Crosstalk of PIF4 and DELLA modulates CBF transcript and hormone homeostasis in cold response in tomato

Summary The ability to interpret daily and seasonal fluctuations, latitudinal and vegetation canopy variations in light and temperature signals is essential for plant survival. However, the precise molecular mechanisms transducing the signals from light and temperature perception to maintain plant growth and adaptation remain elusive. We show that far‐red light induces PHYTOCHROME‐INTERACTING TRANSCRIPTION 4 (SlPIF4) accumulation under low‐temperature conditions via phytochrome A in Solanum lycopersicum (tomato). Reverse genetic approaches revealed that knocking out SlPIF4 increases cold susceptibility, while overexpressing SlPIF4 enhances cold tolerance in tomato plants. SlPIF4 not only directly binds to the promoters of the C‐REPEAT BINDING FACTOR (SlCBF) genes and activates their expression but also regulates plant hormone biosynthesis and signals, including abscisic acid, jasmonate and gibberellin (GA), in response to low temperature. Moreover, SlPIF4 directly activates the SlDELLA gene (GA‐INSENSITIVE 4, SlGAI4) under cold stress, and SlGAI4 positively regulates cold tolerance. Additionally, SlGAI4 represses accumulation of the SlPIF4 protein, thus forming multiple coherent feed‐forward loops. Our results reveal that plants integrate light and temperature signals to better adapt to cold stress through shared hormone pathways and transcriptional regulators, which may provide a comprehensive understanding of plant growth and survival in a changing environment.


Introduction
As sessile organisms, plants must integrate multiple environmental and endogenous signals to adjust their growth patterns and developmental transitions to withstand adverse environments and minimize damage. Low temperature is a major environmental stress that severely impairs plant growth and productivity and affects the geographical distribution of plants. To survive under cold stress, plants trigger a suite of sophisticated physiological and biochemical processes. Previous studies have revealed that the ICE-CBF/DREB1 regulatory pathway plays a critical role in cold stress response (Chinnusamy et al., 2007;Thomashow, 1999). The cold stress-induced CBF transcription factors directly activate the expression of downstream COLD-REGULATED (COR) genes and enhance plant cold tolerance. Knocking out all three CBF genes leads to an extreme sensitivity to cold stress (Jia et al., 2016;Zhao et al., 2016), while overexpressing CBFs leads to constitutively enhanced cold tolerance (Gilmour et al., 2000;Jaglo-Ottosen et al., 1998).
Light and temperature are not completely independent under natural plant growth conditions; they provide vital immediate and predictive cues for plants to ensure optimal growth and development (Franklin, 2009;Legris et al., 2017). It has been reported that light is essential for the development of cold acclimation in plants (Kim et al., 2002). Furthermore, the circadian clock, photoperiod and light quality also regulate plant cold tolerance (Dong et al., 2011;Franklin and Whitelam, 2007;Lee and Thomashow, 2012;Li et al., 2016b;Wang et al., 2016Wang et al., , 2018Wang et al., , 2019. CIRCADIAN CLOCK-ASSOCIATED 1 (CCA1)-mediated and LATE-ELONGATED HYPOCOTYL (LHY)-mediated outputs from the circadian clock positively regulate plant cold tolerance through the CBF pathway in Arabidopsis (Dong et al., 2011). Meanwhile, the CBF pathway is actively repressed by PHYTOCHROME-INTERACTING FACTOR 4 (PIF4) and PIF7 during the warm longday season in Arabidopsis (Lee and Thomashow, 2012). Blue light and low temperature-induced COR27 and COR28 negatively regulate freezing tolerance in Arabidopsis (Li et al., 2016b), whereas a low red/far-red light ratio (L-R/FR) induces cold tolerance in both Arabidopsis and Solanum lycopersicum (Franklin and Whitelam, 2007;Wang et al., 2016Wang et al., , 2018. Intriguingly, recent work has demonstrated that Arabidopsis phytochrome B (phyB) acts as a thermosensor (Jung et al., 2016;Legris et al., 2016), and it negatively regulates cold tolerance in both Arabidopsis and tomato (Franklin and Whitelam, 2007;Wang et al., 2016Wang et al., , 2018. Although we found that phytochrome A (phyA) and phyB function antagonistically to regulate cold tolerance in tomato , whether phyA or other photoreceptors are sensors for low temperature remains to be investigated. LONG HYPOCOTYL 5 (HY5), a bZIP transcription factor, acts downstream of phytochromes and integrates light and cold signalling to optimize plant survival under cold stress (Catal a et al., 2011;Wang et al., 2018Wang et al., , 2019. Therefore, plants have evolved a delicate system that perceives light and temperature signals, allowing them to exquisitely detect and predict changes in the natural environment (Franklin, 2009). PIFs are basic helix-loop-helix transcription factors and have key roles in light-regulated plant development and plant responses to multiple environmental signals (Leivar and Monte, 2014;Leivar and Quail, 2011;Pham et al., 2018). PIFs, particularly PIF4, have emerged as a central signalling hub controlling the thermosensory activation of flowering (Kumar et al., 2012) and thermosensory growth in Arabidopsis (Delker et al., 2014;Gangappa and Kumar, 2017). However, the evening-expressed clock component AtTOC1 interacts with and inactivates AtPIF4 to suppress thermoresponsive growth in the evening, which may serve to increase fitness by matching thermoresponsiveness with the day-night cycles of fluctuating temperature and light conditions . AtPIFs also coordinate light and temperature to regulate the transcription of photosynthesis and photoprotection genes . In the natural environment, there is a significant drop in the R/FR ratio during shade and twilight periods in autumn months (Casal and Q€ uesta, 2018;Ross et al., 1986). L-R/FR stabilized the AtPIF4 protein during shade avoidance , and we previously demonstrated that L-R/FR induced cold tolerance in tomato shade leaves (Wang et al., 2018). Therefore, whether PIF4 is the central signalling hub that integrates light and temperature to regulate cold tolerance in tomato remains to be explored.
PIFs are emerging as integrators of signals from different hormone pathways during growth and development (Leivar and Monte, 2014). Recent studies have demonstrated that AtPIF4/PIF5 induces ethylene and abscisic acid (ABA) signalling in leaf senescence (Sakuraba et al., 2014). Hormones in the gibberellin (GA) and brassinosteroid (BR) classes are also involved in AtPIF4mediated light and temperature signalling (Franklin et al., 2014). DELLA proteins are the key repressors of almost all GA responses (Ueguchi-Tanaka et al., 2007). DELLAs interact with PIFs and have a dual role in modulating PIFs by both sequestration and degradation (Li et al., 2016a). It has been demonstrated that plant hormones, such as ABA, jasmonate (JA), BR and GA, are involved in plant cold tolerance (Achard et al., 2008;Li et al., 2017;Wang et al., 2016;Zhou et al., 2017). Therefore, plants enhance the capacity of perception and prediction of seasonal changes by the multiple integration of light and temperature signals with hormone-signalling pathways and transcriptional regulators (Franklin, 2009).
In this study, we show that far-red light (FR) induces SlPIF4 accumulation dependent on phyA under low-temperature conditions in Solanum lycopersicum. SlPIF4 positively regulates plant cold tolerance in tomato by directly binding to the promoters of the SlCBF genes and activating their expression, while promoting ABA and JA signalling under cold stress. SlPIF4 also directly associates with the promoter sequence of SlGAI4, which encodes a DELLA protein in tomato, and activates its expression under cold stress. Interestingly, when large amounts of SlGAI4 protein accumulated during cold stress, it repressed SlPIF4 accumulation in a negative feedback manner. Thus, our results suggest that SlPIF4 is a pivotal component of light and temperature cues and integrates environmental stimuli with plant hormones to coordinate tomato plant growth with impending cold temperatures.

