Adaptation of iCLIP to plants determines the binding landscape of the clock-regulated RNA-binding protein AtGRP7

Background Functions for RNA-binding proteins in orchestrating plant development and environmental responses are well established. However, the lack of a genome-wide view of their in vivo binding targets and binding landscapes represents a gap in understanding the mode of action of plant RNA-binding proteins. Here, we adapt individual nucleotide resolution crosslinking and immunoprecipitation (iCLIP) genome-wide to determine the binding repertoire of the circadian clock-regulated Arabidopsis thaliana glycine-rich RNA-binding protein AtGRP7. Results iCLIP identifies 858 transcripts with significantly enriched crosslink sites in plants expressing AtGRP7-GFP that are absent in plants expressing an RNA-binding-dead AtGRP7 variant or GFP alone. To independently validate the targets, we performed RNA immunoprecipitation (RIP)-sequencing of AtGRP7-GFP plants subjected to formaldehyde fixation. Of the iCLIP targets, 452 were also identified by RIP-seq and represent a set of high-confidence binders. AtGRP7 can bind to all transcript regions, with a preference for 3′ untranslated regions. In the vicinity of crosslink sites, U/C-rich motifs are overrepresented. Cross-referencing the targets against transcriptome changes in AtGRP7 loss-of-function mutants or AtGRP7-overexpressing plants reveals a predominantly negative effect of AtGRP7 on its targets. In particular, elevated AtGRP7 levels lead to damping of circadian oscillations of transcripts, including DORMANCY/AUXIN ASSOCIATED FAMILY PROTEIN2 and CCR-LIKE. Furthermore, several targets show changes in alternative splicing or polyadenylation in response to altered AtGRP7 levels. Conclusions We have established iCLIP for plants to identify target transcripts of the RNA-binding protein AtGRP7. This paves the way to investigate the dynamics of posttranscriptional networks in response to exogenous and endogenous cues. Electronic supplementary material The online version of this article (doi:10.1186/s13059-017-1332-x) contains supplementary material, which is available to authorized users.


Figure S1
Figure S1 Monitoring for UV stress upon UV crosslinking Two-week-old seedlings grown in 12 h light-12 h dark cycles at 20 o C were exposed to 500 mJ/cm 2 UV-C light (254 nm) and harvested immediately after the irradiation or transferred back to the growth chamber and harvested after 30 min (30') or 60 min (60').
The UV marker transcripts MC8 and HYH were detected by RT-PCR (A) and qRT-PCR (B).
Significance was determined in three biological replicates by a Mann-Whitney Test (significance indicated by asterisk, p<0.1). PP2A served as control. n.s., not significant. C) Immunoblot analysis of AtGRP7-GFP fusion protein detected with a-GFP antibody (top).
Amido black staining of the membrane served as loading control (bottom). D) Photographs of three representative seedlings before UV crosslinking (day 0) and on days 2, 5, 6, and 8 after UV crosslinking. Arrows indicate newly formed leaves. 3

Figure S2
Figure S2 iCLIP of AtGRP7-GFP A) Scheme of iCLIP strategy. Arabidopsis thaliana seedlings were irradiated with 254 nm UV-C light, the cells were lysed and the protein of interest was immunoprecipitated by its GFP tag. An adapter was ligated to the 3' end of the co-precipitated RNA, whereas the 5' end was radioactively labeled. The precipitated complexes were separated on an SDS-PAGE, transferred onto a membrane and visualized by autoradiography. The RNA was isolated from the membrane and the crosslinked protein was digested by Proteinase K, leaving a short peptide behind. The reverse transcriptase was stopped at this position. cDNAs were size separated, eluted from the gel in three fractions (high H, medium M, low L) and free primers were removed. The cDNAs of the separate fractions were circularized and an oligonucleotide was annealed to generate a restriction site. After linearization, both ends of the cDNA contained adapters. These were used for PCR amplification and high throughput sequencing. B) Representative autoradiograms of the RNA-protein complexes precipitated from  Grey nucleotides represent the synthetic oligonucleotide 7-UTR_WT used as binding substrate for recombinant AtGRP7 in electrophoretic mobility shift assays [2]. Red nucleotides represent point mutations in the corresponding oligonucleotide 7-UTR_G 4 mut that impaired in vitro binding of recombinant AtGRP7 in electrophoretic mobility shift assays [2]. Light blue boxes correspond to significant crosslink sites identified in this study. The dark blue box represents a minimal binding sequence for recombinant AtGRP7 determined by 14 fluorescence correlation spectroscopy [3]. The black boxes represent the significant MEME 3´UTR motifs at LL36 (identified using the FIMO tool from the MEME suite; cf. Additional file 1: Fig. S6). The green boxes represent the enriched pentamers (cf . Table S6) Grey nucleotides represent the synthetic oligonucleotide 7-intron_WT used as binding substrate for recombinant AtGRP7 in electrophoretic mobility shift assays [2]. Red nucleotides represent point mutations in the corresponding oligonucleotide 7-intron_G 6 mut that impaired in vitro binding of recombinant AtGRP7 [2]. Blue nucleotides correspond to mutations in the synthetic oligonucleotides 7-intron_mut4, 7-intron_mut6, and 7-intron_mut13 that impaired binding of recombinant AtGRP7 in fluorescence correlation spectroscopy [4]. Light blue boxes correspond to significant crosslink sites identified in this study. The black boxes represent the significant MEME intron motif at LL36 (identified using the FIMO tool from the MEME suite; cf. Additional file 1: Fig. S6).     AtGRP7-ox, E grp7-1 8i), as well as of iCLIP targets with a significant differential AtGRP7-ox, F grp7-1 8i), RIP targets with a significant differential expression AtGRP7-ox, G grp7-1 8i) and high-confidence binders identified by both iCLIP and RIP (D , AtGRP7-ox, H grp7-1 8i). The distribution of all identified DEGs in RNA-seq (A, E) was tested pairwise against all target groups. The resulting p-value is displayed accordingly.   At LL24 the intron retention of TCH3 (A) in grp7-1 8i is not significantly different from the wild type. FNR2 (B) shows enhanced intron retention in AtGRP7-ox compared to the wild type. A GDSL-Like Lipase/Acylhydrolase superfamily protein (C) also retains intron 1 more often in AtGRP7-ox than in the wild type. In contrast to LL36 the increase of the fully spliced isoform of Transducin/WD40 repeat-like superfamily protein (D) in AtGRP7-ox does not meet our criteria of ΔPSI > 0.1 (0.095), but is statistically significant. Again, this is accompanied by elevated steady-state abundance in AtGRP7-ox. AIF3 (E) shows higher levels of intron retention in AtGRP7-ox than in the wild type, whereas an exon of FAX4 (F) is skipped more 27 often in AtGRP7-ox than in the wild type. A thioredoxin superfamily protein (G) shows a differential usage of an alternative 5' splice site in AtGRP7-ox.
Events were validated by RT-PCR in two independent biological replicates taken at LL24, shown are representative results from grp7-1 8i, Col-2 and AtGRP7-ox (7-ox) samples (left to right). Gene models are shown for the relevant isoforms with thin bars representing the 5'UTR (left) and 3'UTR (right), thick bars denoting exons and lines depicting introns. Arrows indicate the position of the PCR primers. The bar plots represent the expected relative amount of the indicated isoform calculated from the RNA-seq data using SUPPA in three independent biological replicates. The RNA coverage plots visualize the read coverage for the respective samples at LL24. Information on the transcripts is taken from TAIR 10 and ARAPORT [6].