GIGANTEA gene expression influence leaf senescence in Populus in two different ways

GIGANTEA (GI) genes have a central role in plant development and influence several processes such as light signaling, circadian rhythm and abiotic stress tolerance. Hybrid aspen T89 (Populus tremula x tremuloides) trees with low GI expression through RNAi have a severely compromised growth. In order to study the effect of reduced GI expression on leaf traits with special emphasis on leaf senescence, we grafted GI-RNAi scions onto wild type (WT) rootstocks and managed to restore scions’ growth. The RNAi line had distorted leaf shape and reduced photosynthesis, probably caused by modulating phloem or stomatal function, increased starch accumulation, higher carbon-to-nitrogen (C/N) ratio and a reduced capacity to withstand moderate light stress. GI-RNAi also induced senescence under long day (LD) and moderate light conditions. Furthermore, the GI-RNAi lines were affected in their capacity to respond to “autumn environmental cues” inducing senescence, a type of leaf senescence with characteristics different from senescence induced directly by stress under LD conditions. Whereas Overexpression of GI delayed senescence. The two different effects on leaf senescence were not affected by the expression of FT (Flowering locus T), were “local” – they followed the genotype of the branch independent on the position in the tree – and trees with modified gene expression grown in the field were affected in a similar way as under controlled conditions. Taken together, GI plays a central role to sense the environmental changes during autumn and determine the appropriate timing for leaf senescence in Populus. One sentence summary Leaf senescence is a complex process that is not well understood, but this paper shows that changing the expression of one gene could influence leaf senescence in Populus trees in two separate ways.

8 (compare Fig. 1C and Fig. 2B; C). This prompted us to perform a detailed analysis of the 183 photosynthetic performance of leaves of WT and GI-RNAi grafted plants. Trees were kept 184 under LD 18h conditions and net CO 2 assimilation rate (A n ), stomatal conductance (g s ), and the 185 internal CO 2 of the leaf (Ci) were measured weekly after the grafting using a LI-COR 186 instrument. The same set of leavesthe first established after grafting -was measured 187 throughout the experiment. Three weeks after grafting, WT and GI-RNAi leaves had 188 relatively similar A n (Fig. 3A). However, A n decreased rapidly with time in GI-RNAi and 189 after six weeks was only ca 20 % of WT (Fig. 3A). g s was already lower in GI-RNAi three 190 weeks after grafting and decreased even further with time ( Fig. 3B) but Ci did not differ 191 much either between the lines or over time (Fig. 3C). To further study the relationship 192 between GI expression level and these photosynthetic parameters (A n , g s , and Ci), we also 193 analysed a second GI-RNAi line (1-1a, grafted onto WT), where GI mRNA levels are only 194 moderately decreased (Fig. S2), and a lines overexpressing GI (GI-ox, Ding et al. 2018). Line 195 1-1a showed little or no growth phenotype, no abnormal leaf shapes, and no signs of 196 photooxidative stress under the conditions employed here, but they formed buds after two 197 months already under LD 18h conditions. These lines did not differ from WT in gas exchange 198 levels than WT leaves (both below and above the GI-RNAi scion); this was evident both in 217 growing and mature leaves ( Fig. 4A; Fig. S4) particularly in exposed conditions; shaded 218 leaves contained as expected less starch and the difference between GI-RNAi and WT was 219 also less pronounced. The starch accumulation was also reflected in an increased fresh weight 220 per area of the leaves by 7% (Fig. 4B) and their C/N ratio that was increased in exposed and 221 shaded leaves but more pronounced in the exposed leaves of GI-RNAi leaves (Fig. 4C). It 222 should also be pointed out that the starch accumulation of GI-RNAi leaves was less uniform 223 than in WT, little starch accumulated along main and secondary veins, consistent with a role 224 of GI in the function of stomata and/or veins (Fig. 4A). Taken together, GI-RNAi (line 8-2) 225 leaves had lowered CO 2 assimilation rate but higher C/N ratio and starch accumulation in 226 exposed leaves also when growing between two sections of WT stems. Therefore, the effect 227 is an intrinsic property of the GI-RNAi leaves, not of the sink/source activities. Furthermore, 228 starch accumulation did not explain the lowering A n and g s in shaded leaves, indicating that 229 stomatal, rather than phloem loading, malfunction was more important in explaining the 230 lower A n . 