To examine the effect of temperature under positive DIF on plant growth, tomato seedlings (Managua RZ) were grown under control (25°C/20°C, CT) and high temperatures (30°C/25°C, HT) (Supplementary Figures 1A,B). The stem, hypocotyl, and epicotyl lengths and diameters were significantly greater under HT than those under CT (Figures 1A–C). Histological analysis of hypocotyl cross sections showed that the number of xylem vessels (diameter > 50 μm) were significantly greater under HT than that under CT (Figures 1D,E), indicating that higher temperatures under positive DIF promote stem elongation, stem thickening, and vascular development in tomato seedlings.

To examine the effect of negative DIF on tomato seedling growth, tomato seedlings were grown under HT (30°C/25°C, positive DIF) or with night and day temperatures reversed (25°C/30°C, negative DIF) (Supplementary Figures 1B,C). Under the negative DIF treatment, temperature-dependent stem elongation was inhibited both in the hypocotyl and epicotyl (Figures 2A,B). The difference in hypocotyl and epicotyl elongation between DIF treatments could be detected from 3 d after the onset of the treatments, and it became significant at 5 d (Supplementary Figures 2A–C). Conversely, hypocotyl and epicotyl thickness were similar under both DIF treatments (Figure 2C and Supplementary Figures 2D,E). The number of xylem vessels (diameter > 50 μm) was also similar (Figures 2D,E), indicating that the negative DIF treatment inhibited stem elongation without any negative effect on stem thickening. The size of cotyledons and true leaves was also similar under both treatments (Supplementary Figure 2F).

To determine whether these responses were cultivar-specific, we tested the effect of the same DIF treatments on four other tomato cultivars. Similar to Managua RZ, the negative DIF treatment decreased stem length in all four tomato cultivars, but it did not affect stem thickness (Supplementary Figure 3). This suggests that the growth responses observed are a common feature among tomato seedlings.

To examine the mechanisms underlying stem growth regulation under the DIF treatments, we analyzed epicotyl transcriptomes in seedlings grown under positive or negative DIF for 7 d (Supplementary Figures 1B,C) and explored differentially expressed genes (DEGs). Microarray analysis revealed more than 5000 DEGs, with some upregulated and downregulated by negative DIF treatment (Figure 3A). We performed gene ontology (GO) analysis using the top 300 upregulated and downregulated genes (P < 0.05) (Supplementary Tables 1, 2). Enriched GO terms in biological process were found in cell wall macromolecule catabolic/metabolic process and (programmed) cell death (Figure 3B). Since stem elongation is related to cell wall modification, we focused on cell wall-related genes for further analyses. In addition, hormone-related genes could be involved in temperature-dependent regulation of epicotyl elongation; consequently, GA and IAA-related genes were also included in further analyses. In total, seven DEGs associated with the cell wall, GA, and IAA, as well as a PIF4 homolog possibly related to temperature-dependent hypocotyl elongation, were chosen for further analyses (Figure 3C).

Among the identified DEGs, a key gene of de novo GA biosynthesis, namely GA20-oxidase (SlGA20ox1: Solyc03g006880), was downregulated (Figure 3C). Using reverse transcriptase real-time PCR, we confirmed that the expression of SlGA20ox1 was significantly downregulated in both hypocotyls and epicotyls under the negative DIF treatment (Figure 4A). The following genes involved in IAA biosynthesis were also among the downregulated DEGs: YUCCA (YUC), which encodes a flavin monooxygenase-like enzyme; TRYPTOPHAN AMINOTRANSFERASE OF ARABIDOPSIS (TAA); and IAA responsive factor SMALL AUXIN UP RNA (SAUR) (Figure 3C). Our reverse transcriptase real-time PCR analysis confirmed that SlTAA1 (Solyc05g031600) and SlYUC (Solyc08g068160) were downregulated in tomato epicotyl tissues under the negative DIF treatment (Figures 4B,C). SlSAUR47 (Solyc04g053000) was significantly downregulated in both hypocotyl and epicotyl tissues under the negative DIF treatment (Figure 4D).

In our transcriptome analysis, expansins (SlEXP2; Solyc06g049050 and SlEXP1; Solyc06g051800) and Xyloglucan endotransglucosylase/hydrolase (SlXTH2; Solyc07g009380) were downregulated (Figure 3C). Our reverse transcriptase real-time PCR analysis revealed that SlEXP2 and SlXTH2 were significantly downregulated in both hypocotyls and epicotyls under the negative DIF treatment (Figures 4E,G). SlEXP1 was also downregulated in epicotyls under the negative DIF treatment (Figure 4F). These results are in line with the negative DIF treatment-driven repression of stem elongation.

We further examined the expression of PIF4, but the expression of SlPIF4 (Solyc07g043580), a homolog of A. thaliana PIF4, was not affected by DIF treatment (Figure 4H).

