Elevated temperatures impose transcriptional constraints on coffee genotypes and elicit intraspecific differences in thermoregulation

The projected impact of global warming on coffee production may require the heat-adapted genotypes in the next decades. To identify thermotolerance cellular strategies, we compared the effect of elevated temperature on two commercial Coffea arabica L. genotypes exploring leaf physiology, transcriptome and carbohydrate/protein composition. Growth temperatures were 23/19°C (day/night), as optimal condition (OpT), and 30/26°C (day/night) as a possible warmer scenario (WaT). The cv. Acauã showed lower levels of leaf temperature under both conditions compared to cv. Catuaí, whereas slightly or no differences for other leaf physiological parameters. Therefore, to explore thermoregulatory pathways the leaf transcriptome was examined using RNAseq. Genotypes showed a marked number of differentially-expressed genes (DEGs) under OpT, however DEGs strongly decrease in both at WaT condition indicating a transcriptional constraint. DEGs responsive to WaT revealed shared and genotype-specific genes mostly related to carbohydrate metabolism. Under OpT, leaf starch content was greater in cv. Acauã although the levels of leaf starch, sucrose, and leaf protein decreased in both genotypes as WaT was imposed. These findings indicate that genotypes with a greater capacity to maintain carbohydrate homeostasis under temperature fluctuations could be more thermotolerant and which may be useful in breeding for a changing climate. HIGHLIGHT In response to warming, transcriptional differences decrease in coffee genotypes hampering breeding programs. Differences in gene expression and sugar levels confirm intraspecific variation associating thermotolerance to maintenance of energetic homeostasis.

2 (OpT), and 30/26°C (day/night) as a possible warmer scenario (WaT). The cv. Acauã showed lower levels of leaf temperature under both conditions compared to cv. Catuaí, whereas slightly or no differences for other leaf physiological parameters. Therefore, to explore thermoregulatory pathways the leaf transcriptome was examined using RNAseq. Genotypes showed a marked number of differentially-expressed genes (DEGs) under OpT, however DEGs strongly decrease in both at WaT condition indicating a transcriptional constraint. DEGs responsive to WaT revealed shared and genotype-specific genes mostly related to carbohydrate metabolism. Under OpT, leaf starch content was greater in cv. Acauã although the levels of leaf starch, sucrose, and leaf protein decreased in both genotypes as WaT was imposed. These findings indicate that genotypes with a greater capacity to maintain carbohydrate homeostasis under temperature fluctuations could be more thermotolerant and which may be useful in breeding for a changing climate.

INTRODUCTION
1 Climate change is multifaceted and despite measurable impacts of elevated 2 temperatures on agriculture (Lobell et al., 2011;Zhao et al. 2017), there remains 3 considerable gaps on how coffee systems will be affected by both short-and long-term 4 changes in the environment. Several studies on the impact of climate change on coffee 5 systems have projected marked negative effects on yield, berry quality, suitable 6 planting areas, and incidence of disease and insects. Collectively, these environmental 7 stresses will likely impose both economic and social problems within many coffee 8 producing regions (reviewed by DaMatta et al., 2019). Despite attenuating factors 9 associated with increasing global CO2 levels that could partially mitigate the negative 10 production trends described above  and numerous studies 11 demonstrating the impact of temperature on coffee physiology (Drinnan and Menzel, 12 1995;DaMatta and Ramalho, 2006;Läderach et al., 2017), a detailed understanding of 13 the molecular thermoregulatory mechanisms is lacking. 14 C. arabica L. is a tropical tree responsible for the major worldwide production 15 of coffee (ICO, 2019) and its optimal growth temperature is considered between 18 to 16 23°C (Camargo, 1985;Teketay, 1999). The coffee tree has a periodicity growth habit 17 that closely follows rainfall patterns and, historically, it is considered highly sensitive 18 to climatic changes, especially temperature and drought (DaMatta and Ramalho, 2006; 19 Camargo, 2010;DaMatta, 2018). Mean temperatures are projected to increase by 2.6 -20 4.8 °C (IPCC, 2013;IPCC, 2014), which may have serious repercussions on coffee 21 production. Considering these changing temperatures, select genotypes were identified 22 that outperformed others when exposed to higher annual mean temperatures. This 23 suggests there is potentially useful intraspecific variability of thermotolerance in some 24 genotypes and investigation into the molecular mechanisms underlying this variability 25 is warranted (DaMatta, 2018). 