Results
Far-red light and low temperature induce SlPIF4 accumulation via a phytochrome-dependent pathway in tomato To investigate the possible involvement of SlPIFs in the plant response to cold stress, we identified the eight tomato SlPIF genes through phylogenetic analysis ( Figure S1) and investigated the expression of these genes in tomato plants exposed to cold stress. We found that the expression of SlPIF4 was the highest among the eight SlPIF genes after the plants were exposed to 4°C (~2fold than other genes; Figure 1a). To clarify which SlPIF gene is the major gene in response to cold stress, we silenced SlPIF family genes by tobacco rattle virus-induced gene silencing (VIGS). After cold stress, the relative electrolyte leakage (REL) in SlPIF4-silenced plants (pTRV-PIF4) was higher than those of the other SlPIF genesilenced plants (Figure 1b), which further demonstrated that SlPIF4 was the predominant gene among the SlPIF family genes in response to cold stress. Since low temperature induced SlPHYA gene expression, but inhibited transcripts of SlPHYB1 and SlPHYB2 compared with those in plants grown at 25°C (Figure 1c), we then wanted to know whether the regulation of SlPIF4 by low temperature was dependent on phytochrome. The results showed that SlPIF4 expression was higher in tomato phyB mutants than in wild-type (WT) plants, while its expression was lower in tomato phyA and phyAB1B2 mutants than in WT plants under cold stress (Figure 1d), which indicated that low temperature regulated SlPIF4 via the phytochrome pathway. Since FR enhanced cold tolerance via phyA, and R inhibited cold tolerance via phyB in tomato plants , we then asked whether SlPIF4 was regulated by light quality during cold stress. We examined the gene expression of SlPIF4 and its protein accumulation in tomato plants under different light conditions, such as white light (WL), red light (R), FR light and dark (D). Compared with plants grown at 25°C, plants grown at a low temperature had markedly induced SlPIF4 gene expression and protein accumulation, especially in combination with FR conditions (Figure 1e,f). The gene expression and protein accumulation of SlPIF4 increased and decreased in plants under FR and R conditions, respectively, compared with those in plants under WL conditions at 4°C (Figure 1e,f). These results suggest that R and FR function antagonistically to regulate SlPIF4 accumulation via a phytochrome-dependent pathway in tomato plants.
SlPIF4 is a positive regulator in L-R/FR-induced plant cold tolerance and directly activates CBF gene expression Since there is a significant drop in the R/FR ratio during twilight periods in autumn months, we used white light supplemented with R and/or FR to obtain different R/FR ratios and examined the effects of different R/FR ratios on cold tolerance in tomato. We found that low R/FR light ratios (L-R/FR) could alleviate coldinduced leaf wilted, the increased REL and the decreased maximum photochemical efficiency of PSII (Fv/Fm) compared with high R/FR light ratios (H-R/FR; Figure S2). To determine the role of SlPIF4 in L-R/FR-induced cold tolerance, we generated pif4 mutant and SlPIF4-overexpressing (SlPIF4-OE) transgenic tomato plants. Two independent pif4 mutants (pif4#3 and pif4#10) and two independent overexpression lines of tomato (OE#87 and OE#89) were used for further analysis ( Figure S3), along with the corresponding untransformed WT. We found that D induced the tomato hypocotyl length compared with WL, but the hypocotyl length of pif4 mutant and SlPIF4-OE was the same in tomato under WL and D conditions ( Figure S4). However, we found that the pif4 mutant exhibited increased sensitivity to cold stress, while the SlPIF4-OE plants exhibited decreased sensitivity to cold stress, as indicated by the changes in Fv/Fm and REL (Figures 2a,b and S5a). The Fv/Fm values and REL were lower and higher in the pif4 mutant, respectively, than those in WT, while the Fv/Fm values and REL in the SlPIF4-OE plants were higher and lower, respectively, than those of the WT plants after cold stress (Figure 2a,b). Meanwhile, the leaves were more wilted in the pif4 mutant, but less wilted in the SlPIF4-OE plants, respectively, than those in WT after cold stress ( Figure S5a). These results indicate that SlPIF4 positively regulates cold tolerance in tomato plants. In addition, we found L-R/FR decreased REL and increased the Fv/Fm values in both the WT and SlPIF4-OE plants under cold stress, but these positive effects on cold tolerance were almost abolished in the tomato pif4 mutant. Meanwhile, L-R/FR induced the transcription of SlCBF1 and SlCOR413-like in the WT and SlPIF4-OE plants was mostly abolished in the tomato pif4 mutant plants (Figures 2c and S5b). These results indicate that L-R/FR-induced cold tolerance is partially dependent on SlPIF4.
Next, we investigated whether SlPIF4 directly regulated the transcription of SlCBF1. A previous study showed that the PIF proteins recognize the G-box, E-box and PBE-box motifs (Kim et al., 2003), and promoter analysis revealed the presence of Gbox, E-box and PBE-box motifs in the SlCBF1 gene promoter ( Figure 2d). Thus, we performed electrophoretic mobility shift assays (EMSAs) to test whether SlPIF4 could directly bind to these fragments of the SlCBF1 gene promoter in vitro. EMSA showed that the His-SlPIF4 protein bound directly to the biotin-labelled Gbox-containing probe (G2-box) of the SlCBF1 promoter (nucleotides À2041 to À2016) and caused a mobility shift (Figure 2e, f). Mutation of the core sequence of the G-box motif in the SlCBF1 probes (SlCBF1-G2-mut) resulted in the loss of the capacity of SlPIF4 to bind the probes. Then, we performed chromatin immunoprecipitation (ChIP) assays to test whether SlPIF4 was associated with the CBF promoters in vivo. The qPCR data showed that the promoters of SlCBF1, SlCBF2 and SlCBF3 were significantly enriched in the 35S:SlPIF4-HA samples compared with the WT control, whereas the IgG control was not enriched (Figures 2g and S6). These results indicate that SlPIF4 directly binds to the G-box motifs in the SlCBF1 gene promoters. Collectively, these data demonstrate that SlPIF4 positively regulates cold tolerance by directly binding to the promoters of the SlCBF genes and activating their transcription in response to cold stress.
SlPIF4 promotes ABA and JA biosynthesis but inhibits GA biosynthesis in response to cold stress Abscisic acid and JA have been shown to enhance plant cold tolerance, while GA inhibits plant cold tolerance (Achard et al., 2008;Wang et al., 2016). We found that the transcription levels of genes involved in ABA biosynthesis (SlNCED6) and signalling (SlAREB), and JA biosynthesis (SlAOS2) and signalling (SlCOI1) were not significantly different between the WT, pif4 mutant and SlPIF4-OE plants at 25°C (Figures 3a,c and S7). In contrast, these genes were markedly up-regulated after cold stress, especially in the SlPIF4-OE plants. The transcript levels of these genes were lower in the pif4 mutant plants than in the WT plants under cold stress. We also noted greater increases in ABA and JA accumulation in the leaves of the WT and SlPIF4-OE plants than the pif4 mutant plants after cold stress, especially in the SlPIF4-OE plants Expression of SlPIF4 in WT plants after exposure to 25°C or 4°C for 6 h, which grown under dark (D), white light (WL), red light (R) or FR light conditions. The light intensity is 120 µmol m À2 s À1 . (f) Accumulation of SlPIF4 protein in tomato SlPIF4-overexpressing (SlPIF4-OE) plants after exposure to 25°C or 4°C for 12 h, which grown under D, WL, R or FR conditions. The light intensity is 120 µmol m À2 s À1 . Data are presented as the means of three biological replicates (AESD). Different letters indicate significant differences (P < 0.05) according to Tukey's test. In contrast, the transcript levels of GA biosynthesis genes (SlGA3ox2 and SlGA20ox1) and the levels of active GAs (GA 1 , GA 3 and GA 4 ) and their precursors (GA 9 , GA 19 and GA 20 ) significantly decreased after cold stress ( Figure 4). Meanwhile, the transcript levels of these GA biosynthesis genes and the accumulation levels of these GAs were higher in the pif4 mutant plants and lower in the SlPIF4-OE plants than in the WT plants at 4°C (Figure 4). These results indicate that SlPIF4 enhances cold tolerance in tomato plants partially by inducing ABA and JA, and repressing GA biosynthesis.
SlPIF4 directly binds the promoter of SlGAI4 and activates its transcription during cold stress DELLA proteins are the key repressors of almost all GA responses (Daviere and Achard, 2016) and play critical roles in the plant cold response (Achard et al., 2008;Zhou et al., 2017). There are ten DELLA genes (GA-INSENSITIVE, SlGAIs) in tomato, which were identified by phylogenetic analysis of tomato DELLA family genes ( Figure S8a). VIGS experiments showed that the REL in SlGAI4silenced plants (pTRV-GAI4) was higher than those of the other SlGAI-silenced plants (Figure 5a), which demonstrated that SlGAI4 was the predominant gene among the SlGAI family genes responsible for plant cold tolerance. To determine whether SlPIF4 participated in the regulation of SlGAI4, we analysed the expression levels of the SlGAI4 gene in the WT, pif4 mutant and SlPIF4-OE plants at 25 and 4°C ( Figure 5b). The transcription of SlGAI4 was induced by low temperatures, especially under L-R/ FR conditions, with the pif4 and the SlPIF4-OE plants exhibiting lower and higher transcript levels of SlGAI4, respectively, than the WT plants ( Figure 5b). Promoter analysis revealed that there were two G-box-, two E-box-and two PBE-box-containing fragments in the promoter of the SlGAI4 gene ( Figure 5c). EMSA showed  d) and (e) G-, E-and PBE-box elements in the promoter of tomato SlCBF1 gene (d) and oligonucleotide used in the EMSA (e). Numbering is from predicted transcriptional start sites. The G-box core sequence was mutated in the CBF1-G2-mut probe. The wt and mutated G-box sequences are underlined. The mutated bases were indicated in red. (f) EMSA. The His-SlPIF4 recombinant protein was incubated with biotin-labelled wild-type (CBF1-G2-wt) or mutant (CBF1-G2-mut) oligos. The protein purified from the empty vector was used as a negative control. (g) ChIP-qPCR assay. WT and 35S:SlPIF4-HA tomato plants were grown at 4°C under L-R/FR light for 6 h, and samples were precipitated with an anti-HA antibody. A control reaction was processed simultaneously using mouse IgG. The ChIP results are presented as percentages of the input DNA. For light-quality treatments, plants were maintained at R conditions (120 µmol m À2 s À1 ) and supplemented with different intensities of FR. Data are presented as the means of three biological replicates (AESD). Different letters indicate significant differences (P < 0.05) according to Tukey's test. that SlPIF4 bound to the biotin-labelled probes containing Gboxes (nucleotides À553 to À533), leading to a mobility shift, but the capacity to bind to the SlGAI4 promoter was lost when the promoter was mutated in the G-box elements (GAI4-G1/2-mut; Figure 5c). Dual-luciferase assays indicated that SlPIF4 significantly activated the promoter of SlGAI4 under low-temperature conditions (Figures 5d and S8b). The results were further verified with ChIP assays, which showed that the SlGAI4 promoter sequence was significantly enriched in the 35S:SlPIF4-HA (SlPIF4-OE) samples pulled down by the anti-HA antibody compared with the WT control samples. No enrichment of the IgG control was observed ( Figure 5e). Therefore, SlPIF4 directly binds to the promoter sequence of SlGAI4 and activates its gene expression during cold stress.