231 The chlorophyll and starch staining data indicated that there were spatial differences 232 in chlorophyll fluorescence properties within GI-RNAi leaves. Analysing chlorophyll 233 fluorescence using SPEEDZEN imaging system is a powerful way that can give information 234 about different photosynthetic properties with spatial resolution, therefore we studied leaves 235 of WT and GI-RNAi (line 8-2) exposed and shaded leaves with the age of two and seven 236 weeks after flushing (Fig. 5) using imaging SPEEDZEN. In general, Fm in was, overall, 237 lower in GI-RNAi leaves ( Fig 5A) indicating that PSII was inhibited or quenched. Fv/Fm 238 ( Fig. 5B) tended also to be lower, indicating that PSII activity was somehow reduced as a 239 consequence of quenching or photoinhibition. Some differences could also be noted in the 240 amount of the fast (qE, Fig 5C) or slow (qZ/qI, Fig 5D)  RNAi and their spatial distribution in the leaf could be caused by a reduced Calvin cycle 243 activity, due to for example lowered activity of phloem transfer cells or stomatal 244 malfunctions, Taken together, our data suggest that lowered GI expression primarily affects 245 the photosynthetic dark reaction and that the effects on the light reaction is indirect. 246 girdled trunks on the same tree, been able to separate different patterns of leaf senescence in 251 aspen. Girdling resulted in the early onset of senescence, overriding the "normal phenological 252 control" of autumn senescence, i.e., that each tree genotype induces senescence at 253 approximately the same date every year (Lihavainen et al. 2020). However, as we saw more 254 similarities such as high C/N ratio and anthocyanin level between the stress-induced 255 senescence in GI-RNAi (line 8-2) leaves growing in LD 18h and the senescence in girdled 256 aspens, rather than in aspens undergoing typical autumn senescence in the field, we set out to 257 induce autumn senescence in a growth chamber by changing light conditions and 258 temperature. We included both grafted trees from their first and second growth cycle and the 259 "graft-on-grafted" trees in this experiment. In addition, we included trees grafted with the 260 weaker GI-RNAi line (1-1a), GI-ox and non-grafted trees of each genotype. To simulate 261 autumn, the trees were moved from LD 18h to SD 14h and after two weeks night temperature 262 was lowered (18/5°C day/night; T 18/5°C ) (the experimental setup is shown in Fig. 2A Under these conditions, GI-RNAi (line 8-2) leaves were always senescent earlier than 265 WT leaves, as was evident in exposed and shaded leaves and independent of the position in 266 the tree (Fig. 6A). GI-RNAi leaves were typically shed two weeks before WT leaves. 267 Furthermore, when ungrafted GI-RNAi line 1-1a trees were also subjected SD 14h T 18/5°C , the 268 leaves also became senescent much earlier than the WT (Fig. 6B), and the phenotype was 269 reproduced when 1-1a was grafted as scion or rootstock with WT ( Fig. 6C and Fig. S6A; B). 270 Obviously, although this line grew and photosynthesized like WT (with no sign of 271 photooxidative stress), senescence induced by SD and lowering temperature was affected, so 272 this senescence traitautumn senescencewas more sensitive to decreased expression of GI 273 than the other type of senescence, which we hereafter will name "premature senescence". On 274 the other hand, GI-ox delayed the senescence by at least six weeks under SD 14h T 18/5°C , both 275 ungrafted and grafted branches displayed delayed senescence phenotype ( We also observed that GI-RNAi (line 8-2) leaves, that senesced earlier than WT, 278 seemed to indirectly affect the senescence behaviour of WT leaves; WT leaves that share the 279 same tree with GI-RNAi leaves in the graft-on-graft plants had higher pre-senescence 280 chlorophyll levels and entered senescence later than control grafted WT trees (Fig. 7). We 281 believe this to be a consequence of improved nutrient status; as GI-RNAi leaves accumulate on the same tree during growth and even more so when the GI-RNAi leaves have started to 284 senescence, and the remobilized mineral nutrients become available for the rest of the plant. 