GENOMES UNCOUPLED PROTEIN1 binds to plastid RNAs and promotes their maturation

Abstract

Plastid biogenesis and the coordination of plastid and nuclear genome expression through anterograde and retrograde signaling are essential for plant development. GENOMES UNCOUPLED1 (GUN1) plays a central role in retrograde signaling during early plant development. The putative function of GUN1 has been extensively studied, but its molecular function remains controversial. Here, we evaluate published transcriptome data and generate our own data from gun1 mutants grown under signaling-relevant conditions to show that editing and splicing are not relevant for GUN1-dependent retrograde signaling. Our study of the plastid (post)transcriptome of gun1 seedlings with white and pale cotyledons demonstrates that GUN1 deficiency significantly alters the entire plastid transcriptome. By combining this result with a pentatricopeptide repeat code-based prediction and experimental validation by RNA immunoprecipitation experiments, we identified several putative targets of GUN1, including tRNAs and RNAs derived from ycf1.2, rpoC1, and rpoC2 and the ndhH-ndhA-ndhI-ndhG-ndhE-psaC-ndhD gene cluster. The absence of plastid rRNAs and the significant reduction of almost all plastid transcripts in white gun1 mutants account for the cotyledon phenotype. Our study provides evidence for RNA binding and maturation as the long-sought molecular function of GUN1 and resolves long-standing controversies. We anticipate that our findings will serve as a basis for subsequent studies on mechanisms of plastid gene expression and will help to elucidate the function of GUN1 in retrograde signaling.

Keywords: GUN1; MORF2; RIP-seq; RNA binding protein; plastid (post)transcriptome; retrograde signaling.

Figure 1

GUN1 does not play a significant role in plastid RNA editing or splicing during retrograde signaling. (A) RNA editing efficiencies of 4-day-old Col-0 and gun1-102 seedlings grown on MS and norflurazon (NF) were determined using previously published RNA-seq data (Habermann et al., 2020). These sequencing data were generated to allow for the detection of organellar transcripts. Mean values ± standard deviations were obtained from three independent experiments. Statistically significant differences between Col-0 NF and gun1-102 NF are indicated (post hoc Tukey’s HSD [honestly significant difference] test; ∗P < 0.05 and ∗∗P < 0.01). A graph showing the statistical differences between Col-0 MS and gun1-102 MS can be found in Supplemental Figure 1. The efficiency of editing sites labeled in magenta and turquoise was found to be elevated and reduced, respectively, by Zhao et al. (2019). We also identified an unexpected increase in editing of rpoC1 in both WT and gun1-102 under NF treatment. Our results may vary due to the use of different analysis methods—Sanger sequencing versus lncRNA-seq data analysis—as well as discrepancies in growth media and conditions. Notably, Zhao et al. (2019) cultivated 5-day-old seedlings on MS plates without sucrose, whereas Habermann et al. (2020) used MS plates with 1.5% sucrose. Thus, to account for these variations, we repeated the experiment for selected editing sites in two distinct laboratories as shown in (B) and (C). (B) Col-0 and gun1-102 seedlings were grown in laboratory 1 for 5 days under continuous light conditions as reported by Zhao et al. (2019). The editing efficiency of the selected sites was visualized by Sanger sequencing for two biological replicates. (C) Col-0, gun1-1, and gun1-102 seedlings were grown in laboratory 2 for 5 days under continuous light conditions as reported by Zhao et al. (2019). The editing efficiency of the selected sites was determined by amplicon sequencing. Mean values with their standard deviations are shown. Statistically significant differences between Col-0 and gun1 seedlings are indicated (post hoc Tukey’s HSD test; ∗P < 0.05 and ∗∗P < 0.01). (D) Overexpression of MORF2 does not result in a significant gun phenotype. Steady-state levels of LHCB1.2 transcripts in 5-day-old seedlings grown under NF conditions are shown. Col-0 serves as the WT control for gun1 and sgs3-1 as a control for oeMORF2 (35S:MORF2-YFP) lines. For each genotype, the total RNA was fractionated on a formaldehyde-containing denaturing gel, transferred to a nylon membrane, and probed with [α-³²P]dCTP-labeled complementary DNA (cDNA) fragments specific for the transcripts encoding LHCB1.2. rRNA was visualized by staining the membrane with methylene blue (M.B.) and served as a loading control. Quantification of signals relative to the WT (=100) is provided below each lane. (E) Snapshots of reanalyzed RNA-seq data published by Zhao et al. (2019) and Habermann et al. (2020). The read depths were visualized with the Integrated Genome Browser. Whereas reads from Habermann et al. (2020) are evenly distributed across LHCB1.2, reads generated by Zhao et al. (2019) exhibit a prominent peak of 16 nucleotides (red arrow). The sequence of the peak (5′-GCTACAGAGTCGCAGG-3′) is also present in LHCB1.3 and from the third nucleotide in LHCB1.1. The sequence of this peak coincides with the sequence of the “LHB1.2” forward primer (actually detecting LHCB1.3 in combination with the given reverse primer) used by Zhao et al. (2019) for RT–qPCR.

