PRC2-mediated H3K27me3 modulates shoot iron homeostasis in Arabidopsis thaliana

ABSTRACT

Plants use intricate mechanisms to adapt to changing iron conditions because iron is essential and also one of the most limiting nutrients for plant growth. Furthermore, iron is potentially toxic in excess and must be tightly regulated. Previously, we showed that chromatin remodeling via histone 3 lysine 27 trimethylation (H3K27me3) modulates the expression of FIT-dependent genes under iron deficiency in roots. This study builds on our previous findings, showing that H3K27me3 also modulates iron regulation in shoots. In the clf mutant, which lacks the predominant H3K27 tri-methyltransferase, we detected increased iron translocation to shoots under iron deficiency as compared to wild type. Transcriptomic analysis of shoots also revealed differential expression of genes consistent with higher iron levels in clf shoots than wild type shoots under iron-deficient conditions. In addition, we verify that YSL1 and IMA1, two genes involved in signaling iron status from shoots to roots, are direct targets of H3K27me3 and reveal iron-dependent deposition of H3K27me3 on these loci. This study contributes to a better understanding of the molecular mechanisms behind iron regulation in plants, as the effect of PRC2-mediated H3K27me3 on iron homeostasis genes expressed in the shoots has not been previously reported to our knowledge.

Introduction

As photosynthetic organisms, plants have a unique need for iron because iron is used as an essential cofactor for chlorophyll biosynthesis and photosynthesis, in addition to playing essential roles in metabolic processes conserved across photosynthetic and non-photosynthetic organisms. However, iron is one of the most limiting nutrients for plant growth. At the same time, plants must prevent or counteract cytotoxic reactive oxygen species (ROS) that can be generated by dysregulated iron homeostasis. Therefore, plants have evolved elaborate strategies to tightly regulate iron. Understanding the molecular mechanisms of iron homeostasis in plants is of high significance for agriculture and human health, as it will provide us with the knowledge to help improve plant growth and enhance the content of bioavailable iron in crops.

Upon iron deficiency, a large number of genes involved in iron acquisition and translocation are induced in plants. This so-called iron-deficiency response is elaborately regulated at the transcriptional through post-translational levels.^1-4 In Arabidopsis, at least 16 basic-helix-loop-helix (bHLH) transcription factors are involved in transcriptional regulation of the iron-deficiency response.³ FER-LIKE IRON DEFICIENCY-INDUCED TRANSCRIPTION FACTOR (FIT) positively regulates iron-acquisition genes via interacting with the bHLH subgroup Ib transcription factors bHLH38, bHLH39, bHLH100, or bHLH101 to induce Strategy I iron-acquisition genes.^5-9 Strategy I is a reduction-based mechanism dicotyledonous plants use to uptake iron, which involves the efflux of protons by P-type H⁺-ATPases¹⁰ and coumarins by PDR9,^11,12 reduction of ferric-chelates to ferrous iron by FERRIC REDUCTASE OXIDASE 2 (FRO2),¹³ and transport of ferrous iron into the root epidermal cell by Iron-Regulated Transporter 1 (IRT1).^14-17IRT1 can directly sense excess metals to regulate its own degradation.¹⁸ FIT has been considered to be the master regulator of Strategy I, but remains subjected to complex iron-dependent regulation at the transcriptional and post-translational levels.^5,19-21 Recently, Upstream Regulator of IRT1 (URI)/bHLH121 has been identified as a positive regulator that acts upstream of FIT.^22,23 URI/bHLH12 is ubiquitously expressed in various tissues regardless of iron conditions but is phosphorylated under iron deficiency and induces bHLH38/39/100/101, which encode transcription factors that heterodimerize with FIT.^22,23