We next examined the expression of the DEGs under the CT (25°C/20°C) and HT (30°C/25°C) positive DIF treatments (Supplementary Figures 1A,B) using reverse transcriptase real-time PCR. SlGA20ox1, SlYUC, SlEXP2, and SlXTH2 were upregulated in epicotyls under the HT treatment (Figures 5A,C,E,G). SlSAUR47 and SlEXP1 were also upregulated in both hypocotyls and epicotyls under the HT treatment (Figures 5D,F). These results are consistent with the positive DIF treatment-driven promotion of stem elongation. In addition, SlPIF4 was upregulated in epicotyls under the HT treatment (Figure 5H).

We further investigated the effect of growth temperature on phytohormone concentrations in hypocotyls and epicotyls (Figure 6). Under positive and negative DIF treatments, GA₁ and GA₄, bioactive forms, could only be quantified in epicotyls under the positive DIF treatment, whereas they were below the quantification limit in all other tissues (Figures 6A,B). Another bioactive form, GA₇, was also below the quantification limit (Figure 6C). The concentrations of the precursors, including GA₉, GA₁₉, GA₂₀, GA₂₄, and GA₄₄, were lower in hypocotyls or epicotyls, while that of GA₅₃ was higher, under the negative DIF treatment compared with those under the positive DIF treatment (Figures 6D,F–J).

When we analyzed hormone species under the CT and HT positive DIF treatments (Figure 7), GA₁ and GA₄ were only detected in epicotyls under the HT treatment (Figures 7A,B). Concentrations of the inactive precursors GA₁₅, GA₂₄, and GA₄₄ in epicotyls were higher under the HT treatment than those under the CT treatment (Figures 7E,H,I). Conversely, concentrations of GA₅₃ were lower under the HT treatment those under the CT treatment (Figure 7J). These patterns were opposite to the changes that occurred following negative DIF treatment (Figure 6).

IAA concentrations in negative DIF-treated hypocotyls and epicotyls were slightly but significantly lower than those in positive DIF-treated tissues (Figure 6K). In contrast, the concentrations of IAA in hypocotyls and epicotyls were higher under HT than those under CT (Figure 7K). These results were consistent with the expression pattern of IAA biosynthesis and signaling genes (Figures 4, 5).

We also quantified the concentration of cytokinins (CKs) because the involvement of this hormone in vascular development is well-documented (Kieber and Schaller, 2014). Several N⁶-(Δ²-isopentenyl) adenine (iP)-type and trans-zeatin-type (tZ-type) species had higher concentrations in negative DIF-treated epicotyls than positive DIF-treated epicotyls (Supplementary Figure 4). Conversely, an opposite effect was found under the CT and HT treatments (Supplementary Figure 5).

The tomato (S. lycopersicum) cultivar used in this study was Managua RZ (RIJK ZWAAN, Netherlands), except for the growth comparison of tomato cultivars under DIF treatments (Supplementary Figure 3). In the comparison experiment, we used CF Momotaro-York and Daiki B Baria (Takii Seed, Kyoto, Japan), and Rinka and Reiyo (Sakata Seed, Kanagawa, Japan).

Tomato seeds were imbibed in water in a petri dish at 28°C in the dark for 2 d. After imbibition, germinated seeds were transferred to a wet rockwool block (Nippon Rockwool Corporation, Japan) and were grown at 23°C/23°C (DT/NT, 16 h photoperiod) under 130 to 140 μmol photons m^–2 s^–1 of fluorescent illumination for 5–6 d. Next, the young seedlings, whose cotyledons were fully opened, were further grown on the rockwool block with liquid culture medium under the following two temperature-condition pairs: 25°C/20°C and 30°C/25°C, or 30°C/25°C and 25°C/30°C (DT/NT, 16 h photoperiod) under 300 μmol photons m^–2 s^–1 of light for 7 d (Supplementary Figure S1). The liquid culture medium contained nutrients at the following concentrations: 5 mM KNO₃, 1 mM NH₄H₂PO₄, 0.5 mM MgSO₄, 5.5 mM Ca(NO₃)₂, 27 μM Fe-EDTA, 25 μM KCl, 10 μM H₃BO₃, 1 μM MnSO₄, 1 μM ZnSO₄, 0.25 μM CuSO₄, and 0.04 μM Na₂MoO₄.

The middle section of the tomato seedling hypocotyls was harvested and fixed in a 4% paraformaldehyde phosphate buffer solution (Nacalai Tesque, Kyoto, Japan). The fixed hypocotyls were thinly sliced (∼0.5 mm) with a razor blade, and cross sections were observed with fluorescence microscopy (Mirror unit with U-FUW, Olympus BX53, Olympus, Japan). The diameter of vessels was measured using the ImageJ software (Abramoff et al., 2004).

