Functional relationship between mTERF4 and GUN1 in retrograde signaling

Highlight

Identification and characterization of mitochondrial transcription termination factor 4 (mTERF4), which cooperates with GUN1 to regulate plastid gene expression and plastid retrograde signaling.

Abstract

Plastid-to-nucleus retrograde signaling plays an important role in regulating the expression of photosynthesis-associated nuclear genes (PhANGs) in accordance with physiological demands on chloroplast biogenesis and function. Despite its fundamental importance, little is known about the molecular nature of the plastid gene expression (PGE)-dependent type of retrograde signaling. PGE is a multifaceted process, and several factors, including pentatricopeptide repeat (PPR) proteins, are involved in its regulation. The PPR protein GUN1 plays a central role in PGE-dependent retrograde signaling. In this study, we isolated a mutant exhibiting up-regulation of CHLOROPHYLL A/B-BINDING PROTEIN (CAB) under normal growth conditions (named coe1 for CAB overexpression 1). The coe1 mutant has a single-base mutation in the gene for mitochondrial transcription termination factor 4 (mTERF4)/BSM/RUG2, which plays a role in regulating the processing of certain plastid transcripts. Defects in GUN1 or mTERF4 de-repressed the expression of specific plastid mRNAs in the presence of lincomycin (LIN). In wild-type plants, treatment with LIN or spectinomycin (SPE) inhibited processing of plastid transcripts. Comparative analysis revealed that in gun1 and coe1/mterf4, but not in wild-type, gun4, or gun5 plants, the processing of plastid transcripts and expression levels of Lhcb1 mRNA were affected in opposite ways when plants were grown in the presence of LIN or SPE. In addition, the coe1 mutation affected the intracellular accumulation and distribution of GUN1, as well as its plastid signaling activity. Taken together, these results suggest that GUN1 and COE1 cooperate in PGE and retrograde signaling.

Introduction

Chloroplasts evolved from a free-living cyanobacterium, following its endosymbiotic integration into a non-photosynthetic eukaryotic host (Douglas and Raven, 2003). However, a large amount of genetic information has been transferred to the nuclear genome during the evolution of chloroplasts (Abdallah et al., 2000; Rujan and Martin, 2001). At present, the plastid genome of higher plants contains only about 100 genes, which encode proteins for plastid gene expression (PGE) and photosynthesis (Abdallah et al., 2000). In contrast, more than 95% of chloroplastic proteins are now encoded in the nucleus, translated in the cytoplasm, and then imported into the organelle (Barbrook et al., 2006). Having been imported into the chloroplast, some nucleus-encoded plastid proteins interact with chloroplast-encoded proteins to form multi-subunit complexes involved in, for instance, the replication and expression of plastid-encoded genes, or in photosynthesis. To ensure correct stoichiometric assembly of these multi-subunit complexes, and enable their reorganization in response to developmental or environmental cues, the activities of the nuclear and chloroplast genomes must be coordinated through an intracellular signaling network (Kleine et al., 2009; Grimm et al., 2014). This network includes signaling pathways that originate in the nucleus (anterograde signaling) and the plastids (retrograde signaling) (Grimm et al., 2014). Anterograde signaling is involved mainly in the regulation of PGE (Jiao et al., 2007). By contrast, retrograde signaling conveys information about the developmental and metabolic state of the chloroplast to the nucleus, modifying nuclear gene expression in accordance with the current status of the organelle (Nott et al., 2006; Kleine et al., 2009; Pfannschmidt, 2010; Chi, et al., 2013). Plastid signals are classified into five distinct groups, depending on their source: (i) PGE; (ii) tetrapyrrole biosynthesis; (iii) reactive oxygen species generation; (iv) plastid redox state; and (v) metabolites, such as 3-phosphoadenosine 5′-phosphate, methylerythritol cyclodiphosphate, and β-cyclocitral (Gray et al., 2003; Baier and Dietz, 2005; Pesaresi et al., 2007; Mochizuki et al., 2008; Moulin et al., 2008; von Gromoff et al., 2008; Pfannschmidt et al., 2009; Bräutigam et al., 2009; Voigt et al., 2010; Galvez-Valdivieso and Mullineaux, 2010; Estavillo et al., 2011; Woodson et al., 2011; Ramel et al., 2012; Xiao et al., 2012; Kim and Apel, 2013; Terry and Smith, 2013).

Tetrapyrrole biosynthesis and PGE-dependent signaling are the best characterized types of plastid signaling (Nott et al., 2006). Much information on their operation has been obtained from studies on gun (genomes uncoupled) mutants of Arabidopsis thaliana (Nott et al., 2006). Six independent gun mutants have been identified. Five of them (gun2–gun6) are defective in different steps in the tetrapyrrole biosynthetic pathway (Susek et al., 1993; Mochizuki et al., 2001; Larkin et al., 2003; Strand et al., 2003; Woodson et al., 2011). GUN2 (heme oxygenase), GUN3 (phytochromobilin synthase), and GUN6 (FC1) operate in the heme branch of tetrapyrrole synthesis (Mochizuki et al., 2001). GUN4 and GUN5 are involved in the addition of Mg to protoporphyrin IX to produce Mg–protoporphyrin IX, which is the first dedicated step in chlorophyll biosynthesis. Unlike the other GUN proteins (Vinti et al., 2000; Mochizuki et al., 2001), GUN1 is a chloroplast-localized pentatricopeptide repeat (PPR) protein that integrates signals from multiple sources (Koussevitzky et al., 2007), although how it performs this function is unclear. Because most characterized PPR proteins are targeted to mitochondria or plastids and are involved in organellar gene expression, with known functions in RNA editing, processing, and translation (Delannoy et al., 2007), GUN1 might integrate plastid signaling on the basis of a function associated with PGE.

Besides PPR proteins, proteins for the mitochondrial Transcription tERmination Factor (mTERF) family have been found to play important roles in regulating the processing of plastid transcripts (reviewed by Kleine and Leister, 2015). For example, the Arabidopsis bsm mutant, which is defective for an mTERF, is albinotic and displays global defects in PGE and embryo development (Babiychuk et al., 2011). More specifically, BSM is required for group II intron splicing of some plastid transcripts, suggesting that defects in the processing of plastid gene transcripts can globally suppress PGE (Babiychuk et al., 2011). Similarly, complete loss of other PGE regulators, such as translation initiation factor 2 (Miura et al., 2007), elongation factor G (Albrecht et al., 2006), and peptide release factors 1 and 2 (Meurer et al., 2002; Motohashi et al., 2007), results in severe suppression of PGE, leading to an albinotic or even embryo-lethal phenotype. In addition, a drastic fall in PGE can trigger PGE-dependent signaling and lead to the inhibition of photosynthesis-associated nuclear gene (PhANG) expression, even under normal growth conditions, as shown by phenotypic analysis of the Arabidopsis prors1 mutant, which is defective for a prolyl-tRNA synthetase (Pesaresi et al., 2006). However, which specific steps in PGE lead to PGE-dependent signaling is still unknown. GUN1 seems to link PGE with retrograde signaling, and contains an SMR domain found in proteins involved in DNA repair and recombination (Koussevitzky et al., 2007), in addition to its PPR domain. In fact, the domain of GUN1 that contains the PPR and SMR motifs binds DNA in vitro (Koussevitzky et al., 2007). However, the gun1 mutation does not significantly affect plastid mRNA profiles or PGE under normal growth conditions (Woodson et al., 2013), suggesting that GUN1 is either not directly involved in the regulation of PGE or that its function in PGE becomes manifest only under certain conditions. PGE signaling normally represses nuclear Lhcb expression in response to perturbations in chloroplast protein production. However, in gun1 plants, expression of Lhcb is slightly higher than in the wild type (WT), even in the absence of overt inhibition of PGE, implying that GUN1 has a subtle effect on PhANG expression and possibly also PGE under normal growth conditions (Sun et al., 2011).