POPEYE (PYE), another bHLH protein, is induced in the pericycle and negatively regulates genes involved in iron mobilization and translocation.²⁴ Some of its targets include NICOTIANAMINE SYNTHASE 4 (NAS4), which encodes an enzyme that synthesizes nicotianamine (NA) that chelates iron for vascular translocation, ²⁵ and FRO3, which encodes a mitochondrial ferric-chelate reductase expressed in the vasculature.²⁶ PYE is tightly co-regulated with BRUTUS (BTS), which encodes a RING E3 ligase with hemerythrin/HHE iron-binding domains.²⁴ BTS is stable under low iron and negatively regulates group IVc transcription factors that interact with PYE.²⁷ Two BTS paralogs, BTS-LIKE 1 (BTSL1) and BTSL2, have also been shown to negatively regulate the iron-deficiency response.²⁸ BTS regulates the abundance of bHLH104 and bHLH105²⁷ as well as phosphorylated URI upon iron re-supply.²² The heterodimer of URI/bHLH121 and IAA-LEUCINE RESISTANT 3 (ILR3), or its close homologs, directly regulates multiple iron homeostasis genes, such as BTS, PYE, and bHLH38/39/100/101, and indirectly controls FIT.^22,23

Iron acquired by the roots is translocated to the shoots through the xylem as an iron-citrate complex; Ferroportin 1/Iron-Regulated Protein 1 (FPN1/IREG1) effluxes iron from the pericycle to the xylem, ²⁹ where iron forms a complex with citrate and is released to the apoplastic space by FERRIC REDUCTASE DEFECTIVE 3 (FRD3).³⁰ Apoplastic iron ions are transported into the phloem by Oligo Peptide Transporter 3 (OPT3) for long-distance transport.³¹ NA is a non-proteinogenic amino acid that binds to iron and other transition metals.³² Iron-NA complexes move laterally from the vasculature to neighboring parenchymal cells via Yellow Stripe-Like (YSL) family transporters such as YSL1, YSL2, and YSL3.^33-35

In the leaves, iron is required for chlorophyll biosynthesis and is in high demand in the photosynthetic electron transport chain. To maintain appropriate iron levels while preventing iron-induced oxidative stress, plants must be able to coordinately respond to the needs in the leaves and the availability of iron in the rest of the plant. Iron sensing and signaling mechanisms are not as well understood as iron acquisition. However, multiple studies have provided evidence that, in addition to local signals in the roots, long-distance signals generated based on the shoot iron status control iron uptake in the roots.^36-41 Notably, many molecular players of iron translocation in Arabidopsis are critical in long-distance iron signaling. For example, the ysl1ysl3 double mutant is unable to induce iron-deficiency response in the roots, ⁴² whereas opt3-2, frd3/man1, and the quadruple nicotianamine synthase mutant nas4x-1 constitutively express iron-acquisition genes.^31,43,44 Small peptides of the IRON MAN (IMA) or FE-UPTAKE-INDUCING PEPTIDE (FEP) family have also been proposed to signal iron from shoots to roots. These peptides are highly expressed in the phloem and positively regulate the iron-deficiency response.^45,46

Our recent work established that Histone 3 lysine 27 tri-methylation (H3K27me3) contributes to the regulation of the FIT-dependent iron-deficiency response through direct targeting of FIT, FRO2, and IRT1.⁴⁷ H3K27me3 is a widespread repressive epigenetic modification, which is regulated by polycomb-group protein complexes and is critical in gene regulation and multicellular development.^48,49 In Arabidopsis, H3K27me3 is established by polycomb repressive complex 2 (PRC2), which contains the methyltransferase subunits CURLY LEAF (CLF) and SWINGER (SWN). CLF and SWN are partially redundant, but CLF predominantly accounts for the H3K27 tri-methyltransferase activity of PRC2.^50,51 There is a strong correlation between the sites of PRC2 binding and H3K27me3 deposition in Arabidopsis, ⁵² unlike in Drosophila where spreading of H3K27me3 can occur far beyond PRC2 binding sites.⁵³ Using the clf mutants, in which PRC2-mediated H3K27me3 deposition is significantly reduced, we identified that the loci of FIT, FRO2, and IRT1 are H3K27me3-enriched under iron-sufficient conditions. However, once exposed to iron-deficient conditions, H3K27me3 is partially removed to allow the induction of these FIT-dependent genes. Additionally, in clf mutants, the induction of FIT and its target genes was pronouncedly higher than in wild type. Based on these observations, we proposed that PRC2-mediated H3K27me3 attenuates the induction of iron-acquisition genes in the roots to protect against iron toxicity.

