Research Article

Signals from Chloroplasts Converge to Regulate Nuclear Gene Expression

See allHide authors and affiliations

Science  04 May 2007:
Vol. 316, Issue 5825, pp. 715-719
DOI: 10.1126/science. 1140516

This article has a correction. Please see:


Plastid-to-nucleus retrograde signaling coordinates nuclear gene expression with chloroplast function and is essential for the photoautotrophic life-style of plants. Three retrograde signals have been described, but little is known of their signaling pathways. We show here that GUN1, a chloroplast-localized pentatricopeptide-repeat protein, and ABI4, an Apetala 2 (AP2)–type transcription factor, are common to all three pathways. ABI4 binds the promoter of a retrograde-regulated gene through a conserved motif found in close proximity to a light-regulatory element. We propose a model in which multiple indicators of aberrant plastid function in Arabidopsis are integrated upstream of GUN1 within plastids, which leads to ABI4-mediated repression of nuclear-encoded genes.

Chloroplasts contain about 3000 proteins, of which more than 95% are encoded by nuclear genes. As a result, enzymatic and photosynthetic complexes within chloroplasts contain subunits encoded by two spatially separated genomes (1). This necessitates a tight coordination of gene expression that involves two-way signaling between the chloroplasts and the nucleus. While plastid development is largely under nuclear control, developmentally arrested or damaged plastids can regulate nuclear gene expression via retrograde signaling pathways. Mitochondria also emit signals that regulate the expression of nuclear genes in response to its condition; this signaling is involved in important processes including senescence and tumor progression (2).

Three independent plastid-to-nucleus retrograde signaling pathways have been described (35). In the best-defined pathway, accumulation of Mg–protoporphyrin IX (Mg-protoPIX), a chlorophyll biosynthetic intermediate, leads to down-regulation of hundreds of genes in Arabidopsis and has been shown to be involved in gene regulation in Chlamydomonas reinhardtii (6). Gene products of four Arabidopsis GENOMES UNCOUPLED (GUN) loci, GUN2, 3, 4, and 5 (6), are involved in modulating Mg-protoPIX levels (69). A second pathway represses Lhcb expression in response to inhibition of plastid gene expression (PGE) and requires GUN1 (3, 5, 10). The third signaling pathway mediates signals derived from the reduction/oxidation (redox) state of the photosynthetic electron transfer chain (PET) and affects both photosynthesis-related and stress-related genes (3, 5). The redox state of the plastoquinone pool and other PET components, as well as reactive oxygen species (ROS), play a role in PET-derived signaling (3, 5). In this work, we present evidence in support of a single retrograde signaling pathway that integrates information from multiple indicators of plastid functional state.

GUN1 encodes a chloroplast-localized pentatricopeptide-repeat protein. We identified seven new gun1 alleles in two genetic screens (11) by using a transgenic Arabidopsis line carrying a luciferase reporter under the control of the Lhcb (encoding a light-harvesting chlorophyll a/b–binding protein) promoter, which is known to respond to retrograde signals (Col 6-3, Fig. 1A). We used either norflurazon (NF), an inhibitor of carotenoid biosynthesis that causes photo-oxidative damage and accumulation of Mg-protoPIX in the light, or lincomycin, an inhibitor of plastid protein synthesis. We mapped GUN1 to a ∼93-kilobase region on the bottom arm of chromosome 2. When we sequenced candidate genes, we found mutations in At2g31400 in all gun1 alleles (Fig. 1B). Genomic fragments, encompassing the predicted open reading frame of At2g31400, fused at its C terminus to either green fluorescent protein (GFP) or six-hemagglutinin (6xHA) tags, were able to rescue the gun1 phenotype (Fig. 1C).

Fig. 1.

GUN1 is a pentatricopeptide-repeat protein. (A) New alleles of gun1 show enhanced derepression of Lhcb. Seedlings were treated with lincomycin for 6 days. Lhcb RNA was detected by Northern blotting. We used 25S ribosomal RNA (rRNA) as a loading control. (B) GUN1 encodes a pentatricopeptide-repeat protein. Schematic representation of GUN1 shows the position and nature of the mutations in various alleles. Boxes and lines indicate exons and introns, respectively. (C) Genomic fragments of At2g31400 complement gun1. Total RNA was isolated from homozygous transgenic lines, and Lhcb RNA was detected by Northern blotting. Ethidium bromide–stained rRNA is shown as a loading control. (D and E) GUN1 and pTAC2 colocalize within chloroplasts. N. benthamiana leaves were infiltrated with Agrobacterium carrying the indicated constructs, and protoplasts were prepared 4 days postinfiltration. Fluorescence was imaged by confocal microscopy, and a representative image from a single protoplast is shown. Scale bar, 10 μM.