In order to identify novel components of retrograde signaling, we developed an ethyl methanesulfonate (EMS) screen for mutants that displayed enhanced activity of the promoter of the Lhcb1 gene for chlorophyll a/b-binding protein (CAB) under normal growth conditions. A series of coe (CAB overexpression) mutants was isolated, and the causative mutation in one of them (coe1) was localized to position 844 of the AT4G02990 gene, thus demonstrating that COE1 codes for BSM/mTERF4. Like gun1, coe1 showed increased Lhcb mRNA expression under normal growth conditions and displayed a weak gun phenotype in the presence of the herbicide norflurazon (NF), which inhibits carotenoid synthesis and causes photo-oxidative damage. Defects in GUN1 or mTERF4 decreased the expression of certain plastid mRNAs in the presence of the antibiotic lincomycin (LIN) which, like spectinomycin (SPE), inhibits protein synthesis in the chloroplast. Comparative analysis revealed that in gun1 and coe1/mterf4, but not in WT, gun4, or gun5 plants, the processing of plastid mRNAs and expression levels of Lhcb1 were affected in opposite ways when plants were grown in the presence of LIN or SPE. In addition, COE1 has an impact on the intracellular accumulation and distribution of GUN1, as well as on its plastid signaling activity.

Materials and methods

Plant materials and growth conditions

The following A. thaliana mutants in the Columbia (Col-1) ecotype were obtained from the Arabidopsis Biological Resource Center: gun1 (SAIL_742_A11, a T-DNA insertion mutant; Sun et al., 2011) and gun4 (SALK_011461, a T-DNA insertion mutant). Homozygous lines were identified by PCR using gene-specific and T-DNA-specific primers (Supplementary Table S1 at JXB online). Mutants were backcrossed to WT plants three times before generating double mutants to segregate out additional mutations. To generate oeCOE1/coe1 and oeGUN1-GFP/coe1 strains, pK7FWG2-COE1 and pB7FWG2-GUN1 (both driven by the cauliflower mosaic virus 35S promoter), respectively, were introduced into lines homozygous for coe1, and gun1 was crossed with coe1 to obtain the double mutant coe1 gun1.

All mutant and WT plants were grown in climate chambers at 22 °C and 120 µmol photons m^–2 s^–1 on a 12h light/12h dark regime. For the NF, LIN and SPE treatments, surface-sterilized mutant and WT seeds were plated on 1/2 Murashige and Skoog (1/2 MS; Murashige and Skoog, 1962) medium (PhytoTechnology Laboratories, LLC™, USA) containing 1% sucrose and 0.8% agar supplemented with either 5 µM NF (Sandoz Pharmaceuticals, Vienna, Austria), 220 µg ml^–1 of LIN (Sigma, St Louis, MO, USA), or 500 µg ml^–1 of spectinomycin (Sigma).

RNA extraction, Northern blotting, and quantitative real-time reverse transcription PCR (qRT-PCR)

RNA was extracted with a Maxwell^® 16 LEV simplyRNA Purification Kit (Promega, WI, USA). Northern blot analysis was performed under stringent conditions, according to Sambrook and Russell (2001). Probes complementary to nuclear or chloroplast genes were used for the hybridization experiments. Primers used to amplify the probes are listed in Supplementary Table S1. All probes used were cDNA fragments labeled with ³²P. Signals were detected with a phosphoimager (Typhoon; GE Healthcare, Munich, Germany) and quantified using the program ImageJ. For qRT-PCR, 1 μg aliquots of total RNA, treated with DNase I (Roche Applied Science) for at least 30min, were utilized for first-strand cDNA synthesis using iScript reverse transcriptase (Bio-Rad) according to the supplier’s instructions. The qRT-PCR profiling was carried out on an iCycler Q5 real-time PCR system (Bio-Rad), using the oligonucleotide sequences listed in Supplementary Table S1. Actin was used as an internal standard. Data from three biological and three technical replicates were analyzed with Bio-Rad iQ5 software (version2.0).

Polysome analysis

For polysome analysis, polysomes were isolated from 5-d-old seedlings according to Barkan (1993), with certain modifications. Approximately 0.3g of seedlings were frozen and ground in liquid nitrogen to a fine powder, 1ml of polysome extraction buffer [0.2M Tris/HCl (pH 9.0), 0.2M KCl, 35mM MgCl₂, 25mM EGTA, 0.2M sucrose, 1% Triton X-100, 2% polyoxyethylene-10-tridecyl ether, 0.5mg ml^–1 of heparin, 100mM β-mercaptoethanol, 100 µg ml^–1 of chloramphenicol, and 25 µg ml^–1 of cycloheximide] was added, and the tissue was ground until thawed. The samples were incubated on ice for 10min and pelleted by centrifugation for 7min at 14 000g. Sodium deoxycholate was added to the supernatant to a final concentration of 0.5%, after which the samples were kept on ice for 5min and then centrifuged at 12 000g for 15min. Next, 0.5ml samples of the supernatant were layered onto 4.4ml sucrose gradients that were prepared, centrifuged, and fractionated as described previously (Barkan, 1993). The samples were kept at 4 °C during preparation. The RNA in each fraction was isolated, separated, and transferred onto nylon membranes (Amersham Phamacia Biotech), which were probed with ³²P-labeled probes. Signals were detected with a phosphoimager (Typhoon; GE Healthcare).

Run-on analysis

Run-on analysis was performed according to Zoschke et al. (2007). Intact chloroplasts from 3g of leaves were isolated in a 40/70% Percoll step-gradient. Chloroplasts (5×10⁷) were used in in vitro transcription experiments, performed at 25 °C for 15min in 50mM Tris/HCl (pH 8.0), 10mM MgCl₂, 10mM β-mercaptoethanol, 20U RNase inhibitor, and 0.2mM ATP, GTP, and CTP, in the presence of α-³²P-UTP (10 μCi μl^–1). Newly synthesized, labeled RNA was extracted and hybridized overnight at 42 °C to DNA fragments (1 μg) dot blotted in duplicate onto nylon membranes. The primers used for the generation of DNA probes are listed in Supplementary Table S1. Signals were detected with a phosphorimager (Typhoon; GE Healthcare) and quantified using the program ImageJ.

Constructs for plant transformation and yeast one-hybrid assays

To generate the pK7FWG2-COE1 and pB7FWG2-GUN1 plasmids (in which green fluorescent protein (GFP) is fused to c-terminus of GUN1), COE1 and GUN1 cDNAs were amplified using the primer pair COE1-GFPs and COE1-GFPa for COE1, and GUN1-GFPs and GUN1-GFPa for GUN1, respectively (Supplementary Table S1). The PCR products were purified, and BP and LR Clonase reactions (GATEWAY Cloning; Invitrogen) were performed according to the manufacturer’s instructions to yield the final constructs pK7FWG2-COE1 and pB7FWG2-GUN1.