In this study, we build on our prior work by investigating the response to iron deficiency in clf shoot tissue. To our knowledge, the role of H3K27me3 on iron homeostasis in shoots has not been reported. We provide evidence that iron homeostasis genes highly expressed in the shoots, such as YSL1 and IMA1, are direct targets of PRC2-mediated H3K27me3. Our results suggest that H3K27me3 contributes to regulating iron homeostasis genes in the shoots by modulating iron translocation from roots to shoots.

Materials and methods

Plant materials and growth conditions

Arabidopsis plants were grown at 22°C under a 16/8 h light/dark cycle. For the treatment in Figure 1(a), plants were germinated and grown on iron-deficient media (-Fe) of Murashige and Skoog (MS) medium with no iron (Caisson MSP33) supplemented with 300 µM ferrozine or iron-sufficient media (+Fe) of MS medium with 100 µM FeNa-EDTA (Caisson MSP34) for up to 12 d. The concentration of 100 µM iron is comparable to that of multiple classical plant media formulations such as Gamborg’s B5 and MS basal salts.^54,55 For the iron deficiency treatment in Figure 2(a), plants were germinated and grown on Gamborg’s B5 medium without sucrose for up to 12 d, then transferred to -Fe or +Fe media and grown for 3 d. The Arabidopsis thaliana ecotype Columbia (Col-0) was used as the wild type, and the clf mutant used was clf-29 (SALK_21003) obtained from the Arabidopsis Biological Resource Center.

Figure 1.

Iron content of clf and wild type plants germinated in low iron or iron-sufficient conditions. (a) Schematic overview of the sample growth conditions (b) Iron content of root tissue from plants germinated and grown in media with 5 µM Fe (low iron) or 100 µM Fe (sufficient iron). (c) Iron content of shoot tissue from the same plants analyzed in (B). Mean values of five pooled samples are shown with error bars (SE). Significant differences compared to wild type based on t-tests are denoted (*: p < .05; **: p < .01).

Figure 2.

Iron content of clf and wild type plants treated in iron-deficient or sufficient media for 3 d. (a) Schematic overview of the sample growth conditions (b) Iron content of roots from 12-d-old plants grown on B5 without sucrose and then transferred to iron sufficient (+Fe; 100 µM Fe) or iron-deficient (-Fe; 300 µM ferrozine) media for 3 d. (c) Iron content of shoots from the same batch of plants analyzed in (B). (d) Relative level of iron translocation as measure by the ratio of iron content in shoots versus roots. Mean values of five pooled samples are shown with error bars (SE). Significant differences compared to wild type based on t-tests are denoted (*: p < .05; **: p < .01).

ICP-MS analysis

Iron content of wild type and clf roots and shoots was measured by ICP-MS. Plants were grown as indicated, and dried tissue was weighed in a microbalance and digested with nitric acid. Elemental content was analyzed with the Perkin-Elmer NexION 350D ICP-MS at the Mass Spectrometry Core, Institute for Applied Life Science at University of Massachusetts, Amherst. Metal content was normalized to the dry mass (mg) of each sample.