GUN1 encodes a 918–amino acid polypeptide that is a member of the P subfamily of pentatricopeptide repeat (PPR)–containing proteins (12). PPR-protein coding genes are a vastly expanded family in land plants, with ∼441 and >650 members in Arabidopsis and rice, respectively (12). Putative GUN1 orthologs can be readily identified in rice (Os12g37550) and poplar (fig. S1). A majority of the PPR proteins are targeted to either mitochondria or plastids, where they have been proposed to function in processing, editing, stability, and translation of RNA molecules (1216). In addition to the PPR motifs, GUN1 also has a small mutS–related (SMR) domain (17). The SMR domain is highly conserved across species and is found in small single-domain proteins, such as Escherichia coli Smr, as part of MutS2 (17), or in association with PPR motifs in higher plants. There are five other Arabidopsis proteins that have a domain structure similar to GUN1, i.e., several PPR motifs followed by a C-terminal SMR domain and one of those, pTAC2, is part of transcriptionally active complexes isolated from Arabidopsis and mustard (18). PPR motifs have been proposed to mediate interactions with nucleic acids (12), including DNA (19, 20), and the SMR domain is found in proteins involved in DNA repair and recombination. We observed that a fragment of GUN1 containing both the PPR and SMR domains could bind DNA (fig. S2A). The SMR domain alone bound very strongly, whereas a fragment containing the PPR motifs alone showed weak binding. This suggests that GUN1 function may be distinct from the known PPR proteins, which are thought to be involved in RNA processing activities. GUN1 has a putative plastid-targeting signal sequence at its N terminus, and a construct lacking this sequence failed to rescue a gun1 mutant plant. Full-length GUN1-GFP fusion constructs rescued the gun1 phenotype, but we did not observe GFP fluorescence in several lines that we tested. Using Agrobacterium-mediated transient expression in Nicotiana benthamiana leaves (21), we found that GUN1– and pTAC2–yellow fluorescent protein (YFP) fusions localized to distinct foci within chloroplasts of mesophyll cells (Fig. 1D). When GUN1-YFP and pTAC2–cyan fluorescent protein (CFP) were coexpressed, we could clearly observe colocalization of the YFP and CFP signals (Fig. 1E), indicating that, like pTAC2, GUN1 is also associated with sites of active transcription on plastid DNA.

GUN1 is insensitive to accumulation of Mg-protoPIX. Among the known gun mutants only gun1 has a gun phenotype when grown in the presence of either lincomycin or NF. However, NF treatment also results in photo-oxidative damage to plastid ribosomes, and it has been suggested that this could indirectly trigger the PGE pathway (22). Because the PGE pathway is active only during the first 3 days after germination (10), we treated older seedlings with NF to test whether GUN1 plays a role in the Mg-protoPIX pathway. Similar to gun5 seedlings, 4-day-old light-grown gun1 seedlings were unable to repress Lhcb expression when treated with NF (Fig. 2A). We then tested to see if gun1 mutants were impaired in their response to elevated Mg-protoPIX levels induced by dipiridyl (DP), which is known to restore the repression of Lhcb genes in gun2 and gun5 seedlings grown on NF (6). DP treatment did not restore Lhcb repression in gun1 seedlings grown on NF or lincomycin (Fig. 2B). The results of the late NF treatment and the inability of DP to restore repression of Lhcb in gun1 demonstrate that GUN1 plays a role in the Mg-protoPIX pathway. Unlike gun2, 3, 4, and 5, gun1 does not seem to be involved in chlorophyll biosynthesis, which indicates that it most likely acts downstream of Mg-protoPIX accumulation.

Fig. 2.

GUN1 and GUN5 are components of the same signaling pathway. (A) gun1 seedlings are impaired in retrograde regulation triggered by NF. Four-day-old green seedlings were treated with NF and harvested 9 days later. Methylene blue–stained membrane shows rRNA as a loading control. (B) gun1 is insensitive to DP treatment. gun1-9 and gun5 seedlings were treated with lincomycin or NF for 5 days and harvested 24 hours after addition of 1 mM DP and 0.5 mM MgCl2 (final concentration). Methylene blue–stained membrane shows rRNA as a loading control. (C) Microarray experiments indicate strong correlation between genes misregulated in gun1-9 and gun5. Average log fold changes for 7778 nuclear genes are plotted. (D) gun1 and gun5 have a similar pattern of nuclear gene expression. Heat map depicts 1150 genes differentially expressed in gun1-9 or gun5 compared with wild type. Average values were calculated using Col 6-3-nf as reference.