To construct the luciferase (LUC) reporter, the promoter region of Lhcb1.1 (–976 to –30 nt) was amplified by PCR using gene-specific primers (Supplementary Table S1). The resulting DNA fragments were digested with HindIII and BamHI, and inserted into the corresponding sites in the 35S:LUC vector (Hellens et al., 2000), replacing the 35S promoter to produce PUC-P _Lhcb1.1 :LUC. The vector was subsequently digested with HindIII and SacI and inserted into the cognate sites of the binary vector pCAMBIA1301, thus generating the pCAMBIA1301- P _Lhcb1.1 :LUC.

To generate pGBKT7-GUN1 and pGADT7-COE1, the GUN1 and COE1 cDNAs were PCR amplified from Arabidopsis cDNA using PrimeSTAR^® HS DNA polymerase and then inserted into the EcoRI and BamHI sites of the pGBKT7 and pGADT7 vectors, respectively.

Plant transformation

The pK7FWG2-COE1 and pB7FWG2-GUN1 constructs were transformed into Agrobacterium tumefaciens strain GV3105 via electroporation, and the resulting strains bearing the pK7FWG2-COE1 (expressing Kan resistance in planta) and pB7FWG2-GUN1 (Basta resistance) constructs were introduced into coe1. T1 transgenic plants were selected by screening on Basta for GUN1 and on kanamycin for COE1. Homozygous transgenic plants were used in all experiments.

Yeast two-hybrid assay

The plasmids pGBKT7-GUN1 and pGADT7-COE1 were co-transformed into the yeast strain AH109 using standard techniques. Growth of diploid yeast colonies on SD–His–Leu–Trp–Ade plates supplemented with 40 μg ml^–1 of X-α-Gal would reveal a GUN1–COE1 interaction.

Mutagenesis and mutant isolation

P _Lhcb1.1 :LUC seeds were mutagenized with EMS (Redei and Koncz, 1992). F2 seeds were sterilized and planted individually in 100 x 10mm plates (150 to 200 seeds per plate) containing 1/2 MS, 1% sucrose, and 0.8% agar (pH 5.7). Five-day-old seedlings grown under light were sprayed with luciferin and immediately placed in the dark (see below) to remove the chlorophyll fluorescence, which was monitored with a CCD camera. After 5min in the dark, an LUC image was acquired with a 5min exposure to identify coe mutants. Putative coe mutants were also transferred to soil. To eliminate false positives, putative mutants were rescreened.

LUC analysis by CCD imaging

Imaging of the activity of the firefly LUC reporter requires application of the exogenous substrate luciferin. Luciferin (Promega) was dissolved in sterile water and stored frozen in small aliquots as a 100mM stock solution. A working solution of 1mM luciferin in 0.01% Triton X-100 was applied uniformly to seedlings by spraying five times. For LUC imaging, the seedlings were kept for 5min in the dark after application of luciferin. The imaging system consisted of a high-performance CCD camera mounted in a dark chamber, a camera controller, and a computer. Image acquisition and processing were performed with the WinView software. Exposure time was 5min, unless stated otherwise.

Positional cloning

To generate the mapping population for COE1, coe1 mutant plants were crossed to WT Arabidopsis plants of the Landsberg erecta (Ler) ecotype. A total of 1600 coe1 mutant plants were selected from the segregating F2 population based on high luminescence (expression of P _Lhcb1.1 :LUC) and a pale green phenotype. Genomic DNA from these plants was extracted and analyzed for co-segregation with respect to simple sequence length polymorphism (SSLP) markers. These markers were developed according to Lukowitz et al. (2000). Primer pairs for fine mapping of COE1 are listed in Supplementary Table S1. In addition, nucleotide differences between Ler and Col ecotypes were identified by direct sequencing of the ORF of T4I9.

Protein extraction and immunoblot analysis

Five-day-old seedlings were harvested from 1/2 MS plates, and total proteins were prepared according to Sun et al. (2011). For immunoblot analyses, the proteins were fractionated by SDS-PAGE (15% acrylamide) (Schägger and von Jagow, 1987). Subsequently, proteins were transferred to polyvinylidene difluoride membranes (Ihnatowicz et al., 2004) and probed with appropriate antibodies. Signals were detected by enhanced chemiluminescence (GE Healthcare).

Thylakoid membrane preparation and blue native polyacrylamide gel electrophoresis (BN-PAGE)

Thylakoid membranes were prepared as described by Zhang et al. (1999). Arabidopsis leaves were ground in an ice-cold isolation buffer containing 400mM sucrose, 50mM HEPES/KOH (pH 7.8), 10mM NaCl, and 2mM MgCl₂, filtered through two layers of cheesecloth, and centrifuged at 5000g for 10min. The thylakoid pellets were washed with isolation buffer, recentrifuged, and finally suspended in isolation buffer. The chlorophyll content was determined spectrophotometrically according to the method described by Porra et al. (1989). BN-PAGE was carried out as described previously (Schägger and Cramer, 1994). The thylakoid membranes were solubilized with 1% (w/v) dodecyl-β-maltoside in 20% glycerol, 25mM BisTris/HCl (pH 7.0), at 0.5mg chlorophyll ml^–1 for 10min at 4 °C, and unsolubilized material was removed by centrifugation at 12 000g for 10min. The supernatant was combined with 0.1 vols of 5% Serva blue G in 100mM BisTris/HCl (pH 7.0), 0.5M 6-amino-n-caproic acid, 30% (w/v) glycerol, and loaded onto 6–12% acrylamide gradient BN gels.

Chlorophyll fluorescence analysis

In vivo chlorophyll a fluorescence of whole seedlings was recorded using an imaging chlorophyll fluorimeter (ImagingPAM; Walz, Germany). Dark-adapted plants were exposed to a pulsed, blue measuring beam (1 Hz, intensity 4; F0) and a saturating light flash (intensity 4) to obtain Fv/Fm. A 10min exposure to actinic light (80 μmol photons m⁻² s⁻¹) was then used to calculate the steady-state magnitudes of the quantum yield of photosystem II (PSII) (ɸ_II), non-photochemical quenching of chlorophyll fluorescence (NPQ), and the fraction of open PSII centers (qL).

Accession numbers

Arabidopsis Genome Initiative locus identifiers for the genes mentioned in this article are AT4G02990 (mTERF4/COE1/BSM/RUG2), AT2G31400 (GUN1), AT3G59400 (GUN4), and AT5G13630 (GUN5).

Results

Identification of the coe1 mutant

To identify components of plastid signaling under physiological conditions, we transformed Arabidopsis plants with a construct containing the Lhcb1.1 [LIGHT HARVESTING CHLOROPHYLL A/B-BINDING PROTEIN 1.1 or CHLOROPHYLL A/B-BINDING PROTEIN 2 (CAB2)] promoter fused to the firefly LUC coding sequence. The resulting P _Lhcb1.1 :LUC plants (and the P _35S :LUC transgenics used as controls) emitted luminescence under normal growth conditions (Supplementary Fig. S1 at JXB online). In contrast to P _35S :LUC plants, P _Lhcb1.1 :LUC plants displayed significantly lower luminescence levels when grown in the presence of either LIN or NF (Supplementary Fig. S1), as expected, because both agents are known to activate plastid signaling (Dong et al., 2007; Sun et al., 2011; Kindgren et al., 2012).

