HEAT SHOCK TRANSCRIPTION FACTOR B2b acts as a transcriptional repressor of VIN3, a gene induced by long-term cold for flowering

To identify the causative mutation, hov1 was crossed with Ler for positional cloning. A total of 156 F2 plants with enhanced GUS signals in the leaves were selected for mapping analysis. We mapped the mutation to the 590 kilobase pair interval on chromosome 4, which contained 142 genes (Supplementary Fig. S1). The genomes of hov1 and parental − 0.2 kb pVIN3_U_I::GUS were sequenced using the Illumina sequencing method for comparison. Analysis of the sequence data revealed nine potentially disruptive point mutations, including one mutation within the At4g11660 gene (G to A, causing a nonsense mutation from Trp89 to the stop codon) (Fig. 1d). At4g11660 encodes the class B heat shock transcription factor, HEAT SHOCK TRANSCRIPTION FACTOR B2b (HsfB2b).

Finally, we introduced HsfB2b::HsfB2b-myc into the hov1 mutant to determine whether HsfB2b can rescue the hov1 mutation. The transcript levels of HsfB2b were found to be overexpressed in all transgenic lines we obtained (Figs. 2c, 6a, and d). Here, we used two representative lines of HsfB2b:HsfB2b-myc hov1, #1, and #2. As expected, the phenotype of the GUS signal in hov1 was complemented by HsfB2b::HsfB2b-myc, such that the GUS signal was barely detected after 3 d of cold exposure (Fig. 2b). Moreover, the endogenous VIN3 transcript levels in both transgenic lines were lower than those in Col-0, as well as the hov1 mutant, whereas HsfB2b transcript levels in the transgenic lines were higher than those in Col-0 (Fig. 2c and d). Taken together, our results indicate that HsfB2b is a causative gene that reduces VIN3 level in hov1 and acts as a transcriptional repressor of VIN3.

In a previous report, VIN3 was found to gradually increase by long-term cold exposure from the first day of cold treatment¹³. Thus, we determined whether HsfB2b affects VIN3 expression during the initial stage of vernalization treatment (Fig. 2e and f). As shown, hsfb2b caused strong derepression of VIN3 from the initial phase, and the effect was strongest after 3 h of cold treatment. Of note, the derepression effect of hsfb2b is stronger at short-term cold and 40VT1 (1 d at room temperature after returning from 40 days of vernalization treatment) than during vernalization treatment (Fig. 2e and f).

To confirm whether the retarded migration of HsfB2b protein is due to the phosphorylation, phosphatase assay was conducted using total protein extracted from vernalized or non-vernalized pHsfB2b::HsfB2b-myc seedlings. Subsequent immunoblot assay showed that the retarded migration of HsfB2b-myc from vernalized seedlings was abolished by phosphatase treatment, while migration of HsfB2b-myc from non-vernalized seedling was not changed by the treatment (Supplementary Fig. S3). These results indicate that retarded migration of HsfB2b-myc from vernalized seedlings was due to the phosphorylation.

As the cellular localization of other Hsf is changed by post-translational modification³², we checked whether cold treatment can change that of HsfB2b (Fig. 3c and d). Using the pHsfB2b::HsfB2b-eGFP transgenic lines, we observed the root tissue before and after 5 days of cold. In both cases, GFP signals were observed in the nucleus, indicating that neither cold treatment nor protein modification altered the cellular localization of HsfB2b. This result is consistent with the fact that the protein sequence of HsfB2b has a nuclear localization signal (NLS), but lacks a nuclear export signal (NES) motif²⁵. Taken together, HsfB2b is neither transcriptionally induced nor is the subcellular localization of the proteins altered by vernalization treatment.

VIN3 expression is reported to show a circadian rhythm and HsfB2b acts as a negative regulator of the circadian clock regulator, PSEUDO RESPONSE REGULATOR7 (PRR7)^15,16,31. Thus, we verified whether the VIN3 rhythm was affected by hsfb2b during cold treatment. Although the amplitude of the circadian rhythm was increased by hsfb2b mutation due to the increase in VIN3 level, the rhythmic pattern was not significantly different (Fig. 3e and f). Therefore, HsfB2b seems to constitutively repress VIN3, and this repression is independent of the HsfB2b-regulated circadian clock.

