The GENOMES UNCOUPLED1 protein has an ancient, highly conserved role but not in retrograde signalling

Summary

• The pentatricopeptide repeat protein GENOMES UNCOUPLED1 (GUN1) is required for chloroplast‐to‐nucleus signalling when plastid translation becomes inhibited during chloroplast development in Arabidopsis thaliana, but its exact molecular function remains unknown.

• We analysed GUN1 sequences in land plants and streptophyte algae. We tested functional conservation by complementation of the Arabidopsis gun1 mutant with GUN1 genes from the streptophyte alga Coleochate orbicularis or the liverwort Marchantia polymorpha. We also analysed the transcriptomes of M. polymorpha gun1 knockout mutant lines during chloroplast development.

• GUN1 evolved within the streptophyte algal ancestors of land plants and is highly conserved among land plants but missing from the Rafflesiaceae that lack chloroplast genomes. GUN1 genes from C. orbicularis and M. polymorpha suppress the cold‐sensitive phenotype of the Arabidopsis gun1 mutant and restore typical retrograde responses to treatments with inhibitors of plastid translation, even though M. polymorpha responds very differently to such treatments.

• Our findings suggest that GUN1 is an ancient protein that evolved within the streptophyte algal ancestors of land plants before the first plants colonized land more than 470 million years ago. Its primary role is likely to be in chloroplast gene expression and its role in chloroplast retrograde signalling probably evolved more recently.

Introduction

Plant chloroplasts evolved from endosymbiotic cyanobacteria. While chloroplasts retain their own genomes, the majority of the cyanobacterial genes were lost or have moved into the nuclear genome during evolution (Timmis et al., ²⁰⁰⁴). Consequently, many chloroplast protein complexes contain both nucleus‐ and chloroplast‐encoded components (Rolland et al., ²⁰¹²). Therefore, expression of the chloroplast and nuclear genes needs to be tightly coordinated to ensure optimal chloroplast function during development and under changing environmental conditions (Chan et al., ²⁰¹⁶). This coordination of chloroplast and nuclear gene expression requires both nucleus‐to‐chloroplast signals (anterograde signals) and chloroplast‐to‐nucleus signals (retrograde signals). Chloroplast‐to‐nucleus retrograde signalling mechanisms fall into two categories: biogenic retrograde signals that operate during early chloroplast development, and operational retrograde signals that are emitted by mature chloroplasts (reviewed in Pogson et al., ²⁰⁰⁸; Chan et al., ²⁰¹⁶; Hernández‐Verdeja & Strand, ²⁰¹⁸; Wu & Bock, ²⁰²¹).

Chloroplasts develop from small, undifferentiated plastids called proplastids. In Arabidopsis, proplastid‐to‐chloroplast differentiation takes place in young cotyledons and in specific cells of the shoot apical meristem (Charuvi et al., ²⁰¹²; Wu et al., ²⁰¹⁸). Inhibition of chloroplast gene expression during proplastid‐to‐chloroplast differentiation results in strong transcriptional repression of photosynthesis‐associated nuclear genes (PhANGs), including genes coding for light‐harvesting Chl a/b binding (LHCB) proteins and rubisco small subunit (RbcS) proteins (Harpster et al., ¹⁹⁸⁴; Mayfield & Taylor, ¹⁹⁸⁴).

While the exact identity of the chloroplast‐emitted retrograde signal or signals is still being debated (reviewed in Hernández‐Verdeja & Strand, ²⁰¹⁸; Wu & Bock, ²⁰²¹), the retrograde signalling mechanism that operates during proplastid‐to‐chloroplast differentiation in Arabidopsis involves the GENOMES UNCOUPLED (GUN) genes. The genomes uncoupled mutants were isolated based on their inability to repress transcription of the nuclear LIGHT HARVESTING CHLOROPHYLL A/B‐BINDING PROTEIN 1.2 (LHCB1.2) gene in response to inhibition of chloroplast translation imposed by norflurazon treatment (Susek et al., ¹⁹⁹³; Mochizuki et al., ²⁰⁰¹; Woodson et al., ²⁰¹¹). Norflurazon inhibits carotenoid biosynthesis, which results in oxidative stress to the chloroplasts and loss of chloroplast ribosomes (Bartels & Watson, 1978; Breitenbach et al., ²⁰⁰¹). While five out of the six GUN genes (GUN2–GUN6) code for proteins involved in tetrapyrrole metabolism, the molecular mechanism of GUN1 function has remained unresolved. GUN1 is unique among the other GUN genes in that it is also required for chloroplast‐to‐nucleus signalling in response to treatments with chemicals that specifically inhibit plastid gene expression (Susek et al., ¹⁹⁹³; Koussevitzky et al., ²⁰⁰⁷).

GUN1 codes for a chloroplast‐localized pentatricopeptide peptide repeat (PPR) protein with a C‐terminal small MutS‐related (SMR) domain (Koussevitzky et al., ²⁰⁰⁷). PPR proteins constitute one of the largest protein families in seed plants, with 450 members in Arabidopsis (Barkan & Small, 2014). PPR proteins are characterized by tandem arrays of 31–36 amino‐acid‐long PPR motifs (Lurin et al., ²⁰⁰⁴). Most previously characterized PPR proteins function as sequence‐specific RNA binding proteins. PPR proteins are encoded by the nuclear genome and almost exclusively localize to chloroplasts or mitochondria, where they bind their target RNAs. PPR proteins have been described to alter the expression of their target RNAs in a variety of ways, including RNA cleavage, splicing, editing, stabilization and translational activation (Pfalz et al., ²⁰⁰⁹; Prikryl et al., ²⁰¹¹; Wu et al., ²⁰¹⁶; Aryamanesh et al., ²⁰¹⁷; Zhou et al., ²⁰¹⁷; Rojas et al., ²⁰¹⁸; Lee et al., ²⁰¹⁹).

A subset of land plant PPR proteins, the PPR‐SMR proteins, contain a C‐terminal SMR domain. The SMR domain was first characterized in the MutS2 protein of the cyanobacterium Synechocystis, where it was reported to have DNA endonuclease activity (Moreira & Philippe, 1999; Fukui et al., ²⁰⁰⁷). SMR‐domain‐containing proteins are commonly found in both prokaryotic and eukaryotic organisms, although PPR proteins with SMR domains are (apart from a few exceptions) restricted to land plants and green algae (Liu et al., ²⁰¹³). The Arabidopsis genome encodes for eight PPR‐SMR proteins, five of which locate to plastids. Three of these, PLASTID TRANSCRIPTIONALLY ACTIVE2 (pTAC2), SUPPRESSOR OF THYLAKOID FORMATION1 (SOT1) and SUPPRESSOR OF VARIEGATION7 (SVR7), are among the most abundant PPR proteins in plastids (Liu et al., ²⁰¹³; Mergner et al., ²⁰²⁰). pTAC2 is an essential component of the plastid‐encoded RNA polymerase complex and ptac2 mutants are only able to grow on sucrose‐supplemented media (Pfalz et al., ²⁰⁰⁶). SOT1 facilitates proper assembly and maturation of the plastid ribosome by processing the plastid 23S–4.5S ribosomal RNA (rRNA) precursor (Wu et al., ²⁰¹⁶; Zhou et al., ²⁰¹⁷). svr7 mutants also possess plastid rRNA processing and gene expression defects, most notably impaired expression of the plastid ATP synthase (Liu et al., ²⁰¹⁰; Zoschke et al., ²⁰¹³). Therefore, PPR‐SMR proteins are required for appropriate plastid gene expression.