P _Lhcb1.1 :LUC plants, designated in the following as WT*, were then mutagenized with EMS, and mutants with increased luminescence under normal growth conditions were identified and classified as putative chlorophyll a/b-binding protein overexpression (coe) mutants (Fig. 1A, B). We identified more than 100 coe mutants, and one of them, designated coe1, was chosen for detailed characterization. Compared with WT*, coe1 mutants showed increased luciferase expression under normal growth conditions (Fig. 1C, D). Interestingly, coe1 leaves were pale yellow in color, and this trait first appeared in young seedlings (Supplementary Fig. S2 at JXB online). As suggested by the pale yellow phenotype, the maximum quantum yield of PSII (Fv/Fm) at very early developmental stages was substantially reduced in the coe1 strain relative to WT* but increased gradually as the seedlings got older (Fig. 2A and Supplementary Fig. S2). The higher level of luciferase expression in coe1 was particularly evident on days 3 and 4 after germination (Fig. 2B). This phenotype suggested that COE1 may play a special role during early seedling development.

Fig. 1.

Identification of coe mutants. (A) Four-day-old F2 seedlings of EMS-mutagenized P _Lhcb :LUC (WT*) plants. (B) Luminescence image of the seedlings shown in (A). A putative coe mutant is highlighted by a circle. (C) Plate with 4-d-old WT* and coe1 seedlings. (D) Luminescence image of the seedlings shown in (C). Note that the plate shown was the initial plate of the M2 generation from which coe1 was isolated, and this plate was by coincidence one that contained many M2 coe1 mutants, while on many other plates we did not identify coe mutants.

Fig. 2.

Characterization of the developmental changes in Fv/Fm and LUC activity in coe1 and WT* seedlings. (A) Changes in Fv/Fm from day 3 to day 6 after germination. (B) Changes in LUC activity were measured from day 1 to day 6 after germination. Each data point represents the mean value for a sample of 48 seedlings. Data were expressed as mean±SD of three independent experiments. Asterisks indicate P<0.05 (Student’s t-test).

The coe1 mutant was also backcrossed to the WT*. The resulting F1 plants all exhibited a WT* phenotype (Table 1). The F2 progeny of the selfed F1 segregated at approximately 3:1 for WT:mutant (Table 1), indicating that coe1 is a recessive mutation in a single nuclear gene.

Table 1.

Analysis of coe1 (WT*×coe1)^a

Generation	Seedlings tested	Mutant	WT*	P value
F1	23		23
F2	465	113	352	˂0.05b

^a Female×male.

^b For 3:1 segregation.

PhANG expression in coe1

To determine whether the coe1 mutation also affected endogenous Lhcb1.1 gene expression, we extracted total RNA from coe1 and WT* seedlings grown on standard 1/2 MS plates, and performed Northern blotting and real-time PCR analyses. Fig. 3A shows that the steady-state levels of Lhcb1.1 mRNA were higher in coe1 than in WT* under normal growth conditions. Analysis of the changes in Lhcb1.1 expression during development showed that enhanced Lhcb1.1 transcript accumulation was seen at all time points between days 3 and 7 after germination, although the difference was especially evident on days 3 and 4 (Supplementary Fig. S3 at JXB online). In agreement with results obtained for LUC activity and Fv/Fm values, Lhcb1.1 transcript levels in 7-d-old coe1 were close to those of the WT* (Supplementary Fig. S3). We therefore used 3- to 5-d-old seedlings for all subsequent experiments.

Fig. 3.

The coe1 mutant displays a gun phenotype. (A–D) Levels of Lhcb1.1 mRNA in 5-d-old WT* and mutant (gun1, gun4, gun5, coe1, and rug2-1) and complemented lines (COM.-1 and COM.-2) seedlings grown under standard conditions (Control; A) or in the presence of LIN (B), NF (C), or SPE (D). (E, H) Signals were quantified using ImageJ software. Levels in mutants are expressed as a percentage of the WT* value. Data were expressed as means±SD of three independent experiments. *P<0.05, **P<0.01 (Student’s t-test versus WT*). (F, G) Levels of rbcL mRNA in WT* and mutant (gun1, gun4, gun5, and coe1) seedlings after treatment with LIN (F) or NF (G). The levels of Lhcb and rbcL mRNA were determined by RNA gel blots and real-time PCR analyses. The relative expression values obtained with real-time PCR analyses are indicated in the upper panel. Coomassie Brilliant Blue-stained ribosomal RNA served as the loading control (CBB).

To further examine the effects of the coe1 mutation on PhANG expression, we analyzed the expression of Lhcb1.1 in the presence of LIN, NF, or SPE (Fig. 3B–D). While after LIN treatment the expression of Lhcb1.1 in the gun1 and coe1 mutants was about 750 and 250% of WT* levels, respectively, coe1 plants treated with NF displayed almost gun1-like levels of Lhcb1.1 mRNA (Fig. 3C, E). In addition, the expression of other PhANGs (Lhcb2, Lhcb3, Lhcb4, RbcS1a, and CA1) was also slightly higher in coe1 than in WT* in the presence of NF (Supplementary Fig. S4 at JXB online), suggesting that COE1 may be involved in modulating retrograde signaling. In gun1, Lhcb1.1 levels were significantly higher than in WT* in the presence of LIN, SPE, or NF. Unlike Lhcb1.1, accumulation of the rbcL transcript was strongly inhibited in all genotypes by LIN (Fig. 3F, H). Interestingly, levels of rbcL mRNA were significantly higher in gun4 and gun5 than in the other genotypes in the presence of NF (Fig. 3G, H), suggesting that gun4 and gun5 mitigate the effects of NF on the expression of plastid mRNAs, as has been shown for gun5 by Ankele et al. (2007).

