arising from: P. Zhu etal. Nature https://doi.org/10.1038/s41586-021-04062-5 (2021)
In over-wintering annuals of Arabidopsis thaliana grown in temperate regions, prolonged cold exposure in winter, through the physiological process of vernalization, represses the expression of the potent floral repressor FLOWERING LOCUS C (FLC) to enable the transition to flowering in spring1,2. Recently, Zhu et al.3 reported that cold induced nuclear condensation of FRIGIDA (FRI) for FLC repression and that cold-induced antisense RNA COOLAIR promoted FRI condensation during prolonged cold exposure. Here we report that the cold-induced formation of nuclear FRI condensates is independent of COOLAIR.
Before exposure to cold, FRI activates FLC expression, and long-term continuous winter cold exposure (typically lasting more than a month) results in FLC repression in FRI-bearing winter annuals grown at high latitudes1,4. Prolonged cold exposure induces the expression of COOLAIR, a group of non-coding antisense RNAs initiating from a region downstream of the 3′ end of FLC that is composed of proximally polyadenylated class I and distally polyadenylated class II transcripts5. COOLAIR expression reaches a high level after around 3 weeks of cold exposure and subsequently declines under constant cold temperature6. Using a CRISPR–Cas9 system, we previously constructed several lines in which a large part of the core COOLAIR promoter region was removed7, resulting in the elimination of both class I and class II COOLAIR transcripts before cold exposure (Fig. 1a,b and Extended Data Fig. 1a,b). Furthermore, consistent with a recent study8, cold induction of COOLAIR expression was eliminated in these core promoter deletion lines (Fig. 1b), partly because the cis-acting cold-responsive elements located in the promoter region have been removed. To examine the role of COOLAIR in FRI condensation, we introduced a functional FRI-GFP into two lines in which the COOLAIR promoter was deleted—ΔCOOLAIR-1 and ΔCOOLAIR-2—in the rapid-cycling accession Col-0 (bearing a loss-of-function fri allele9) by genetic transformation. Subsequently, independent FRI-GFPΔCOOLAIR lines (numbers 2, 5 and 7) were backcrossed to Col-0 and ΔCOOLAIR-1 or ΔCOOLAIR-2, respectively, resulting in F1 progeny of FRI-GFPΔCOOLAIR−/− and FRI-GFPΔCOOLAIR+/−. In these lines, FRI–GFP is fully functional and acts to activate FLC expression before cold exposure (Extended Data Fig. 1c–f).
We next determined whether the loss of COOLAIR expression might reduce nuclear FRI condensation. We measured the fluorescence intensity of FRI–GFP in the root tips and the size and number of FRI–GFP condensates in root tip nuclei of cold-treated FRI-GFPΔCOOLAIR+/− and FRI-GFPΔCOOLAIR−/− seedlings and found that there was no difference between these two genotypes (Fig. 1e–i). Furthermore, we crossed a FRI-GFP line3 to both ΔCOOLAIR-1 and ΔCOOLAIR-2 and obtained homozygous FRI-GFPΔCOOLAIR lines. Subsequently, we measured the size and number of FRI–GFP condensates in root tip nuclei in these lines after cold exposure and found that there was no statistically significant difference (Fig. 2a–f). Together, these results show that the cold-induced formation of nuclear FRI condensates is independent of COOLAIR expression, given that, before and during cold exposure, COOLAIR expression (including class I and class II transcripts) was eliminated in both the ΔCOOLAIR-1 and ΔCOOLAIR-2 lines.
Zhu et al.3 reported that, in the cold, the FRI protein enriched class II.ii COOLAIR transcripts and that the FRI–class II.ii interaction was closely connected with cold-induced FRI condensation. Using a transgenic FLC terminator exchange (TEX) line in which the COOLAIR promoter was replaced with an RBCS3B (encoding Rubisco small subunit 3B) terminator, Zhu et al.3 found that the size and number per nucleus of FRI–GFP condensates were reduced in the TEX line compared with those in the non-transgenic background (Col-0). In the ΔCOOLAIR lines that we used in this study, the class II transcripts are eliminated before cold exposure and in the cold. Thus, we conclude that the FRI-class II.ii interaction and COOLAIR expression are not involved in cold-induced FRI condensation. The cause for the discrepancy between the two studies is unclear.