In contrast to the other chloroplast‐localized Arabidopsis PPR‐SMR proteins, the GUN1 protein only accumulates in specific tissues (Mergner et al., ²⁰²⁰). While the GUN1 transcript is abundantly expressed, GUN1 protein accumulation coincides with proplastid‐to‐chloroplast differentiation in the cotyledons during early seedling development and in specific cells of the shoot apical meristem, based on GUN1‐green fluorescent protein (GFP) fusion signal (Wu et al., ²⁰¹⁸). In addition, the GUN1‐GFP protein accumulates in response to inhibitors that induce plastid retrograde signalling (Wu et al., ²⁰¹⁸). Several recent studies have proposed roles for the Arabidopsis GUN1 protein based on its interactions with other chloroplast‐localized proteins. These include interaction with the chaperone cpHSC70‐1 to control chloroplast protein import (Tadini et al., ²⁰¹⁶; Wu et al., ²⁰¹⁹), interaction with nucleus‐encoded plastid RNA polymerase (NEP) to control the accumulation of NEP‐encoded transcripts (Tadini et al., ²⁰²⁰), interaction with multiple tetrapyrrole biosynthesis enzymes, haem and porphyrins to control the flux through the tetrapyrrole biosynthesis pathway (Shimizu et al., ²⁰¹⁹), and interaction with MULTIPLE ORGANELLAR RNA EDITING FACTOR2 (MORF2) to control RNA editing (Zhao et al., ²⁰¹⁹).

The numerous proposed GUN1 interaction partners and mechanisms of function are intriguing but confusing. Current data on GUN1 function are limited to Arabidopsis, so we set out to investigate the origin and evolution of the land plant GUN1 proteins. Here we identify GUN1 as an ancient protein that is highly conserved across land plants. The pattern of amino acid conservation along the GUN1 protein is consistent with the hypothesis that GUN1, like other characterized PPR proteins, codes for a nucleic acid binding protein. Finally, the retrograde signalling cascade that functions downstream of the conserved core GUN1 mechanism is not conserved and may have evolved more recently.

Materials and Methods

Identification of GUN1 sequences

Arabidopsis and Marchantia GUN1 sequences were retrieved from TAIR (https://www.arabidopsis.org/) and MarpoIBase (https://marchantia.info/), respectively. Full‐length GUN1 sequences were obtained from a representative set of land plants by Blast searches using the Arabidopsis sequence to search GenBank and selected translated sequence sets (whole genome shotgun or transcriptome shotgun assemblies) via the NCBI Sequence Set Browser (https://www.ncbi.nlm.nih.gov/Traces/wgs/). A set of 76 phylogenetically diverse GUN1 sequences (including representatives from algae, bryophytes, lycophytes, ferns, gymnosperms and angiosperms) were aligned using the G‐INS‐i algorithm in Mafft v.7 (Katoh & Standley, 2013). The most highly conserved region of this alignment (876 positions) was used to generate a GUN1 sequence profile with hmmbuild from the hmmer package (v.3.3.1) (http://hmmer.org; Eddy, ²⁰¹¹), which in turn was used to search for GUN1 sequences (using hmmsearch with default parameters) in translations of various transcriptome datasets, most notably putative PPR protein sequences compiled by Gutmann et al. (2020) from the 1KP dataset (Carpenter et al., ²⁰¹⁹; Leebens‐Mack et al., ²⁰¹⁹). The 1KP transcriptomes were filtered to remove those encoding fewer than 10 000 distinct proteins to avoid trivial false negatives due to low coverage and those from organisms other than green algae and land plants. This resulted in 1128 analysable samples from 894 plant species. Specific searches were also made in datasets of particular interest, or where GUN1 could not be found in the corresponding 1KP samples. These additional datasets included whole genome shotgun data from Sapria himalayana (Cai et al., ²⁰²¹) and whole transcriptome data from Rafflesia cantleyi (Lee et al., ²⁰¹⁶), both holoparasites from the Rafflesiaceae.

Plant lines and growth conditions

Arabidopsis thaliana wild‐type Columbia (Col‐0) and the Atgun1 T‐DNA insertion line gun1‐102 SAIL_290_D09 (Sessions et al., ²⁰⁰²; Tadini et al., ²⁰¹⁶) were used in this study. Arabidopsis seeds were sterilized with chlorine gas for 2–4 h and grown on sterile half‐strength Murashige & Skoog medium (½MS medium; PhytoTech Labs, Lenexa, KS, USA) under long‐day conditions (16 h : 8 h, light : dark) at 100 μmol m⁻² s⁻¹ at 22°C unless otherwise mentioned.

Marchantia polymorpha wild‐type accessions Takaragaike‐1 (Tak‐1) male and Tagarakaike‐2 (Tak‐2) female were kindly provided by Prof. John Bowman, Monash University. Marchantia polymorpha CRISPR knockout lines Mpgun1‐1 and Mpgun1‐2 were generated in this study. Marchantia polymorpha plants were grown on sterile half‐strength Gamborg's medium (Duchefa Biochemie, Haarlem, the Netherlands) supplemented with 1.2% agar under long‐day conditions (16 h : 8 h, light : dark) unless otherwise mentioned. Crossing and spore sterilization were carried out as described in Supporting Information Methods S1.

Generation of A. thaliana gun1 mutant complementation lines

All primer sequences are provided in Table S1. Annotated sequences of the complementation constructs are provided in Notes S1. AtGUN1, MpGUN1 and CoGUN1 protein coding sequences were aligned using Geneious (Kearse et al., ²⁰¹²). The region containing chloroplast transit peptides was estimated based on lack of conservation between the protein sequences (Fig. S1). The MpGUN1 coding sequence excluding the putative chloroplast transit peptide‐encoding region (amino acids 1–114) was amplified from M. polymorpha cDNA using PrimeSTAR HS DNA polymerase (Takara Bio, Shiga, Japan) with primers CDS_MpGUN1_At_gib_F and CDS_MpGUN1_At_gib_R. The CoGUN1 coding sequence excluding the chloroplast transit peptide‐encoding region (amino acids 1–145) was ordered as a gene synthesis fragment from Integrated DNA Technologies (IDT, Coralville, IA, USA). A fragment containing the AtGUN1 promoter, 5′ untranslated region and chloroplast transit peptide (amino acids 1–135) was amplified from Arabidopsis Col‐0 genomic DNA with primers pro_AtGUN1_gib_F and tp_AtGUN1_Mp_gib_R for the MpGUN1 construct, and primers pro_AtGUN1_gib_F and tp_AtGUN1_Co_gib_R for the CoGUN1 construct. The AtGUN1 terminator fragment was amplified with primers term_AtGUN1_Mp_gib_F and term_AtGUN1_gib_R for the MpGUN1 construct and with primers term_AtGUN1_Co_gib_F and term_AtGUN1_gib_R for the CoGUN1 construct. The three fragments for each construct were assembled on a XbaI‐ and SacI‐digested gel‐purified MpGWB101 vector (Ishizaki et al., ²⁰¹⁵) using Gibson assembly. To create a control construct for complementation of the Atgun1 mutant with AtGUN1 genomic fragment, the wild‐type AtGUN1 gene sequence was amplified from Col‐0 DNA using primers pro_AtGUN1_gib_F and term_AtGUN1_gib_R and cloned onto the XbaI‐ and SacI‐linearized MpGWB101 vector backbone.