26 Increasing temperature impacts plant physiology from the cellular to the whole 27 plant level and changes photoassimilate allocation to repair and recovery processes 28 (Bita and Gerats, 2013;Bokszczanin et al., 2013;Marias et al., 2017a). However, the 29 stress severity depends on intensity and duration of exposure beyond the plant species 30 and within genotypes (Teskey et al., 2015). Therefore, beyond particular characteristics 31 and growth conditions, a comprehensive effect of increasing temperature on plants 32 2 needs first to differentiate data from a moderate long-term change to more drastic ones 33 such as short-duration heat-waves (Thornton et al., 2014). Both phenomena are 34 predicted to be more frequent in the future and may occur singly or concomitantly (Hao 35 et al., 2013;IPCC, 2014) highlighting the need for independent and overlapping 36 studies. 37 To understand the limits of coffee thermotolerance, recent studies have explored 38 the effect of a gradual increasing temperature or extreme heat stress on select 39 physiological processes (reviewed by DaMatta et al., 2019). Minimal impact on 40 photosynthetic-related parameters was observed when various coffee genotypes were 41 exposed to temperatures up to 37°C whereas maximum photosynthetic damage 42 occurred at 42°C for all coffee genotypes . Although coffee 43 presents moderate thermotolerance of photosynthetic-related processes, most 44 genotypes produced abnormal reproductive structures at these elevated temperatures 45 (DaMatta et al., 2019). Accordingly, coffee plants subjected to 45°C for 1-to-1.5 hours 46 showed leaf age-related differences in physiological recovery and did not bear flowers 47 or fruits (Marias et al., 2017a). These results demonstrate that, depending on the tissue 48 and stage of plant development, coffee thermotolerance may be substantial regarding 49 physiological parameters. 50 From the molecular point of view, temperature is perceived by multiple 51 pathways in model plants and crops (Wigge, 2013;Hasanuzzaman et al., 2013;Jagadish 52 et al., 2016;Ibañez et al., 2017). A general and critical cellular response to heat stress 53 is the activation of heat shock proteins (HSPs), which function as chaperones ensuring 54 proper folding of proteins (Ohama et al., 2016). Importantly, phytochromes act as 55 thermosensors joining the related processes of light perception to temperature (Jung et 56 al., 2016). 57 The impact of elevated temperature on gene expression and associated 58 thermotolerance is highly heterogeneous in plant species ,59 2007; Ohama et al., 2017). Therefore, extrapolation of molecular mechanisms relating 60 to thermotolerance in model species is unreliable and will require direct validation. At 61 present, molecular studies examining the effect of elevated temperature on Coffea sp. 62 are limited when compared to molecular-based drought studies in this crop (Moffato et 63 al., 2016;DaMatta et al., 2019). To the best of our knowledge, a large-scale analysis of 64 3 the intraspecific transcriptional variation in response to elevated temperature has not 65 been reported in Coffee arabica L. and could reveal thermotolerance molecular 66 pathways and important strategies towards breeding programs. 67 Thermotolerance is acquired via protective cellular machinery gained 68 throughout coffee plant maturation (Marias et al., 2017a;2017b (Carvalho, 2008). Physiological parameters were evaluated as well as a global 76 transcriptional analysis in conjunction with an initial metabolomics investigation of 77 photo-assimilates, sugars and protein.  81 Two Coffea arabica genotypes, cvs. Acauã and Catuaí IAC 144, hypothesized 82 to differ in heat tolerance (Carvalho, 2008) were examined in the present study. Coffee 83 plants were cultivated in growth chambers with 12 hours of light, 60% humidity and 84 either 23/19ºC or 30/26ºC (day/night temperatures) that are considered the optimal 85 (OpT) and warm temperatures (WaT), respectively (DaMatta and Ramalho, 2006 between two chambers at either OpT or WaT conditions and maintained for four weeks. 96 Each biological repetition was comprised of five excised leaves that were harvested 97 immediately and placed in liquid nitrogen, pulverized with a mortar and pestle, and 98 subsequently stored at -80 ⁰C until analysis. 