Errata: Corrigendum Volume 200, Issue 3, 933, Article first published online: 12 September 2013
Iron–sulfur proteins have iron–sulfur clusters as a prosthetic group and are responsible for various cellular processes, including general transcriptional regulation, photosynthesis and respiration. The cytosolic iron–sulfur assembly (CIA) pathway of yeast has been shown to be responsible for regulation of iron–sulfur cluster assembly in both the cytosol and the nucleus. However, little is known about the roles of this pathway in multicellular organisms.
In a forward genetic screen, we identified an Arabidopsis thaliana mutant with impaired expression of the endosperm-specific gene Flowering Wageningen (FWA). To characterize this mutant, we carried out detailed phenotypic and genetic analyses during reproductive and vegetative development.
The mutation affects NAR1, which encodes a homolog of a yeast CIA pathway component. Comparison of embryo development in nar1-3 and other A. thaliana mutants affected in the CIA pathway showed that the embryos aborted at a similar stage, suggesting that this pathway potentially functions in early seed development. Transcriptome analysis of homozygous viable nar1-4 seedlings showed transcriptional repression of a subset of genes involved in ‘iron ion transport’ and ‘response to nitrate’. nar1-4 also exhibited resistance to the herbicide paraquat.
Our results indicate that A. thaliana NAR1 has various functions including transcriptional regulation in gametophytes and abiotic stress responses in vegetative tissues.
Proteins containing an iron–sulfur (Fe–S) cluster(s) as a prosthetic group play pivotal roles in diverse cellular processes, including respiration, photosynthesis, response to oxidative conditions, and genome stability (Rouault & Klausner, 1996; Balk & Pilon, 2011). Several Fe–S cluster biogenesis pathways exist in bacteria, such as the sulfur mobilization (SUF), iron–sulfur cluster (ISC) and nitrogen fixation (Nif) systems; these systems have also been found in different cellular compartments of eukaryotes (Lill & Mühlenhoff, 2006). The SUF machinery is conserved from archaea to chloroplasts in plants (Takahashi & Tokumoto, 2002; Godman & Balk, 2008), and mitochondrial ISC systems have been identified in eukaryotic cells (Lill & Mühlenhoff, 2005).
Several studies in yeast identified a novel Fe–S cluster assembly system, which has been named the cytosolic iron–sulfur cluster assembly (CIA) pathway (Lill, 2009). Unlike the SUF and ISC systems, the CIA system was found mainly in eukaryotes. This system is essential for the maturation of both cytosolic and nuclear Fe–S proteins (Lill & Mühlenhoff, 2008). For example, the yeast ATP-binding cassette transporters RNase L Inhibitor (Rli1) and isopropyl malate isomerase (Leu1p) are representative Fe–S proteins in the cytosol (Kispal et al., 1999, 2005), while many nuclear Fe–S proteins, such as an excision repair protein XPD helicase and DNA polymerase delta, are involved in DNA metabolism (Kispal et al., 1999, 2005; Rundolf et al., 2006). In addition to identification of many cytosolic and nuclear Fe–S proteins, recent studies have also found that the CIA pathway consists of multiple proteins. The cytosolic Fe–S cluster assembly factor Nuclear Architecture Related 1 (NAR1) seems to be responsible for incorporation of an Fe–S group into the apoprotein (Balk et al., 2004). Cytosolic Fe–S cluster Deficient 1 (Cfd1) and Nucleotide Binding Protein 35 (Nbp35) provide a scaffold for assembly of Fe–S clusters (Netz et al., 2007), and Top1T722A mutant Hypersensitive 18 (Tah18) and Derepressed for Ribosomal protein S14 Expression 2 (Dre2) are involved in electron-transfer processes in an early step of Fe–S cluster assembly in the cytoplasm (Netz et al., 2010). Interestingly, Nar1, Nbp35 and Dre2 proteins also contain binding motifs for Fe–S clusters (Balk et al., 2004; Hausmann et al., 2005; Zhang et al., 2008). The protein Cia1, which has a WD40 repeat motif, is believed to serve as a platform for protein–protein interactions in the CIA pathway (Balk et al., 2005). In addition, recent studies have reported that Cia2 (YHR122w) and Mms19 (gene YIL128w) proteins are components of the CIA pathway, required for DNA metabolism and maintenance of genome integrity (Gari et al., 2012; Stehling et al., 2012). Thus, over the last decade, remarkable progress has been made in elucidation of the yeast CIA pathway.