285 During the experiments when senescence was induced by SD and lowered 286 temperature (SD 14-12h T 18/5°C )simulating autumn senescencewe noted differences in the 287 patterns of leaf senescence in GI-RNAi leaves (in the stronger mutant lines; 8-2) compared to 288 when senescence was induced under LD 18h , "premature senescence". "Autumn senescence" 289 of GI-RNAi was rather uniform but started along the main and secondary veins in both 290 exposed and shaded leaves than interveinal regions, while in some cases the veins stayed 291 green similar to intervein than vein's close regions. These regions were associated with the 292 previously mentioned phenotypes such as reduced starch accumulation or photooxidative 293 stress in the exposed leaves ( Fig. 2; Fig. 4A). WT and line 1-1a senescence under "simulated 294 autumn conditions" was more uniform. In LD 18h , "premature senescence" in GI-RNAi line 8-295 2 started in intervein regions mainly in the exposed leaves. Under LD 18h conditions, leaves 296 also produced more anthocyanin and flavonol compounds ( Fig. 8B; C). The differences were 297 obvious not only in the coloration of the leaves when visually compared but also when the 298 chlorophyll fluorescence patterns were compared (Fig. 8D). Taken together, these 299 observations show that a reduced expression of GI in Populus leaves affected two different 300 pathways leading to leaf senescence, one that made the strong RNAi line (8-2) leaves more 301 vulnerable to (photooxidative) stress already in the stage of active growth of the tree and 302 associated with starch accumulation in the intervein regions. The second pathway was 303 induced by shortened photoperiod and lowering the temperature, i.e., resembling conditions 304 that induce the autumnal senescence that we study in the field. We also subjected WT and 305 GI-RNAi (line 8-2) leaves, grafted onto a WT rootstock, to a SPEEDZEN time-course 306 analysis over 15-weeks (LD 18h to SD 14h -to-SD 14h T 18/5°C ). The results (Fig. S7) were 307 consistent with lowering of GI expression both reduced the leaf's ability to withstand 308 photooxidative stress and changed their senescence behaviour when days became shorter and 309 nights colder. 310 311 GI expression influence senescence also under field conditions, whereas FT expression 312 did not 313 In the laboratory, under LD 18h conditions, GI-RNAi trees were able to grow if 314 provided with a mobile signal from a WT rootstock (i.e grafted on WT) but were still affected 315 in the progression of leaf senescence. To see if the same was true under field conditions we 316 set up an outdoor experiment where trees were grown in pots but exposed to natural changes 317 in light and temperature (in Umeå). Several different trees were used (in several replicates), 318 a) GI-RNAi (line 8-2) grafted to WT and b) control grafted WT, both in their second growth 319 cycle. For these trees WT buds of the rootstock were removed, producing trees with only one 320 type of leaf after bud flush to check the senescence phenotype when the tree has only GI-321 RNAi leaves. In addition, we included trees with GI-RNAi and WT scions on the same WT 322 rootstock, resulting in a "Y grafting" (Fig. 1B; Fig. S8B). In these trees, the WT scion in the 323 Y graft grew faster and growth of GI-RNAi branches were supressed already under LD 324 conditions, in contrast to the situation in simple grafted trees. Obviously, the WT scion was 325 able to impose apical dominance over the GI-RNAi scion. After a cycle of 326 dormancy/dormancy break, all trees were flushed in LD 18h T 20/15°C and moved to the outdoor 327 in July (when the photoperiod was ca 20 hours). The GI-RNAi parts of all trees set bud 328 already in July, while the WT in the two groups including the trees that had GI-RNAi scions 329 did not form buds until October. GI-RNAi leaves senescent and were shed earlier than WT 330 leaves ( Fig. S8C; D) and under these conditions senescence also started around the veins of 331 GI-RNAi leaves in September (Fig. S8C). The effect of GI on aspen leaf senescence under 332 field conditions appeared not to be mediated through FT. When FT-RNAi, WT, and FT-ox 333 (overexpressing FT) aspen trees were grown under outdoor conditions, FT-RNAi set bud 334 more than one month before the other lines, but senescence was not affected (Fig. S9A). This 335 was also confirmed by measurements of senescence in two-year old trees of FT-RNAi, WT, 336 and FT-ox in a field experiment in southern Sweden (Fig. S9B), also in this case bud set was 337 affected in FT-RNAi but senescence was not different from WT. 338 Discussion 341