We proceeded to assess whether HsfB2b bound to HSE_VIN3 using yeast one-hybrid assay (Fig. 4b). Among the nine Hsf proteins analyzed, HsfA1, HsfA6a, HsfB1, HsfB2b, and HsfC1 were found to interact with HSE_VIN3. To confirm in planta binding, we also performed chromatin immunoprecipitation-qPCR using transgenic pHsfB2B::HsfB2b-eGFP, grown under long days without cold treatment (Fig. 4a and c). HsfB2b-eGFP proteins were found to be enriched in the promoter region near the HSE location even without cold treatment, which is consistent with the elevated VIN3 level in hsfb2b. We also conducted electrophoretic mobility-shift assay (EMSA) to assess whether HsfB2b specifically binds to HSE_VIN3 DNA element in vitro. The purified recombinant protein, MBP-HsfB2b^DBD, DNA binding domain of HsfB2b fused with maltose binding protein (MBP), from E. coli indeed binds the HSE_VIN3 probe but fails to bind the mutated version of HSE_VIN3 probe (Fig. 4d). Taken together, these data strongly support the hypothesis that HsfB2b directly regulates VIN3 repression.

We proceeded to verify whether HsfB2b overexpression caused any changes in the vernalization response. Briefly, we introduced pHsfB2b::HsfB2b-eGFP into FRI Col by transformation. As expected, all 10 transgenic lines showed overexpression of HsfB2b (3 ~ 10 folds) based on the level after 20 days of vernalization (20V) (Fig. 6d). In these transgenic lines, VIN3 levels were lower than those in both FRI Col and hsfb2b FRI after 20V, which supports the hypothesis that HsfB2b overexpression causes the repression of VIN3 in FRI Col plants (Fig. 6d). Consistent with the fact that VIN3 is required for the suppression of FLC⁸, the FLC levels in pHsfB2b::HsfB2b-eGFP FRI lines were slightly higher than those in FRI Col after 20V (Fig. 6d). Finally, the flowering time of the transgenic lines, pHsfB2b::HsfB2b-eGFP FRI, was less accelerated than that of both FRI Col and hsfb2b FRI by 20V (Fig. 6e and f). Therefore, in contrast to hsfb2b mutation, HsfB2b overexpression causes defects in the vernalization response under normal growth conditions.

All Arabidopsis thaliana lines used were in the Columbia (Col-0) background except Ler ecotype used to generate mapping population for map-based gene cloning. The wild-type, Col:FRI^Sf2 (FRI Col) have been previously described³⁴. hsfb1, hsfb2b, and hsfb1 hsfb2b mutants have been previously described²⁹.

To produce pHsfB2b::HsfB2b-eGFP construct, the genomic sequences including 2624 bp upstream of the promoter and the whole coding sequence of HsfB2b were amplified by polymerase chain reaction. The fragment was cloned into pCR2.1-TOPO vector, then fused in-frame to pCAMBIA1300 vector containing eGFP. The construct was transformed into the hov1 mutant. To produce pHsfB2b::Hsfb2b-myc, the 3 kb HsfB2b promoter and the HsfB2b-coding sequence were amplified and fused in-frame to pPZP221 vector containing 4xmyc (EQKLISEEDL). The construct was transformed into the indicated lines using Agrobacterium (Agrobacterium tumefaciens)-mediated Arabidopsis floral dip method⁵⁵.

The plants were grown under 16 h/8 h light/dark cycle (long day) or 8 h/16 h light/dark cycle (short day) (22 °C/20 °C) in a controlled growth room with cool white fluorescent lights (125 μmol m⁻² s⁻¹). Vernalization treatments were done as previously described³⁶. Nonvernalized seedlings were grown for 11 d. For 10V, 20V, 40V treatments, seedlings were grown for 10, 9, 7 d under short days respectively after germination, then transferred to vernalization chamber at 4 °C. After vernalization treatment, seedlings were sampled or transplanted to the soil. Flowering time was measured by counting the number of rosette leaves when the first flower opened using at least 20 plants.