Accumulation of photosynthetic proteins is differentially affected in coe1

The pale yellow leaf coloration and the photosynthetic defect (Fig. 1C, Supplementary Fig. S5 at JXB online) suggested that chloroplast development in coe1 is impaired. When grown on soil, the lower photosynthetic efficiency of the coe1 mutant resulted in a decrease in growth rate (Supplementary Fig. S6 at JXB online). The decrease in Fv/Fm levels observed in the coe1 mutant during early development (Fig. 2A, Supplementary Fig. S5) might be a consequence of altered thylakoid protein levels. To address this, the steady-state levels of chloroplast proteins in 5-d-old plants was determined by immunoblot analyses with antibodies raised against representative chloroplast proteins. In fact, semi-quantitative determination of protein levels by applying a dilution series and quantification with ImageJ showed that the levels of the plastid-encoded PSI reaction center protein psaA and subunits of the PSII core subunits D1, D2, CP47, and CP43 were reduced to about 15–25% of WT* levels in coe1 (Fig. 4A). Nucleus-encoded subunits of the oxygen-evolving complex (PsbO), light-harvesting complex II (LhcB1), and ferredoxin:NADP(H) oxidoreductase (FNR) accumulated in coe1 to ~75%, ~30%, and ~15% of WT* levels, respectively (Fig. 4A). In contrast, the level of chloroplast ATPase protein CF₁β was virtually equivalent to that of WT* (Fig. 4A). In some respects, coe1 behaves like the known gun mutants. Like those of coe1, the cotyledons of the gun4 and gun5 mutants were pale yellow in color (Supplementary Fig. S5). Interestingly, the effect of gun4 on the accumulation of chloroplast proteins was very similar to that of coe1: levels of D1, D2, CP43, CP47, PsaA, and Lhcb1 were reduced to about 10–30% of WT* levels (Fig. 4A), and amounts of FNR, RbcS1a and RbcL declined to about 10, 15, and 40% of WT* levels, respectively, in gun4 (Fig. 4A). The gun5 mutant showed less pronounced effects on the accumulation of the chloroplast proteins. In gun5, D1, D2, CP43, CP47, Lhcb1, PsbO, FNR, PsaA, RbcS1a, and RbcL accumulated to about 35–95% of WT* levels (Fig. 4A), while amounts of CF₁β in gun5 were almost identical to WT* (Fig. 4A). Furthermore, in the gun1 mutant, all investigated photosynthesis proteins accumulated to or nearly to WT* levels (Fig. 4A). The significant decrease of thylakoid membrane proteins suggested that thylakoid membrane complexes are also altered in coe1. To investigate this, Blue Native (BN) gel analysis was performed. Indeed, the levels of thylakoid membrane complexes, such as PSI and PSII, were clearly reduced in coe1 and rug2-1 compared with WT* and gun1 (Fig. 4B, C).

Fig. 4.

Effects of the coe1 mutation on levels of thylakoid proteins and thylakoid complexes. (A) Selected thylakoid proteins were quantified on immunoblots. Total proteins were extracted from 5-d-old mutants (gun1, gun4, gun5, coe1, and rug2-1), complemented lines (COM.-1 and COM.-2), and WT* seedlings, and fractionated by SDS/urea-PAGE. Blots were probed with anti-D1, -D2, -CP43, -CP47, -PsbO, -Lhcb1, -PsaA/B, -FNR, -RbcL, -RbcS, and -CF₁β antibodies. Signal intensities (quantified by the ImageJ software), expressed relative to those of the WT* (=100), are indicated below each panel. (B) Thylakoid membranes (10 µg of chlorophyll) from WT, gun1, coe1, and rug2-1 leaves were solubilized and separated by BN-gel electrophoresis. The positions of protein complexes representing monomeric PSI and dimeric PSII (band I: PSI-M and PSII-D), Cytb6f/ATPase/monomeric PSII (band II), light-harvesting complex II (band III: LHCII), CP43 minus PSII (band IV: CP43-PSII), trimeric LHCII (band V: LHCII-T), and monomeric LHCII (band VI: LHCII-M) are indicated. (C) The BN-gel was stained with Coomassie Brilliant Blue (CBB) to show the level of protein in each of the protein complexes.

The dramatic reductions in chloroplast proteins observed in coe1 could be due to decreased transcription of the corresponding genes. To assess this possibility, plastid-encoded transcripts were detected by RNA gel-blot hybridization, and levels of the transcripts were determined semi-quantitatively by applying a dilution series followed by quantification with ImageJ. This confirmed that levels of psbA (encoding the D1 subunit of PSII), psbB (encoding the CP47 subunit of PSII), psbC (encoding the CP43 subunit of PSII), psbD (encoding the D2 subunit of PSII), and rbcL mRNAs in the coe1 mutant were almost identical to that of WT* (Fig. 5A, B). In contrast, levels of atpB, psaB, and petA transcripts (which encode the β-subunit of ATP synthase, the B subunit of PSI, and cytochrome f, respectively) were increased in coe1 relative to WT* (Fig. 5A, B). In gun4, psbA, psbB, psbC, psbD, psaB, petA, and rbcL transcripts were under-represented, which was consistent with the changes in protein levels (Fig. 5A, B), whereas amounts of atpB and psaB in gun4 were comparable to WT* (Fig. 5A, B). In gun1 and gun5, levels of psbA, psbB, psbC, psbD, rbcL, and psaB RNAs were slightly decreased (Fig. 5A, B), but atpB and petA transcripts were increased relative to WT* (Fig. 5A, B). Amounts of nucleus-encoded PsbO mRNA in gun1, gun4, gun5, and coe1 were similar to WT*, while nucleus-encoded RbcS1a mRNA levels were slightly increased in gun5, coe1, and rug2-1 (Fig. 5A, B). Taken together, these results suggested that COE1 plays an important role in PGE.

Fig. 5.

Quantification of plastid gene transcripts. (A) Aliquots (10 µg) of total RNA from WT*, mutant (gun1, gun4, gun5, coe1, and rug2-1), and complemented line (COM.-1 and COM.-2) plants were size fractionated by agarose-gel electrophoresis, transferred to a nylon membrane, and probed with ³²P-labeled cDNA probes derived from psbA, psbB, psbC, psbD, atpB, petA, psaB, rbcL, RbcS1a, and PsbO. Coomassie Brilliant Blue-stained ribosomal RNA served as the loading control (CBB). (B) The level of each transcript was quantified using ImageJ software, and expressed relative to that of WT*(=100%). Data were expressed as means±SD of three independent experiments. *P<0.05, ** P<0.01 (Student’s t-test versus WT*).

Compared with WT*, the levels of chloroplast psaB, petA, and atpB transcripts were elevated (Fig. 5); however, the levels of the encoded proteins were reduced in coe1 (Fig. 4), suggesting that a defect in translation of chloroplast transcripts might be responsible for reduced accumulation of the corresponding proteins. To test this, the association of psaB, petA, and atpB mRNA with polysomes was analyzed (Supplementary Fig. S7 at JXB online). To this end, plant extracts were fractionated in sucrose gradients under conditions that preserve polysome integrity, and mRNAs were identified by hybridization with specific probes. As shown in Supplementary Fig. S7, the amounts of petA, psaB, and atpB mRNA assembled with ribosomes (fractions 7–12) was generally similar in WT* and coe1, whereas mRNA accumulation of polycistronic versions of these transcripts, especially the transcript of petA, was higher in the non-polysomal fractions (fractions 1–6) in coe1. These results suggested that the majority of these mRNAs in coe1 chloroplasts were not engaged in translation, which accounts for the reduction in synthesis of chloroplast proteins. The distribution of petA, psaB, and atpB mRNAs in non-polysomal and polysomal fractions of gun1 was similar to that of WT* (Supplementary Fig. S7), suggesting that gun1, as expected, does not affect the translation of these proteins.