Arabidopsis thaliana wild‐type and gun1 mutant plants were transformed using the floral dip method (Zhang et al., ²⁰⁰⁶) with binary vectors described above using Agrobacterium strain GV3101. Seeds were surface‐sterilized and sown on sterile plates supplemented with 15 μg ml⁻¹ hygromycin. After 1 wk, hygromycin‐resistant plants were transferred to soil. The genotype of Arabidopsis wild‐type, gun1 and gun1 complemented with AtGUN1, MpGUN1 and CoGUN1 was PCR‐verified using the Phire plant direct PCR kit (Thermo Fisher, Waltham, MA, USA) as recommended by the manufacturer. The presence of the T‐DNA insertion was verified using primers LBb1.3 and SAIL_280_D09_LP, whereas the presence/absence of intact wild‐type GUN1 sequence was verified with primers SAIL_280_D09_LP and SAIL_280_D09_RP. Seeds from 10 to 15 independent transformant lines were observed for the initial phenotypic assessment. All final phenotype and gene expression analyses were carried out on nonsegregating T2 or T3 seeds of three independent homozygous transformant lines for each construct.

Marchantia polymorpha transformation and generation of transgenic Mpgun1 CRISPR/Cas9 knockout lines

Marchantia polymorpha transformation and generation or transgenic Mpgun1 CRISPR/Cas9 knockout lines are described in Methods S1 and S2.

Plant growth experiments

Lincomycin treatment of A. thaliana

Arabidopsis seeds were sterilized and plated on sterile ½MS medium pH 5.7, 2% sucrose and 0.8% agar. For the lincomycin treatment, the growth medium was supplemented with 220 μg ml⁻¹ lincomycin (Sigma‐Aldrich). The seeds were stratified at 4°C for 2 d, after which the plants were grown at 22°C under continuous light at 100 μmol m⁻² s⁻¹ for 5 d before imaging or harvesting the tissue for RNA extraction.

Combined low light and lincomycin treatment of A. thaliana

Arabidopsis seeds were sterilized and plated on sterile ½MS medium pH 5.7 (without sucrose) and 0.8% agar. For the lincomycin treatment the growth medium was supplemented with 220 μg ml⁻¹ lincomycin (Sigma‐Aldrich). The seeds were stratified at 4°C for 4 d, and then grown at 22°C under continuous light at 1 μmol m⁻² s⁻¹ for 3 d before imaging.

Arabidopsis thaliana cold growth experiment

Arabidopsis seeds were sterilized and plated on ½MS medium pH 5.7 (without sucrose) and 0.8% agar. The plates were placed directly at 4°C in a growth room under long‐day conditions (8 h : 16 h, dark : light) at 100 μmol m⁻² s⁻¹. Plants were imaged after 7 wk.

Spectinomycin and norflurazon treatments of M. polymorpha spores

Marchantia polymorpha spores were sterilized and plated on ½ Gamborg's medium (Duchefa Biochemie) supplemented with 1.2% agar and 500 μg ml⁻¹ spectinomycin or 5 μM norflurazon. The plates were placed under long‐day conditions for 48 h, after which the spores were resuspended in 1 ml of sterile water, transferred to a microcentrifuge tube and spun down at 3400  g for 1 min. The water was removed, and the spore pellet wasflash‐frozen in liquid nitrogen.

RNA extraction

Extraction of total RNA from Arabidopsis seedlings and Marchantia spores was carried out using the Direct‐Zol RNA MINIprep kit (Zymo Research, Irvine, CA, USA) following the manufacturer's protocol. Three independent biological replicates were extracted for each line. RNA was quantified using a NanoDrop spectrophotometer (Thermo Fisher Scientific) and diluted to 250 ng μl⁻¹ before DNase treatment using Turbo DNase (Ambion, Austin, TX, USA) as recommended by the manufacturer.

cDNA synthesis and quantitative real‐time PCR (qPCR)

One microgram of DNase‐treated RNA was used as a template for cDNA synthesis. cDNA was generated with oligo dT18 primer using the Protoscript II reverse transcriptase (NEB, Ipswich, MA, USA) in the presence of Murine RNase inhibitor (NEB). cDNA was diluted 1 : 5 before qPCR. qPCR was carried out using SYBR Premix Ex Taq II (Takara Bio) qPCR reagent. Each 5 μl reaction contained 1 μl primer mix, 1.5 μl cDNA dilution and 2.5 μl Sybr II master mix. Each biological replicate sample was run in three technical replicates. Amplification was carried out in a LightCycler480 instrument (Roche) using the following cycling conditions: initial denaturation for 1 min at 95, then 40 cycles of 10 s at 95°C, 10 s at 60°C and 20 s at 72°C. For primer sequences see Table S1. The data were analysed using LinRegPCR (Ruijter et al., ²⁰⁰⁹) v.2017.1. The expression of each gene of interest was first separately normalized against the reference genes AtPP2AA3 and AtUBQ10 for Arabidopsis and MpEF1α and MpACT for M. polymorpha. The geometric mean of the two normalized values was then recorded as the expression level.

RNA sequencing

Transcriptome libraries of M. polymorpha wild‐type and Mpgun1 mutant spores were prepared using 200 ng of DNase‐treated total RNA as a template for the TruSeq Stranded Total RNA kit with Ribo‐Zero Plant (Illumina, San Diego, CA, USA). The libraries were sequenced on an Illumina HiSeq 4000 platform (150 nt paired‐end reads) at Novogene, Hong Kong. At least 5.7 GB of raw data was obtained for each replicate library. Sequencing read data were deposited at the Short Read Archive database at the National Center for Biotechnology Information under project nos. PRJNA800059 and PRJNA838206.

Optical duplicate reads were first removed with clumpify (parameters: dedupe optical dist = 40) from the bbmap package (https://sourceforge.net/projects/bbmap/) and adapters were trimmed with bbduk (parameters: ktrim = r k = 23 mink = 11 hdist = 1 tpe tbo ftm = 5). The reads were then assigned to transcripts using Salmon v.1.3.0 (Patro et al., ²⁰¹⁷) (parameters: ‐l A ‐‐validateMappings) against an index prepared with the M. polymorpha MpTak_v.6.1 reference genome and cDNA assemblies (https://marchantia.info/). Differential expression analyses were carried out using DESeq2 (Love et al., ²⁰¹⁴). Functional annotations for the MpTak_v.5.1 genome release were used to annotate differentially expressed genes (log₂ fold‐change > 1 or < −1 and P _adj < 0.01) and to identify M. polymorpha photosynthesis‐associated nuclear genes. Gene Ontology (GO) enrichment analyses were performed on the Dicots Plaza 4.5 platform (Van Bel et al., ²⁰¹⁸) using standard settings with differentially expressed genes showing log₂ fold‐change > 1 or < −1 and P _adj < 0.01 as an input.