Common targets for gun1 and gun5. To identify a comprehensive set of nuclear genes under the control of retrograde pathways, we analyzed the global gene expression response of wild-type, gun1, and gun5 seedlings to NF, using the Affymetrix Arabidopsis ATH1 genome arrays. From these arrays, 7778 genes showed differential expression between either the condition (control versus NF) or the genotype (wild-type versus gun1,gun5 and gun1 versus gun5). A striking linear correlation was observed between the multiple of change (fold increase or decrease, logarithmic scale, hereafter, log fold change), in gene expression observed in NF-treated gun1 or gun5 compared with wild type (Fig. 2C). Clustering analysis of 1150 genes that were differentially expressed in gun1 or gun5 compared with wild type on NF showed a very similar pattern between gun1 and gun5 (Fig. 2D). Of these, 488 nuclear genes were repressed by NF in wild type [P < 0.01 RankProd (23)], and 370 of those had expression in either gun1 or gun5 on NF greater than twice what was seen in the wild type. Of these 370 genes, 329 (89%) were derepressed in both gun1 and gun5, which strongly implicated GUN1 and GUN5 in the same pathway. The lack of an additive effect in a gun1-9,gun5 double mutant grown on NF compared with individual mutants lends further support to this conclusion (fig. S2C).

Identification of ABI4 as a nuclear component of retrograde signaling. To identify regulatory elements enriched in the promoters of target genes of the retrograde signaling pathways, we analyzed 500–base pair (bp) regions of sequence upstream of the 329 genes derepressed in both gun1 and gun5. An ACGT motif, the core of both the light-responsive G box (CACGTG) (24) and the abscisic acid (ABA) response element (ABRE) (2527), appeared 944 times in the promoters of these genes compared with the expected 239 times (11); a Z score of 3.2 indicated that it is significantly overrepresented (P <0.01).

To test whether components of the ABA signaling pathway play a role in retrograde signaling, we analyzed the retrograde response in ABA-deficient and ABA-insensitive mutants (25, 28). Only one ABA-related mutant, abi4, showed Lhcb mRNA accumulation even when chloroplast function was impaired (Fig. 3A). Lincomycin treatment resulted in a reduction in Lhcb transcript levels to 1/10th those in the untreated seedlings, in either wild-type or gun5 seedlings (Fig. 3B and fig. S3A) (3, 29). In contrast, transcription in gun1-1 or abi4 (Fig. 3B) was reduced to one-third of controls, which indicated that, like GUN1, ABI4 plays a role in PGE-dependent signaling.

Fig. 3.

abi4 is a gun mutant. (A) Lhcb1 mRNA accumulation in 6-day-old lincomycin-treated wild type (of three Arabidopsis ecotypes Col-0, Ler, and Ws), gun1, and ABA mutants. gun1 and abi4 are in the Col-0 background, aba1, abi1, abi2, and abi3 are in Ler, and abi5 is in Ws. (B) Quantitative analysis of three replicates showing the average ratio of Lhcb/rRNA between lincomycin-treated and untreated wild type, gun1, and abi4. (C) Lhcb1 mRNA accumulation in 8-day-old ABA-treated wild type, gun1, and abi4. (D) Luciferase and GFP activity in lincomycin-treated seedlings of wild type, gun1, and gun1 overexpressing an ABI4-GFP fusion protein.

The identification of abi4 as a retrograde signaling mutant suggested that ABA, a known stress hormone, might be a plastid-derived retrograde signal. This does not appear to be the case, because an ABA-deficient mutant (aba1) did not accumulate any Lhcb mRNA when grown on lincomycin (Fig. 3A). Furthermore, addition of 5 μM ABA, with or without NF, inhibited the accumulation of Lhcb transcripts inwildtype and gun1, but had no effect on abi4 seedlings (Fig. 3C, fig. S3B). gun1 was also distinguished from abi4 in germination assays (fig. S3C). Collectively, these results indicate that ABA is unlikely to be the actual retrograde signal. It is noteworthy that GUN5 has recently been shown to be an ABA receptor, but its Mg-chelatase activity appears to be distinct from its role in ABA signaling (9, 30).