Associated with this recent progress in yeast, studies have begun to uncover the Fe–S cluster assembly pathway in plants and to determine its biological roles. Comparative genomic analyses in green algae and other species suggest that plants show conservation of proteins of the Fe–S cluster assembly pathway (Godman & Balk, 2008). According to these analyses, plant genomes do not have a Cfd1 homolog; instead, NBP35 seems to form a homodimer that is responsible for the necessary scaffold (Bych et al., 2008). Studies using recombinant homologs of the NAR1 and NBP35 proteins in Arabidopsis thaliana revealed that these proteins are able to coordinate the [4Fe–4S] cluster(s) (Bych et al., 2008; Cavazza et al., 2008; Kohbushi et al., 2009). In addition, deficiencies in NBP35 or Arabidopsis Thaliana P450 Reductase; ATR3 (the A. thaliana homolog of yeast Tah18) cause seed abortion (Bych et al., 2008; Cavazza et al., 2008; Kohbushi et al., 2009). A new component of the CIA pathway in A. thaliana has recently been reported: ASYMMETRIC LEAVES1/2 ENHANCER7 (AE7), which is probably the plant counterpart of CIA2, is required for activity of the Fe–S protein cytosolic aconitase; it is also essential to the proper activity of the nuclear Fe–S protein ROS1 (a DNA demethylase) based on the observation of hypermethylation of Repressor Of Silencing 1 (ROS1) target loci (Luo et al., 2012). However, much work still needs to be done to understand the physiological properties of the CIA pathway in higher plants throughout the life cycle.
Here, we isolated a nar1-3 mutant of the CIA pathway in A. thaliana and showed that it had impaired expression of FWA-GFP in the central cell and endosperm. FWA encodes an Homeodomain leucine zipper IV (HD-ZipIV) class homeodomain transcription factor. When the FWA gene is overexpressed in vegetative tissues of DNA methylation-deficient mutants (e.g. methyltransferase 1 (met1) and decreased DNA methylation 1 (ddm1)), it results in a late-flowering phenotype (Soppe et al., 2000; Kankel et al., 2003). However, in the wild-type, FWA is always silent during vegetative growth and is expressed specifically in endosperm tissue (Kinoshita et al., 2004). The FWA-GFP reporter mimics the imprinted gene expression pattern of endogenous FWA. This expression is controlled by the DNA glycosylase DEMETER (DME) (Kinoshita et al., 2004), which has a motif associated with an Fe–S cluster that is important for 5-methylcytosine excision activity (Mok et al., 2010). In addition, genetic analyses in the reproductive phase showed that NAR1 is required for both gametophytic and zygotic functions. We also determined the transcription profile of the weak nar1-4/− allele in vegetative tissues and compared this profile to that of nbp35-3/−, a mutant of a component of the Fe–S scaffold in the CIA pathway. The genetic analyses of these different nar1 alleles demonstrated that NAR1 is required for the transcriptional activation of some imprinted genes in female gametophytes and for proper abiotic stress responses in vegetative tissues.
Materials and Methods
Plant material and growth conditions
Arabidopsis thaliana (L.) Heynh Columbia-0 (Col-0) was used as the wild-type and also as the genetic background of the mutants in this study. Plants were grown under 16 h : 8 h, light : dark at 22°C. Seeds of the T-DNA insertion lines nar1-1 (GK-674D01), nar1-4 (SALK_151803), nbp35-1 (SALK_056204), nbp35-3 (SALK_054678), cia1-1 (SALK_060584: CS24064), atr3-3/tah18-3 (GK-004E05-014807: CS323950) and aconitase 1 (aco1) (GK-138A08) were obtained from the Salk Institute (http://signal.salk.eud/cgi-bin/tdnaexpress) or GABI-KAT (Kleinboelting et al., 2012) via the Arabidopsis Biological Resource Center (ABRC) or the Nottingham Arabidopsis Stock Centre (NASC). The MEDEA-GFP reporter line was kindly provided by R. Fischer (Choi et al., 2002).
Sequencing of the T-DNA flanking region
The inserted T-DNA alleles were amplified by genomic PCR using the gene-specific primers listed in Supporting Information Table S1. T-DNA insertion sites were confirmed by sequencing flanking genomic fragments amplified by the left border primer and the gene-specific primer.
Isolation and mapping of the nar1-3 allele
Several mutants defective in FWA-GFP expression in the endosperm were isolated from an ethyl methanesulfonate-mutagenized M1 population carrying the FWA-GFP marker (Ikeda et al., 2011). The following criteria were used to identify mutants: about half of the ovules or seeds in the silique showed a defect in GFP fluorescence both before and after fertilization (see Fig. 1b,d); and maturation of the female gametophyte was not impaired before fertilization. The m218 mutant line was selected for further investigation in this study. After a fifth backcross to wild-type Col-0 plants carrying FWA-GFP, the m218 mutant line was crossed to Ler (Landsberg erecta), also carrying FWA-GFP. For map-based cloning, the F2 population was examined using cleaved-amplified polymorphic sequence markers that were designed using information on single nucleotide polymorphisms (SNPs) (http://polymorph-clark20.weigelworld.org/; J. Fitz, S. Ossowski, N. Warthmann, R. M. Clark, K. Schneeberger and D. Weigel, pers. comm.). The locus between the SGCSNP257 and MASC03242 SNP markers was tightly linked to the impaired expression of FWA-GFP, and a mutation site was determined by sequencing of this region.