Figure 2

The nuclear transcriptome of white and marbled gun1 seedlings is significantly affected. (A) Phenotypes of 10-day-old Col-0, gun1-1, gun1-102, and gun1-103 seedlings grown on MS without inhibitor supplementation under 16-h light/8-h dark conditions. Zoomed-in images were taken of white seedlings, denoted by the circles below the overview pictures. The percentages of abnormal seedlings (white and marbled cotyledons) were calculated for three different seed batches. (B) Phenotypes of Col-0, gun1G, gun1M, and gun1W seedlings (derived from gun1-102). (C) Analysis of transcriptome changes in white (gun1W), marbled (gun1M), and green (gun1G) gun1-102 mutant seedlings. The numbers represent genes with at least a two-fold reduction (down) or elevation (up) compared with the Col-0 WT control. (D) Venn diagrams depicting the degree of overlap between the sets of genes whose expression levels were altered at least two-fold in gun1W, gun1M, and gun1G compared with the Col-0 control. (E) Heatmap showing transcript accumulation of genes encoding chlorophyll a/b binding proteins.

Figure 3

Heatmap illustrating the impact of GUN1 deficiency and NF and LIN treatment on plastid-encoded transcripts (Z scores). Low to high expression is represented by the blue to red transition. Note that Z scores are calculated for each individual transcript over the different genotypes. NEP is a single-subunit enzyme, whereas PEP consists of core subunits that are encoded by the plastid genes rpoA, rpoB, rpoC1, and rpoC2 (which are transcribed by NEP) and additional protein factors (sigma factors and polymerase-associated proteins [PAPs]) encoded by the nuclear genome (Borner et al., 2015; Liebers et al., 2018). The general picture has been that only PEP transcribes photosystem I and II genes (psa and psb), most other genes have both NEP and PEP promoters, and NEP alone transcribes a few housekeeping genes (rpoB, accD, ycf2) (Hajdukiewicz et al., 1997). However, more recent analyses have shown that the division of labor between NEPs and PEPs is more complex (Legen et al., 2002; Borner et al., 2015), and no clear conclusion can be drawn about PEP- or NEP-dependent transcription in gun1W: the so-called PEP-dependent genes had lower expression in gun1W than in Col-0, as did the genes transcribed by PEP and NEP, although to a lesser extent. NEP-dependent gene expression was also reduced or in the range of Col-0. The transcriptome changes in lincomycin (LIN)-treated (Xu et al., 2020) and NF-treated (Habermann et al., 2020) seedlings were reanalyzed in the same way as the sequencing data generated for this publication. NEP, nuclear-encoded RNA polymerase; PEP, plastid-encoded RNA polymerase.