99 For gas exchange and sugar content analyses, the experiment was repeated using  (Robinson et al., 2010;Huber et al., 2015). Benjamini and 151 Hochberg's false discovery rate (FDR) below 0.05 and a minimum log2 fold change of 152 one were the parameters used to consider a gene differentially expressed between the 153 two conditions. To improve the quality of functional characterization of the DEGs, their 154 respective protein sequences were subjected to homology searches with BLASTP 155 version 2.7.1+ (Camacho et al., 2009) against all plant proteins in the NCBI non-156 6 redundant protein database (nr). In addition, we enriched our DEG results by mapping 157 those proteins against the KEGG database with the BlastKOALA tool (Kanehisa et al.,158 2016) in order to find the pathways that the DEGs were related to.  (Table S2)  soluble sugars were quantified as described by Dische (1962), and the level of reducing 195 sugars was quantified according to Miller (1959). Protein was quantified as described 196 by Bradford (1976) and analyzed in a spectrophotometer at 570 nm comparing results 197 with a standard curve of 0.1 μmol/mL Bovine Serum Albumin (BSA). 200 The modeling approach was carried out by Linear Mixed Models (LMM) using 201 the "lmer" function from the lme4 R package (Bates et al., 2015) for the statistical 202 analysis of IRGA physiological data, metabolic parameters and RT-qPCR expression. 203 In all experiments individuals were used as random factors to deal with the dependence 204 between observations at the same individual across different weeks or conditions. 205 Additionally, the models were fitted by maximum likelihood. The treatments were 206 coded as a factor level and used as fixed effects including temperature conditions (WaT 207 or OpT), cultivar (Catuaí or Acauã) and weeks (only for the physiological analyses), in 208 cases of interest, the interactions between the fixed effects were accounted. Residuals 209 normality and variance homogeneity were assessed by Shapiro-Wilk test and residuals 210 versus fitted plots, respectively. The post hoc pairwise contrasts between factor levels 211 were obtained by "lsmeans" function from "lsmeans" package (Lenth, 2016) using 212 Tukey adjust method. Statistical significance was assessed using Satterthwaite 213 approximation, by the package lmerTest (Kuznetsova et al., 2015). For the RT-qPCR  Table S1). These differences in Tleaf were not correlated with leaf transpiration as both 235 genotypes showed similar values at OpT (Fig. 1B). From this, we concluded that plants 236 of cv. Catuaí, in general, presented a basal temperature higher than cv. Acauã. As plants 237 were subjected to WaT conditions, Tleaf gradually increased for both coffee cultivars 238 during the first three weeks of elevated temperatures and then decreased for both 239 cultivars (Fig. 1A). As was observed under OpT conditions, cv. Catuaí showed higher 240 Tleaf values compared to cv. Acauã. Catuaí always remained above the 30°C imposed 241 by chamber ambient temperature whereas cv. Acauã showed Tleaf mostly below 30°C. 242 In examining leaf transpiration at WaT, both cultivars displayed an increase in 243 transpiration over that observed at OpT (Fig. 1B). A transient difference in transpiration 244 was observed at week 2 with cv. Catuaí showing a marked increase, but this difference 245 was not significant and did not persist into the subsequent weeks under elevated 246 temperature (Table S1). Regardless, Tleaf and transpiration values did not correlate 247 well since, in general, lower leaf temperatures are associated with evaporative cooling 248 9 driven by higher transpiration rates. This poor correlation is especially apparent when 249 examining the results observed at week 2 under WaT conditions; transpiration rate and 250 leaf temperature for cv. Catuaí was markedly higher compared to cv. Acauã (Fig. 1A, 251 1B and Table S1). We propose that cv. Catuaí juvenile trees have a lower 252 thermoregulatory efficiency because Tleaf was higher than cv. Acauã under both OpT 253 and WaT growth conditions. 254 Additionally, photosynthetic rates and stomatal conductance did not show 255 consistent differences across time points between the coffee genotypes (Fig. 1C, 1D   256 and Table S1). For instance, comparing the two genotypes in each condition a similar 257 trend can be noted, except at week 4 for OpT conditions where a significant difference 258 was observed (Fig 1C, Table S1). For stomatal conductance, consistent differences 259 between the two genotypes were not apparent since stomatal conductance varied only 260 at week 4 in OpT ( Fig. 1D and Table S1). These results suggest that, despite apparent 261 differences in their thermoregulatory efficiency, the coffee genotypes examined did not 262 show consistent differences in photosynthesis or stomatal conductance during the 4 263 weeks of elevated growth temperature. within and between the two cultivars in response to warming (Fig. 2). 