Microscopy and image analysis

Arabidopsis seedlings and M. polymorpha gemmae were imaged using an Olympus Camedia C 7070 camera mounted on a Leica SZ61 dissecting microscope or with an Apple iPad (7^th generation). Transmitted light micrographs of M. polymorpha spores were obtained using an Olympus Camedia C 7070 camera mounted on a Olympus BX51 microscope. Image analysis to quantify phenotypic differences was carried out using Fiji (Schindelin et al., ²⁰¹²). Images were adjusted using Adobe Photoshop 2020.

Accession numbers

The following Arabidopsis genes were mentioned in this article: AtGUN1, AT2G31400; AtLHCB1.2, AT1G29910; AtLHCB2.2, AT2G05070; AtCA1, AT3G01500; and AtCP12, AT3G62410.

The following M. polymorpha genes were mentioned in this article: MpGUN1, Mp1g08430; MpLHCB2, Mp7g06790; MpRBCS, Mp4g10850; and MpFBPase, Mp6g13790.

Results

GUN1 evolved within the streptophyte algae and is conserved among land plants

To discover when GUN1 arose and its distribution within the plant lineage, we searched for GUN1‐like sequences using the Arabidopsis GUN1 (AtGUN1) sequence as a query in Blastp (Altschul et al., ¹⁹⁹⁰) searches of the GenBank nonredundant protein database. Probable GUN1 sequences were verified by alignment to AtGUN1, by reciprocal Blastp to Arabidopsis protein sequences and by identification of the expected PPR motifs and C‐terminal SMR domain. Land plants form a monophyletic group that evolved from the streptophyte algae (Wickett et al., ²⁰¹⁴). We identified putative GUN1 sequences in the streptophyte algal groups most closely related to land plants (Zygnematales, Coleochaetales and Charales), but not in other streptophyte algal groups (Klebsormidiales, Mesostigmatales and Chlorokybales) nor in other green algae (Chlamydomonadales) (Fig. S2) This suggests the GUN1 protein is ancient and evolved in the streptophyte algal ancestors of land plants before the first plants colonized land.

To systematically assess the conservation of GUN1 among plants, we selected 76 phylogenetically diverse full‐length GUN1 sequences from a representative set of land plants and streptophyte algae (Figs 1, S2), aligned them and retained the most conserved region of the alignment (which included all of the PPR motifs and the SMR domain). From this alignment we developed a sequence profile using hmmbuild (http://hmmer.org; Eddy, ²⁰¹¹) and used it to screen c. 500 000 PPR sequences derived from the 1KP plant transcriptome dataset (Carpenter et al., ²⁰¹⁹; Leebens‐Mack et al., ²⁰¹⁹; Gutmann et al., ²⁰²⁰) with hmmsearch. Putative GUN1 orthologues could be distinguished from other PPR‐SMR sequences based on the domain score reported for each match by hmmsearch (Fig. S3; Table S2). GUN1 transcripts were detected in 824 of the 894 species analysed. We identified conserved GUN1 sequences in 345 of the 366 plant families represented in this dataset. Notably, we found that transcriptomes of nonphotosynthetic parasitic plants with highly reduced plastid genomes, such as Pilostyles thunbergii (Bellot & Renner, 2016), Balanophora fungosa (Su et al., ²⁰¹⁹) and Conopholis americana (Wicke et al., ²⁰¹³), encode GUN1. Of the 21 families apparently lacking GUN1, 14 are green algal families outside the streptophyte clades closely related to land plants, and thus not expected to contain GUN1. One of the remaining seven is represented in the original set of 76 sequences used to build the GUN1 profile, and thus does contain GUN1, even though no GUN1 transcripts were detected in the 1KP sample. This left six land plant families expected to contain GUN1 (Agapanthaceae, Cyrillaceae, Hymenophyllaceae, Juglandaceae, Monocleaceae, Thelypteridaceae) for which the data so far did not provide clear evidence of GUN1 sequences (Table S2). After searching in additional whole transcriptome or whole genome shotgun sequencing datasets, we were able to identify putative GUN1 sequences from Agapanthus, Hymenophyllum and Juglans. For the three remaining families, each represented by only a single sample in our 1KP dataset, we were unable to find additional data to search. We also searched whole transcriptome shotgun sequences from R. cantleyi (Lee et al., ²⁰¹⁶) and whole genome shotgun sequences from S. himalayana (Cai et al., ²⁰²¹) (both Rafflesiaceae). These two closely related genera are reported to lack a plastid genome. We found no putative GUN1 sequences in either dataset. Sapria himalayana is the only embryophyte for which we were unable to find a putative GUN1 gene in its genome, and R. cantleyi one of the few for which we were unable to find putative GUN1 transcripts in its transcriptome.

Fig. 1.

GUN1 is conserved in most streptophytes. The top part of the figure represents a multiple alignment of 76 GUN1 protein sequences from diverse streptophyte algae and land plants. The alignment was constructed using Mafft and visualized with Geneious. Only the central, most conserved region of the alignment is shown (corresponding to positions 135–888 in AtGUN1). Darker shading indicates higher similarity to the consensus. The sequences are grouped (approximately) by phylogenetic relationship as indicated; for the species names and amino acid sequences, see the alignment in Supporting Information Fig. S2. Below the alignment, the annotation tracks show helix–turn–helix motifs (grey arrows) and SMR domain (red arrow) predicted by AlphaFold and PPR motifs (brown arrows) predicted by hmmsearch using a P‐type PPR motif HMM. Dark brown motifs have higher confidence. The final PPR motif is interrupted by a long insertion but alphafold predicts that the N‐ and C‐terminal helices nevertheless interact together as in a typical PPR motif. Below these motif predictions, the fifth and last amino acids of the conserved PPR motifs are indicated as sequence logos generated by Skylign (Wheeler et al., ²⁰¹⁴) (letter height represents information content) from the alignment of 893 GUN1 sequences (the 76 sequences shown here plus 817 sequences from the 1KP PPR dataset). Finally, at the bottom of the figure, the expected RNA binding specificities of the conserved PPR motifs are indicated, again as a Skylign sequence logo. The HMM profile and full alignment of all 893 identified GUN1 sequences are obtainable from Dryad (10.5061/dryad.x0k6djhmk).

The predicted binding specificity of GUN1 proteins is conserved

Conservation of protein function is likely to depend on the level of sequence conservation in the functionally important regions of the protein. The GUN1 protein consists of 12–16 PPR motifs (depending on the approach used to recognize and define the motifs) and a C‐terminal SMR domain. Fig. 1 shows the alignment of the conserved regions of the GUN1 sequences used to generate the HMM profile, and the structural motifs that correspond, as predicted by AlphaFold (Jumper et al., ²⁰²¹; Varadi et al., ²⁰²²) or hmmsearch (Eddy, 2011). AlphaFold predicts 16 helix–turn–helix motifs (Fig. S4) of which 12 are recognized as PPR motifs by hmmsearch with a P‐type PPR HMM (Cheng et al., ²⁰¹⁶). The first helix–turn–helix motif predicted by AlphaFold is poorly conserved between species, but the final three are highly conserved within GUN1 sequences although quite divergent in sequence and length from typical PPR motifs.