SlGAI4 is a positive regulator in L-R/FR-induced plant cold tolerance
To substantiate the role of SlGAI4 in cold tolerance, we obtained SlGAI4-silenced tomato plants (pTRV-SlGAI4) and SlGAI4-overexpressing plants (SlGAI4-OE#54, SlGAI4-OE#56) and analysed the expression levels of SlGAI4 in these plants by qRT-PCR. The results showed that SlGAI4 gene transcription was suppressed by 62% and induced by~20-fold in SlGAI4-silenced plants and SlGAI4overexpressing plants, respectively ( Figure S9). Then, the SlGAI4silenced plants (pTRV-SlGAI4), pTRV plants, SlGAI4-overexpressing plants and WT plants were exposed to cold stress under different light conditions (H-R/FR or L-R/FR). No differences in REL were observed between the pTRV-SlGAI4 and pTRV plants and between the WT and SlGAI4-overexpressing plants grown under optimal growth conditions (Figure 6a,d). However, the pTRV-SlGAI4 plants showed increased sensitivity to cold stress at 4°C, with wilted leaf phenotypes, and L-R/FR-induced cold tolerance was also significantly decreased at 4°C in the pTRV-SlGAI4 plants compared with the pTRV plants, as indicated by the increased REL and decreased Fv/Fm values (Figures 6a,b and S10a). In contrast, the REL of the SlGAI4-overexpressing plants decreased significantly, and Fv/Fm values increased consistently compared with those of the WT plants at 4°C (Figures 6d,e and S10a). These results demonstrate that SlGAI4 is a positive regulator in L-R/FRinduced cold tolerance in tomatoes.
We further examined the expression levels of the cold stressresponsive genes, such as SlCBF1 and SlCOR413-like, in the SlGAI4-silenced plants (pTRV-SlGAI4), SlGAI4-overexpressing plants and the WT plants (pTRV and WT) via qRT-PCR. The expression levels of SlCBF1 and SlCOR413-like were significantly higher in the WT and SlGAI4-overexpressing plants under cold stress, especially in the SlGAI4-overexpressing plants, whereas those of SlCBF1 and SlCOR413-like were lower in the pTRV-SlGAI4 plants than in the pTRV plants (Figures 6c,f and S10b,c). Furthermore, L-R/FR-induced SlCBF1 and SlCOR413-like gene expression levels in the WT and SlGAI4-overexpressing plants were markedly decreased in the SlGAI4-silenced plants. These results indicate that SlGAI4 positively regulates the CBF-pathway genes under cold stress.
SlGAI4 acts downstream of SlPIF4 to positively regulate ABA and JA biosynthesis and signalling under cold stress The results described above suggested that SlPIF4 enhanced cold tolerance in tomato plants by directly activating SlGAI4 and inducing ABA and JA biosynthesis. We then asked whether SlGAI4 regulated the levels of ABA and JA under cold stress. To this end, we examined the ABA biosynthesis (SlNCED6) and signalling (SlAREB and SlRD22-like) genes, JA biosynthesis (SlLOXD and SlAOC) and signalling genes (SlCOI1), and the levels of ABA and JA in WT plants (WT-pTRV), SlGAI4-silenced plants (WT-pTRV-SlGAI4) and SlGAI4-overexpressing plants (OE-SlGAI4-pTRV) under cold stress. The results showed that low temperature induced the transcription of ABA-and JA-related genes and their contents (Figures 7 and S11). Meanwhile, the levels of ABA, JA and their related gene expression in the SlGAI4-overexpressing To further explore the exact role of GA signalling in SlPIF4regulated cold tolerance in tomato plants, we tested whether the cold tolerance of the WT, pif4 mutant and SlPIF4-OE plants was affected by altered GA levels. We observed that exogenous GA 3 significantly repressed cold tolerance, with wilted leaves, decreased Fv/Fm values and CBF-pathway gene (CBF1, CBF3, COR47-like and COR413-like) transcript levels and increased REL in the SlPIF4-OE and WT plants at 4°C (Figures 8a-c and S12). In contrast, application of the GA biosynthesis inhibitor paclobutrazol (PAC) dramatically enhanced the cold tolerance of both the WT and pif4 mutant plants, with increased Fv/Fm values and CBFpathway gene (CBF1, CBF3, COR47-like and COR413-like) transcript levels and decreased REL. These results suggest that GA signalling functions downstream of SlPIF4 in the cold response in tomato plants.
To establish whether the actions of GA and ABA in the cold response occur in a linear sequence, we analysed the changes in cold tolerance after foliar application of exogenous GA 3 and PAC in WT and ABA-deficient mutant (notabilis, not) tomato plants. We observed that PAC clearly enhanced the cold tolerance of WT plants, with increased Fv/Fm values and CBF-pathway gene (CBF1, CBF3, COR47-like and COR413-like) transcript levels and decreased REL (Figures 8d-f and S13). However, this induction of plant cold tolerance by PAC was significantly inhibited in the not mutant. These results suggest that GA signalling functions upstream of ABA in the cold response in tomato plants. Collectively, these results indicate that SlGAI4, a key repressor of GA signalling, acts downstream of SlPIF4 to promote cold tolerance in tomato plants partially through activating ABA and JA signalling.  Lucas et al., 2008). To investigate whether tomato DELLAs also promoted SlPIF4 degradation under cold stress, 35S:SlPIF4-HA plants were treated with GA 3 or PAC under optimal temperature conditions and cold stress, and the SlPIF4-HA protein levels were determined. As shown in Figure 9a, the SlPIF4-HA protein accumulated at increased levels at 4°C, especially after GA 3 application. In contrast, the SlPIF4-HA protein abundance significantly decreased when PAC was applied compared with the effect of mock treatment at 4°C (Figure 9a). These data indicate that DELLAs promote the degradation of the SlPIF4 protein. Next, we tested the transcription of SlPIF4 and its protein abundance in SlGAI4-silenced plants and SlGAI4overexpressing plants at 4°C under H-R/FR or L-R/FR conditions. The results showed that SlPIF4 transcription and its protein levels were higher in the SlGAI4-silenced plants and lower in the SlGAI4-overexpressing plants than in the WT plants (pTRV or WT) during cold stress (Figures 9b,c and S14). These results demonstrate that SlGAI4 negative feedback regulates SlPIF4 at both transcriptional and post-translational levels under cold stress.