For complementation analysis of nar1-3, a 3.6-kb genomic fragment including the AT4G16440 gene was amplified using the following primers: 5′-GAA AGG AAA GGT ACC GTT TAG TAA GCC TC-3′ and 5′-GCC AAC CAA AAA GAA TTC CAG CTT AG-3′. The amplified fragment was cloned into pENTR/dTOPO (Invitrogen) and subsequently recombined into pBGW (Karimi et al., 2002) using LR clonase (Invitrogen). The vector construct was then introduced into the EHA105 strain of Agrobacterium tumefaciens by electroporation. For the promoter–β-glucuronidase (GUS) construct, a c. 1.2-kb upstream region of the NAR1 gene was amplified using the primers 5′-CGG ATC TAG AAC TTT GAC TTG AGA TAG ATT-3′, containing an XbaI restriction site, and 5′-TCT CTG ACG GAT CCC TTT CTA TTC AA-3′, containing a BamHI restriction site. The PCR fragment of this promoter region was subcloned into the pT7 blue T vector (Novagen, Madison, WI, USA). Finally, the fragment was cloned into the binary vector pBI101.2 (Clontech, Palo Alto, CA, USA) after digestion with XbaI and BamHI. For the NAR1 promoter NAR1::GFP construct, a 3.3-kb genomic fragment of AT4G16440 including the 1.1-kb upstream region was amplified using the primers 5′-CTC AAG CTA AGC TTG GAA AGG AAA GGT TTT GTT TAG TAA GC-3′ and 5′-TTG CTC ACC ATT GAT AAC CAG TTG TTG AGC TGC GAC GTA ACA C-3′. Subsequently, this fragment was cloned into SacI and EcoRV sites in the binary vector pB7FWG2 (Karimi et al., 2002) using an In-Fusion PCR cloning kit (Clontech) according to the manufacturer's instructions. For construction of the RNAi vector, a partial NAR1 coding sequence from the conserved amino acid region was amplified using the primers 5′-CAC CCT CTC AAG GAT TGC TTG GCT TGC AG-3′ and 5′-GAT TGA GAA TTT TCA CCA TCA TCA GAA TTA-3′. The resultant PCR fragment was cloned into pENTR/dTOPO (Invitrogen), and then transferred to the binary vector pHELLSGATE12 using LR clonase (Invitrogen). Agrobacterium tumefaciens-mediated transformation of A. thaliana plants was performed by the floral dip technique (Clough & Bent, 1998).
Tissues were prefixed in ice-cold 90% (v/v) acetone for 20 min at room temperature, rinsed twice with 50 mM sodium phosphate buffer (pH 7.2) containing 0.05% triton X-100, 2 mM K3Fe(CN)6 and 2 mM K4Fe(CN)6, infiltrated with the staining solution (2 mM 5-bromo-4-chloro-3-indolyl-β-GlcUA in the buffer) under vacuum on ice for 15 min, and incubated at 37°C overnight. After staining, chlorophyll was removed using 70% ethanol. Samples were mounted on slides and examined under a light microscope.
For whole-mount seed clearing, dissected seeds were soaked in a mixture of chloral hydrate, glycerol, and water (8 g : 1 ml : 2 ml). Bright field images were captured using an Axioimager M1 microscope equipped with an AxioCamMRc5 (Carl Zeiss, Jena, Germany). GFP fluorescence images and GUS staining images were captured using an MVX10MacroView microscope equipped with a DP71 camera (Olympus, Tokyo, Japan) or an Axioimager M1 microscope equipped with an AxioCam MRc5 (Carl Zeiss).
