (i.e. GO and KEGG analysis) were carried out by a commercial gene sequencing 590 company (Gene Denovo Corporation, Guangzhou, China) and were the same as described in 591 our previous study (Xu et al., 2019). The sequencing data are available from the NCBI 592 bioproject (accession No. PRJNA756535). 593

qRT-PCR analysis 594
The total RNA was isolated using Plant RNA Kit (Omega Bio-tek, Georgia, USA). After DNA 595 digestion with Perfect Real Time gDNA Eraser Kit (TaKaRa, Otsu, Japan), first strand cDNA 596 was synthesized using PrimeScript TM RT reagent Kit (TaKaRa). The qRT-PCR reaction was 597 performed in a 20 μl reaction volume using SYBR green master mix (Thermo Scientific) on a 598 Roche LightCycler ® 480 II Real-Time PCR machine. The qRT-PCR analysis was performed 599 with four biological replicates and relative expression levels of target genes were calculated using the 2 −ΔΔCT method with LpeIHF4A as the reference gene . Detailed 601 information of primers used in this study are listed in Table S1. 602

Statistical analysis 603
Data from all samples were statistically analyzed using SPSS software (Version 12, SPSS Inc.,