Arabidopsis thaliana Col-0 accession was used as the wild-type plant. Seeds were surface-sterilized in bleach solution containing 0.01% (v/v) Triton X-100 for 3 min, washed 3 times with sterile water, and sown onto MS medium (Murashige and Skoog, 1962) containing 3% sucrose and 0.8% agar. Plants were grown under 16 h light and 8 h dark at 22°C. For thermospermine treatment, seedlings were incubated for 24 h in liquid MS medium with 100 μM thermospermine-4HCl, which was purchased from Santa Cruz Biotechnology.

Arabidopsis genomic DNA was prepared as described previously (Imai et al., 2006). Genomic DNAs of broccoli (Brassica oleracea), soybean (Glycine max), poplar (Populus tremula × alba), and rice (Oryza sativa) were prepared from each seedling by using NucleoSpin Plant II kit (Macherey-Nagel) following the manufacturer’s instruction.

For the SAC51 promoter-driven expression of the GUS reporter gene, a 990-bp genomic fragment upstream from the first exon of SAC51 was amplified by PCR with primers, SAC51-proFCl, and SAC51-proRBg (Supplementary Table S1), cloned into a pGEM-T easy vector (Promega), and then transferred as a ClaI-BglII fragment to ClaI-BamlHI sites of Ti-plasmid vector pBI101 (Clontech). The GUS gene construct fused to the SAC51 promoter with the first untranslated exon and intron was similarly made using primers, SAC51-proFCl, and SAC51-ex2RBg (Supplementary Table S1). The GUS construct containing the SAC51 promoter and the whole 5′ leader region has been described previously (Imai et al., 2006). The mutant versions were generated by PCR-based site-directed mutagenesis as follows. First-round PCR was performed in 10 cycles of two separate reactions using the wild-type construct as a template with a primer pair of SAC51-proFCl and SAC51-mXR or with a primer pair of SAC51-mXF and SAC51-5RBg (Supplementary Table S1). The PCR products were mixed and subsequently amplified by 10 cycles of PCR with a primer pair of SAC51-proFCl and SAC51-5RBg. The product from a second round of PCR was cloned into pGEM-T Easy and transferred as a ClaI-BglII fragment to pBI101. The no-uORF version was generated sequentially by introducing a point mutation in the start codon of each uORF.

To generate the CaMV 35S promoter-GUS construct containing a genomic or cDNA fragment of the 5′ leader region of SAC51, the fragment was amplified from genomic DNA or reverse-transcribed cDNA by PCR with primers, SAC51-5FSp and SAC51-5RBg (Supplementary Table S1), cloned into pGEM-T easy, and then inserted as a SpeI- BglII fragment between the 35S promoter and the GUS coding region of pBI121 (Clontech). Point mutations were sequentially introduced into the start codon of each uORF as described above. The other 35S-5′-GUS constructs containing a 5′ leader region of SAC51 homologs from different plant species were made in a similar way. All PCR was performed using Ex Taq DNA polymerase (Takara) according to the manufacturers protocol. PCR conditions were 50 cycles of 94°C for 30 s, 55°C for 30 s, and 72°C for 90 s, unless otherwise stated. The primers used were shown in Supplementary Table S1.

Ti plasmid constructs were introduced into Agrobacterium tumefaciens C58C1 by electroporation (Mersereau et al., 1990). Arabidopsis plants were transformed using the floral dip method (Clough and Bent, 1998). Transgenic lines were selected on kanamycin and confirmed by PCR using PBI-Cl and GUS primers (Supplementary Table S1). At least five independent homozygous lines carrying single copy of the transgene were further selected based on the segregation ratio of kanamycin-resistant plants in subsequent generations.