ABI4 acts downstream of GUN1 in retrograde signaling. Several lines of evidence indicate that GUN1 and ABI4 act in the same signaling pathway. First, gun1 is epistatic to abi4; Lhcb mRNA levels in a gun1-1,abi4 double mutant were similar to a gun1-1 single mutant (fig. S3D). Second, we observed significant overlap in nuclear genes misregulated in gun1-1 and abi4, 57% and 46% for genes that were either derepressed or repressed, respectively (fig. S4). Finally, overexpression of ABI4-GFP suppressed the gun1-1 phenotype. We transformed Lhcb::luciferase gun1-1 seedlings with an ABI4-GFP fusion driven by the cauliflower mosaic virus CaMV 35S promoter. Individual second-generation (T2) seedlings were imaged for ABI4-GFP and luciferase expression. We detected no luciferase activity in lincomycin-treated T2 seedlings that overexpressed the fusion protein (Fig. 3D).

The GUN1-ABI4 pathway integrates multiple chloroplast-derived signals. We then tested whether ABI4 is also involved in the Mg-protoPIX pathway. Wild-type, gun1, gun5, and abi4 were grown on NF for 4 days in the dark and then transferred to light. After 3 hours of illumination, all seedlings accumulated similar levels of Lhcb mRNA (fig. S3E), which confirmed that neither the Mg-protoPIX nor the PGE-dependent pathway had been triggered at this stage. With a prolonged light treatment, gun1, gun5, and abi4 accumulated higher levels of Lhcb mRNA than wild type; this was also true when we measured luciferase activity in Lhcb::luciferase seedlings grown under these conditions (Fig. 4A). These results indicate that, like GUN1 and GUN5, ABI4, also plays a role in the Mg-protoPIX pathway.

Fig. 4.

GUN1 and ABI4 mediate several chloroplast-derived signals. (A) Lhcb1 mRNA accumulation (left) and relative luciferase activity (right) in 7-day-old Norflurazon-treated wild type, gun1, gun5, and abi4. Plants were grown in the dark for 4 days and transferred to light for 3 more days. (B) Lhcb1 mRNA levels in 4-week-old wild type, gun1, and abi4 seedlings grown under low light and exposed to high light intensities for the indicated time. (C) Zat12 and Zat10 mRNA levels in seedlings described in (B). (D) Lhcb1 mRNA accumulation in wild type, gun1, and abi4 seedlings treated with 7% glucose.

We then tested if GUN1 and ABI4 are also involved in PET-derived signaling. Lhcb transcript levels were reduced in wild-type seedlings in response to intense light; this reduction was evident even after a 15-min exposure (Fig. 4B). Lhcb mRNA levels in both gun1 and abi4 were less affected by such exposure, although Lhcb levels declined in both mutants after 3 to 6 hours (Fig. 4B). We then studied the expression of Zat12 and Zat10, two genes that encode zinc-finger DNA binding proteins, in response to intense light. Zat12 is required for the expression of APX1 under oxidative stress (31), and protein products of both these genes mediate the response of Arabidopsis to several abiotic stresses (32). Rapid, transient accumulation of both transcripts was observed within 15 min of exposure to intense light; a more sustainable accumulation was observed after 1 to 3 hours (Fig. 4C). The latter was delayed in both gun1 and abi4 seedlings, which suggests that not only repression, but also induction, of gene expression in response to a PET-derived signal, requires GUN1 and ABI4. The role of both GUN1 and ABI4 in all three retrograde signaling pathways is in agreement with a previously proposed “master switch,” which controls the expression of a large number of nuclear genes in response to plastid-derived signals (33).

abi4 has also been identified as a sugar-insensitive mutant (28), and sugar signaling through ABI4 has been linked to chloroplast retrograde signaling (34). PET-induced gene expression is inhibited by sugar and this inhibition requires ABI4 (34). In sucrose uncoupled 6 (an allele of abi4), exogenous sucrose does not inhibit the expression of nuclear-encoded photosynthetic genes (35). To understand the connection between sugar and retrograde signaling, we examined Lhcb expression in 4-day-old seedlings treated with exogenous glucose. Lhcb expression in wild type was significantly reduced in response to 7% glucose in the growth medium (Fig. 4D), but only slightly decreased in abi4 and largely unchanged in gun1 seedlings. Sorbitol did not affect Lhcb expression, which suggested that this is not a response to osmotic stress. Glucose also enhanced the greening of etiolated wild-type seedlings, whereas the greening of gun1 and abi4 seedlings was less affected (fig. S5). These results suggest that in addition to retrograde signaling, GUN1 and ABI4 are involved in glucose-mediated repression of photosynthetic gene expression and the transition from heterotrophic to photoautotrophic growth. Unlike abi4, germination of gun1 was inhibited by 7% glucose, which suggested that it is not strictly a glucose-insensitive mutant. However, because it is required for the glucose-mediated Lhcb repression, it is possible that this represents a fourth, previously unidentified, plastid-derived signal.