276 Two types of RNAseq analyses were made. One analysis compared gene 277 expression between the two coffee cultivars at a select temperature (OpT; WaT), which 278 revealed DEGs related to transcriptional differences between genotypes at a given 279 growth temperature (Fig. 2A). The second analysis examined DEGs within a genotype 280 in response to different temperature conditions, which revealed genotypic-dependent 281 DEGs in response to WaT ( Fig. 2A). A heat map shows the expression levels of DEGs 282 ranging between -10 to +10 fold changes in expression, and presents two visible 283 patterns; most of the DEGs were down-regulated in cv. Acauã in relation to cv. Catuaí 284 at a fixed growth temperature, whereas most of DEGs responsive to WaT were up-285 regulated in both genotypes (Fig. 2B). The annotation and fold expression details of all 286 DEGs are provided in Table S2. In total, 186 DEGs were found when comparing gene 287 expression of cv. Acauã to cv. Catuaí ( Fig. 2C and Table S2 Trehalose (EC 3.1.3.12). Thus, coffee genotypes at one-year of age presented 300 differences in gene transcription at optimal growth temperatures that are related to 301 energy metabolism. In contrast, the number of DEGs between the two genotypes were 302 reduced drastically when cultivars were placed under WaT (Fig. 2C) suggesting that 303 many of the intraspecific transcriptional differences were restricted to optimal 304 temperature conditions. 305 Our analyses of coffee genotypes revealed a total of 52 DEGs in response to 306 WaT ( Fig. 2D and Table S2), in which 16 DEGs occurred exclusively in cv. Catuaí (9 307 up-and 7 down-regulated) and 30 in cv. Acauã (26 up-and 4 down-regulated) while 6 308 DEGs were in common between the two genotypes (5 up-and 1 down-regulated). We 309 performed gene annotation and GO analyses of DEGs (Table S2 and Fig. S3) which 310 revealed that the most enriched GO category and biological process was related to 311 11 carbohydrate metabolism (Fig. S3). Indeed, three of these DEGs represent enzymes that 312 are part of the carbohydrate pathway of starch and sucrose metabolism (Fig. S4) Table S3 for statistics).

332
DEGs between coffee genotypes (Fig. 2D) suggest the existence of a conserved 333 mechanism in response to WaT and also exclusive pathways, both mainly related to 334 carbohydrate metabolism control ( Fig. S1 and S3). For example, Granule-bound starch 335 synthase 1 (Cc08_g16970) was up-regulated in both coffee genotypes in response to 336 WaT ( Fig. 3 and S4) (Bertrand et al., 2015). Warmer temperature affects sugar and protein content of coffee genotypes 344 Based on previous results from transcriptional analyses, we investigated 345 whether warmer growth temperatures could affect the sugar content in a genotype-346 dependent manner (Fig. 4). With several noted exceptions, i.e. soluble and reducing 347 sugars in cv. Acauã leaves, both coffee genotypes showed similar patterns with higher 348 carbohydrate and protein content in leaves at OpT compared to WaT growth conditions 349 (Fig. 4). Statistical analyses revealed specific differences in response to WaT (see Table   350 S4), including a significant drop in leaf starch content in cv. Acauã at WaT (p<.0001; roots did not exhibit a difference in sucrose or soluble sugars content due to growth 363 temperature or genotype identity ( Fig. 4F and 4G whereas photosynthetic rates did not differ between the coffee cultivars ( Fig. 1 and   379 Table S1 ). The present findings are in agreement with recent studies that describe 380 photosynthetic stability, as well as thermoregulatory differences, between coffee 381 genotypes in response to increased growth temperatures (Bertrand et al., 2015;Martins 382 et al., 2016, DaMatta et al., 2019. Comparative studies identifying warm tolerant 383 coffee genotypes are scarce and our results show that, based on the established 384 experimental conditions, cv. Acauã appears to better regulate temperature than cv. 385 Catuai for some of the physiological parameters analyzed. 