Total RNA was extracted from tissues using the RNeasy Plant Mini Kit (Qiagen) and treated with DNaseI (Qiagen) and RQ1 DNase (Promega) to eliminate DNA contamination. cDNA was prepared using Primescript Reverse Transcriptase (TaKaRa, Shiga, Japan) with both oligo(dT) and random-hexamer primers. The synthesized cDNA was subjected to real-time quantitative PCR (qPCR) or semiquantitative PCR. For semiquantitative PCR analysis, 160 ng of total RNA from different organs was used for the reverse transcription reaction. PCR cycles were adjusted depending on transcript levels (32 cycles for NAR1; 26 cycles for UBQ10). The PCR products were loaded on a 1.5% agarose gel and stained with ethidium bromide. qPCR was performed using a Thermal Cycler Dice Real Time System (TaKaRa) using SYBR Green I. The following qRT-PCR conditions were used: 98°C (for 1 min); up to 40 cycles of 98°C (for 30 s) and 60°C (for 30 s except for UBIQUITIN 10 (UBQ10), 20 s). The primer sequences for RT-PCR are listed in Table S1.
nar1-4/−, nbp35-3/− and wild-type plants were grown on Murashige and Skoog (MS) plates and were harvested at 2 wk after germination. Total RNA was extracted using the RNeasy Plant Mini Kit (Qiagen). RNA was labeled using the GeneChip® 3′ IVT Express Kit (Affymetrix, Santa Clara, CA, USA) and was hybridized to Affymetrix A. thaliana ATH1 GeneChips. Raw data were obtained from two biologically independent experiments. Signal values were obtained from two biologically independent experiments via RMA (robust multi-array analysis) processing and then analyzed using the statistical procedures provided by R and Bioconductor (http://www.r-project.org/). Functional category analysis was performed using the TAIR Gene Ontology description (Berardini et al., 2004) Microarray expression data are available in GEO under Series accession GSE44116 (Edgar et al., 2002).
Aconitase activity assay
Pools of 0.3 g of 2-wk-old whole plants were ground into fine powder in liquid nitrogen. The powder was extracted with 0.5 ml of extraction buffer (100 mM potassium phosphate, pH 8.0, 1 mM EDTA, 0.5 mM phenylmethylsulfonyl fluoride, and 1 μg ml−1 each of antipain, leupeptin and pepstatin), and centrifuged at 20 000 g for 30 min. The supernatant was purified using a BioSpin 6 column (Bio-Rad, Hercules, CA, USA). Enzyme activity was monitored by measuring the absorbance at 240 nm, which is associated with the formation of the cis-aconitate (Racker, 1950; Courtois-Verniquet & Douce, 1993). The assay was performed at room temperature in 50 mM potassium phosphate buffer (pH 7.4), containing 30 mM DL-isocitrate.
Paraquat resistance assay
Seeds from the nar1 mutant series and wild-type plants were surface-sterilized and sown on 1/2 × MS-agar plates. At 5 d after germination, the seedlings were transferred to MS-agar plates containing 0, 1 or 3 μM paraquat. Survival rates were estimated as the proportion (%) of plants with young green leaves at 0, 1, 2 and 3 wk after transplantation.
Identification of a gene involved in FWA-GFP activation
We isolated a series of mutants that showed defects in the expression of the endosperm-specific imprinted gene FWA (see the 'Materials and Methods' section). One of the mutants was a new allele of nar1, a component of the CIA pathway. As the role of the CIA pathway in specific gene regulation is unknown, we focused on this nar1 mutant for further analysis. No homozygous plant for this allele was recovered from self-pollinated heterozygous plants. In the heterozygous mutant, expression of FWA-GFP was reduced in c. 50% of the central cells before fertilization compared with the wild-type. When the mutant plants were pollinated with wild type Col-0 pollen, less than half of the developing seeds 3 d after pollination (DAP) had a lower level of GFP expression than the wild-type; however, expression was almost equal to that of wild-type at 5 DAP (Fig. 1a–f). This phenotype suggests that NAR1 has a gametophytic function. We determined that the mutation involved a G to A base substitution generating a premature stop codon within the fifth exon of the At4G16440 gene (Fig. S1a), which encodes a protein with a conserved Fe-hydrogenase domain (Fig. S1b). The At4G16440 gene shows similarity to yeast Nar1. The NAR1 genes are present as a single copy in plant genomes and in yeast, while mammals have two copies (Fig. S1b). Two other alleles of NAR1 (nar1-1, GK_674D01, and nar1-2, GK_462G04) have been identified in T-DNA insertion lines and used to investigate the genetic interaction with another plant CIA component, AE7 (Luo et al., 2012). Therefore, the mutant allele isolated here has been named nar1-3. We also analyzed another allele with T-DNA insertion in the NAR1 gene locus, named nar1-4 (SALK_151803). Unlike the other alleles, homozygous nar1-4 was viable. A genomic fragment including the At4G16440 gene was able to complement the defect in FWA-GFP expression and embryonic lethality in the nar1-3 mutant (Fig. S1c,d). The heterozygous nar1-1 (nar1-1/+) mutant also showed impaired FWA-GFP expression and seed abortion at an early stage (data not shown), similar to nar1-3 (Fig. 1a–f); however, homozygous nar1-4 (nar1-4/−) was not affected in FWA-GFP expression and produced seeds with a normal appearance, even though the level of NAR1 transcripts was reduced (Fig. S2). Therefore, nar1-4 is a weak, but not a null allele.