Discussion
Light and temperature are arguably two of the most important environmental factors that coordinately control plant growth and survival. Light and temperature signals change with daily and seasonal fluctuations and according to latitudinal and vegetation canopy variations. For example, the R/FR ratios are reduced Numbering is from predicted transcriptional start sites. The G-box core sequence was mutated in the GAI4-G1/2-mut probe. The wt and mutated G-box sequences are underlined. The mutated bases were indicated in red. The His-SlPIF4 recombinant protein was incubated with biotin-labelled wild-type (GAI4-G1/2-wt) or mutant (GAI4-G1/2-mut) oligos. The protein purified from the empty vector was used as a negative control. (d) Dual-LUC assay showing the effects of SlPIF4 on SlGAI4 promoter activation under cold stress. The SlGAI4 promoter was fused to the luciferase (LUC) reporter (pGAI4::LUC), and promoter activity was determined by transient expression of it with empty vector (EV) or 35S:SlPIF4 (PIF4) in tobacco. The tobacco plants were exposed to 25°C or 4°C for 24 h after infiltration at 25°C for 24 h. Relative LUC activity was normalized to the Renilla (REN) luciferase. (e) ChIP-qPCR assay. WT and 35S:SlPIF4-HA tomato plants were grown at 4°C under L-R/FR light for 6 h, and samples were precipitated with an anti-HA antibody. A control reaction was processed simultaneously using mouse IgG. The ChIP results are presented as percentages of the input DNA. For light-quality treatments, plants were maintained at R conditions (120 µmol m À2 s À1 ) and supplemented with different intensities of FR. Data are presented as the means of three biological replicates (AESD). Different letters indicate significant differences (P < 0.05) according to Tukey's test. naturally with vegetative shading and twilight durations at northern latitudes in cool seasons (Franklin and Whitelam, 2007). We previously found that L-R/FR ratios induced plant cold tolerance via a phyA-dependent pathway in tomato . Intriguingly, recent work has demonstrated that phyB acts as a thermosensor in the regulation of elongation growth in Arabidopsis at warm temperatures via regulation of AtPIF4 (Jung et al., 2016;Legris et al., 2016). AtPIF4 and AtPIF7 repress CBF gene transcription to decrease the unnecessary activation of the cold acclimation pathway during the warm long-day season (Lee and Thomashow, 2012), which suggests the possibility that PIFs integrate light and low-temperature signalling. Recently, AtPIF3 was identified as a negative regulator of freezing tolerance by directly repressing the CBF gene in Arabidopsis (Jiang et al., 2017). OsPIF14 directly bound the promoter of DREB1B and repressed its expression in Oryza sativa (Cordeiro et al., 2016). However, ZmPIF3 directly activated DREB2A gene expression in Zea mays under drought (Gao et al., 2015), and OsPIL16 positively regulated plant cold tolerance by directly binding the promoter of the DREB1 gene and activating its transcription in Oryza sativa (He et al., 2016). These results indicate that PIFs are functionally diverse among different plant species. Here, we provide several lines of evidence showing that SlPIF4 is involved in L-R/FR-induced cold tolerance in tomato plants ( Figure 10). First, FR induced SlPIF4 accumulation at low temperatures via phyA (Figure 1d-f). Second, the L-R/FR-induced transcript levels of CBF1 under cold stress decreased in pif4 mutant plants, which displayed impaired cold tolerance, but its transcript levels increased in the SlPIF4overexpression lines, which exhibited enhanced cold tolerance (Figures 2a-c and S5). Third, SlPIF4 directly bound to the G-box of the CBF promoters in vitro and was associated with the promoters of the CBF genes in vivo (Figures 2d-g and S6). Finally, SlPIF4 promoted ABA and JA accumulation in tomato plants under cold stress (Figures 3 and S7), which were positive regulators in cold stress , and enhanced the degradation of GA (Figures 3 and S7), which was a negative regulator in cold tolerance (Achard et al., 2008;Zhou et al., 2017). Collectively, our results indicate that SlPIF4 works as a positive regulator of L-R/FR-induced cold tolerance in tomato plants that directly activates the transcription of CBF genes and regulates phytohormone homeostasis.
We observed that GA 3 significantly repressed cold tolerance in the SlPIF4-OE and WT plants at 4°C (Figures 8a-c and S12); in contrast, application of PAC dramatically enhanced the cold Previous studies showed that PHYTOCHROME-INTERACTING FACTOR3-LIKE5 (PIL5) inhibited seed germination by directly binding the promoters of two DELLA genes (GAI and RGA) and promoting their transcription in Arabidopsis (Oh et al., 2007). Consistent with these observations, our ChIP analyses and EMSA showed that SlPIF4 directly and specifically bound G-box elements of the SlGAI4 promoters in vitro and in vivo (Figure 5c,e). Dual-luciferase assays and transcript analyses in the pif4 mutant and SlPIF4-overexpressing plants further confirmed that SlPIF4 directly activated SlGAI4 gene expression under cold stress, especially under L-R/FR conditions (Figures 5b,c and S8b). In addition, we provided evidence that SlGAI4 was a positive regulator that modulates cold tolerance in tomato plants. We showed that multiple transgenic lines overexpressing SlGAI4 displayed a decrease in REL and increases in Fv/ Fm values and CBF-pathway gene (CBF1 and COR413-like) expression under cold stress (Figures 6d,e and S10a,c). However, L-R/FR failed to induce Fv/Fm values and CBF-pathway gene (CBF1 and COR413-like) expression in the SlGAI4-silenced plants (Figures 6b,c and S10b). Moreover, it has been reported that CBF and DELLAs positively regulate each other in response to low temperature in Arabidopsis (Achard et al., 2008;Zhou et al., 2017). Taken together, these results demonstrate that SlPIF4 also directly activated SlGAI4, which acts as a positive regulator in the regulation of CBF1 gene expression and L-R/FR-induced cold tolerance in tomato plants during cold stress.
DELLA proteins serve as integrators of various hormonal and environmental signals (Achard et al., 2006;Daviere and Achard, 2016); thus, we investigated the levels of ABA-related genes and their contents in the SlGAI4-silenced and SlGAI4-overexpressing plants under cold stress. Our results indicated that SlGAI4 positively regulated the levels of ABA-related genes and ABA under cold stress (Figures 7a,b and S11a,b). Foliar application of exogenous GA 3 and PAC significantly decreased and increased cold tolerance in tomato plants, respectively, while foliar application of PAC failed to fully rescue the changes in CBF-related gene transcription and cold tolerance in the ABA-deficient not plants (Figures 8d,f and S13). These results suggest that low temperature induces a decrease in GA levels and promotes SlGAI4 gene expression, which may result in an increase in ABA levels in tomato plants. Consistent with this finding, DELLA proteins interact with ABI3 and ABI5 and form DELLA/ABI3/ABI5 complexes under unfavourable conditions (e.g. high or low temperature), which positively regulate ABA biosynthesis in Arabidopsis (Kim et al., 2008;Lim et al., 2013;Park et al., 2011). Conversely, a previous report showed that ABA decreased GA levels by repressing GA biosynthetic genes in seeds (Seo et al., 2006). Thus, it appears that ABA and GA antagonistically regulate each other. In addition, recent studies have shown that jasmonate ZIM-domain proteins (JAZs), major repressors in JA signalling, directly target ICE1 to inhibit the activation of CBFs, while DELLAs competitively bind to JAZs to release MYC2 and activate the JA response (Hou et al., 2010;Hu et al., 2013;Wild et al., 2012). Meanwhile, MYC2 also interacts with ICE1 to enhance CBF gene transcription in cold conditions (Zhao et al., 2012). Here, we showed that the levels of JA-related gene transcription and JA decreased in the SlGAI4-silenced plants but increased in the SlGAI4-overexpressing plants under cold stress (Figures 7c,d and S11c,d). Thus, SlGAI4 could promote ABA and JA accumulation under cold stress (Figures 7 and S11), which would positively regulate plant cold tolerance via CBF-dependent and CBFindependent pathways (Eremina et al., 2016;Hu et al., 2013;Shinozaki and Yamaguchi-Shinozaki, 2000;Wang et al., 2016).
Previous studies demonstrated that DELLAs interact with AtPIF3 and AtPIF4 and inhibit their activities by sequestering their DNArecognition domains, ultimately results in the inhibition of hypocotyl elongation (Feng et al., 2008;de Lucas et al., 2008). Excitingly, recent studies have revealed that DELLA proteins negatively regulate PIF accumulation by inducing rapid degradation of PIFs through the 26S proteasome pathway (Li et al., 2016a;Pham et al., 2018). Indeed, our work showed that GA 3 promotes SlPIF4 protein accumulation, while PAC inhibits SlPIF4 protein accumulation (Figure 9a). Furthermore, we found that the SlPIF4 protein and its transcription increased in the SlGAI4silenced plants, while its protein abundance and gene expression decreased in SlGAI4-overexpressing plants compared with WT plants (Figures 9b,c and S14). These results support a role for SlGAI4 in the negative feedback regulation of SlPIF4 at both transcriptional and post-translational levels under cold stress. Therefore, SlPIF4 acts as a central hub that integrates light and temperature signals to orchestrate the regulation of the transcriptional network that drives multiple facets of downstream cold response. During cold stress, low temperature and FR signals induced the accumulation of SlPIF4, which directly activated SlCBFs and SlGAI4 to enhance cold tolerance. SlGAI4 -induced ABA and JA signalling enhanced plant cold tolerance by regulating CBF-dependent or CBF-independent pathways. Since increased expression of CBFs and SlGAI4 would result in plant growth cessation, SlGAI4 forms a negative feedback loop with SlPIF4. This feedback modulation and redundant cold response pathways likely contribute to maintain appropriate levels of SlPIF4 to balance plant growth and cold tolerance. SlCBF1 gene expression (c) in tomato WT, pif4 and SlPIF4-OE plants after exposure to 25°C or 4°C under L-R/FR light for 7 days and 6 h, respectively, which pretreated with H 2 O, GA 3 or PAC for 12 h prior to exposure to cold conditions at 4°C. (d) Fv/Fm in tomato wild-type (WT) and ABA-deficient mutant (not) plants after exposure to 4°C under L-R/FR light for 7 days, which pretreated with water (H 2 O), GA 3 or PAC for 12 h prior to exposure to cold conditions at 4°C. (e) and (f) REL (e) and SlCBF1 gene expression (f) in tomato WT and not plants after exposure to 25°C or 4°C under L-R/FR light for 7 days and 6 h, respectively, which pretreated with H 2 O, GA 3 or PAC for 12 h prior to exposure to cold conditions at 4°C. For light-quality treatments, plants were maintained at R conditions (120 µmol m À2 s À1 ) and supplemented with different intensities of FR. Data are presented as the means of three biological replicates (AESD). Different letters indicate significant differences (P < 0.05) according to Tukey's test.

