Arabidopsis thaliana (Columbia-0) plants were grown in soil with a 16/8-h light/dark cycle at 22°C in the plant growth room. Seeds were sterilized and plated onto 1/2 Murashige and Skoog (MS) basal salt medium with 0.8% agar and 1.5% (w/v) sucrose in the dark for 2 days at 4°C. Then, the plates were grown in the growth chamber MLR-352H-PC (Panasonic, Japan) at a 16/8-h light/dark cycle at 22°C for 7 days, and the seedlings were transplanted in the soil for continual growth. The siliques were collected at the exact developmental stages for the next analysis.

Arabidopsis genomic DNA was isolated from 4-week-old leaves using the extraction buffer [200 mM Tris-HCl pH 7.5, 250 mM NaCl, 25 mM EDTA, 0.5% sodium dodecyl sulfate (SDS)] as described (Edwards et al., 1991). Total RNA was extracted by TRIZOL Reagent (Thermo Fisher Scientific, USA) from Arabidopsis seeds according to the manufacturer's instructions (Thermo Fisher Scientific) and quantified using a Nanodrop ND-1000 spectrophotometer (LabTech, USA). Total RNA (5 μg) was used for cDNA synthesis with Superscript II RNase H-Reverse Transcriptase (Life Technologies, USA) and oligo(dT)₁₈ primers according to the manufacturer's instructions.

Two T-DNA insertion lines of AtERG2: aterg2-1 (SALK_032115) and aterg2-2 (SALK_032124) were purchased from Arabidopsis stock centers (http://www.arabidopsis.org/). The genotype of T-DNA lines was confirmed by PCR using the gene-specific primers and a T-DNA–specific primer (Table S1).

The coding sequence of AtERG2 was amplified by PCR and cloned into the Gateway Cloning vector pE6c or pE3c (Addgene, USA) with Bam HI and Not I sites. To generate the various truncated AtERG2 constructs, such as AtERG2_N (containing only N-terminal), AtERG2_C [containing GTPase domain (GD) and KH domain (KH)], AtERG2_GD (containing only GD), and AtERG2_KH (containing only KH), the corresponding AtERG2 fragments were amplified and subcloned into pE6c. Construct maps containing different parts of AtERG2 are shown in Figure S2. The different constructs were used in combination with the destination vector pMDC32 (http://www.arabidopsis.org/) with the Gateway LR II kit (Life Technologies, USA) to generate the plant expression vectors. A genomic fragment of 1.8 kb in the promoter region of AtERG2 was amplified by PCR and cloned into a GUS binary vector pCAMBIA 1305.1 with EcoR I and Bgl II sites. The primers used in these constructs are listed in Table S1. All vectors were confirmed by sequencing.

The resulting 35S::AtERG2-6XMyc and P_AtERG2::GUS constructs were transformed into Agrobacterium tumefaciens GV3101. Transgenic plants were generated by the floral dip method (Clough and Bent, 1998) and selected on MS medium containing hygromycin B (25 mg/L) or kanamycin (50 mg/L). More than 100 AtERG2-6XMyc or P_AtERG2::GUS lines were recovered. Independent homozygous transformants carrying a single insertion in the T3 generation were further analyzed.

The vectors containing AtERG2-eYFPs were co-transformed with a mCherry-tagged organelle marker gene and transiently expressed in Arabidopsis mesophyll protoplasts, as described previously (Liang et al., 2015). After incubation for 16 h at 22°C, fluorescence was visualized using an LSM 700 confocal microscope (Zeiss, Germany). Observations were made using a 63 × objective under oil immersion. eYFP fluorescence was excited at 488 nm and collected at SP 550 IR. The mCherry fluorescence was stimulated at 555 nm and collected at SP 630 IR.