Positional cloning of COE1

To map coe1 genetically, homozygous coe1 mutant plants (Col ecotype) were crossed with WT plants of the Ler ecotype. The resulting F1 plants were selfed and homozygous coe1 mutant plants were selected from the segregating F2 population based on their pale yellow phenotype. A survey of representative molecular markers from each of the five Arabidopsis chromosomes localized COE1 to chromosome IV (Fig. 6). Further analysis showed that COE1 is closely linked with the SSLP marker nga8. Several new SSLP markers were selected from the Arabidopsis Mapping Platform (http://amp.genomics.org.cn/) and TAIR (http://www.arabidopsis.org/) between the markers ciw6 and nga8. Fine-scale mapping using these new markers delimited COE1 to the BAC clone T4I9 (Fig. 6). Candidate ORFs on T4I9 were sequenced in WT* and coe1 mutant plants, revealing a single-nucleotide substitution in AT4G02990 in the coe1 mutant. This mutation was predicted to lead to a change from Arg to Trp at position 282 of the protein (Fig. 6).

Fig. 6.

Positional cloning of the COE1 gene. (A) Physical mapping of COE1. Genetic mapping delimited COE1 to the bacterial artificial chromosome (BAC) clone T4I9. The coe1 mutation was identified by sequencing all predicted genes on this BAC in the coe1 mutant and comparing them with their counterparts from WT plants. (B) Structure of COE1 and position of the coe1 mutation. Positions are indicated relative to the initiation codon. The filled box indicates the ORF, and lines between boxes indicate introns. The nucleotide substitution in AT4G02990 created by the coe1 mutation is marked with an asterisk. Arrows show the T-DNA insertion positions of rug2-1 and rug2-2.

To prove that the alteration in AT4G02990 was responsible for the phenotypes observed in the coe1 mutant, the isolated full-length AT4G02990 cDNA was fused to the 35S promoter in the plant transformation vector pK7FWG2. The construct was introduced into coe1 mutant plants via Agrobacterium-mediated transformation. Transformants were selected based on kanamycin resistance. Both kanamycin resistance and pale yellow leaf coloration segregated in the T2 generation. All kanamycin-resistant plants exhibited the WT* phenotype. A deeper analysis of two complemented coe1 mutant plants in the T2 generation, COM.-1 and COM.-2, demonstrated the full restoration of the WT phenotype: Luminescence of P _Lhcb1.1 :LUC, Fv/Fm, ɸ_II, and NPQ, as well as growth of COM.-1 and COM.-2 were the same as in WT plants (Supplementary Fig. S8 at JXB online).

The coe1 mutation is a new allele of mterf4/bsm/rug2

Arabidopsis AT4G02990 encodes BELAYA SMERT (BSM; Babiychuk et al., 2011)/RUGOSA2 (RUG2; Quesada et al., 2011), a plastid-localized mTERF protein, which has been designated mTERF4 in the systematic nomenclature of Kleine (2012). The mTERF4 protein is essential for normal plant development and for maintenance of adequate levels of transcripts in both mitochondria and chloroplasts (Babiychuk et al., 2011; Quesada et al., 2011). In rug2-1, the conserved proline residue at position 420 is replaced by leucine, and the mutant shows a variegated phenotype similar to var1 and var2 (Quesada et al., 2011). The maize RUG2 ortholog ZmmTERF4 is localized to the chloroplast, and an allelic series of Zmmterf4 mutants showed pale yellow/green and albino phenotypes (Hammani and Barkan, 2014). mTERF4 contains 10 mTERF motifs between aa 98 and 444, and the Arg282 to Trp mutation in COE lies in the fifth of these. In order to compare the effects of coe1 and rug2 on AT4G02290/BSM/RUG2 function, we analyzed growth and photosynthesis in these genotypes (Quesada et al., 2011). In fact, growth rate, cotyledon coloration, and photosynthesis parameters (Fv/Fm, ɸ_II, NPQ, and qL) were very similar in coe1, rug2-1, and rug2-2 (Fig. 7). In addition, as in coe1, the levels of chloroplast proteins D1, CP43, and CP47 in rug2-1 were equivalent to only about 25% of WT amounts, while PsbO in rug2-1 was reduced to about 60% of WT levels, but RbcL protein amounts were not affected in rug2-1 (Fig. 4). Furthermore, analysis of the expression of Lhcb1.1 revealed that Lhcb mRNAs were also slightly up-regulated in rug2-1, both under normal growth conditions and in the presence of NF and LIN (Fig. 3C, E). These results confirmed that coe1 is a new allele of bsm/rug2.

Fig. 7.

The coe1, rug2-1, and rug2-2 mutants show similar photosynthetic defects. (A) Diagram of the plate layout showing the position of each genotype. (B) Bright-field image of coe1, rug2-1, rug2-2, and WT* seedlings grown on a 1/2 MS plate. (C–F) Fv/Fm (C), ɸ_II (D), NPQ (E), and qL (F) values for the seedlings shown in (B) were determined as described in Materials and methods. Signal intensities correspond to the color scale at the bottom of the panel. Results are shown as means±SD for Fv/Fm, ɸ_II, NPQ, and qL in coe1, rug2-1, rug2-2 and WT* seedlings.

mTERF4 plays an important role in group II intron splicing of certain plastid transcripts, and the null bsm mutant seriously affected the global expression of plastid genes (Figs 5 and 8A–D) (Babiychuk et al., 2011). In this respect, our coe1 mutant showed a relatively weak bsm phenotype (Figs 5 and 8A–D) but, like rug2-1, displayed a gun phenotype in the presence of NF and LIN (Fig. 3). This suggested a possible connection between processing of plastid transcripts and retrograde signaling. To address this possibility, the processing of plastid-encoded transcripts was investigated in gun1, gun4, gun5, and coe1 plants. As shown in Fig. 8E–H and Supplementary Fig. S9 at JXB online, under standard growth conditions, the processing of atpF, clpP, rpl2, and rps12 was normal in gun1, gun4, and gun5, but was strongly perturbed in coe1 compared with WT*. Similarly, exposure to LIN, like treatment with SPE (Babiychuk et al., 2011), strongly inhibited the processing of atpF, rpl2, and rps12 in WT* seedlings (Fig. 8E–H and Supplementary Fig. S9). Strikingly, gun5 markedly mitigated the effects of SPE on the processing of atpF and clpP transcripts (Fig. 8E–H and Supplementary Fig. S9). In contrast, gun1 did not alter the inhibitory effect of LIN and SPE on the processing of atpF, clpP, rpl2, and rps12 RNAs (Fig. 8E–H and Supplementary Fig. S9), and further repressed the mRNA levels of atpF, clpP, rpl2, and rps12 in the presence of LIN but not SPE (Fig. 8A–D and Supplementary Fig. S9). In fact, in this context, coe1 behaved like gun1 with respect to atpF and rpl2. In WT*, gun4, and gun5 seedlings grown in the presence of LIN and SPE, levels of unprocessed chloroplast transcripts and of Lhcb mRNA followed opposing trends (Fig. 8I), i.e. the more unprocessed chloroplast transcripts present, the less Lhcb1.1 mRNA was detected. However, in gun1 and coe1 grown with either antibiotic, disruption of plastid RNA processing was accompanied by a rise in levels of Lhcb1.1 mRNA. These results suggested a link between the accumulation of non-processed transcripts and PGE-dependent signaling.

Fig. 8.

Accumulation of total and unprocessed plastid RNAs. (A–D) Relative levels of total mRNA for rps12 (A), atpF (B), clpP (C), and rpl2 (D) in mutants versus WT*. (E–H) Relative levels of unprocessed mRNAs for rps12 (E), atpF (F), clpP (G) and rpl2 (H). (I) Scatter plot for relative Lhcb1.1 mRNA level and the relative non-processed mRNA level (means±SD) of three independent experiments for rps12, atpF, clpP, and rpl2 in LIN- or SPE-treated seedlings/control. *P<0.05, ** P<0.01 (Student’s t-test versus WT*).