The binding specificity of PPR proteins can be predicted based on the 5^th and last amino acid of each PPR motif (Barkan et al., ²⁰¹²). The amino acids at these positions of PPR motifs 2–12 are highly conserved between all land plant and streptophyte algae GUN1 protein sequences (Figs 1, S2). When these amino acid combinations are used to predict the nucleotide most likely to be bound by each motif, this conservation is even more striking, with the only predicted variation being at motifs 6 and 9 where different combinations are predicted to bind either C or U. This may not actually reflect functional divergence, as any of the combinations are likely to have high affinity for both C and U (Yan et al., ²⁰¹⁹). The structure predicted by AlphaFold forms a contiguous solenoid similar to that of other PPR proteins that are known to bind RNA, albeit with some idiosyncracies. The conservation of the PPR motifs, and in particular the residues that in other PPR proteins define RNA binding specificity, is consistent with the hypothesis that GUN1 codes for an RNA binding protein with a highly conserved target.

GUN1 SMR domains are conserved

The GUN1 protein contains a C‐terminal SMR domain that is highly conserved between all the full‐length GUN1 sequences in our dataset. Four SMR‐domain‐containing proteins have been reported to have endonuclease activity in other organisms (Liu et al., ²⁰¹³). Subfamily 2 SMR proteins that have endonuclease cleavage activity have two conserved motifs: the LDXH motif in the N‐terminus of the SMR domain and a central TGXG motif (Watanabe et al., ²⁰⁰³; Bhandari et al., ²⁰¹¹). The central TGXG motif is perfectly conserved as TGWG in all full‐length GUN1 proteins in our dataset. The N‐terminal LDXH motif is also almost perfectly conserved as LDLH, with a scattering of exceptions that have VDLH in this position. These observations show that the SMR domains of land plant GUN1 proteins are conserved and probably contribute to GUN1 function.

GUN1 genes from the streptophyte alga Coleochate orbicularis and the liverwort M. polymorpha restore chloroplast retrograde signalling in the Atgun1 mutant

As the predicted binding specificity of GUN1 proteins is conserved among land plants and in the streptophyte algae, we hypothesized that GUN1 proteins are also functionally conserved. To test this hypothesis, we assessed whether GUN1 genes from the streptophyte alga C. orbicularis and the liverwort M. polymorpha can complement the Arabidopsis gun1 knockout mutant. Coleochate orbicularis and M. polymorpha each contain a single‐copy GUN1 gene. The C. orbicularis GUN1 protein shares 42.5% amino acid identity (57.2% match using the BLOSUM62 substitution matrix) with AtGUN1, while the M. polymorpha GUN1 protein is 48.4% identical (66.4% match using BLOSUM62) with AtGUN1. Sequences corresponding to chloroplast transit peptides in CoGUN1 and MpGUN1 were estimated based on lack of sequence conservation (Fig. S1) and replaced with the Arabidopsis GUN1 transit peptide. These coding sequences were then expressed from the AtGUN1 promoter in the Atgun1 (SAIL_290_D09) mutant background. As a control the Atgun1 mutant was complemented with an AtGUN1 genomic fragment.

In wild‐type Arabidopsis, inhibition of plastid translation during chloroplast development activates chloroplast retrograde signalling that results in strong transcriptional repression of PhANGs. In the Atgun1 mutant this transcriptional repression is defective, although not completely abolished (Fig. 2; Koussevitzky et al., ²⁰⁰⁷). We hypothesized that CoGUN1 and MpGUN1 can restore the GUN1‐dependent repression of PhANGs in the Atgun1 mutant when plastid translation is inhibited. To test this hypothesis, we germinated the Atgun1 seeds expressing CoGUN1, MpGUN1 or AtGUN1 on media supplemented with 200 μg ml⁻¹ lincomycin, an inhibitor of plastid translation. As expected, the inhibition of plastid translation by lincomycin resulted in strong reduction in steady‐state transcript levels of four PhANGs, namely AtLHCB1.2, AtLHCB2.2, AtCA1 and AtACP12 in wild‐type Arabidopsis, but less so in the Atgun1 mutant (Fig. 2). Expression of CoGUN1 and MpGUN1 in the Atgun1 mutant background restored the repression of these four PhANG transcripts to wild‐type levels (Fig. 2). These results indicate that CoGUN1 and MpGUN1 can activate the AtGUN1‐dependent retrograde signalling cascade that leads to transcriptional repression of PhANGs in Arabidopsis.

Fig. 2.

CoGUN1 or MpGUN1 expression in the Atgun1 mutant restores the repression of photosynthesis‐associated nuclear genes (PhANGs) in response to inhibition of plastid translation by lincomycin treatment. qPCR quantification of steady‐state transcript levels of four PhANGs, AtLHCB1.2, AtLHCB2.2, AtCA1 and AtACP12, in wild‐type Arabidopsis, the Atgun1 mutant, and Atgun1 mutant complemented with AtGUN1 (control), MpGUN1 or CoGUN1. The transcript levels were normalized against AtPP2AA3 and AtUBQ10. Each dot represents an independent biological replicate, and bars illustrate the average of the three biological replicates.

C. orbicularis and M. polymorpha GUN1 restore most aspects of the Atgun1 mutant phenotype

As CoGUN1 and MpGUN1 can complement the AtGUN1‐mediated repression of PhANGs when expressed in the Atgun1 mutant, we investigated whether other aspects that involve GUN1‐mediated signalling are also complemented.

Wild‐type Arabidopsis seedlings germinated on media supplemented with lincomycin in the presence of 2% sucrose develop purple cotyledons due to accumulation of anthocyanins. This anthocyanin accumulation is defective in the Atgun1 mutant (Fig. 3; Cottage et al., ²⁰¹⁰). Atgun1 seedlings complemented with CoGUN1 or MpGUN1 develop similar purple coloration to wild‐type seedlings when germinated on lincomycin‐containing media (Fig. 3a). Treatment with lincomycin also inhibits cotyledon expansion in wild‐type Arabidopsis seedlings more than in Atgun1 mutant seedlings. CoGUN1 and MpGUN1 restore the inhibition of cotyledon expansion in response to inhibition of plastid translation by lincomycin (Fig. 3b). These findings are consistent with the hypothesis that CoGUN1 and MpGUN1 can activate AtGUN1‐dependent downstream signalling that leads to upregulation of anthocyanin biosynthesis and inhibition of cotyledon expansion when plastid translation is inhibited.

Fig. 3.

CoGUN1 and MpGUN1 expression in the Atgun1 mutant restores anthocyanin accumulation and repression of cotyledon expansion in response to lincomycin treatment. (a) Five‐day‐old seedlings of wild‐type Arabidopsis, Atgun1 mutant, and Atgun1 mutant complemented with AtGUN1 (control), MpGUN1 or CoGUN1 grown under constant light on sterile plates for 7 d (top row, Bar, 4 mm) or on media supplemented with 220 μg ml⁻¹ lincomycin for 5 d (bottom row, Bar, 3 mm). (b) Quantification of cotyledon width of seedlings grown on media supplemented with 200 μg ml⁻¹ lincomycin for 5 d. Centre line indicates the mean, box limits indicate the upper and lower quartiles, and whiskers indicate the data range. Dots represent individual measurements. All other lines were significantly different from gun1 (Tukey's HSD: P < 0.001).