The 2-week-old transgenic Arabidopsis seedlings overexpressing AtERG2-6Xmyc were ground in liquid nitrogen, and the protein/RNA complexes were extracted using equal volumes of lysis buffer (IP buffer (50 mM NaCl, 2 mM MgCl₂, 10 mM HEPES pH 7.5, 2 mM DTT, 2 mM EDTA, 100 U/mL RNasin RNase inhibitor [Promega, USA], and 100 U/mL protease inhibitor cocktail [Sigma-Aldrich, USA] plus 1% TritonX-100 and 0.5% SDS) and incubated for 1 h on ice, then centrifuged twice at 13,000 g for 10 min at 4°C to remove the insoluble debris. The supernatant was diluted with nine volumes of IP Buffer and incubated with 10 μL anti-Myc antibody (Sigma-Aldrich) for 4 h on ice with occasional gentle mixing. Next, the anti-Myc-decorated extracts were incubated with 10 μl protein G agarose beads pre-washed with IP buffer for 1 h at 4°C with rotation. Then, the beads were spun down at 10,000 rpm in a microfuge for 1 min at 4°C, and washed eight times with 1 mL washing buffer (100 mM NaCl, 2 mM MgCl₂, 10 mM HEPES pH 7.5, 2 mM DTT, 2 mM EDTA, 100 U/mL RNasin RNase inhibitor, and 100 U/mL protease inhibitor cocktail). After elution with TRIZOL reagent, co-immunoprecipitated RNA was isolated and analyzed by quantitative real-time RT-PCR (qRT-PCR).

qRT-PCR was performed on the 7500 Fast Real-Time PCR System (Applied Biosystems, USA) using Power SYBR® Green PCR Master Mix (Applied Biosystems). The thermal program was 2 min at 50°C, 10 min at 95°C, followed by 40 cycles of 15 s at 95°C and 60 s at 60°C. The data were normalized to the expression of Arabidopsis actin. The dissociation curve program was used to confirm the specificity of the target amplification product. All primers used in this study are listed in Table S1. At least three independent biological replicates were performed for qRT-PCR analysis.

Seeds were removed from siliques at different developmental stages, cleared in Hoyers solution (7.5 g gum arabic, 100 g chloral hydrate, 5 mL glycerol in 30 mL water) and examined using a Zeiss Axio Observer A1 microscope equipped with DIC optics as described by Liu and Meinke (1998). The ROS detection was performed according to a modified version of a previously described method (Daudi et al., 2012) by staining with 1 mg/mL 3,3′-diaminobenzidine (DAB, Sigma) containing Triton X-100 (0.1% v/v) and 10 mM sodium phosphate buffer (pH 7.0). As previously described (Yu et al., 2005), the GUS staining was performed for 4 h with seedlings and leaves or overnight with flowers and seeds at 37°C in GUS staining solution (50 mM sodium phosphate buffer, pH 7.0, 10 mM EDTA, 0.1% Triton X-100, 2 mM potassium ferricyanide, 2 mM potassium ferrocyanide, 1 mg/mL X-Gluc) before the clearing procedure.

The seeds from 1.5- and 2.0-DAP siliques were harvested and fixed overnight in FAA solution (90 mL 50% acetic acid; 5 mL glycerine; 5 mL 37% formaldehyde). After gradual rehydration (100, 90, 70, 50, 30, and 10% ethanol) for 20 mins, the materials were stained twice in 10 mg/mL Eosin B (2 h each). Subsequently, the sample was gradually dehydrated (10, 30, 50, 70, 90, and 100% ethanol) for 20 mins until the solvent was replaced by 100% ethanol. Finally, seeds were cleared with methyl salicylate and observed using an LSM 700 confocal microscope (Zeiss, Germany).

The total proteins were extracted from 2.0-DAP siliques in extraction buffer (50 mM Tris-HCl [PH 8.0], 200 mM NaCl, 1 mM β-mercaptoethanol, 2% SDS, 10% glycerin, and 0.5% Triton-X 100), and the protein concentration was determined using a Bio-Rad Protein Assay Kit with bovine serum albumin (BSA) as the standard. The samples were separated by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to a polyvinylidene difluoride (PVDF) membrane (Millipore, USA) under a 200-mA constant current for 45 min in transfer buffer (12.5 mM Tris-HCl, 192 mM glycine, and 10% methanol; pH 8.3). Anti-ATPase Subunit 6 (1:1,000, Thermo Fisher Scientific) and anti-actin (1:1,000, ZSGB-BIO, China) antibodies were used as the primary antibodies, and horseradish peroxidase-conjugated goat anti-rabbit IgG antibodies (1:2,000, Cell Signaling Technology, USA) were used as the secondary antibodies. Signals were detected using the automatic chemiluminescence image analysis system Tanon 5200 Multi (Tanon, China).

All statistical analyses were performed using the Data Processing System (Tang and Zhang, 2013). One-way analysis of variance (ANOVA) and Tukey's multiple range test were conducted to determine statistical significance. In the study, the letters a–e or ^* indicate statistical significance (P < 0.05), and ^*** indicates high levels of statistical significance (P < 0.001).