RNA extraction

Total RNA was isolated from roots and shoots using the Plant RNA Isolation Kit (Agilent) following the manufacturer’s instructions. The quantity and purity of RNA were assessed using NanoDrop One (Thermo Scientific), and the integrity of RNA was examined by electrophoresis in a bleach gel.⁵⁶

RNA-sequencing and data analysis

RNA was extracted from three biological replicates of wild type and clf shoots treated under iron deficient or sufficient conditions for 3 d as described above. cDNA library construction, 150 bp paired-end sequencing on Illumina NovaSeq6000 platform, and standard RNA-seq analysis were conducted by Novogene. Gene expression was quantified using the union mode of the HTSeq software and reported as FPKM (Fragments Per Kilobase of transcript sequence per Millions of base pairs sequenced). Sequencing reads were aligned to the TAIR10 Arabidopsis thaliana reference genome (www.arabidopsis.org), and Integrative Genomics Viewer (IGV), IGV 2.4^57,58 was used to generate read count histograms. Differential gene expression analysis was performed with DESeq (padj < 0.05) using negative binomial distribution for p-value estimation and BH for FDR estimation.⁵⁹ The enrichment analysis tool⁶⁰ available from the Gene Ontology Consortium^61,62 was used for Gene Ontology (GO) term enrichment analysis.

ChIP-qPCR

ChIP was conducted with shoot tissues of wild type and clf plants following the procedure described.⁴⁷ Briefly, shoot tissue was infiltrated with 37% formaldehyde in GB buffer (0.4 M sucrose, 10 mM Tris pH 8.0, 1 mM EDTA) and 100 mM PMSF to crosslink proteins with gDNA, ground in liquid nitrogen, and then sonicated in lysis buffer (500 mM HEPES pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.1% deoxycholate, 0.1% SDS) using Bioruptor Pico (Diagenode). Crosslinked DNA was precleared with Protein A Agarose/Salmon Sperm DNA beads (Millipore Sigma, 16–157) for 1 h at 4°C with gentle rotation and then incubated with anti-IgG (Millipore Sigma 12–370), anti-H3 (AbCam1791), or anti-H3K27me3 antibodies (Millipore Sigma07-449) overnight at 4°C with gentle rotation. After washing with a series of lysis buffer, LNDET buffer (0.25 M LiCl, 1% NP-40, 1% deoxycholate, 1 mM EDTA, 10 mM Tris-HCl pH 8.0), and TE buffer, the crosslinked DNA was eluted with elution buffer (1% SDS, 0.1 M NaHCO₃). Proteinase K was used to reverse crosslinking, and chromatin DNA was purified using the Zymo-Spin ChIP Kit per manufacturer’s instructions. qPCR was conducted using Power SYBR Green PCR Master Mix (Applied Biosystems) in QuantStudio 3 Real-Time PCR System (Applied Biosystems), and gene enrichment was normalized to 10% of input.⁶³ Primers were designed using QuantPrime.⁶⁴ The primer sequences used for qPCR are IMA1/FEP forward: 5ʹ-GGCCATCAAGAGATTTGACCATGC- 3ʹ; IMA1/FEP3 reverse: 5ʹ-TGCCACTCGAGAACTATCTACCAC-3ʹ; YSL1 forward: 5ʹ-CTGCGGTTCTC-ATCAATGGCTTCC-3ʹ; YSL1 reverse: 5ʹ-GAAACCACGCACTTGTTTCCTGCAAATC-3ʹ.

Results

Iron content is higher in roots and shoots of clf than in wild type germinated and grown in low iron

Our previous study showed that clf loss-of-function mutants expressed higher levels of FIT-regulated genes involved in iron acquisition, such as FIT, IRT1, FRO2, and F6ʹH1, and exhibited better growth with longer roots under prolonged iron deficiency, i.e., when plants were germinated and grown in low iron conditions compared to wild type plants.⁴⁷ Despite the partial redundancy between CLF and SWN, we did not use the clf swn double mutant for our experiments due to its severe phenotypes and development into a mass of callus-like tissue without distinct roots or shoots.⁵⁰ The clf mutant used in this study, clf-29, is a well-established H3K27me3 mutant with a remarkably reduced level of H3K27me3 and has been used in multiple studies, ^65–70 including our previous report.⁴⁷ To follow up with the low iron growth phenotype and to test if iron uptake was enhanced in clf plants, we used inductively coupled plasma mass spectrometry (ICP-MS) to quantify iron content in roots and shoots of wild type and clf seedlings germinated in media with low iron (5 µM Fe) or sufficient levels of iron (100 µM Fe), the same conditions for germination and growth tests with clf mutants in our prior study.⁴⁷ Iron content was significantly higher in both roots and shoots of clf plants germinated and grown in low iron media (Figure 1(b,c)), which is consistent with both the enhanced induction of iron uptake genes and better growth in clf under iron-deficient conditions as previously reported.⁴⁷ As predicted, the iron content of clf roots or shoots was not significantly different from that of wild type plants when germinated and grown in iron-sufficient media (Figure 1(b,c)).