Furthermore, the higher accumulation of atpF, clpP, rpl2, and rps12 in coe1 and also partly in gun1 (Fig. 8) and psaB, atpB, and petA transcripts (Fig. 5) in coe1 could be the result of higher transcription rates of these transcripts. To test this, run-on transcription assays were carried out on isolated chloroplasts of 2-week-old WT*, coe1, and gun1 seedlings. As shown in Supplementary Fig. S10 at JXB online, no signals were detected in the controls with the nucleus-encoded RbcS1a gene and the mitochondrial-encoded atp1 gene, but the transcripts of the chloroplast-encoded genes atpB, clpP, atpF, petA, psbC, psaB, rpl2, and rps12 were detected. In general, in all genotypes, psbC, psaB, and rpl12 showed relatively strong transcription rates, while the other genes were relatively weakly transcribed (Supplementary Fig. S10). Compared with those of WT*, the transcription rate for psaB was 3-fold higher in coe1, and those for atpF and clpP were 2-fold higher in coe1 but only slightly elevated in gun1, while the other transcription rates were all about the same in WT*, coe1, and gun1 (Supplementary Fig. S10). These results suggested that elevated mRNA levels of some chloroplast-encoded genes in coe1 might be caused by higher transcription rates.

COE1 genetically interacts with GUN1

Mutations in mTERF4 and GUN1 have similar effects on atpF and rpl2 RNA processing (Fig. 8). Moreover, both proteins potentially interact with nucleic acids (Koussevitzky et al., 2007; Babiychuk et al., 2011, and this study). We therefore tested whether these two proteins functionally interacted with each other. To this end, we first investigated the genetic relationship between GUN1 and mTERF4 (Fig. 9). In fact, overexpression of GUN1 (oeGUN1-GFP/coe1) could partially rescue the pale green phenotype of coe1 under normal growth conditions (Fig. 9B). Levels of luminescence of P _Lhcb1.1 :LUC were also slightly lower in oe-GUN1-GFP/coe1 than in coe1 (Fig. 9C). In parallel, the value of Fv/Fm was slightly higher in oe-GUN1-GFP/coe1 than in coe1 (Fig. 9D). These results suggested that overexpression of GUN1 may partially compensate for the defect of coe1 in the regulation of chloroplast biogenesis with respect to leaf coloration and Fv/Fm. In addition, we also generated a gun1 coe1 double mutant. Compared with coe1, the gun1 coe1 mutant showed a more severe leaf color phenotype and grew more slowly (Fig. 9B), but levels of P _Lhcb1.1 :LUC luminescence were not affected in the gun1 coe1 double mutant (Fig. 9C). The value of Fv/Fm of gun1 coe1 was also lower than that in coe1 (Fig. 9D). These results indicated that GUN1 is required to maintain chloroplast biogenesis and function when COE1 is impaired.

Fig. 9.

Genetic interactions between GUN1 and mTERF4. (A) Diagram of the plate layout showing the position of each genotype. (B) Bright-field images of coe1, GUN1-GFP/coe1, gun1 coe1, and WT* plants grown on a 1/2 MS plate. (C) Luminescence of the seedlings shown in (B). Note that the luminescence in some of the F2 gun1 coe1 seedlings may be due to segregation of the P _Lhcb1.1 :LUC transgene. (D–F) Fv/Fm (D), ɸ_II (E), and NPQ (F) of the seedlings shown in (B) was determined as described in Materials and Methods. Signal intensities for Fv/Fm are indicated according to the color scale. Results are shown as means±SD for luminescence, Fv/Fm, ɸ_II, and NPQ.

The potential for functional interaction between GUN1 and COE1 prompted us to test whether GUN1 could physically interact with COE1 in a yeast two-hybrid system. The coding sequence of GUN1 was cloned into the yeast bait vector pGBKT7 to generate GUN1-BD and the coding sequence of COE1 was cloned into the prey vector pGADT7 to produce COE1-AD. The resulting co-transformants expressing both GUN1-BD and COE1-AD were analyzed for growth on plates lacking His, Leu, Ade, and Trp (SD–His–Leu–Trp-Ade), and their α-galactosidase activities were assayed. As in the negative control, no interaction between GUN1 and COE1 was detectable, suggesting that the two proteins did not physically interact (Supplementary Fig. S11).

Alternatively, COE1 might affect the function of GUN1 by modulating its distribution in the plant. Analysis of the tissue localization of GUN1 fused to green fluorescent protein (GFP) indicated that the GFP fluorescence was prominent primarily in guard cells and in leaf-vein cells of cotyledons in WT* under normal growth conditions (Fig. 10A). Interestingly, in the presence of NF, enhanced GFP fluorescence was observed in all epicotyls and hypocotyls of the seedlings (Fig. 10B). In the coe1 genetic background, the GUN1–GFP fluorescence showed a similar trend to that seen in NF-treated WT seedlings (Fig. 10C). These results suggest that coe1 may affect the accumulation or distribution of GUN1. Alternatively, the defects in processing of plastid transcripts in coe1 may cause a stress syndrome similar to that induced by NF, thereby changing the behavior of GUN1–GFP.

Fig. 10.

The coe1 mutation alters the accumulation and distribution of GUN1-GFP. (A) Accumulation and distribution of GUN1–GFP in WT* under normal growth conditions. (B) Accumulation and distribution of GUN1–GFP in WT* in the presence of NF. (C) Accumulation and distribution of GUN1–GFP in coe1 under normal growth conditions. Bars, 200 µm.

Discussion

Plastid retrograde signaling is essential for the biogenesis and development of chloroplasts because of its impact on the expression of PhANGs (Mochizuki et al., 1996; Larkin et al., 2003; Koussevitzky et al., 2007; Kleine et al., 2009). However, in spite of its fundamental importance, little is known about the molecular nature of the process. In order to dissect the mechanism of retrograde signaling, we analyzed the expression of plastid and nuclear genes in gun mutants and WT seedlings. Interestingly, under normal growth conditions, levels of nucleus-encoded Lhcb transcripts were slightly higher in gun mutants than in WT at early stages of seedling development (Fig. 3A, E). Furthermore, gun mutants showed a subtle decrease in the expression of the plastid-encoded rbcL mRNA during the same developmental phase (Fig. 3F, G). These results clearly indicated that the activity of PGE is impaired in gun mutants. However, how the PGE-dependent signal is produced, and how it is modulated by GUN1, is unknown. In this study, we isolated a novel mutant, coe1, which shows up-regulation of Lhcb1.1 expression under normal growth conditions (Figs 1 and 3A, E). The coe1 mutant is also characterized by slightly higher PhANG expression than WT in the presence of NF and LIN (Fig. 3B, C, E and Supplementary Fig. S4), suggesting that COE1 plays a role in modulating retrograde plastid signaling.