A previous study showed that there are two homologs of ERG in the Arabidopsis genome: AtERG1 (At5g66470), localized in the chloroplast, and AtERG2 (At1g30960), localized in the mitochondria (Suwastika et al., 2014). We used the sequence of ERA, ERAL1, ERG, AtERG1, and AtERG2 to search NCBI public databases (http://blast.ncbi.nlm.nih.gov) at an E-value of 1e-10. The presence of conserved GTPase and KH domains is the exclusive criterion for confirmation of 16 ERA homologous proteins in 14 species, which include MmERA from Mus musculus, DrERA from Danio rerio, DmERA from Drosophila melanogaster; ScERA from Saccharomyces cerevisiae, OsERG1 and OsERG2 from Oryza sativa, CsERG from Cucumis sativus, ZmERG from Zea mays, NtERG from Nicotiana tabacum, StERG from Solanum tuberosum, and SlERG from Solanum lycopersicum. Additionally, Jeon et al. (2014) found a chloroplast-localized double Era-like GTPase in Nicotiana benthamiana (NbDER), which binds to chloroplast 23S and 16S ribosomal RNAs. We further identified the homologs of NbDER, such as AtDER from Arabidopsis thaliana, EcDER from E. coli, ZmDER from Zea mays, and OsDER from Oryza sativa. Then, a phylogenetic tree was generated using the CDS (sequence coding for amino acids in protein) of 21 ERA homologs, and we found that members of the ERA homologs were separated into four distinct clades, designated I, II, III, and IV (Figure S1). Clade I included 12 members: ERG, AtERG2, CsERG, OsERG2, ZmERG, NtERG, SIERG, StERG, ERAL1, MmERAL1, DrERAL1, and DmERAL1, which were predicted and/or identified as mitochondria-localized. Clade II contained three members: ERA, AtERG1, and OsERG1; this suggested that ERA is most closely related to chloroplast-localized AtERG1 and OsERG1. Clade III included one member: ScERA. Clade IV had five DERs. These eukaryotic ERALs and ERGs were nuclear-encoded and organelle-localized genes, indicating that they were transferred to the nucleus after organelle symbiogenesis.

To further investigate the subcellular localization of AtERG2, we developed various constructs containing different parts of AtERG2 fused with yellow fluorescent protein (eYFP) at their C-terminal end (Figure S2) and transiently co-transformed cells with a mCherry-labeled organelle marker in the Arabidopsis mesophyll protoplasts. The full-length AtERG2 was localized in the mitochondria as previously reported (Figure 1A; Suwastika et al., 2014). We further found that the N-terminal end (1–150 aa) of AtERG2 was localized in the mitochondria. However, the C-terminal end, GTPase domain, and KH domain alone diffused into the cytoplasm (Figure 1A). These results indicated that the N-terminal end of AtERG2 is critical for its localization to the mitochondria.

We next used RNA immunoprecipitation (RNA-IP) assays to detect AtERG2 binding ribosome rRNA. The AtERG2-Myc:rRNA complex was extracted from the transgenic Arabidopsis stably expressed AtERG2-Myc fusion protein, preincubated with Myc antibody, and purified by protein G agarose beads. Then, the rRNA was isolated with Trizol solution and detected by RT-PCR (Figure 1B). The results showed that AtERG2 binds mitochondria ribosome small subunit 18S rRNA but not large subunit component 26S rRNA, suggesting that AtERG2 plays the same role as its homologs ERA and ERAL1 for the maturation of 16S rRNA and assembly of ribosomes.

To detect the transcription pattern of AtERG2 in Arabidopsis, we first generated the independent transgenic Arabidopsis lines, which stably expressed P_AtERG2::GUS (Figure 2). In the F2 generation, GUS activity was observed in the leaf vein of 1-week-old seedling cytoledons (Figure 2A), trichomes of 3-week-old rosette leaves (Figures 2B,C), pollen grains inside the anthers (Figures 2D,E), and ovules during the period of pollination and early-stage seed development (Figures 2F–I). In anthers, GUS activity was relatively weak the day before pollination (Figure 2D), but it increased in mature pollen at day 0.5 after pollination (DAP) (Figure 2E). We also found that GUS staining was detected at a low level in the pre-matured embryo sac before pollination (Figure 2F); it then reached the highest level in the ovule of 0.5- and 1.0-DAP seeds during the zygotic stage (Figures 2G,H) and quickly dropped in the 1.5-DAP seeds as soon as it reached the early-globe stage (Figure 2I). These results were further confirmed by qRT-PCR assays to evaluate the expression of AtERG2 in different tissues and in the development of Arabidopsis. Figure 2J showed that AtERG2 transcripts were highly expressed in the 0.5-, 1.0-, and 1.5-DAP seeds.