386 The pathways that integrate temperature perception with physiological and 387 metabolic regulation in plants largely depend on complex transcriptional networks 388 (Ruan et al., 2010;Bita and Gerats, 2013). Given this, together with observed 389 physiological differences between cv. Acauã and Catuaí, we conducted an RNAseq 390 analyses of coffee leaves in response to elevated growth temperature. The results 391 showed a number of DEGs (Table S1) including HSPs and genes related to 392 photosynthesis and carbohydrate regulation, which agree with published literature from 393 other species (Jagadish et al., 2016;Ohama et al., 2016). bottlenecks) in molecular signaling networks (Dietz et al, 2010), and the genotype-408 dependent effect of ambient temperature in plant plasticity (Ibañez et al., 2017;Zhu et 409 al. 2018). Thus, stress-imposed constraints impact the energy costs for plant 410 development limiting phenotypic plasticity (Auld et al., 2010;Murren et al., 2015) and 411 bringing a grand challenge to select crop genotypes more resilient to climate change 412 (Pereira, 2016). 413 In response to warm temperature, we found six DEGs shared by the two coffee 414 genotypes examined ( Fig. 2D and Table S2), whose expression trends were validated 415 by RT-qPCR (Fig. S7). These results indicate a core conservative thermoregulatory 416 mechanism within the coffee genotypes. The homologues of Small Heat shock Proteins 417 (Cc11_g16360) are triggered in response to stress and during the ripening process, 418 acting as chaperones presenting a complex expression pattern (Ohama et al., 2016;Arce 419 et al., 2018). RuBisCO activase (Cc04_g14500), whose homologues promote RuBisCO 420 activity (Salvucci and Ogren, 1996), is up-regulated by WaT in coffee. RuBisCO is 421 involved in CO2 fixation during photosynthesis and it is negatively affected by 422 increased growth temperature (Salvucci et al., 2001;Crafts-Brandner and Salvucci, 423 2000). Thus, the increase of RuBisCO activase in coffee leaves was interpreted as a 424 compensatory mechanism at WaT agreeing with photosynthetic rates that were 425 unaffected by warmer temperature (Fig. 1C). Homologues of Isoflavone reductase 426 (Cc10_g02660) are involved in isoflavonoid synthesis, which are secondary 427 metabolites related to lignin biosynthesis and pathogen defense (Shoji et al., 2002;428 Cheng et al., 2015). However, the direct relationship between Isoflavone reductase and 429 temperature stress has not been previously established (Wang et al., 2006b). The DEGs 430 UP-9A (Cc02_g00580) and TPR-like (Cc01_g17230) appear to be a stress response 431 related to sulfur deficiency. In Arabidopsis, homologues of UP-9A are putative 432 interactors with ADP-glucose, which plays a key role in starch metabolism by 433 converting glucose 1-phosphate to ADP-glucose (Crevillén et al., 2005). Homologues 434 of Granule-bound starch synthase (Cc08_g16970) are involved in starch and sucrose 435 metabolism pathways and in thermotolerance acquisition (Wang et al., 2006a;Tian et 436 al., 2018). 437 A series of DEGs responsive to warmer temperatures were not shared between 438 the two cultivars suggesting the possible existence of genotype-specific 439 thermoregulatory mechanisms (Fig. 2D). We observed that many of these DEGs are 440 involved with carbohydrates and carbon regulatory pathways (Fig. S1 to S6). Within 441 cv. Acauã, nearly twice as many differentially expressed genes were found in WaT 442 versus OpT in comparison to WaT versus OpT DEGs in cv. Catuaí (Fig. 2D) Our results demonstrated that there are differences in leaves regarding the sugar 458 content, such as starch, sucrose and total soluble sugars (TSS), and in total protein 459 content in response to warmer temperatures (Fig. 4). These negative correlations 460 between sugar content and temperature stress are in agreement with the report for C. 461 arabica L. by Bertrand et al. (2015), and reinforces our conclusion that warmer 462 temperature has a genetic and physiologic impact on coffee leaves in a genotype-  Table S4 -Statistical analyses of sugars and protein analyses presented in Fig. 4.  Table S2. Labels: OpT -optimal temperature (23/19°C, day/night); WaT -warm temperature (30/26°C, day/night); CA -cv. Catuaí; AC -cv. Acauã.
28 Figure 3 -RNAseq expression analysis of warm-responsive DEGs related to energy metabolism and thermotolerance. Based on analysis of DEGs ( Fig. 2D and Table S2), we evaluated RNAseq expression of ten DEGs related to energy metabolism and thermotolerance; six DEGs that were shared between the two genotypes and four that were exclusive to one of the two genotypes, three in cv. Acauã and one in cv.
Catuaí, respectively: the TPR-like (Cc01_g17230), Isoflavone reductase analyses were performed comparing the same coffee genotype at different temperatures (capital letters) and comparing different genotypes at the same temperature (small letters). Differences were considered significant at p<0.05 (see Table S2 for details).