NAR1 is not a specific FWA-GFP regulator and affects endogenous imprinted genes
As the nar1-1 and nar1-3 mutants were affected in expression of the transgenic imprinted FWA-GFP gene, we determined the level of transcripts of two endogenous imprinted genes, MEDEA (MEA) and FERTILIZATION INDEPENDENT SEED 2 (FIS2), in seeds. For this analysis, we crossed the mutants to wild-type plants and separated seeds with strong or weak GFP signals at 3 DAP under a fluorescence microscope (Fig. 2a). The level of GFP transcripts was considerably reduced in nar1-1 and nar1-3 mutants, validating our sampling method. Importantly, the mutants also showed a consistently reduced level of endogenous FWA and FIS2 transcripts, although the effect was not as great as for GFP (Fig. 2b). The mutants did not show any change in the level of MEA transcripts (Fig. 2b), a result that was consistent with MEA-GFP reporter expression (Figs 3c,d, S3a,b).
Genetic analyses of the nar1-3 mutant allele
Endosperm development was delayed in the silique of self-pollinated nar1-3 heterozygotes, as demonstrated in the MEA-GFP line (Fig. 3c,d). In wild-type plants, the majority of seeds reached the endosperm developmental stage with > 32 nuclei at 2 DAP (Fig. 3c); by contrast, putative nar1-3/− seeds were developmentally delayed and showed arrest at the four or eight nuclei stage (Fig. 3d, Table 1). Many of the previously identified A. thaliana mutants of the CIA pathway are homozygous lethal (Bych et al., 2008; Kohbushi et al., 2009; Varadarajan et al., 2010; Luo et al., 2012). As in the nar1-1 mutant (Luo et al., 2012), self-pollination of nar1-3 heterozygotes also failed to produce homozygous mutant progeny. If NAR1 functions in embryo development as a CIA component, nar1-3 could influence embryo development at the same stage that was affected in other CIA mutants (nbp35-1, cia1-1 and atr3-3/tah18-3). To confirm this, we compared their respective embryo phenotypes (Fig. S4a). Self-pollination of the heterozygous mutants yielded a 3 : 1 segregation ratio of normal to arrested seeds at an early stage of seed development, similar to nar1-3/+ (Fig. S4b, Table 2). These ratios are consistent with a recessive trait at a single locus. As expected, the nar1-3, nbp35-1, cia1-1 and atr3-3/tah18-3 heterozygous plants showed no obvious abnormalities in morphology during the vegetative phase (data not shown), and microscopic examination of the aborted seeds showed that mutant embryos failed to grow beyond the octant cell stage (Fig. S4c–h), although an additional division was seen in a small population of cia1-1 seeds (data not shown). In the nar1-3 abnormal seeds (and also in those of the other mutants examined), embryos were arrested at the pre-globular stage; in seeds with a normal appearance, the embryo had continued to grow to the globular or heart stage (Fig. 3e–h). Thus, based on the results here and those published previously (Bych et al., 2008; Kohbushi et al., 2009; Varadarajan et al., 2010; Luo et al., 2012), embryonic growth is arrested at a similar stage of development in nar1-3, nbp35-1, cia1-1 and atr3-3/tah18-3 mutants, supporting the proposal that the CIA pathway is essential for early seed development.
Table 1. Endosperm development in self-pollinated and outcrossed nar1 and wild-type plants of Arabidopsis thaliana
No. of endosperm nuclei (2 d after pollination)
n, number of siliques examined.
413 (n =9)
439 (n =9)
507 (n =9)
469 (n =9)
Table 2. Seed phenotype in mature siliques at 10–14 d after pollination (DAP) in Arabidopsis thaliana mutants of the cytosolic iron–sulfur assembly (CIA) pathway
P-value (χ2 test)
Next, we assessed the gametophytic effect of the mutant alleles by determining their transmission rates in reciprocal crosses with wild-type Col-0. Genotypes of F1 plants were identified by PCR. No evidence of reduced transmission of the mutant allele was found in nbp35-1, cia1-1 or atr3-3/tah18-3. However, the transmission rate of the nar1-1 and nar1-3 mutant alleles through the male was significantly decreased compared with the expected 50% transmission; transmission through the female was not affected. When we used the Ler strain instead of wild-type Col-0 in the crosses, the male transmission defect for the nar1 allele was still observed (Table 3). Thus, we conclude that loss of NAR1 caused a male-specific transmission defect that was not seen for other mutants affected in the CIA pathway.