Conclusions
In summary, we propose a model that illustrates how light and temperature signals are integrated to regulate cold tolerance in tomato plants ( Figure 10). Briefly, L-R/FR and low temperature induce SlPIF4 accumulation via phyA, while phyB represses SlPIF4 accumulation under cold stress. SlPIF4 not only directly activates CBF expression but also associates with the promoter of the SlGAI4 gene and activates its transcript, which promotes ABA and JA biosynthesis and CBF expression. Thus, SlPIF4 is a positive regulator in L-R/FR-induced cold tolerance in tomato plants. SlGAI4, a DELLA protein, acts downstream of SlPIF4 and positively regulates L-R/FR-induced cold tolerance. Interestingly, when large amounts of SlGAI4 protein accumulate during cold stress, it represses SlPIF4 accumulation in a negative feedback manner. In the natural environment, a decrease in temperature is often associated with longer twilight durations during autumn at northern latitudes, which are characterized by a significant drop in the R/FR ratio. Monitoring of R/FR ratio signals would provide an early warning and confer some protection to plants subject to a sudden decrease in temperature during night and seasonal variations. This study unveils a novel mechanism by which tomato plants have evolved a phytochrome-dependent, SlPIF4-mediated adaptation strategy by sensing and integrating environmental cues with plant hormone signals.