Fluorometric assay of GUS activity was performed as described previously (Jefferson et al., 1987). The fluorescence was measured with an RF-5300PC spectrofluorophotometer (Shimadzu, Japan). Total protein content was measured by using the Bradford assay (BioRad). For histochemical staining of GUS activity, samples were prefixed for 20 min in ice-cold 90% (v/v) acetone under vacuum, rinsed three times with water, and incubated in GUS staining buffer containing 50 mM Na₂HPO₄/NaH₂PO₄ (pH7.0), 2 mM K₃Fe(CN)₆, 2 mM K₄Fe(CN)₆, 0.1% Triton-X100, and 1 mM 5-bromo-4-chloro-3-indolyl glucuronide, at 37°C overnight. Samples were then treated with 70% ethanol to remove chlorophyll.

Total RNA was extracted from Arabidopsis seedlings by the SDS-phenol method (Hanzawa et al., 1997) and reverse transcribed by using the PrimeScript II first strand cDNA Synthesis Kit (Takara). Quantitative RT-PCR was performed on the Thermal Cycler Dice TP760 (Takara) using the KAPA SYBR FAST Universal (KAPA Biosystems). ACTIN8 (ACT8) was used as a reference gene for normalization. Means of expression levels were calculated from three technical replicates. Primers used were ACT8-F (5′-GTGAG CCAGA TCTTC ATTCG TC-3′) and ACT8-R (5′-TCTCT TGCTC GTAGT CGACA G-3′) for ACT8 and SAC51-FF (5′-CATTC CTTTC TAAGA TACTA AAG-3′) and GUS (Supplementary Table S1) for the SAC51 5′-fused GUS, respectively.

The GFP reporter gene was amplified by PCR from pEGFP (Clontech) using primers, GFP-ATG and GFP-3 (Supplementary Table S1), and cloned as a BamHI-XhoI fragment into pT7 Blue-2 (Promega) to generate a control plasmid, pSI020. For the SAC51 5′ leader-GFP transcriptional fusion construct, an 860-bp cDNA fragment of the SAC51 5′ leader region was amplified by PCR with primers, SAC51-5FBal, and SAC51-5RBal (Supplementary Table S1), and inserted into the BalI restriction site of pSI020 just upstream of the GFP coding sequence. The no-uORF version of the SAC51 5′ leader was generated by PCR from that of the 35S-SAC51 5′-GUS T-DNA and similarly cloned into pSI020.

These plasmids were digested with XbaI to generate linear DNA templates for transcription and transcribed using T7 RNA polymerase in the presence of Ribo m7G cap analog (Promega). The resulting capped RNAs were translated in wheat germ extract with Transcend biotinylated lysine-tRNA (Promega) in the presence or absence of thermospermine according to the manufacturer’s instructions.

In vitro-translated proteins were subjected to SDS–polyacrylamide gel electrophoresis, transferred to PVDF membrane, and detected using Transcend non-radioactive translation detection systems (Promega). The gel images were visualized using a LAS-4000 mini luminescent imaging analyzer (Fujifilm).

All statistical analyses were performed using the EZR software (Kanda, 2013), which is a graphical user interface for R (The R Foundation for Statistical Computing). Significance of differences between groups was estimated by one-way or two-way ANOVA with Tukey-Kramer multiple comparisons test.

There are six uORFs in the 5′ leader region of the SAC51 mRNA, including the sixth uORF, which presumptively starts from an in-frame start codon, namely, the second methionine codon of the fourth uORF (Yamamoto and Takahashi, 2017). These uORFs are encoded in the second and third exons of SAC51. To confirm whether more upstream regions are responsive to thermospermine or not, we generated transgenic lines carrying the GUS reporter gene under the control of the SAC51 promoter or that followed by the first exon and intron with its splice acceptor site (Figure 1A) and examined the GUS activity. The results showed that the SAC51 promoter directed the GUS expression sharply to vascular cells (Figure 1B) and neither only the promoter nor the promoter followed by the first exon and intron was responsive to thermospermine (Figure 1C). However, when the SAC51 upstream regions containing the whole 5′ leader sequence was used, weak GUS expression was detected in additional tissues to the vasculature (Figure 1B) and the GUS activity was increased by 24-h treatment with thermospermine (Figure 1C), as described previously (Kakehi et al., 2008). The sac51-d mutant construct in which the fourth uORF contains a premature stop codon (Figure 1A) shows much higher GUS expression than the wild-type construct (Imai et al., 2006) but still retained the responsiveness to thermospermine (Figure 1C).