Devernalization treatment was performed following the previously described method with some modifications⁵⁶. Seeds were sown on a round plate containing half-strength Murashige and Skoog medium with 1% sucrose in 1% agar. The plates were wrapped with aluminum foils for complete darkness and placed at 4 °C for vernalization treatment. After 40 days of vernalization treatment, plates were transferred to heating incubator (30 °C) for additional 7 days, or placed at room temperature. Then, aluminum foils were peeled from the plates, and the plates were placed at room temperature. After 10 days of growth, seedlings were sampled or transplanted to the soil for further growth. All the plant materials and methods used in the current study were carried out following relevant institutional, national, and international guidelines and legislation.

EMS mutagenesis was performed as previously described⁵⁷. For the positional cloning of the causative gene of hov1, F2 progenies were obtained by crossing hov1 to Ler. Mapping procedure was followed using 135 GUS-hypersensitive F2 plants and molecular makers described as previous reports^58,59. After rough mapping, the genomes of hov1 and the parental − 0.2 kb pVIN3_U_I::GUS were sequenced and compared by illumina Hiseq2000 platform (illumina) sequencing to find mutant-specific SNPs in hov1 using BGI services.

GUS staining was done following the standard methods that have been previously described⁶⁰. Photographs were taken with a USB digital‐microscope Dimis‐M (Siwon Optical Technology, South Korea).

For real-time quantitative PCR, total RNA was isolated using TRIzol solution (Sigma). Four micrograms of total RNA were treated with recombinant DNase I (TaKaRa, 2270A) to eliminate genomic DNA. cDNA was generated using the RNA with reverse transcriptase (Thermo scientific, EP0441) and oligo(dT). Quantitative PCR was performed using the 2 × SYBR Green SuperMix (Bio-Rad 170-8882) and monitored by the CFX96 real-time PCR detection system. The relative transcript level of each gene was determined by normalization of the resulting expression levels compared to that of UBC. The primer sequences used in real-time RT-PCR analyses were shown in Supplementary Table. S1

For immunoblot assay, the seedlings of pHsfB2b::HsfB2b-eGFP were harvested at each time point. Total proteins were prepared from 100 mg of harvested samples in protein extraction buffer (50 mM Tris–Cl pH 7.5, 150 mM NaCl, 10 mM MgCl₂, 1 mM ethylenediaminetetraacetic acid (EDTA), 1% Triton X-100, 1 mM phenylmethylsulfonyl fluoride (PMSF), 1 mM 1,4-Dithiothreitol (DTT), 1Χ complete Mini, and EDTA-free protease inhibitor cocktail (Roche). Total proteins were separated by sodium-dodecyl sulfate (SDS)-PAGE. For phosphatase assay, total proteins were treated with or without alkaline phosphatase (Thermo scientific, EF0652) for 1 h at 37 °C, then separated by SDS-PAGE. The proteins were transferred to PVDF membranes (Amersham Biosciences) and probed with anti-GFP (Clontech, JL-8, 1:10,000 dilution) or anti-myc (Santa Cruz Biotechnology, sc-40, 1:10,000 dilution) antibodies overnight at 4 ℃. The samples were then probed with horseradish peroxidase-conjugated anti-mouse IgG (Cell Signaling, #7076, 1:10,000 dilution) antibodies at room temperature. The signals were detected using ImageQuant LAS 4000 (GE Healthcare) with WesternBrightTM Sirius ECL solution (Advansta).

For microscopic observations, 5-day-old pHsfB2b::HsfB2b-eGFP seedlings with or without 5 days of additional cold treatment were prepared. Seedlings were pre-stained with propidium iodide (PI), mounted on glass slides, and observed using confocal microscopy (LSM700, Zeiss) following the manufacturer's instructions.

The promoter sequences from plant species were downloaded from GBrowse at Phytozome (phytozome-next.jgi.doe.gov↗). The following VIN3 loci (At5g57380) were identified using BLAST Search: Arabidopsis lyrata (AL8G33360), Boechera stricta (Bostr.26833s0518) and Capsella rubella (Carub.0008s1790). The sequences were processed and aligned in T-coffee (tcoffee.crg.eu).