Plant materials and constructs
The tomato phyA, phyB1B2 and phyAB1B2 mutants were obtained from the Tomato Genetics Resource Center (http:// tgrc.ucdavis.edu). Tobacco rattle virus (TRV)-based vectors (pTRV1/2) were used for VIGS of the SlPIF and SlGAI family genes, and VIGS was performed as described previously . The tomato pif4 mutant in the Ailsa Craig ecotype was obtained by using the CRISPR/Cas9 technique (Pan et al., 2011;Wang et al., 2018). The target sequence (AGGTCATCCAATGTGCAGCT) and its complementary sequence were annealed and inserted into the BbsI site of the AtU6-sgRNA-AtUBQ-Cas9 vector, and the AtU6-sgRNA-AtUBQ-Cas9 cassette was inserted into the HindIII and KpnI sites of the pCAMBIA1301 binary vector. Transgenic plants overexpressing HA-tagged SlPIF4 and SlGAI4 were generated by cloning the full-length SlPIF4 and SlGAI4 cDNAs into the pFGC1008-HA vector, which contains a CaMV 35S promoter, transforming the vectors into Agrobacterium tumefaciens strain EHA105, and then introducing them into tomato seeds of ecotype Ailsa Craig via a previously described method (Fillatti et al., 1987). All primers used for plasmid construction are listed in Table S1. Two independent homozygous lines of the F2 generation in SlPIF4-and SlGAI4-overexpressing plants and two independent pif4 lines, which were mutated at the first base of the protospacer adjacent motif (PAM) to stop translation immediately, were used for the study. In tomato pif4#3 and pif4#10 mutants, there was a single nucleotide (T) insertion at 3-bp upstream of the PAM sites and 2-bp (GC) deletion at 4-bp upstream of the PAM sites in the sgRNA, respectively, as showed in Figure S3a-c. The growth conditions for the tomato mutants and overexpressing seedlings were as follows: temperature of 25°C/20°C (day/night), 12-h light/dark cycles and photosynthetic photon flux density (PPFD) of 600 µmol m À2 s À1 . VIGS plants were grown at 21°C/20°C under 12-h light/dark cycles.