342
We still only have a shallow understanding of leaf senescence. Although it is a strictly 343 regulated developmental program, much effort has been expended to identify genes 344 regulating leaf senescence, how environmental conditions trigger the process, and which 345 hormones are involved. Moreover, there isn't much consensus on which genes, metabolites or 346 pathways that provides the keys to the process. One obvious possibility is that if senescence 347 is the default pathway for leaf development, which has to be prevented from being executed 348 by a set of "blocking factors", those that are removed could vary between conditions and 349 species. Comparing genetically identical plant individuals experiencing different conditions is 350 a useful way to understand complex phenomena, and has also been widely used to understand 351 senescence, although alternative approaches could give additional information. We have, for phloem export also in Populus. Therefore, the early senescence we observed already under 365 LD conditions in GI-RNAi (the stronger line; 8-2) could be analogous to the early senescence 366 that can be induced by disrupted transport of photosynthates due to stem girdling (Lihavainen 367 et al. 2020); lowered GI expression could disrupt transport out of the leaf. Another type of 368 leaf senescence was observed when we exposed the trees to conditions that simulated autumn 369 conditions: GI-RNAi leaves senescent earlier than WT leaves, but here the senescence 370 process resembled more the "typical autumn senescence" that every autumn turns boreal 371 forests colourful in a very consistent fashion. However, low GI expression also influenced the 372 way how the leaf experienced "autumn conditions" that initiate leaf senescence in a manner year in a given genotype and largely unaffected by weather. This type of senescence is quite 377 different on a macroscopic level: leaves senesce in a more uniform way in exposed and 378 shaded leaves, and with less obvious signs of photooxidative damages. We believe that these 379 two modes of leaf senescence have a wide physiological relevance but under most conditions 380 are hard to study since they could occur in parallel, and our experimental manipulations -381 affecting the expression of GI or girdling experimentsjust has made it possible to better 382 distinguish between the two modes. A better distinction between different aspects of leaf 383 senescence would be useful for the community, and the mere fact that we are uncertain how 384 to name them illustrates the lack of a conceptual framework. We decided to use the term 385 "premature senescence" although "stress-induced senescence" could be an alternative name. 386

387
We also use our results to draw other conclusions about autumn senescence. First and 388 foremost, although premature/stress-induced senescence clearly could be local, it has 389 previously not been clear whether autumnal senescence in a mature tree is local or systemic. 390 Early observations that street lights may delay leaf fall (Matzke, 1936) have led to 391 speculations that illumination of one part of the crown of a tree could keep that part green. 392 The fact that this has not been possible to replicate by others (e.g. Sarala et al, 2013) or us 393 (unpublished results) could be interpreted as if there is a systemic "senescence signal" that 394 coordinates autumn senescence within a tree. However, the non-uniform senescence 395 behaviour in our grafted trees showed unequivocally that, at least in Populus, senescence 396 timing was strictly dependent on the genetic background of the branch. Secondly, the effect 397 of GI on autumn-induced senescence is, in contrast to the effect on bud set, not mediated by 398 FT whose expression level did not influence autumnal senescence, not even under field 399 conditions. FT is probably indirectly involved, as growth cessation and/or bud set somehow 400 predisposes the tree for sensing the illusive "autumn signal". This is consistent with our 401 previous findings that the light signal that is triggering induction of autumnal senescence in To conclude, a reduced expression of GI in Populus leaves had dramatic changes in 445 leaf physiology causing changes in for example leaf shape, C/N ratio, photosynthesis, and, 446 most notably, leaf senescence through two different pathways, that appears to be independent 447 of the expression of FT and bud set. The ungrafted and grafted trees were transferred from LD to SD (14/10 h; 20/15°C day/night) 471 ( Fig. 2A). After two weeks, the trees were subjected to cold night conditions (18/5 °C 472 day/night), then after a further two weeks the photoperiod decreased to 12/12 h (day/night). 