Figure 4

GUN1 deficiency has a significant impact on the chloroplast transcriptome. (A) Venn diagrams depicting the degree of overlap between the sets of plastid protein-coding genes whose RNA expression levels were reduced by at least two-fold in gun1W relative to Col-0, as well as in LIN- and NF-treated seedlings compared with Col-0 grown on medium without inhibitor (MS). The transcripts of inverted repeat B have been omitted. Note that for the transcripts downregulated by LIN or NF, the adjusted P value may also be higher than 0.05. (B) RT–qPCR was used to determine expression levels of selected chloroplast transcripts. The results were normalized to the expression of AT4G36800, which encodes a RUB1-conjugating enzyme (RCE1). Expression values are reported relative to the corresponding transcript levels in Col-0, which were set to 1. Mean values ± SE were derived from three independent experiments, each performed with three technical replicates per sample. Statistically significant differences (post hoc Tukey’s HSD test; ∗P < 0.05 and ∗∗P < 0.01) between Col-0 (batch grown together with gun1 seedlings), gun1 mutants, and Col-0 seedlings grown on MS, NF, or LIN are indicated by black asterisks. Transcripts marked in bold were downregulated exclusively in gun1W but not under NF or LIN treatment. (C) Coverage plots depict the accumulation of reads across the ycf1.2–rps15–ndhH–ndhA–ndhI–ndhG–ndhE–psaC–ndhD gene cluster. Vertical arrows point to predicted GUN1 binding sites (see Figure 6; Supplemental Table 6). (D) Analysis of ndhG and ycf1.2 transcript accumulation by northern blotting. Total RNA was isolated from 4-day-old Col-0 and gun1-102 white, marbled, and green seedlings, as well as from Col-0 seedlings grown on medium supplemented with NF or LIN. The samples were run on the same gel but rearranged for clarity. As a loading control and for visualization of rRNAs, the membrane was stained with M.B. The arrows point to bands representing chloroplast rRNAs.

Figure 5

Predicted GUN1 binding sites. (A) Predicted ambiguous GUN1 target sequence. The numbers in the first row depict the PPR motif number, whereas the second row displays the amino acids in each PPR motif that are crucial for prediction of target nucleotides. For some amino acid combinations, the predicted target nucleotide is unique (such as ST and SN), whereas for others (such as ND), multiple nucleotides are predicted with descending preference. Subsequent rows indicate the prospective target sequences dependent on the stringency applied to the predicted nucleotides. For example, using only the first nucleotide of each of the predicted nucleotides results in 0 target sites. Allowing U, C, or G for the ambiguous B and G or C for “G>>C” results in 78 potential target sites. Allowing U, C, or G for the ambiguous B and only G for “G>>C” results in 25 potential target sites (here and in the following, marked in magenta). Allowing only U or C for the ambiguous Y and only G for “G>>C” results in 9 potential target sites (here and in the following, marked in blue). Highly conserved regions in GUN1 are highlighted in bold letters, according to Honkanen and Small (2022). In addition, representative predicted binding sites at ndhG, ndhE, and rrn23S are shown. wo IR, without inverted repeat. (B) Table showing the nine sites in the “U and C (Y)” category.