Phytohormones were extracted and semi-purified as previously described (Kojima et al., 2009; Kojima and Sakakibara, 2012). CKs were quantified using an ultra-performance liquid chromatography (UPLC)-electrospray interface (ESI) tandem quadrupole mass spectrometer (qMS/MS) (AQUITY UPLC^TM System/Xevo-TQS; Waters, Milford, MA, United States) as described previously (Kojima et al., 2009) with an ODS column (AQUITY UPLC HSS T3, 1.8mm, 2.1 x 100 mm; Waters). IAA and GAs were quantified using an ultra-high performance liquid chromatography (UHPLC)-ESI quadrupole-orbitrap mass spectrometer (UHPLC/Q-Exactive^TM; Thermo Fisher Scientific, United States) as described previously (Kojima and Sakakibara, 2012; Shinozaki et al., 2015) with an ODS column (AQUITY UPLC HSS T3, 1.8mm, 2.1 x 100 mm; Waters).

Hypocotyls and epicotyls were frozen in liquid nitrogen and ground to a fine powder with a mortar and pestle. Total RNA was extracted using an RNeasy Mini kit with an RNase-Free DNase Set (Cat. No. 74104/79254, Qiagen, Hilden, Germany). The RNA in samples was quantified using a NanoDrop-1000 spectrophotometer and the quality was monitored using an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, United States).

Total RNA was extracted from epicotyls of young seedlings grown for 7 d under 30°C/25°C (positive DIF) and 25°C/30°C (negative DIF) (DT/NT, 16 h photoperiod) as described in the preceding subsection. Three biological replicates were used for microarray analysis. Target labeling was performed according to the manual of the Low Input Quick Amp Labeling Kit, One-Color (Agilent Technologies). We used a tomato custom-designed microarray (platform ID “GPL21511”). Hybridization was performed according to the manufacturer’s instructions. We scanned the microarray images using an Agilent DNA Microarray Scanner G2565CA (Agilent Technologies). Scanned images were converted to signal data using Feature Extraction software (Agilent Technologies). Value definition in the data matrix was Log2. GO enrichment analysis among DEGs (top 300 upregulated and top 300 downregulated genes; P < 0.05) was performed using the GO Analysis Toolkit and Database for Agricultural Community (AgriGO, http://systemsbiology.cau.edu.cn/agriGOv2/) (Tian et al., 2017).

First-strand cDNA was synthesized using SuperScript^TM III First-Strand Synthesis SuperMix (Invitrogen, Waltham, MA, United States). Real-time PCR was performed using a StepOnePlus Real Time PCR system (Applied Biosystems, Waltham, MA, United States) and a KAPA SYBR FAST qPCR Master Mix (2×) Kit (Kapa Biosystems, London, United Kingdom) under the following conditions: 95°C for 3 min; followed by 40 cycles of 95°C for 3 s and 60°C for 20 s. Gene expression was calculated using the ΔΔCT method and normalized to that of the ubiquitin homolog as the housekeeping gene (Solyc01g056940). The following primers were used: for the housekeeping gene (Solyc01g056940), forward primer 5′-CG TGGTGGTGCTAAGAAGAG-3′, reverse primer 5′-ACGAAG CCTCTGAACCTTTC-3′; for SlGA20ox1 (Solyc03g006880), forward primer 5′-TGGCGTTCCATCAGTCCAAA-3′, reverse primer 5′-TTCGAGGGTTGTTGGAGTCC-3′; for SlTAA1 (Solyc05g031600), forward primer 5′-TGAAGCACACCCTGC ATTTG-3′, reverse primer 5′-ACTTCCAAATCTTCTTCCACT CCTT-3′; for SlYUC (Solyc08g068160), forward primer 5′-GC CCTCGTGGCTAAAGGAA-3′, reverse primer 5′-CCACTGCA TAAAGTCCACACTCTC-3′; for SlSAUR47 (Solyc04g053000), forward primer 5′-GAAGAACAGTTTGGCTTCGATTAC-3′, reverse primer 5′-CGGTATGTGAGATCAACAAACA-3′; for SlEXP2 (Solyc06g049050), forward primer 5′-TTCGAAGGGTG CCCTGTAT-3′, reverse primer 5′-TGAATATCACCAGCAC CTCCA-3′; for SlEXP1 (Solyc06g051800), forward primer 5′-CGCTGGCATTGTTCCTGT-3′, reverse primer 5′-CTGC ACCTGCTACATTCGTG-3′; for SlXTH2 (Solyc07g009380), forward primer 5′-TATGCACAAGGCAAGGGAGA-3′, reverse primer 5′-TGTATTGTCTTATTGGTGTCCCATC-3′; for SlPIF4 (Solyc07g043580), forward primer 5′-ATCAAGCAGCTGCAAT GTGC-3′, reverse primer 5′-CTGCTGAGTTTGCTGTGCTG-3′.