In addition to the COOLAIR transcripts initiated downstream of the 3′ end of FLC, there are other antisense transcripts (ASTs) initiated within the FLC locus (see, for example, ref. 10). We examined these ASTs in FRI-Col (a reference winter-annual line11) and FRIΔCOOLAIR seedlings, and found that they were at low levels before cold exposure (Fig. 2g). After cold exposure for 3 days, the expression of ASTs in all three examined regions declined in FRI-Col, and was apparently reduced in two examined regions in the FRIΔCOOLAIR seedlings (Fig. 2g); cold exposure for 14 days strongly suppressed the expression of ASTs in both the FRI-Col and FRIΔCOOLAIR lines (Fig. 2h). Thus, in contrast to COOLAIR, the expression of ASTs is repressed along the early phase of long-term cold exposure or vernalization. The function of ASTs in the vernalization-mediated FLC repression remains to be seen.
Cold induction of COOLAIR expression in the early phase of vernalization was reported to mediate FLC repression5,6. We measured the levels of FLC transcripts (both spliced and unspliced) in the cold-treated FRIΔCOOLAIR seedlings, and found that loss of COOLAIR expression in either FRIΔCOOLAIR-1 or FRIΔCOOLAIR-2 had no effect on the progression of transcriptional shutdown of FLC during cold exposure or on post-cold stable silencing of FLC (Fig. 1c,d and Extended Data Fig. 1g), consistent with observations in a recent study8. Notably, in our vernalization study, like several other studies reporting a role of COOLAIR for FLC repression5,6, seedlings were exposed to a constant cold temperature, the mechanisms uncovered through which may not fully represent FLC regulation by winter cold in the fields with fluctuating cold temperatures.
In summary, our study shows that the cold-induced formation of nuclear FRI condensates is independent of COOLAIR. Moreover, our vernalization study with constant cold temperature shows that COOLAIR is not involved in FLC repression by prolonged cold exposure. Thus, more in-depth experiments would be required to resolve the role of COOLAIR in vernalization.
Table of Contents
Arabidopsis thaliana FRI-Col, Col-0, ΔCOOLAIR-1 and FRIΔCOOLAIR-1 were described previously7. Treatment of the seedlings with constant cold and quantification of the expression of genes of interest using quantitative PCR were performed as previously described12. COOLAIR expression was examined by semiquantitative PCR, after reverse transcription using transcript-specific primers (5′-TGGTTGTTATTTGGTGGTGTGAA-3′ for class I; and 5′-GCCCGACGAAGAAAAAGTAG-3′ for class II10). A list of the PCR primers is provided in Extended Data Table 1.
FRIpro:FRI-GFP was constructed by cloning a 4.8 kb genomic FRI fragment (2.5 kb promoter plus the 2.3 kb entire coding region without the stop codon) upstream of the GFP-coding region in the binary vector pMDC11013. Microscopy analysis and image quantification of nuclear FRI–GFP condensates were performed as follows. Root tips of the seedling samples were imaged using the Zeiss LSM900 confocal microscope with a ×40/1 NA water objective and an Airyscan detector of GaAsP-PMT. GFP fluorescence was excited at a wavelength of 488 nm (argon ion laser and laser power 8.0%), and detected at 490–620 nm in lambda mode. All of the images were obtained with a pixel size of 0.119 μm, and exported using the ZEN3.1 software (Zeiss) for quantitative analysis. The number of spots (with an area of larger than 0.1 μm2) per nucleus and the spot area in the F1 seedlings were obtained by outlining the spots using Graphics from ZEN3.1. Similarly, the fluorescent spots with an area of larger than 0.05 μm2 were scored in the root tips of the seedlings bearing the homozygous FRI-GFP3.
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this Article.