Figure 6

GUN1 binds to RNAs in vivo and in vitro. (A) Schematic presentation of predicted RNA binding sites (indicated by black vertical arrows) in ycf1.2, the rps15–ndhH–ndhA–ndhI–ndhG–ndhE–psaC–ndhD polycistron, and the rrn23S gene. Positions of primers used in (B) are depicted with arrowheads using the color code explained in the legend to Figure 5. (B) Demonstration of co-purification of selected RNAs with GUN1. RNAs that were isolated from the pellet after Co-IP experiments with Col-0 and a GUN1 overexpression line (GUN1–GFP) (IP) and the respective input RNAs (Input) were amplified by RT–qPCR. Ratios of immunoprecipitated versus input RNA levels are reported relative to the corresponding levels in the first Col-0 replicate, which were set to 1. Mean values ± SD were derived from three independent experiments, each performed with three technical replicates per sample. Statistically significant differences (post hoc Tukey’s HSD test; ∗P < 0.05 and ∗∗P < 0.01) between GUN1–GFP and Col-0 lines are indicated by black asterisks. (C) Overexpression and purification of a His-tagged GUN1–PS protein in E. coli. GUN1–PS encompasses all PPR and SMR motifs (PS) spanning amino acids 232 to 918. Left: SDS–PAGE before (−) and after (+) 20 h of induction at 18°C; middle: western blot of the induced protein with an anti-His antibody; right: SDS–PAGE after purification. W, wash fraction with a buffer containing 20 mM imidazole; E1 and E2, elution fractions with a buffer containing 250 mM imidazole; E3 and E4, elution fractions with a buffer containing 500 mM imidazole. (D) The GUN1 protein interacts in vitro with RNA sequences located in ndhG and trnG. EMSAs were performed with purified His-tagged GUN1 protein that was produced in E. coli. Aliquots (0, 100, 200, 400, and 600 nM) of purified GUN1 protein were incubated with Cy5-labeled single-stranded RNA (ssRNA) probes representing the putative target sequences and an nonspecific ssRNA probe. Binding reactions were performed at 23°C, followed by electrophoresis on non-denaturing TBE polyacrylamide gels at 4°C. (E) Aliquots (0, 200, and 400 nM) of purified GUN1 protein were incubated with Cy5-labeled ssRNA probes in the presence of increasing concentrations (5×, 25×, 50×; indicated by the light gray triangle) of the same unlabeled ssRNA (specific) or a nonlabeled ssRNA of unrelated sequence (nonspecific) as competitors. Binding reactions were then subjected to electrophoresis on non-denaturing TBE-polyacrylamide gels as performed in (D).

Figure 7

Identification of putative GUN1 targets by RIP-seq analysis. (A) Libraries were prepared from RNAs co-immunoprecipitated from a GPF-tagged GUN1 line (GUN1–GFP) and, as controls, from a PP7L–GFP line and Col-0 and then sequenced. The experiment was performed with three biological replicates. Coverage plots of reads per kilobase per million (RPKM) values show the accumulation of reads across the chloroplast genome, here shown without inverted repeat B. Vertical arrows indicate examples of regions with higher read accumulation in GUN1–GFP compared with PP7L–GFP and Col-0 and that also contain a match to the predicted GUN1 target code (see Supplemental Tables 4 and 6). The color code is explained in the legend to Figure 5. (B) Coverage plot of RPKM values across the blue light responsive promoter (BLRP) of psbD encompassing the predicted GUN1 binding site. (C) RT–qPCR to determine expression levels of GUN1 and the BLRP region covering the predicted GUN1 binding site. Seedlings were grown under continuous light (100 μmol m⁻² s⁻¹) for 5 days. The results were normalized to AT4G36800, which encodes a RUB1-conjugating enzyme (RCE1). Expression values are reported relative to the corresponding transcript levels in Col-0, which were set to 1. Mean values ± SE were derived from three independent experiments, each performed with three technical replicates per sample. Statistically significant differences (post hoc Tukey’s HSD test; ∗P < 0.05 and ∗∗P < 0.01) between Col-0 and the transgenic lines are shown. (D) Under our conditions, the GUN1 protein does not interact in vitro with the predicted GUN1 binding site located in the BLRP. EMSAs were performed with purified His-tagged GUN1 protein that was produced in E. coli. Aliquots (0 and 800 nM) of purified GUN1 protein were incubated with 2 nM Cy5-labeled ssRNA probes representing the putative target sequences and a BLRP probe containing 10 mutated sites (mut.). Binding reactions were performed at 23°C, followed by electrophoresis on non-denaturing TBE polyacrylamide gels at 4°C. (E) Libraries were prepared from RNAs isolated from the Co-IP experiments described in (A). Relative enrichment ratios (calculated at the exon level) of GUN1–GFP relative to Col-0 and GUN1–GFP relative to PP7L–GFP are shown. Gray shading indicates genes located in a polycistron. Transcripts that also contain a match to the predicted GUN1 target code (see Supplemental Tables 4 and 6) are written in bold. The color code is explained in the legend to Figure 5. (F) Plot of RIP-seq data over two example regions. Relative depth was calculated at each nucleotide (nt) position by relating the number of reads to the total depth of the sequencing output. Black vertical arrows indicate predicted GUN1 RNA-binding sites.