Yeast one-hybrid assay was performed following the previously described method with some modifications¹⁵. For the reporter constructs used in the Y1H analysis, four tandem repeats containing HSE_VIN3 (5′-TTAGAAACATCTAGAAAAAACAAA-3′) were cloned into the pHisi vector. For the effector, the coding sequences of HsfA1a, HsfA2, HsfA4a, HsfA6a, HsfA8, HsfB1, HsfB2b and HsfC1 were cloned in-frame with the sequences of the GAL4 activation domain into pGADT7. The Y1H assay was performed following the manufacturer’s instructions. In brief, the reporter construct and effector construct were transformed into yeast strain YM4271. The yeast cells were spotted on synthetic define (SD) medium lacking Leu, Ura, and His, with or without 5-mM 3-amino-1,2,4-triazole (3-AT).

Coding sequence encoding DNA binding domain of HsfB2b was fused to pMAL-c2 vector. The MBP and MBP-HsfB2b^DBD proteins were expressed in Escherichia coli BL21 strain according to the manufacturer’s instructions using the pMAL Protein Fusion and Purification System (#E8200; New England BioLabs) and purified using MBPtrap HP column (Cytiva) attached to ÄKTA FPLC system (Cytiva). The Cy5-labeled probes (HSE, 5′-Cy5- TTTCCTCCTTAGAAACATCTAGAAAAAACAAAAGGAGAGA-3′; mHSE, 5′-Cy5- TTTCCTCCTTAAAAACATTTAAAAAAAACAAAAGGAGAGA -3′) and unlabeled competitors were generated by annealing 40 bp-length oligonucleotides. 5 μM of purified proteins and 100 nM of Cy5-labeled probe were incubated at room temperature in binding buffer (10 mM Tris–HCl (pH 7.5), 50 mM NaCl, 1 mM EDTA, 5% glycerol and 5 mM DTT). For competition assay, 100-fold molar excess of each competitor was added to the reaction mixture before incubation. The reaction mixtures were resolved by electrophoresis through 6% polyacrylamide gel in 0.5X Tris–borate EDTA buffer at 100 V. The Cy5 signals were detected using WSE-6200H LuminoGraph II (ATTO).

Approximately 4 g of whole Arabidopsis seedlings were collected and cross-linked using 1% (v/v) formaldehyde for 10 min and quenched by 0.125 M glycine for 5 min under vacuum. Seedlings were rinsed with distilled water, frozen in liquid nitrogen, and grounded to fine powder. The powder was resuspended in Nuclei Isolation Buffer [1 M hexylene glycol, 20 mM PIPES-KOH (pH 7.6), 10 mM MgCl₂, 15 mM NaCl, 1 mm EGTA, 1 mM PMSF, complete protease inhibitor mixture tablets (Roche)], and Arabidopsis nuclei were isolated by centrifugation, lysed by Nuclei Lysis Buffer [50 mM TRIS–HCl (pH 7.4), 150 mM NaCl, 1% Triton X-100, 1% SDS], and sonicated using a Branson sonifier to shear the DNA. Sheared chromatin solution was diluted tenfold with a ChIP Dilution Buffer [50 mM TRIS–HCl (pH 7.4), 150 mM NaCl, 1% Triton X-100, 1 mM EDTA]. The beads, chromatins and GFP-Trap A beads (gta-20, ChromoTek, Planegg, Germany) or Binding control agarose (bab-20) were mixed and incubated for overnight at 4 °C. Beads were washed with ChIP dilution buffer for 4 times and DNA extraction was performed using Chelex 100 resin following the manufacturer’s instruction. qPCR analysis was performed using 1% input and immunoprecipitated DNA.

The Arabidopsis Genome Initiative locus identifiers for the genes discussed in this paper are as follows: VIN3 (At5g57380), FLC (At5g10140), FRI (At4g00650), HsfA1a (At4g17750), HsfA2 (At2g26150), HsfA3 (At5g03720), HsfA4a (At4g18880), HsfA6a (At5g43840), HsfA8 (At1g67970), HsfB1 (At4g36990), HsfB2a (At5g62020), HsfB2b (At4g11660), HsfC1 (At3g24520), CCA1 (At2g46830), LHY (At1g01060), and PP2A (At1g13320).

Supplementary Information.