Cold and light treatments
Plants at the 4-leaf stage were used for all experiments, which were carried out in controlled-environment growth chambers (Zhejiang Qiushi Artificial Environment Co., Ltd, China). Plants were grown under dark (D) or white light (WL), red light (R) and FR light conditions with an aerial temperature of 25°C or 4°C for 12-h cold treatment. The light intensities of WL, R and FR were 120 µmol m À2 s À1 . For the different R/FR ratio (high R/FR ratios, 2.5, or low R/FR ratios, 0.5) treatments, R (k max = 660 nm, Philips, Netherlands) light intensity was maintained at 120 µmol m À2 s À1 , and FR (k max = 735 nm, Philips, Netherlands) was added. The R/FR ratio was calculated as the quantum flux density from 655 to 665 nm divided by the quantum flux density from 730 to 740 nm. Plants remained under 12-h light/dark cycles while exposed to the cold treatment. The cold treatment at 4°C lasted for 7 days, unless stated otherwise in the text.

GA 3 and PAC treatment
To unveil the relationship between GA and ABA in cold tolerance, 50 lM GA 3 or 25 lM PAC (GA biosynthesis inhibitor) was applied on WT and not plants 12 h prior to exposure to cold conditions at 4°C under low R/FR (L-R/FR, 0.5) light conditions for 7 days. To determine the effect of DELLAs on SlPIF4 in cold tolerance, WT, pif4 and SlPIF4-OE plants were pretreated with 50 lM GA 3 or 25 lM PAC prior to cold treatment at 4°C under L-R/FR light for 7 days. The GA 3 (Sigma-Aldrich, St. Louis, MO, USA) and PAC (Sigma-Aldrich) solutions were prepared by dissolving the solutes in ethanol followed by dilution with distilled water (ethanol : water [v/v] = 1 : 10 000), respectively. The cold tolerance of tomato plants was analysed after foliar application with 20 mL solution or water on each plant.

Cold tolerance and hypocotyl length assays
The REL, indicating the membrane permeability, was measured as described previously (Cao et al., 2007). The maximum quantum yield of PSII (Fv/Fm) in the leaves was assayed by using the Imaging-PAM set-up (IMAG-MAXI; Heinz Walz, Germany), as previously described (Wang et al., 2018). Hypocotyl length was measured after the germination seeding under white light (12-h light/12-h dark) or dark (24-h dark) conditions for 7 days.

Determination of ABA, JA and GA levels
Endogenous ABA and JA were extracted from tomato leaves and determined by LC/MS-MS on an Agilent 1290 Infinity HPLC system coupled to an Agilent 6460 Triple Quad LC-MS device (Agilent Technologies, Amstelveen, the Netherlands), as described previously . GA was extracted from 1-g samples of tomato leaves and quantified by a derivation approach coupled with nano-LC-ESI-Q-TOF-MS analysis as described previously (Chen et al., 2012;Wang et al., 2019). For the determination of GA levels, D 2 -GA 1 , D 2 -GA 3 , D 2 -GA 4 , D 2 -GA 9 , D 2 -GA 19 and D 2 -GA 20 were added to the extraction solution as internal standards.

Phylogenetic analysis
The amino acid sequences of the eight Arabidopsis thaliana canonical PIF proteins (Leivar and Quail, 2011) were used as queries to perform a BLAST search against Sol Genomics databases (https://solgenomics.net/). Sequence alignment and phylogenetic tree construction were determined with MEGA 6 software using the corrected Nei-Gojobori method. A consensus neighbour-joining tree was obtained from 1000 bootstrap replicates of aligned sequences. The percentage at the branch points represents the posterior probabilities of amino acid sequences.

Isolation of RNA and qRT-PCR
Total RNA was isolated using an RNAprep Pure Plant Kit (Tiangen Biotech Co., Ltd., Beijing, China) from tomato leaves under different conditions as indicated in the figure legend. The extracted RNA was reverse-transcribed using a ReverTra Ace qPCR RT Kit with an enzyme for genomic DNA removal (Toyobo, Osaka, Japan). qRT-PCR was performed with SYBR Green PCR Master Mix (Takara, Japan) using a LightCycler 480 II detection system (Roche, Germany). The PCR procedure was described previously (Wang et al., 2018). The expression levels were normalized to the expression of tomato ACTIN2 gene, which was stably expressed in tomato plants under cold and light stress combined conditions by geNorm algorithm (Livak and Schmittgen, 2001;Løvdal and Lillo, 2009). Primers are listed in Table S2.