Iron translocation from roots to shoots is enhanced in clf after iron-deficiency treatment

To examine iron content and translocation in clf plants upon transition into iron deficiency, we germinated wild type and clf plants in iron-sufficient medium and then treated in iron sufficient or deficient media for 3 d (Figure 2(a)). This 3-d iron treatment was long enough to induce the expression of iron-acquisition genes without causing visible growth phenotypes in clf plants.⁴⁷ We predicted that iron uptake and thus iron content in roots would be greater in clf plants than in wild type due to enhanced expression of FIT-dependent iron-acquisition genes in clf. Unexpectedly, under iron-deficient conditions, the iron content of clf roots was lower than that of wild type roots (Figure 2(b)); however, clf shoot tissue contained 19% more iron than wild type shoots (Figure 2(c)). The relative amount of iron translocated from roots to shoots, as determined by the ratio of shoot to root iron content, was approximately 50% higher in clf plants treated under iron-deficient conditions than in wild type (Figure 2(d)). This increased iron translocation in clf suggests that clf plants might still be transporting more iron into their roots, as we predicted, but not accumulating iron in their roots. Furthermore, these results imply a differential response to iron deficiency in clf shoot tissues.

Transcriptomic analysis of genes differentially regulated by iron in clf and wild type shoots

To examine the effect of H3K27me3 on transcriptional regulation in shoots under iron deficiency and to find clues to explain the increased translocation of iron to shoot tissue detected in clf plants (Figure 2(d)), we performed RNA-seq analysis with the wild type and clf shoot tissue from plants treated in iron deficient and sufficient media for 3 d, i.e., the same growth conditions represented in Figure 2(a), which were used in our previous transcriptomic analyses with root samples.⁴⁷

We conducted hierarchical clustering with 1628 genes differentially expressed in wild type or clf shoots (Figure 3). A large proportion of shoot genes exhibited a genotype-specific expression pattern (Figure 3), contrary to the expression profile in the roots where differentially regulated genes primarily responded to iron regardless of the genotype.⁴⁷ However, we identified 621 and 270 genes differentially regulated by iron in wild type shoots and in clf shoots, respectively, and noted four groups of iron-regulated genes based on the cluster analysis (Figure 3).

Figure 3.

Cluster analysis of genes differentially regulated by iron in wild type and clf shoots. Heat map of genes differentially regulated by iron in wild type or clf shoot tissue. Enriched GO terms and known iron homeostasis genes are indicated. Numbers on the scale bar represent standard deviation from the mean.

Group I

The first group of interest consisted of genes that accumulated less mRNA transcript under iron-sufficient conditions than in iron-deficient conditions in both genotypes, but had higher steady-state mRNA level in clf than in wild type regardless of the iron conditions (Figure 3). Genes involved with glucosinolate biosynthesis were enriched in this region. Glucosinolates are sulfur-rich compounds that are integral to plant defense mechanisms, ⁷¹ and iron serves as a cofactor of specifier proteins involved in glucosinolate breakdown.⁷² However, the implications of differential expression of glucosinolate biosynthetic genes remain to be understood. Ribosomal RNA and ribosome processing genes were also overrepresented in this group. Pan et al. (2015) found post-transcriptionally induced changes in ribosomal protein composition in iron deficiency and proposed that the ribosomes might be specialized in iron-deficient conditions. In their study, the abundance of a large number of ribosomal proteins was pronouncedly increased, but significant changes at the transcriptional level had not been detected.⁷³ Further studies will be necessary to determine if the detected transcriptional response in ribosomal genes under iron-deficient conditions in clf (Figure 3) is consistent with the post-transcriptional changes in ribosomes reported by Pan et al. (2015). Another enriched GO term from this group was oxidoreductase activity, and FRO1 and FRO7, which encode ferric-chelate reductases that belong to the family of oxidoreductases, were found in this category. FRO7 is involved in chloroplast iron acquisition,⁷⁴ but the role of FRO1 remains to be understood.