473 The fertilization was stopped after three weeks in SD condition coinciding with the decrease 474 in growth and the starting of dormancy stage. 475

Second growth cycle 476
The trees that passed the simulation of autumn conditions were subjected to 5 °C temperature 477 and photoperiod (8/16 h; day/night) for two months to break the dormancy. The trees were re-478 potted in ten litres pot and flushed again under LD conditions. A similar experiment setup 479 that mentioned in the previous section was used to study leaf senescence under simulating 480 autumn conditions for these trees. 481 Starch staining and leaf weight measurement iodine solution containing 0.7% KI (W/V) and 0.3% I (W/V) for three minutes. The leaves 485 were rinsed by water and photographed. 486 For weight measurements, several discs were obtained in replicates from ten leaves. Leaf 487 fresh weight and leaf area were measured, then the leaves were dried at 70°C for 48 h to 488 measure the dry weight. 489

Gas exchange and chlorophyll fluorescence measurement 490
The net CO 2 assimilation (A n ), stomatal conductance (g s ), and internal CO 2 concentration 491 (Ci) were measured in the leaves using a portable CO 2 infrared gas analyzer (LI-COR-492 6400XT, LI-COR Environmental, USA), equipped with a chamber that controlled irradiance 493 (1000 µmol photons m −2 s −1 ), temperature (20 °C), CO 2 concentration (400 µmol mol −1 ), and 494 flow rate (250 cm 3 min −1 ). The measurements were tested at zheitgeber (ZT) 3, 6, and 9 h 495 after the light on. The differences between the WT and GI-RNAi leaves were reproducible at 496 the different ZT. Therefore, the data of ZT9 are presented here. 497 For chlorophyll fluorescence imaging of the leaves, we used an SPEEDZEN imaging system, 498 and recorded an induction curve (after 30 mins of dark incubation under ambient O 2 and CO 2 499 conditions) with 2600 µmol actinic light for 3.5 minutes with 6000 µmol saturating pulse at 500 every 30 second intervals to attain maximum NPQ levels without causing artificial damage time PCR (qPCR) was run on a LightCycler® 480 with SYBR Green I Master (Roche). All 519 kits and machines were used according to the manufacturer's instructions. The reaction 520 protocol started with 5 min pre-incubation at 95°C, followed by 50 cycles of amplification 521 consisting of 10 s denaturation at 95°C, 15 s annealing at 60°C and 20 s elongation at 72°C. 522 For the acquisition of a melting curve fluorescence was measured during the step-wise 523 increase in temperature from 65°C to 97°C. Relative expression levels were obtained using 524 For GI, two groups of trees were used in this experiment. The first group is the trees that were 543 simply grafted on WT rootstock as described in Fig. 1B. The second group was grafted using 544 Y method grafting as described in Fig.1B. The trees passed the first growth cycle and after 545 the dormancy break the WT buds were removed from the first group of trees. All trees were 546 repotted in 10 litre pots and flushed in the greenhouse in June 2020. On the beginning of July, 547 the trees were moved out of greenhouse (Umeå, Sweden) to follow the autumnal senescence, 548 exposed to natural conditions such as light and temperature. 549 Field experiment potted in the greenhouse. The trees were the potted in at field site at Våxtorp, southern 553 Sweden, in 2014. Height and chlorophyll content were measured for two year old trees. 554

Statistical analyses 555
The multiple ways analysis of variance (ANOVA) analysis was performed using the Info-556 Stat/Student program (http://www.infostat.com.ar/index.php?mod=page&id=37) and the 557 Fisher's Least Significant Difference (LSD) were calculated for statistical analyses. As well 558 as t-tests were used for two-group statistical analyses. A difference at P <0.05 was considered 559 as significant.