Table 3. Transmission of Arabidopsis thaliana mutations of the cytosolic iron–sulfur assembly (CIA) pathway
Preferential expression of NAR1 in dividing cells and young seeds
Given the fundamental importance of the CIA pathway to a wide range of cellular activities, it is likely that the genes in this pathway are expressed throughout the plant life cycle. To determine whether this is indeed the case, we first examined NAR1 expression in different organs using semiquantitative RT-PCR. NAR1 transcripts were detected in all organs examined, with a relatively uniform level of expression in different tissues and at different developmental stages (Fig. 4a). Next, we also analyzed spatial and temporal expression patterns using NAR1 promoter–GUS or GFP fusion to full-length NAR1 in eight independent transgenic lines for the GUS construct and six for the GFP construct. As expected from the results of the RT-PCR analysis, promoter activities were observed throughout plant development (Fig. 4b–g). However, activities were higher in tissues with proliferating cells (Fig. 4b,d,e). In young seedlings, GUS activity was observed around the shoot apex, in young leaves, in the vascular tissues of leaves and in the root tips (Fig. 4b,d). In inflorescences, the pedicel and carpel showed GUS activity (Fig. 4c). Expression of the promoter GFP was similar in root tips (Fig. 4e). Strong GUS staining and GFP florescence were also found in the developing seed (Fig. 4f,g). These results are consistent with seed lethality in nar1 homozygotes. Our observations suggest that NAR1 function is not ubiquitous but is limited to developing tissues.
Transcriptional profiling of nar1-4 and nbp35-3 homozygous mutants in vegetative tissues
As described herein, the nar1-3 mutation was identified in a screen for abnormal gene expression in the endosperm. The NAR1 gene expression pattern indicated that NAR1 could also function in the vegetative phase (Fig. 4). To investigate whether the mutation affected expression of specific genes in vegetative tissues, we performed a genome-wide analysis of nar1-4/− transcription profiles using the Affymetrix ATH1 GeneChip. We also used nbp35-3 homozygous plants as NBP35 and NAR1 both function in a late step in the CIA pathway according to yeast studies (Lill & Mühlenhoff, 2006). In addition, NBP35 and NAR1 are mutually highly ranked in the A. thaliana co-expression database ATTED-II (Obayashi et al., 2009). Two-week-old whole wild-type (Wt), nar1-4 homozygous and nbp35-3 homozygous plants were sampled. The correlation coefficients (r2) of fold change between nar1-4 and Wt and between nbp35-3 and Wt for probes with both P-values < 0.1 were 0.74134 for replication 1 and 0.72544 for replication 2 (Fig. 5a,b). To identify significant changes in expression levels, we compared with transcriptional profiles of the mutants and wild-type and determined the significance of differences using a P-value of < 0.05 (Student's t-test) from two independent biological replications. When we compared the 541 differentially expressed genes in nar1-4/− with the 451 genes in nbp35-3/−, we found that 24 genes were up-regulated in both mutants (out of 218 genes in nar1-4/− and 248 genes in nbp35-3/−) (Fig. 5c), and 27 genes were down-regulated in both mutants (out of 323 genes in nar1-4/− and 203 genes in nbp35-3/−) (Fig. 5d). The frequency of genes with similar changes in expression was greater than expected from randomly pooled genes. Among 541 differentially expressed genes in nar1-4/−, we selected subsets of genes showing > 1.5-fold or < 0.75-fold difference between the two genotypes. This selection process identified 83 genes (Table S2). We next categorized the function of these genes using the Gene Ontology (GO) Annotations at TAIR (Berardini et al., 2004). The terms ‘transition metal ion transport’ and ‘response to nitrate’ were significantly enriched in the nar1-4/− plants (Table S3). Interestingly, five genes given the GO terms ‘iron ion transport’ and ‘cellular response to iron ion starvation’ were all down-regulated (Table S3, Fig. S5). Among 451 differentially expressed genes in nbp35-3/−, 27 genes conformed to the criterion of at least > 1.5-fold or < 0.75-fold change compared with wild-type (Table S4). The GO analysis of this gene subset indicated overrepresentation of genes for response to environmental stress, including ‘cell redox homeostasis’, ‘response to osmotic stress’ and ‘response to cold’ (Table S5). However, because of the relatively small number of genes in the selected subset, no specialized GO term with more than four genes was identified. Although the nbp35-3 transcriptional profile showed limited changes compared with nar1-4/−, we found that the At5G58980, At2G43590 and At5G44620 loci were down-regulated in both nar1-4/− and nbp35-3/− (P-value < 0.05 and < 0.75-fold change). Thus, the differentially expressed genes in nar1-4/− and nbp35-3/− were positively correlated, suggesting that deficiencies in each mutant have a similar effect on transcriptional profiles.