Group II

The second category was comprised of genes that had a higher steady-state mRNA level under iron deficiency in wild type shoots but were not as responsive to iron deficiency in the clf mutant (Figure 3). Group II was overrepresented with iron or transition metal homeostasis genes and genes involved in secondary metabolite biosynthesis. Multiple known iron homeostasis genes induced by iron deficiency, such as IMA1/2/3,bHLH38/39/100/101, BTS, BTSL1, PYE, NAS3/4, FRO3, OPT3, and YSL2 were found in this group (Figure 3). Many of these genes are highly expressed in the vasculature.⁷⁵ Under iron-sufficient conditions, the Group II genes accumulated more mRNA in clf or had similar mRNA levels in both genotypes (Figure 3). Since H3K27me3 serves as a repression mark, the reduced mRNA level of Group II genes in clf under iron-deficient conditions is more likely due to higher iron content in clf shoots compared to wild type (Figure 2) instead of a direct effect of PRC2-mediated regulation.

Group III

Genes with less mRNA accumulation under iron deficiency in wild type but not in clf, and genes with a higher level of mRNA under iron-sufficient conditions in clf than in wild type shoots were found in the third group (Figure 3). Given the higher mRNA accumulation level in clf under, both iron conditions, the genes identified in this group are more likely to be under H3K27me3-mediated control than genes associated with the other groups. The enriched GO terms in this category were amino acid catabolic processes and secondary metabolite biosynthesis. Iron homeostasis genes that play critical roles in iron translocation were also identified in this group. For example, NAS3, which encodes an enzyme in the biosynthetic pathway of NA, a non-proteinogenic amino acid, ⁷⁶ and YSL1 and YSL7, which encode transporters of iron-NA complexes, ⁷⁷ were found in this region.

Group IV

This category consisted of genes that had increased mRNA accumulation under iron sufficiency and reduced mRNA accumulation under iron limiting conditions in both wild type and clf, but whose differences in steady-state mRNA level were not as robust in clf compared to wild type (Figure 3). Genes involved in tetrapyrrole or heme biosynthesis and photosynthesis were highly enriched in this region (Figure 3). Considering the essential role of iron in these processes, the differential mRNA accumulation of genes in this group is consistent with the greater iron levels in clf shoots than in wild type under iron deficiency (Figure 2). Genes that play roles in oxidative stress response and iron homeostasis were also overrepresented in this category including FER1 and FER4, which encode ferritins that sequester iron to prevent iron-induced oxidative stress, ⁷⁸ VTL5, which is positively regulated by iron supply and encodes a vacuolar iron importer that can buffer cytosolic iron, ⁷⁹ and APX1 and FSD1, which encode enzymes that detoxify reactive oxygen species.^80,81 The enhanced expression of genes that buffer iron and mitigate oxidative stress in Group IV is consistent with the higher iron content in clf shoots under iron-deficient conditions (Figure 3).

H3K27me3 deposition levels on YSL1 are consistent with the iron-dependent expression of YSL1

YSL1 is expressed in the shoots, but its expression is decreased under iron deficiency.⁸² Consistently, in both wild type and clf mutant, YSL1 transcript level was higher in shoot tissue from plants treated in iron-sufficient media compared to those treated in iron-deficient media (Figure 4(a); Table 1). However, we detected markedly enhanced expression of YSL1 in clf compared to wild type, YSL1 expression in clf shoots was increased ~2.5-fold under iron-sufficient conditions and increased ~6-fold under iron-deficient conditions (Figure 4(a); Table 1). The higher expression of YSL1 in clf than in wild type is in accordance with the more robust iron translocation from roots to shoots in clf plants as determined by our ICP-MS analysis (Figure 2).