To characterize the function of AtERG2, the two T-DNA insertion lines (SALK_032115, named as aterg2-1, and SALK_032124, as aterg2-2, respectively, Figure 3A) were obtained from The Arabidopsis Information Resource (TAIR, https://www.arabidopsis.org/), and the genotype of the progeny from self-crossed plants was examined by PCR. In 25 randomly selected mature siliques from self-crossed aterg2-1 +/− plants, we found 687 aborted and 319 available seeds, in which 153 seeds are WT and 166 seeds are hemizygous (Figure 3B, Table 1). These results suggest that the insertion inactivation of AtERG2 severely affect the seed development. During the whole growth period from germination to flowering, aterg2-1+/− showed similar phenotypes to WT, with the exception of silique length and the mature seed number, both of which were significantly lower in the hemizygous lines compared to the WT (Figures 3C–E). Because aterg2-1 +/− and aterg2-2 +/− had similar phenotypes, we selected the data on aterg2-1 +/− for presentation; the subsequent analyses were performed using aterg2-1 +/− as the study material. Moreover, the developed seeds per silique in aterg2-1+/− were about 30% of WT at 6.0 DAP until seed maturation (Figure 3F). These results indicate that seed development was arrested and aborted at the early development of seeds in aterg2-1 +/−.

In order to explore which development stage was abnormal during the reproductive period in aterg2-1 +/− plant, we first backcrossed aterg2-1 +/− with WT plants and found that 21 WT and 10 hemizygous available seeds in 6 backcrossed siliques when using WT as a paternal provider, and 47 WT and 36 hemizygous seeds in 5 backcrossed siliques when using WT as the maternal provider (Table 1). In addition, staining of the mature pollens with iodine and measurement of the in vitro germination of pollens revealed no significant differences in the starch accumulation or the pollen germination ratio of WT and aterg2-1 +/− (Figures 3G,H). Similarly, we did not detect any significant difference in the development of embryo sacs between WT and aterg2-1 +/− plants at the pre-pollination stage or at 0.5 DAP (Figure 3I and Figure S3). Therefore, our results suggest that aterg2-1 +/− exhibited a sporophytic maternal effect (SME) phenotype, suggesting that AtERG2 plays a vital role in the early stage of seed development in Arabidopsis, but it does not affect gametophyte development.

We investigated the silique development of WT and aterg2-1 +/− from the pre-pollination stage (BP) to 4.0 DAP and found that the silique length was significantly shorter from 3.0 DAP in aterg2-1 +/− than it was in WT (Figure 4A). Some seeds were small and arrested at 1.0 DAP, and the arrested ratio of seed development reached 65% at 3.0 DAP in aterg2-1 +/− (Figure 4B), which is almost the same as the ratio of seed development at 6.0 DAP (Figure 3F). Additionally, we dissected the siliques and hyalinized seeds to observe the embryo development at 1.5 and 2.0 DAP to explore the impairment of early seed development in aterg2-1 +/−. There are 50.7 and 64.2% small-sized seeds at 1.5 DAP and 2.0 DAP in aterg2-1 +/− silique, respectively, but only 2.4% were found in WT silique; however, the average silique length was the same in WT and aterg2-1 +/− plants at 2.0 DAP (Figures 4C,D). At 1.5 DAP, there were two types of embryos in WT and aterg2-1 +/− siliques: 31.1% in the zygote stage and 68.9% in the early-globe stage for WT and 56.9% in the zygote stage and 43.1% in the early globe-stage embryos for aterg2-1 +/− siliques, respectively (Figure 4E and Figure S3). Furthermore, we found 9.8% heart-stage, 57.9% globe-stage (eight or more cells of embryos), and 32.2% early globe-stage embryos in 2.0-DAP WT seeds, and 1.2% heat-stage, 21.1% globe-stage, and 23.3% early globe-stage embryos were found in 2.0-DAP aterg2-1 +/− seeds. More interestingly, as much as 38% of seeds stayed in the zygote stage, and 16.5% of these seeds were collapsed seeds without embryos in 2.0-DAP aterg2-1 +/− siliques (Figure 4E and Figure S3). These results suggested that insertional inactivation of AtERG2 causes delay in the development of embryos after pollination and arrested seeds from 1.5 DAP in aterg2-1 +/− plants.