AtGUN1 is also required for cold tolerance during germination; germinating Atgun1 mutant seedlings are hypersensitive to cold (Marino et al., ²⁰¹⁹). This cold‐sensitive phenotype is barely noticeable in the cotyledons but becomes striking when the first true leaves appear after 6–7 wk of growth at 4°C (Fig. 4). True leaves of wild‐type plants germinated at 4°C are green and expanded, while the true leaves of the Atgun1 mutant are small, narrow and bleached. To check if CoGUN1 and MpGUN1 can restore the cold‐sensitive phenotype of the Atgun1 mutant, we germinated the complemented seeds on agar plates at 4°C. Both CoGUN1 and MpGUN1 improved the cold tolerance of the Atgun1 mutant (Fig. 4). However, CoGUN1 only partially restored cold tolerance: the first true leaves that emerged expanded and greened more than those of the gun1 mutant but were variegated. These findings indicate that MpGUN1 and CoGUN1 can at least partially complement the cold‐sensitive phenotype of the Arabidopsis gun1 mutant.

Fig. 4.

MpGUN1 and CoGUN1 complement the cold‐sensitive phenotype of the Atgun1 mutant. Phenotype of 7‐wk‐old seedlings germinated at 4°C under long‐day light regime. Bar, 4 mm.

Arabidopsis seedlings germinated in the absence of light (such as under soil) develop an etiolated phenotype characterized by an elongated hypocotyl, closed cotyledons and differentiation of proplastids into nongreen etioplasts. Upon illumination (or emerging from the soil), etiolated seedlings undergo deetiolation: hypocotyl elongation becomes inhibited, the cotyledons expand and etioplasts differentiate into photosynthetic chloroplasts. AtGUN1‐dependent signalling can be detected when plastid translation is inhibited during germination under low‐light conditions. Wild‐type Arabidopsis seedlings germinated on lincomycin‐containing media under low‐light conditions resemble dark‐grown seedlings: their hypocotyls become elongated and cotyledons remain closed (Fig. 5a–c). The Atgun1 mutant is defective in this response, indicating that inhibition of deetiolation in response to inhibition of plastid translation is dependent on functional GUN1 protein (Martín et al., ²⁰¹⁶). CoGUN1 and MpGUN1 expressed in the Atgun1 mutant background restore the inhibition of deetiolation in response to inhibition of plastid translation by lincomycin (Fig. 5). This finding suggests that CoGUN1 and MpGUN1 can replace AtGUN1 during deetiolation.

Fig. 5.

CoGUN1 and MpGUN1 expression in the Atgun1 mutant restores the repression of deetiolation when plastid translation is inhibited by lincomycin during germination. (a) Wild‐type Arabidopsis, Atgun1, and Atgun1 mutant complemented with AtGUN1 (control), MpGUN1 or CoGUN1 germinated for 3 d under constant low light (1 μmol m⁻² s⁻¹) on sterile plates without (control) or with 220 μg ml⁻¹ lincomycin. (b) Quantification of cotyledon angle in seedlings grown on lincomycin as in (a). (c) Quantification of hypocotyl length in seedlings grown on lincomycin as in (a). In (b) and (c) the centre line indicates the mean, box limits indicate the upper and lower quartiles, and whiskers indicate the data range. Dots represent individual measurements. All other lines were significantly different from gun1 (Tukey's HSD: P < 0.001).

Tgether, our results indicate that GUN1 proteins from streptophyte algae, such as C. orbicularis, and distantly related land plants, such as M. polymorpha, can functionally replace the Arabidopsis GUN1 protein.

Mpgun1 mutants are indistinguishable from wild‐type under nonstressful conditions

To investigate if GUN1 genes are functionally conserved among land plants, we generated two different CRISPR deletion lines of the single‐copy GUN1 gene in the liverwort M. polymorpha (Fig. 6a). The resulting Mpgun1‐1 and Mpgun1‐2 deletion mutant plants were PCR‐genotyped, and PCR products of selected plants were sequenced to confirm the deletion introduced a premature stop codon (Fig. S5). Male and female plants for each line were then crossed together to obtain nonsegregating populations of Mpgun1‐1 and Mpgun1‐2 knockout mutant spores.

Fig. 6.

The Mpgun1 mutant phenotype is indistinguishable from the wild‐type phenotype. (a) Gene model showing the structure of the Marchantia polymorpha GUN1 gene. Exons are indicated as black boxes, and noncoding regions are indicated as grey boxes. The locations of the CRISPR‐induced deletions are shown as white triangles. Blue and red lines indicate the region containing the PPR motifs and the SMR domain, respectively. (b) 40‐h‐old (top row, Bar, 20 μm) or 68‐h‐old (middle row, Bar, 20 μm) germinating spores or 14‐d‐old gemmae (bottom row, Bar, 3 mm) of M. polymorpha wild‐type and Mpgun1 mutants.

To check if GUN1 is involved in chloroplast development in M. polymorpha, we observed germination of wild‐type and Mpgun1 mutant spores. Wild‐type M. polymorpha spores begin germination by cell expansion. After the first 24 h, Chl autofluorescence can be observed from the chloroplasts (Bowman et al., ²⁰¹⁷). By 48 h the spores have become fully expanded and chloroplasts appear fully developed. After 48 h the spores undergo the first asymmetric cell division, followed by the emergence of the first rhizoid from the smaller daughter cell. Sometimes the emergence of the first rhizoid precedes the first cell division. The Mpgun1‐1 and Mpgun1‐2 knockout mutant spores are phenotypically indistinguishable from wild‐type spores during spore germination (Fig. 6b). This suggests that MpGUN1 is not required for chloroplast development under nonstressful conditions.

Interestingly, dark‐grown M. polymorpha spores develop green chloroplasts (Fig. S6). This indicates that unlike in Arabidopsis, proplastid‐to‐chloroplast differentiation in M. polymorpha spores is independent of light signals. Consequently, proplastid‐to‐etioplast and etioplast‐to‐chloroplast transitions do not take place in M. polymorpha spores and therefore we conclude that MpGUN1 is not involved in these processes.

MpGUN1 is not involved in global repression of PhANGs in response to inhibition of plastid translation

The Arabidopsis GUN1 protein is required for chloroplast retrograde signalling and the M. polymorpha GUN1 protein can restore retrograde signalling in the Arabidopsis gun1 mutant. Therefore, we hypothesized that GUN1 might also be involved in chloroplast retrograde signalling in M. polymorpha. To test this hypothesis, we tested if germinating Mpgun1 mutant spores are defective in chloroplast retrograde signalling when plastid translation is inhibited. Marchantia polymorpha spores are almost fully resistant to lincomycin. Therefore, we used another inhibitor of plastid translation, spectinomycin, which effectively blocks M. polymorpha spore development (Fig. S6). Spectinomycin is commonly used as a selectable marker in plastid‐transformation protocols, including in M. polymorpha (Chiyoda et al., ²⁰⁰⁶; Boehm et al., ²⁰¹⁶), and its mechanism of function is well‐characterised (Ellis, 1970; Parker et al., ²⁰¹⁴). Marchantia polymorpha spores that were germinated on media containing 500 μg ml⁻¹ spectinomycin expanded and developed chloroplasts, but never grew a rhizoid or underwent the first cell division (Fig. S6). This is consistent with the hypothesis that spectinomycin effectively inhibits plastid translation in M. polymorpha. We then checked if inhibition of plastid translation by spectinomycin during chloroplast development in M. polymorpha leads to reduced transcript levels of PhANGs or other transcripts that are regulated by plastid retrograde signalling in seed plants. To this end, we prepared transcriptomes of wild‐type and Mpgun1 knockout spores germinated in the absence or presence of spectinomycin. The spores were grown under long‐day conditions and harvested 48 h after plating. Read coverage across the MpGUN1 transcript was checked to verify the lack of intact MpGUN1 sequence in Mpgun1 mutant transcriptomes (Fig. S7). Wild‐type spores germinated on spectinomycin had reduced PhANG transcript levels compared to the control (Fig. 7; Table S3). However, the reduction in PhANG transcript levels was much less profound than that observed in wild‐type Arabidopsis under similar conditions, and Mpgun1 mutant spores were not defective in this response (Fig. 7; Table S3). GO terms associated with photosynthesis were over‐represented among the genes downregulated in response to spectinomycin in both wild‐type and Mpgun1‐1 mutant spores (Table S4). Furthermore, the set of genes that was differentially expressed in the Mpgun1‐1 mutant compared to wild‐type on spectinomycin was not enriched in photosynthesis‐associated GO terms, except for RbcS‐encoding transcripts, which were more abundant in Mpgun1‐1 compared to the wild‐type (Table S4). Under control conditions, the levels of PhANG transcripts did not differ markedly in wild‐type and Mpgun1 mutant spores (Fig. 7). These findings indicate that inhibition of plastid translation does not induce MpGUN1‐dependent global repression of nuclear encoded photosynthesis‐associated genes in M. polymorpha.