Defective NAR1 confers paraquat resistance
Previous studies showed the NAR1 protein contains an Fe–S cluster(s) (Cavazza et al., 2008). It has been reported that some Fe–S proteins are susceptible to oxidative damage (Crack et al., 2012). Hence, we hypothesized that oxidative stress might influence the phenotype of homozygous nar1-4 plants. To test this hypothesis, we compared the rates of survival of nar1-3/+, nar1-4/− and wild-type seedlings on paraquat-containing medium. Paraquat undergoes cellular redox cycling and continuously provides reactive oxygen species in the cell. Therefore, paraquat is commonly used as a responsive assay for oxidative stress. Survival was scored as the absence of bleaching at the shoot apex. The survival rate of wild-type and nar1-3 heterozygous seedlings fell with time, and most did not survive to 3 wk after the start of paraquat treatment (Fig. 6a–d). Unexpectedly, nar1-4 homozygous seedlings showed a nearly 100% survival rate and produced true leaves (Fig. 6d–f). In an RNAi knockdown of NAR1, seedlings of RNAi lines 1 and 2, in which NAR1 expression was effectively suppressed, also exhibited paraquat resistance (Fig. S6). Thus, the NAR1 deficiency produced a significant paraquat-resistance phenotype. These observations indicate that NAR1 functions in the oxidative stress response.
Aconitase activity in nar1-4/−
Defects in the CIA pathway lead to decreased activity of cytosolic Fe–S proteins by diminishing the rate of Fe–S cluster incorporation into apoproteins. Aconitase is a representative Fe–S protein in the cytosol and mitochondria. It was previously demonstrated that aconitase activity was affected by a mutation in NAR1 in yeast (Balk et al., 2004). To determine whether aconitase activity changed in nar1-4/−, we analyzed 2-wk-old wild-type and nar1-4/− whole seedlings for enzymatic aconidase activity. ACO1 is one of the three ACO proteins in A. thaliana and it is predominantly active in the cytosol (Arnaud et al., 2007). Hence, we used the aco1 homozygous mutant as a negative control. We found that aconitase activities were significantly decreased in aco1/−. In nar1-4/−, they were modestly but reproducibly decreased compared with the wild-type (Fig. S7). This result suggests that A. thaliana NAR1 is required for Fe–S protein functions, which is consistent with its role as a component of the CIA pathway.
In view of the number of putative Fe–S proteins in the cytosol and nucleus, it would appear that the CIA pathway is essential to a wide range of cellular processes. Thus, detailed phenotypic analyses of mutant alleles of the component genes will provide valuable information on the roles of the CIA pathway. In this study, we characterized mutations of NAR1, a gene that encodes an Fe-hydrogenase-like protein that appears to function in the CIA pathway (Luo et al., 2012). Our analyses showed that NAR1 is required for the expression of the imprinted FWA gene and for seed development, and is involved in oxidative stress responses in the vegetative tissues.
A possible mechanism for transcriptional activation of endosperm-specific genes
The nar1-3 mutation was identified in a screen for impaired transcriptional activation of the endosperm-specific gene FWA. The CIA pathway seems to control gene transcription, as RNA polymerase subunits and transcription elongation factors have Fe–S cluster binding motifs (Hirata & Murakami, 2009; Okada et al., 2010). However, it is unlikely that NAR1 could influence global transcription as we found that MEA-GFP was normally activated in the central cell and endosperm (Fig. 3c,d). In the qRT-PCR analysis, the normal expression of MEA in seeds also excludes the possibility that impaired FWA and FIS2 expression was attributable to abnormal female gametophyte development (Fig. 2). Overall, our evidence suggests that impaired transcriptional activation of FWA and FIS2 in nar1 mutants is mediated through another more gene-specific system. FWA, FIS2 and MEA are genes known to be imprinted via changes in DNA methylation. However, the nar1 mutation did not affect MEA expression in the endosperm, in contrast to FWA and FIS2. This difference may be explained by the presence of an additional regulatory mechanism at the MEA locus (Baroux et al., 2006; Gehring et al., 2006; Jullien et al., 2006a,b). Therefore, the fact that the nar1 mutation affects FWA and FIS2 but not MEA implies that NAR1 may be involved in regulating gene expression simply via the DNA methylation pattern. It is notable that impaired FWA-GFP expression was observed in half of the ovules before fertilization (Fig. 1a,b). Additionally, an impaired FWA-GFP phenotype was evident even when wild-type pollen was used for pollination (Fig. 1c–f). These observations suggest that the impairment of transcriptional activation of FWA and FIS2 may be a maternal gametophytic effect, as both genes are known to be imprinted. Imprinted genes show a parent-of-origin effect on expression. It is known that transcriptional activation of many imprinted genes requires the DNA demethylase activity of DME. The DME protein has an Fe–S cluster loop motif (Stoesser et al., 2002). Given that DME functions in the nucleus, the protein is a possible target for the CIA pathway. Indeed, it was reported that A. thaliana plants homozygous for the ae7 gene mutation have defects in their CIA pathway and exhibit DNA hypermethylation at specific loci in the vegetative phase (Luo et al., 2012). Taken together, these data suggest that NAR1 could influence FWA-GFP expression through a DNA (de)methylation mechanism.