The coe1 mutant displayed a pale yellow leaf phenotype, suggesting that the biogenesis of chloroplasts is impaired (Supplementary Fig. S2). The maximum quantum efficiency of PSII (Fv/Fm) was indeed substantially reduced in coe1 relative to WT and other gun mutants (Supplementary Figs S2 and S5). The accumulation of chloroplast proteins was also significantly decreased in coe1 and rug2-1 (Fig. 4), and the levels of thylakoid membrane complexes were also clearly decreased in coe1 and rug2-1 but almost unchanged in gun1 compared with WT* (Fig. 4B, C). Furthermore, as a secondary effect of reduced thylakoid complex accumulation in coe1, Lhcb1 protein levels were also decreased (Fig. 4). Like protein levels of Lhcb1, RbcS proteins levels were also reduced in coe1 (Fig. 4), which is likely the consequence of the reduced RbcL levels due to disturbed chloroplast translation in this mutant. Although chloroplast proteins accumulated to lower levels in coe1, the amount of plastid transcripts in coe1 was about the same or even up-regulated compared with WT* (Fig. 5). This can be explained by the fact that the majority of transcripts for petA, psaB, and atpB might not be engaged in translation (Supplementary Fig. S7). Interestingly, in run-on experiments, levels of psaB, atpF, and clpP were increased in coe1 and slightly increased in gun1, suggesting that at least COE1 might modulate the transcription activity of plastid genes (Supplementary Fig. S10). Unlike prors1, which is defective in PGE and associated with a reduction of PhANG expression under normal growth conditions (Pesaresi et al., 2006), transcription of Lhcb1.1 was increased in coe1 (Figs 2B and 3A), while levels of Lhcb1 protein were significantly lower in coe1 than in WT under normal growth conditions (Fig. 4).

Molecular cloning of COE1 revealed that it codes for mTERF4, and is allelic to BSM/RUG2/ZmmTERF4 (Babiychuk et al., 2011; Quesada et al., 2011; Hammani and Barkan, 2014). The mTERFs form a large and complex protein family in both metazoans and plants (Kleine, 2012; Kleine and Leister, 2015). In stark contrast to the case in mammals, the functions of mTERFs in plants are poorly understood (Kleine, 2012). The mTERF family in plants is considerably larger than in Metazoa; for example, A. thaliana and Oryza sativa Japonica contain at least 35 and 48 genes for mTERF proteins, respectively (Babiychuk et al., 2011; Kleine et al., 2012), all of which are predicted or known to localize to mitochondria or chloroplasts (Babiychuk et al., 2011; Kleine, 2012). So far, only four plastid mTERFs—SOLDAT10, BSM, ZmmTERF4, and TWIRT1/mTERF9—have been identified and functionally characterized (Meskauskiene et al., 2009; Babiychuk et al., 2011; Mokry et al., 2011; Quesada et al., 2011). In this study, we described coe1 as a new allele of bsm/rug2/Zmmterf4 that causes similar defects in the accumulation of chloroplast proteins and the biogenesis of chloroplasts (Fig. 4). Expression of PhANGs was slightly up-regulated in coe1 under normal growth conditions. Furthermore, coe1 also showed a gun phenotype in the presence of NF (Fig. 3). Genetic analysis revealed that the effects of coe1 on the expression of Lhcb1 were partially dependent on GUN1 (Fig. 9). Yeast two-hybrid analysis indicated that GUN1 does not interact with COE1 (Supplementary Fig. S11), but overexpression or loss of GUN1 in coe1 can partially rescue or aggravate its defects in the regulation of the biogenesis of chloroplasts. More interestingly, coe1 can also regulate GUN1 function by affecting its accumulation and distribution. For instance, compared with WT, both the level and the distribution of GUN1–GFP fluorescence are altered in coe1 under normal growth conditions (Fig. 10). Taken together, these results suggest that GUN1 and COE1 do interact at some level in regulating the expression of plastid genes and PhANGs under certain physiological conditions.

In WT*, LIN and SPE treatments lead to the accumulation of non-processed RNA (Fig. 8). The coe1 mutant accumulates high levels of unprocessed RNAs even under normal growth conditions (Fig. 8). The gun1 mutation did not dramatically alter RNA processing in the presence of LIN but instead seemed to affect the expression of plastid genes (Fig. 8). Alterations in chlorophyll metabolism might affect the processing of transcript stability/maturation, because for example, the chlorophyll-deficient mutants atcrs1 and atcaf2 (Asakura and Barkan, 2006) and rice white stripe leaf (wst) (Tan et al., 2014) show defects in the processing of plastid transcripts. The gun4 and gun5 mutants, in which chlorophyll metabolism is perturbed, indeed showed subtle alterations in the processing of rps12 and atpF transcripts under normal conditions but not in the presence of LIN and SPE (Fig. 8). Comparative analysis revealed that levels of unprocessed plastid transcripts are negatively correlated with expression levels of Lhcb1.1 in WT plants exposed to LIN and SPE, and in gun4 and gun5 plants, but not in gun1 or coe1. These results suggest that the accumulation of unprocessed plastid transcripts might trigger plastid signaling to inhibit gene expression of nuclear photosynthesis genes. In addition, altered mTERF4 levels affected the intracellular accumulation and distribution of GUN1, as well as its plastid signaling activity. Taken together, these results suggest that GUN1 and COE1 cooperate in PGE and retrograde signaling (Supplementary Fig. S12).

Supplementary data

Supplementary data are available at JXB online.

Table S1. List of oligonucleotides used in this study.

Fig. S1. LUC activity in P _Lhcb1.1 :LUC plants can be suppressed by treatments with LIN or NF.

Fig. S2. The coe1 phenotype is especially prominent during early development of chloroplasts.

Fig. S3. Quantification of steady-state Lhcb1.1 mRNA levels in coe1 and WT* plants during early plant development.

Fig. S4. Transcripts of PhANGs are slightly increased in NF-treated coe1 plants.

Fig. S5. Photosynthetic performance of mutant (gun1, gun4, gun5, coe1, rug2-1 and rug2-2) and WT* plants.

Fig. S6. Growth phenotype of gun1, gun4, gun5, coe1 and WT* on soil. Plants were grown on soil in a climate chamber for 3 weeks, on a 12-h light/12-h dark regime.

Fig. S7. Polysome association analysis for chloroplast transcripts in WT*, coe1 and gun1 plants.

Fig. S8. Complementation of the coe1 mutation by AT4G02990.

Fig. S9. Analysis of plastid transcript processing in gun1, gun4, gun5, coe1, and WT*.

Fig. S10. Transcription rates of plastid genes in WT*, coe1 and gun1 seedlings.

Fig. S11. GUN1 does not interact with COE1 in yeast-two-hybrid experiments.

Fig. S12. A model for the functional relationship of mTERF4 and GUN1.

Glossary

Abbreviations:

BAC

bacterial artificial chromosome

BN

blue native

EMS

ethyl methanesulfonate

GFP

green fluorescent protein

LIN

lincomycin

LUC

luciferase

mTERF

, mitochondrial transcription termination factor

NF

norflurazon

PGE

plastid gene expression

PhANG

photosynthesis-associated nuclear gene

PPR

pentatricopeptide repeat

PS

photosystem

qRT-PCR

quantitative real-time reverse transcription PCR

SPE

spectinomycin

SSLP

simple sequence length polymorphism

WT

wild type.