Immunoblotting assays
Total proteins were extracted from tomato leaves by homogenization in extraction buffer as described previously (Wang et al., 2019). Protein concentrations were measured using Coomassie stain (Bradford, 1976). Equal amounts of total proteins from each sample were subjected to 15% SDS-PAGE and electrotransferred to nitrocellulose membranes (Bio-Rad, Hercules, CA). The proteins were blotted with antibodies against PIF4 (AS163955; Agrisera) or anti-HA (Cat. No. 26183; Pierce) and subsequently with horseradish-peroxidaseconjugated secondary antibody (anti-goat, Invitrogen, Sweden). The signals were visualized with enhanced chemical luminescence (ECL).

Electrophoretic mobility shift assay
The pET-32a-His-SlPIF4 vector was generated using the full-length coding region of SlPIF4 with the primers listed in Table S1. Histagged SlPIF4 protein was expressed in Escherichia coli strain BL21 (DE3) and purified with the manufacturer's instructions of the Novagen pET purification system. EMSA was performed using biotin-labelled probes and the LightShift Chemiluminescent EMSA Kit (Cat. no. 20148;Thermo Fisher Scientific). The SlPIF4 proteins and biotin-labelled probe were incubated together in binding buffer for 20 min at room temperature, the reaction mixture was resolved by 6% non-denaturing polyacrylamide gel in Tris-glycine buffer and electrophoresed at 100 V, then transferred to a positive nylon membrane, and subjected to UV cross-linking. Finally, the protein-DNA signals were detected by chemiluminescence according to the instructions of the LightShift Chemiluminescent EMSA Kit. The sequences of the biotinlabelled are shown in Table S3.

Chromatin immunoprecipitation assay
ChIP-PCR assays were performed following the manufacturer's instructions for the EpiQuik TM Plant ChIP Kit (Cat. No. P-2014; EpiGentek) as previously described (Wang et al., 2018). Approximately 1 g of leaf tissue was harvested from SlPIF4-OE and WT plants, which were grown at 4°C under L-R/FR conditions for 1 day and were treated with formaldehyde to cross-link the protein-DNA complexes. The chromatin complexes containing SlPIF4-3HA fusion protein were immunoprecipitated with an anti-HA antibody (Cat. No. 26183;Pierce) and Protein A Agarose beads (GE). Goat antimouse IgG (Cat. No. AP124P; Millipore) was used as a negative control. The immunoprecipitated DNA was analysed by qPCR using gene-specific primers which are listed in Table S4.

Dual-luciferase assays
SlPIF4 full-length and the GAI4 promoter fragment were cloned into the pGreenII 0029 62-SK and pGreenII 0800-LUC vectors, respectively ( Figure S8b). The recombinant vectors were transformed into Agrobacterium strain GV3101. The pGreenII 0029 62-SK empty vector was used as a negative control, and the 35S promoter-driven Renilla luciferase was used as an internal control. Different combinations of strains were injected into the back of tobacco leaves. After infiltration, the tobacco plants were grown at 25°C for 24 h, and then, one group of these plants was transferred to 4°C for 24 h before taking samples. The tobacco leaves were ground, and the extraction solutions were incubated in buffer at a low temperature. LUC/REN was detected with an enzyme standard instrument (SpectraMax iD5, Tecan, Basel, Switzerland) by using a Modulus Luminometer (Promega, Madison, WI) as previously described (Yin et al., 2016).

Statistical analyses
Three biological replicates for each treatment were used with at least 6 plants for each replicate. The experiments were independently performed three times. To determine statistical significance, we employed Tukey's least significant difference (LSD) test. The difference was considered significant at P < 0.05 and indicated by different letters.

Accession numbers
Sequence data from this article can be found in the Sol Genomics databases (https://solgenomics.net/) under the accession numbers listed in Tables S2, S3 and S4.

Supporting information
Additional supporting information may be found online in the Supporting Information section at the end of the article.
Figure S1 Phylogenetic analysis of tomato PIF family genes (SlPIFs). Figure S2 Low R/FR enhances the cold tolerance in tomato plants. Figure S3 Tomato pif4 mutant and SlPIF4-overexpressing plants. Figure S4 SlPIF4 has no effect on the hypocotyl length in tomato plants. Figure S5 Phenotypes (a) and expression of COR413-like gene (b) in tomato WT, pif4 mutants and SlPIF4-OE plants after exposure to 25°C or 4°C for 7 days or 6 h, respectively, under high R/FR or low R/FR light conditions. Figure S6 ChIP-qPCR assay shows the relative amount of SlCBF2 and SlCBF3 fragments in 35S: SlPIF4-HA and wild-type tomato plants. Figure S7 SlPIF4 positively regulates expression of ABA and JA signalling genes in response to cold stress. Figure S8 Phylogenetic analysis of tomato GAI family genes (SlGAIs) and schematic diagram showing vectors construction in dual-luciferase assays. Figure S9 Expression of SlGAI4 gene in wild-type (WT/pTRV), SlGAI4-silenced plants (pTRV-GAI4) and SlGAI4-overexpressing plants (OE#54, OE#56). Figure S10 Phenotypes (a) and expression of COR413-like gene in tomato SlGAI4-silenced plants (pTRV-GAI4; b) and SlGAI4overexpressing plants (OE#54, OE#56; c) after exposure to 25°C or 4°C for 7 days or 6 h, respectively, under high R/FR or low R/ FR light conditions. Figure S11 SlGAI4 positively regulates expression of ABA and JA signalling genes in response to cold stress. Figure S12 The effects of GA3 and PAC on cold tolerance in tomato WT, pif4 mutant and SlPIF4-OE plants. Figure S13 The effects of GA3 and PAC on cold tolerance in tomato WT and not plants. Figure S14 Expression of SlPIF4 in tomato SlGAI4-silenced plants (a) and SlGAI4-overexpressing plants (b) after exposure to 25°C or 4°C under H-R/FR or L-R/FR conditions for 6 h.