To address which uORFs of the SAC51 mRNA are involved in the response to thermospermine, we generated transgenic lines that carry the GUS fusion constructs with the SAC51 promoter and 5′ leader containing a base substitution in the start codon of each uORF (Figure 2A) and examined the effect of thermospermine on the GUS activity. All constructs in which one of the uORF start codons is mutated were shown to retain the response to thermospermine. As also shown by GUS staining (Figure 2B), the results revealed that the basal GUS activity was rather reduced in uORF1, uORF2, uORF3, and uORF5-disruption constructs and increased in uORF4 and uORF6-disruption constructs compared with that in the wild-type construct, suggesting that uORF4 and uORF6 are inhibitory but uORF1, uORF2, uORF3, and uORF5 are stimulatory to the main ORF translation. We confirmed by RT-PCR experiments that the relative ratio of the GUS activity to the GUS transcript level was reduced in transgenic lines with uORF1, uORF2, uORF3, and uORF5-disruption constructs (Figure 2C). Disruption of uORF4 or uORF6 appeared to reduce the responsiveness to thermospermine. Our previous study suggests the importance of this uORF6 because the 5′ leaders of both SAC51 and SACL1 contain this in-frame uORF and are responsive to thermospermine but those of SACL2 and SACL3 don’t contain it and show no response to thermospermine (Yamamoto and Takahashi, 2017). We thus disrupted start codons of both uORF4 and uORF6 and found that this mutant construct resulted in no response to thermospermine (Figure 2A). Disruption of all of six start codons of the uORFs also caused no response to thermospermine but the construct with the loss of all but the uORF6 start codon showed the response (Figure 2A). These results collectively suggest the requirement for at least one of uORF4 and uORF6 in the response to thermospermine.

We also examined the response of the SAC51 5′ leader region to thermospermine under the control of the constitutive cauliflower mosaic virus (CaMV) 35S promoter. Although the 35S promoter is not responsive to thermospermine, insertion of a genomic or cDNA fragment of the SAC51 5′ leader region between the promoter and the GUS reporter gene conferred the response to thermospermine in the GUS activity (Figure 3). Furthermore, the mutant cDNA construct in which all but uORF6 is disrupted was responsive to thermospermine whereas the 5′ fragment containing the minimum uORF6 alone resulted in no response (Figure 3).

Because the uORF4 of SAC51 is widely conserved in SAC51 family genes of different plant species, we examined whether 5′ leader regions of these mRNAs are responsive to thermospermine or not. The 5′ regions were cloned from broccoli, soybean, poplar, and rice genomic DNA, inserted between the 35S promoter and the GUS gene, and introduced into Arabidopsis. The deduced amino acid sequences of the conserved uORFs, some of which contain in-frame ATG codons, namely, in-frame uORFs, are aligned in Figure 4A, and their phylogenetic relationships are shown in Figure 4B. We detected the response to thermospermine in some constructs including BoSACL1, OsSACL3A, OsSACL3C, and GmSACL3, but not in others including OsSACL2, PtSACL2, GmSACL2, and BoSACL3 (Figure 4C).

We finally tested whether or not the response of the SAC51 transcript to thermospermine can be reproduced in vitro. The cDNA fragment of the SAC51 5′ leader region was fused to the GFP reporter gene (Figure 5A), transcribed in vitro, and then translated by using a wheat germ extract translation system. Detection of the chemiluminescent-labeled GFP protein showed that the efficiency of GFP translation was reduced by adding the SAC51 5′ leader sequence of both wild-type and no-uORF versions to the GFP transcript and further reduced by increasing the amount of the transcript in the translation reaction (Figure 5B). Addition of thermospermine to the translation reaction mixture had no effect on the GFP production (Figure 5C). We found, however, that pretreatment of the 5′-GFP fusion transcript of both wild-type and no-uORF versions with heat at 65°C for 10 min effectively increased the translation efficiency (Figure 5D).