We performed WE-CLSM analysis to monitor the development of WT and aterg2-1+/− seeds at 1.5 and 2.0 DAP. We found that, similar to WT, most aterg2-1+/− seeds at 1.5 DAP developed into the zygote stage and early globe stage (Figure 5 and Table S2). However, some of the hemizygous seeds in the zygote stage presented with abnormal tissue degradation surrounding the endosperm and embryo (Figure 5 and Table S2). Moreover, a number of heterozygous seeds at 2.0 DAP showed severe tissue degradation, revealing the dissolved content inside the embryo sac and seed coat invagination (Figure 5 and Table S2). These data provide a plausible explanation for the observed collapsed phenotype observed in the aterg2-1 +/− siliques at 2.0 DAP (Figure 4E and Figure S3).

Previous studies showed that the knockdown of ERAL1 in human cell lines by siRNA inhibits mitochondrial protein translation and elevates mitochondrial ROS production, leading to cell death (Uchiumi et al., 2010; Xie et al., 2012). Thus, ROS production and the expression of the ROS-related genes may cause the developmental stagnation of embryos and arrested seeds in aterg2-1 +/−. To test this hypothesis, we first performed DAB staining to monitor the ROS content in Arabidopsis seeds after pollination. We found that about 32% zygote-stage seeds in aterg2-1 +/− at 1.0 DAP exhibited more pronounced increases in DAB staining (dark color), but only 10% zygote-stage seeds in WT showed strong DAB staining (Figures 6A,B). At 1.5 DAP, about 42% of aterg2-1 +/− zygote-stage seeds exhibited strong DAB staining compared to the 4% in WT seeds (Figures 6A,B). At 0.5 and 2.0 DAP however, none of the WT or hemizygous seeds showed dark color DAB staining (data not shown). Thus, these data suggest that the accumulation of ROS occurred prior to the abortion of seed development in aterg2-1 +/− plants.

Furthermore, Alternative oxidase1a (AOX1a), a nuclear gene encoding mitochondrial protein, is activated by ROS through ROS-responsible transcription factors WRKY40 and ANAC017 (Ng et al., 2013; Van Aken et al., 2013; Liu et al., 2014). Thus, these three genes can act as the marker genes for monitoring ROS levels. qRT-PCR showed that the transcription levels of these three genes was specifically higher at 1.5 and 2.0 DAP in the arrested seeds of aterg2-1 +/− than in the developed seeds of WT and aterg2-1 +/− (Figure 6C). Additionally, BCL2-associated athanogene (BAG) family, as the calmodulin-binding proteins, serve as co-chaperones to recruit HSP70/HSC70 to a specific target protein regulating diverse cellular pathways, such as programmed cell death (PCD) and stress responses in eukaryotes (Hung et al., 2003; Yan et al., 2003). Kang et al. (2006) showed that overexpression of AtBAG6 induced cell death phenotypes consistent with PCD in yeast and plants. Here, we found that AtBAG6 is extremely up-regulated in 2.0-DAP seeds of aterg2-1 +/− but not in WT plants (Figure 6B), which suggested that inactivation of AtERG2 induces the higher ROS levels and the cell death of embryos after pollination to produce the arrested seeds in aterg2-1 +/− plants.

To confirm whether AtERG2 regulates early seed development via affecting the translation of mitochondrial DNA-encoded genes, we performed western blot assay to detect the protein level of some mitochondria-encoded genes. ATPase subunit 6 (ATPase S6) is encoded by the mitochondrial DNA, transcribed and translated in mitochondria, and assembled into the mitochondrial proton-transporting ATP synthase complex. We collected the 2-DAP (including 1.5- and 2.0-DAP) siliques of WT and aterg2-1 +/− plants and extracted the total proteins to quantify the levels of ATPase S6. Western blot assays showed that the protein level of ATPase S6 in aterg2-1 +/− siliques was significantly lower than that in the WT siliques (Figure 6D). The original images of Figure 6D is shown in Figure S4, suggesting that AtERG2 is essential for mitochondrial protein synthesis and that aterg2-1 deficiency causes ROS-induced cell death due to mitochondria protein dysfunction.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.