Fig. 7.

In Marchantia polymorpha inhibition of chloroplast translation by spectinomycin treatment or chloroplast stress imposed by norflurazon treatment does not result in GUN1‐mediated downregulation of the same set of genes as in seed plants. (a) Transcript levels of M. polymorpha orthologues of seed plant plastid retrograde signalling‐regulated genes LHCB1, LHCB2, RbcS, Beta‐CA (BCA), CHLI, CHLD, CHLH, CP12, FBPase, HEMA, HEMB, NR and PSBS in wild‐type and Mpgun1 mutant M. polymorpha spores germinated on control media or media supplemented with 500 μg ml⁻¹ spectinomycin. Transcript abundances are shown as log₂ fold‐difference compared to wild‐type M. polymorpha spores grown under control conditions. Each column represents an independent biological replicate transcriptome. (b–d) qPCR quantification of PhANG transcript levels in wild‐type and Mpgun1 mutant spores germinated on spectinomycin or norflurazon‐containing media. The transcript levels were normalized against MpEF1α and MpACT. Each dot represents an independent biological replicate, and bars illustrate the average of the three biological replicates. White asterisks indicate statistically significant differences between Mpgun1 and wild‐type for the same treatment (Tukey's HSD: *, P < 0.05; **, P < 0.01). The code and source data for reproducing (a) are obtainable from Dryad (10.5061/dryad.x0k6djhmk).

To independently verify that GUN1 is not required for inhibition of PhANG expression in response to inhibition of plastid translation in M. polymorpha, we carried out qPCR quantification of PhANG transcript levels in wild‐type and Mpgun1 mutant spores germinated on spectinomycin‐ or norflurazon‐containing media. Norflurazon is an inhibitor of carotenoid biosynthesis. Inhibition of carotenoid biosynthesis results in oxidative stress to the chloroplasts and leads to repression of PhANG expression in Arabidopsis. Marchantia polymorpha spores germinated on 5 μM norflurazon became bleached and their development arrested at a single‐cell stage (Fig. S6), suggesting that norflurazon probably also results in oxidative stress to the chloroplasts in M. polymorpha. We quantified the transcript levels of three PhANGs, LHCB2, RBCS and FBPase, in spectinomycin‐ and norflurazon‐treated wild‐type, Mpgun1‐1 and Mpgun1‐2 mutant spores. These PhANGs are present in M. polymorpha as multicopy genes. Therefore, primers were designed to amplify the most abundantly transcribed isoform in each gene group. Transcript levels of these PhANGs were reduced in spores germinated on spectinomycin or norflurazon, but Mpgun1‐1 or Mpgun1‐2 mutant spores were not clearly defective in this response (Fig. 7b–d). RBCS and FBPase transcripts were more abundant in gun1‐1 and gun1‐2 mutant spores germinated on spectinomycin compared to wild‐type spores (Fig. 7b–d). These differences were statistically significant, but we did not observe them in our transcriptome experiment, suggesting they may not be biologically significant. These findings indicate that GUN1 is not involved in strong transcriptional repression of PhANGs in response to inhibition of plastid translation in the liverwort M. polymorpha.

To elucidate the effects of the Mpgun1 mutation on gene expression during chloroplast development, we carried out further transcriptome analysis of germinating M. polymorpha wild‐type and gun1 mutant spores. In Arabidopsis the GUN1 protein accumulates in tissues in which proplastids differentiate into chloroplasts, including young cotyledons and shoot apical meristem 36–72 h after germination (Wu et al., ²⁰¹⁸). As proplastid to chloroplast differentiation in germinating M. polymorpha spores takes place during the first 48 h, we harvested spores 40 h after germination when the effects of the Mpgun1 mutation could be expected to be detectable.

Transcript levels of 575 genes were significantly different between Mpgun1 mutant and wild‐type spores (log₂ fold‐change > 1 or < −1, adjusted P‐value < 0.01, Table S5). Most notably, 11 out of 16 transcripts encoding putative chloroplast Chl a/b binding proteins homologous to Arabidopsis early light‐induced proteins (ELIPs) were less abundant in Mpgun1 spores compared to wild‐type spores. ELIPs are thylakoid‐associated stress‐induced proteins that appear earlier than other photosynthesis‐associated proteins during the early stages of deetiolation, and disappear before chloroplast development is completed (Grimm et al., ¹⁹⁸⁹). By contrast, six out of 10 transcripts encoding putative chloroplast alpha‐carbonic anhydrases were more abundant in Mpgun1 compared to the wild‐type. The homologues of neither of the gene families are significantly differentially expressed in comparisons of deetiolation of Arabidopsis wild‐type and Atgun1 plants (Hernández‐Verdeja et al., ²⁰²²). Conversely, many genes are differentially expressed in the Arabidopsis deetiolation experiment that are not differentially expressed in M. polymorpha (Hernández‐Verdeja et al., ²⁰²²). Thus, while significant transcriptome‐level differences can be observed between Mpgun1 and wild‐type spores, these are not similar to those observed in Arabidopsis.

Discussion

GUN1 has become the most studied PPR protein since the discovery of its connection to plastid retrograde signalling 15 yr ago (Koussevitzky et al., ²⁰⁰⁷). Despite numerous attempts to pin down its function in this process, the molecular action of GUN1 remains unclear. We attempted to take a different approach by using evolutionary conservation of sequence and function to guide our interpretation of how GUN1 might act. We have shown that GUN1 is probably present in every land plant with a chloroplast genome (as well as in the streptophyte algae most closely related to land plants) and equally highly conserved in terms of its structure and sequence. Therefore, it clearly has adaptive significance to almost all plants under natural conditions. This includes many nonphotosynthetic parasitic or mycoheterotrophic plants.