NAR1 is required not only for zygotic development but also for paternal transmission
Inactivation of two endosperm-specific genes, FWA and FIS2, in nar1-3 is attributable to a female gametophytic effect (Fig. 1a); by contrast, embryo defects were only observed when the nar1 heterozygous mutant was self-pollinated (Tables 1, 2). Most mutants of CIA components fail to generate homozygous mutants (Bych et al., 2008; Kohbushi et al., 2009; Varadarajan et al., 2010; Luo et al., 2012). We investigated the transmission rates and compared embryo phenotypes between nar1-3, nbp35-1, cia1-1 and atr3-3/tah18-3 alleles. This analysis showed that embryo lethality was a zygotic effect and that putative homozygous plants were aborted at an early embryo stage (Figs 3e–h, S4c–h, Tables 2, 3). A similar observation has been made in mice, where knockout of the NAR1 homolog IOP1 leads to lethality before embryonic day 10.5 (Song & Lee, 2008). Thus, CIA components are essential to early embryonic development in both plants and animals. Among the four CIA components investigated, nar1 alleles showed a partial male-specific transmission defect (Table 3). In this respect, therefore, NAR1 appears to be different from the other CIA components. In yeast studies, each component showed a somewhat different cellular localization pattern or different maturation process (Balk et al., 2004, 2005; Hausmann et al., 2005; Vernis et al., 2009). These data may suggest that each CIA pathway component functions at a different step of the pathway and not as one complex. This might explain why not all of the already characterized CIA mutants share all the phenotypes. To our knowledge, the additional role of NAR1 in the male gametophyte is first reported here.
The role of NAR1 in iron homeostasis and oxidative stress
To circumvent embryonic lethality and gain further insights into the role of the CIA pathway in vegetative tissues, we selected the weak alleles nar1-4 and nbp35-3 and used these to produce viable homozygotes. A transcriptional profile of nar1-4/− showed that five genes involved in ‘iron ion transport’ and ‘response to nitrate’ were down-regulated in the homozygote mutant. This may reflect a physiological compensation in nar1-4 through iron homeostasis. Previous reports indicate that nitric oxide targets some Fe–S proteins and regulates their activities catalytically or structurally (Butler & Megson, 2002). Therefore, if nitric oxide signaling is mediated through Fe–S proteins, it is possible that defects in the Fe–S cluster assembly in nar1 might cause an impaired response to these signals. Further analyses are necessary to investigate these possibilities. Iron is used in the form of Fe–S clusters or heme in organisms. The dysregulation of iron/sulfur homeostasis might therefore be a source of oxidative stress, indicating that organisms may need to strictly regulate this homeostasis in cellular processes. Hence, differentially expressed genes in seedlings seem to reflect the defects in iron homeostasis in the nar1-4 mutant rather than a specific transcriptional regulation (Table S3).
The Fe–S clusters of some proteins are sensitive to oxidative damage (Crack et al., 2012). The molecular mechanism for resistance to paraquat in nar1-4/− is unknown, but one possibility is that the generation of reactive oxygen species or the induction of the cell death pathway might be influenced by a defective Fe–S protein(s). Interestingly, yeast mutants for another CIA component, Tah18, show resistance to H2O2, an oxidative stress reagent (Vernis et al., 2009). Vernis et al. postulated that the yeast Tah18 protein might be required for cell death signaling to mitochondria. An alternative explanation comes from the report that a plant with decreased aconitase activity showed a reduction in paraquat sensitivity (Moeder et al., 2007). Reduced aconitase activity in nar1-4/− supports the idea that the failure of Fe–S cluster incorporation into aconitase proteins in the absence of the intrinsic CIA pathway leads to paraquat resistance in nar1-4/−.
We thank Dr R. L. Fischer for MEA-GFP seeds. We acknowledge the Salk Institute and the GABI-KAT project for providing T-DNA lines. The technical assistance of Y. Kinoshita, Yuriko Ikeda, M. Yamanaka and M. Yoshino was invaluable. We also thank Dr Yoko Ikeda and Dr A. Ono for helpful advice. This work was supported by grants from the Bio-oriented Technology Research Advancement Institution (BRAIN) and Grant-in-Aid for Scientific Research on Innovative Areas (23113001 and 23113003 to T.K.).