ABI4 can bind the promoter of Lhcb. Our promoter analysis did not reveal any known binding sites for ABI4 (27, 36), yet a yeast one-hybrid assay indicated that ABI4 could directly bind the promoter of Lhcb (Fig. 5A). In CUF1, which is a G-box element required for retrograde signaling (6), two cytosine residues precede the G box and that results in two overlapping elements, the CCAC and the ACGT core. The CCAC was overrepresented in the promoters of the 329 genes derepressed by gun1 and gun5; it appeared 1048 times, although the expected frequency was only 302 (Z score of 11.34). About 70% of these promoters contained a CCAC/ACGT combination, and in the majority, these elements were either overlapping or closely spaced (1 to 3 bp), with about 70% having fewer than 20 bp between the two. A similar enrichment was found in the genes derepressed on lincomycin in gun1 and abi4 (fig. S4). When the CUF1 element was mutated to TTACGT (leaving the G-box core intact), ABI4 could not facilitate growth on His-deficient media. In contrast, when the CUF1 element was mutated to CCACAA, which eliminated the G-box core, ABI4 enabled growth in the absence of His (Fig. 5A); this result suggested that ABI4 binds this CCAC motif. This motif is significantly smaller than previously described ABI4-binding sites, which suggests that it might be a core element required for ABI4 binding. It has been reported that light and retrograde signals are mediated by cis elements found in close proximity (3). It is conceivable that by binding to the CCAC motif, ABI4 inhibits G box–mediated, light-induced expression of photosynthetic genes when chloroplast development is arrested. Proteins that contain Apetala 2 (AP2) have been shown to act as transcriptional repressors in response to ABA, ethylene, and jasmonic acid (37). Not all of the promoters of retrograde-regulated genes contained the CCAC motif, and in some, it was too far from the G box to enable inhibition by competitive binding. Repression of those genes by retrograde signals might occur through different ABI4-binding sites.

Fig. 5.

ABI4 can bind the promoter of Lhcb1*2. (A) A yeast strain with the HIS3 gene driven by a 200-bp Lhcb1*2 promoter containing either wild type (CCACGT) or two mutated (TTACGT and CCACAA) motifs, was transformed with a plasmid encoding the GAL4 activation domain with (ABI4-AD) or without (AD) an ABI4 fusion. Transformants were grown on His-deficient media with or without 5 mM 3-amino-1,2,4-triazole. (B) ABI4 expression is down-regulated in gun1. RNA was extracted from 5-day-old seedlings grown with or without lincomycin. First strand cDNA was used as template with ABI4 or ubiquitin UBQ10 specific primers. Polymerase chain reaction products were transferred to a membrane and probed with a radiolabeled ABI4 probe.

ABI4 is highly expressed in seeds and is barely detectable in seedlings after germination (38); its expression is induced by glucose (39) and possibly other sugars (40). In wild-type seedlings, expression of ABI4 did not change after lincomycin treatment; however, in both treated and untreated gun1 seedlings ABI4 mRNA levels were significantly lower than wild type (Fig. 5B). This suggests that, although ABI4 expression is not directly induced by lincomycin, GUN1 is required for its expression. ABI4 expression was up-regulated by lincomycin treatment of abi4 seedlings (Fig. 5B), possibly in an attempt to compensate for the nonfunctional ABI4. Although originally identified as an ABA-insensitive mutant, abi4 has since been identified in screens for sugar and salt resistance, as well as several others (28). Our data suggest that ABI4 may be the previously proposed master switch (33) required for the regulation of nuclear genes in response to developmental cues, as well as environmental signals, integrated within the chloroplast. Our findings also distinguish between chloroplast and mitochondria retrograde signaling, because neither organellar PPR-containing protein nor an AP2-type transcription factor are known to act in the mitochondria-derived signaling in plants (41), yeast, or mammalian cells (2).

In this study, we have presented genomic, genetic, and biochemical evidence that the three known retrograde signals, previously thought to be mediated by partially redundant signaling pathways, are integrated within the chloroplast upstream of GUN1 and that ABI4 acts downstream in the same pathway. We propose a model in which all three signals are integrated within plastids and GUN1 is required to either generate or transmit a second, common signal to the nucleus (fig. S6). In response to the GUN1-derived signal, ABI4 binds the promoter of Lhcb, which prevents the binding of G box–binding factors required for light-induced expression of nuclear photosynthetic genes (24). A potentially important benefit of multiple plastid-generated signals converging on one common pathway is that this convergence allows the monitoring of key parameters of a functional chloroplast while avoiding the potentially disastrous consequences of conflicting signals exerting control over nuclear gene expression. The chemical nature of this common signal now surfaces as the key question.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S6


References and Notes

View Abstract

Navigate This Article