The question that remains is what is this essential function? Under laboratory conditions, GUN1 is entirely dispensable in Arabidopsis and Marchantia. The only conditions under which the gun1 phenotype is truly dramatic are nonphysiological treatments with inhibitors of chloroplast development such as norflurazon, spectinomycin or lincomycin, leaving its true physiological role unclear (Pogson et al., ²⁰⁰⁸). Nevertheless, the assumption has often been made that GUN1's essential, conserved role is connected to retrograde signalling. For example, the first identification of GUN1 genes in streptophyte algae led to the suggestion that the GUN1‐mediated retrograde signalling pathway evolved before the colonization of land by plants (de Vries et al., ²⁰¹⁸; Nishiyama et al., ²⁰¹⁸). We found that inhibition of chloroplast translation induced much less profound repression of PhANGs in the liverwort M. polymorpha than in Arabidopsis. Furthermore, Mpgun1 mutant spores were not clearly defective in this response. This suggests that, in Marchantia, GUN1 is not involved in a strong global transcriptional repression of PhANGs in response to inhibition of chloroplast translation. Marchantia polymorpha spores develop green chloroplasts when germinated in complete darkness. Similarly, in many gymnosperms and ferns, the differentiation of proplastids into chloroplasts is not light‐dependent (Raghavan, 1993; Ranade et al., ²⁰¹⁹). The ability to synthesize Chl in the dark correlates with the presence of chloroplast chl genes encoding subunits of a light‐independent protochlorophyllide reductase (Fujita, 1996). Knockout of chlB in M. polymorpha prevents protochlorophyllide reduction in the dark and reduces Chl accumulation under short‐day conditions (Ueda et al., ²⁰¹⁴). Flowering plants lack chl genes, and consequently accumulate protochlorophyllide in etioplasts, making deetiolation a critical process due to potential oxidative damage from photoreactive Chl precursors. GUN1 plays a key role in Arabidopsis during this etioplast–chloroplast transition (Hernández‐Verdeja et al., ²⁰²²). We speculate that GUN1‐associated retrograde signalling may have only evolved in flowering plants under this novel selection pressure. Our findings highlight the importance of functional characterization of proteins in divergent model systems when inferring evolutionary conservation of signalling pathways.

It thus seems likely that the conservation of GUN1 from streptophyte algae through bryophytes, lycophytes, ferns and gymnosperms is due to its involvement in some other essential physiological process and the later involvement in retrograde signalling is a secondary role. The distribution of GUN1 across land plants and algae rules out some processes as candidates. It was recently proposed that GUN1 regulates RNA editing via directly interacting with MORF2 protein (Zhao et al., ²⁰¹⁹). RNA editing has not been observed in streptophyte algae and probably only evolved in the lineage leading to land plants (Schallenberg‐Rüdinger & Knoop, 2016). Furthermore, complex thalloid (Marchantiid) liverworts, such as M. polymorpha, have entirely lost RNA editing (Groth‐Malonek et al., ²⁰⁰⁷). Moreover, MORF2 homologues are only present in seed plants (Schallenberg‐Rüdinger & Knoop, 2016; Gutmann et al., ²⁰²⁰). Therefore, the primordial role of GUN1 cannot be in regulating RNA editing. Similarly, GUN1 has been proposed to be involved in feedback regulation of protein import into plastids (Wu et al., ²⁰¹⁹; Tadini et al., ²⁰²⁰), but the loss of GUN1 in Rafflesiaceae (which still import proteins into plastids but cannot synthesize any plastid proteins) rather implies a role in chloroplast gene expression. Thus, while these recent propositions may be relevant to understanding GUN1's secondary role in retrograde signalling, they do not appear to help identify its primary function.

Despite the fact that GUN1 is not involved in a retrograde signalling pathway in Marchantia, MpGUN1 fully complements an Arabidopsis gun1 mutant, including its retrograde signalling phenotype. This surprising result implies to us that it is not the GUN1 protein itself that directly acts in the signalling pathway, but rather the result of GUN1's (conserved) action.

What is the primary (primordial) function of GUN1? The extreme conservation of GUN1's PPR motifs (exceeding even that of the PPR splicing factors EMB2654 and PPR4, which have a similarly long evolutionary history, Lee et al., ²⁰¹⁹) and of its SMR domain are consistent with it functioning as a typical PPR protein, that is as a sequence‐specific RNA binding protein. Based on the proposed relationships between PPR motif sequences and aligned nucleotides, we predicted an 11‐nucleotide binding site sequence putatively recognized by GUN1. This sequence is not found at conserved positions in chloroplast genomes (and is missing entirely from many), so we do not believe this sequence to accurately represent the GUN1 binding site(s), but it may resemble it. We can infer some probable features of the RNA that GUN1 might bind from evolutionary patterns. The presence of GUN1 in all land plants except those lacking a plastid genome indicates that if its target is an RNA, it is one present in all plastid genomes. This rules out almost all protein‐coding transcripts and tRNAs, leaving the rrn16 and rrn23 transcripts as the most likely targets. The extremely high conservation of the specificity‐determining residues would suggest that the GUN1 binding site is also highly conserved, even in species with highly divergent plastid genomes, again consistent with a highly conserved target transcript such as an rRNA. We predict that the essential conserved role of GUN1 will turn out to be in some way involved with the regulation of plastid ribosome biogenesis.

Author contributions

SH and IS designed the research, analysed the data and wrote the manuscript. SH performed the experiments and collected the data.

Supporting information

Fig. S1 Identification of GUN1 chloroplast transit peptide‐containing region.

Fig. S2 Multiple sequence alignment of 76 GUN1 protein sequences from diverse streptophyte algae and land plants.

Fig. S3 Identification of GUN1 sequences by hmmsearch scores.

Fig. S4 AtGUN1 protein structure prediction by AlphaFold.

Fig. S5 Location of DNA deletions in Mpgun1‐1 and Mpgun1‐2 CRISPR lines.

Fig. S6 Phenotype of Marchantia polymorpha wild‐type and Mpgun1 spores germinated under long day conditions in the absence (control) or presence of chemical inhibitors of plastid function (spec = spectinomycin 500 μg ml⁻¹ or nor = norflurazon 5 μM) or in complete darkness.

Fig. S7 Read coverage across the MpGUN1 transcript in Marchantia polymorpha spore transcriptome data.

Methods S1 Agrobacterium‐mediated transformation of the liverwort Marchantia polymorpha.

Methods S2 Generation of transgenic Mpgun1 CRISPR/Cas9 knockout lines.

Notes S1 Sequence maps of plasmids used for complementation of the Arabidopsis gun1 mutant (GenBank format).

Table S1 List of primers used in this study.

Table S2 Identification of GUN1 sequences by hmmsearch scores.

Table S3 Differentially expressed transcripts in wild‐type and Mpgun1 mutant spores grown in the presence or absence of spectinomycin.

Table S4 Gene ontology term enrichment analysis of wild‐type and Mpgun1 spores grown in the presence or absence of spectinomycin.

Table S5 Differentially expressed transcripts in 40‐h‐old Mpgun1 mutant spores compared to wild‐type spores.

Please note: Wiley Blackwell are not responsible for the content or functionality of any Supporting Information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office.

Data availability

Sequencing read data have been deposited at the Short Read Archive database at the National Center for Biotechnology Information under project nos. PRJNA800059 and PRJNA838206. The GUN1 HMM profile and all 893 identified GUN1 sequences aligned in Fasta format are available from Dryad (10.5061/dryad.x0k6djhmk). This Dryad repository also contains the code and source data for reproducing Figs 1, 7, S2 and S3; Tables S2 and S3.