Figure 4.

YSL1 transcript level and H3K27me3 deposition in wild type and clf shoots. (a) RNA-seq read count histograms of YSL1 in wild type and clf shoots under iron deficient and sufficient conditions. Gene diagrams depict introns (line) and exons (boxes) and are aligned with read counts. The read count range is denoted in the upper left corner of each diagram. Read count scale is equivalent for wild type and clf within each gene. (b) H3K27me3 deposition at YSL1 locus. ChIP-qPCR signal was normalized with input DNA, and H3K27me3 enrichment relative to that of wild type grown under iron-sufficient conditions was plotted. Error bars represent standard error of the mean of four biological replicates and significant differences compared to wild type based on t-tests are denoted (n = 4; *: p < .05).

Table 1.

Relative expression (log₂ fold change) of select iron-regulated genes expressed in wild type and clf shoots.

Gene ID	Gene Name	Annotated Function	WT -Fe vs. WT +Fe	clf -Fe vs. clf +Fe
At1g01590	Ferric Reductase Oxidase 1 (FRO1)	reduces ferric-chelate iron to ferrous iron	4.91	1.68
At1g09240	Nicotianamine Synthase 3 (NAS3)	synthesizes nicotianamine	−3.21	−1.24
At1g21140	Vacuolar Iron Transporter-Like 1 (VTL1)	catalyzes Fe transport into the vacuole	−4.11	−1.60
At1g23020	Ferric Reductase Oxidase 3 (FRO3)	reduces ferric-chelate iron to ferrous iron	4.54	3.86
At1g47400	Fe-Uptake Inducing Peptide 3/Ironman 1 (FEP3/IMA1)	regulator of the iron deficiency response	10.60	8.65
At1g58290	HEMA1	protoporphyn biosynthesis	−3.73	−2.27
At2g30766	Fe-Uptake Inducing Peptide 1/Ironman 3 (FEP1/IMA3)	regulator of the iron deficiency response	6.31	5.29
At2g38460	Ferroportin 1/Iron Regulated 1 (FPN1/IREG1)	transmembrane iron transporter	−0.36	−0.15
At2g40300	Ferritin 4 (FER4)	sequesters iron in mitochondria and plastids	−3.17	−1.91
At2g41240	bHLH100	heterodimerizes with FIT to activate iron deficiency response	13.05	9.68
At3g18290	Brutus (BTS)	negative regulator of the iron deficiency response	3.79	3.19
At3g25190	Vacuolar Iron Transporter-Like 5 (VTL5)	catalyzes Fe transport into the vacuole	−4.53	−4.01
At3g56090	Ferritin 3 (FER3)	sequesters iron	−1.87	0.21
At3g56970	bHLH038	heterodimerizes with FIT to activate iron deficiency response	11.48	10.14
At3g56980	bHLH039	heterodimerizes with FIT to activate iron deficiency response	10.01	8.92
At4g16370	Oligopeptide transporter 3 (OPT3)	phloem iron transporter	3.63	2.68
At4g24120	Yellow Stripe Like 1 (YSL1)	iron-nicotianamine transmembrane transporter	−2.41	−1.11
At4g25100	Fe-Superoxide Dismutase 1 (FSD1)	dismutates superoxide during oxidative stress	−2.47	−1.09
At5g01600	Ferritin 1 (FER1)	sequesters iron in plastids	−3.91	−2.04
At5g04150	bHLH101	heterodimerizes with FIT to activate iron deficiency response	7.46	7.15
At5g04950	Nicotianamine Synthase 1 (NAS1)	synthesizes nicotianamine	1.54	0.59
At5g24380	Yellow Stripe-Like 2 (YSL2)	iron-nicotianamine transmembrane transporter	1.57	0.96
At5g51720	NEET	2Fe-2 S cluster binding	−5.30	−3.50

Based on the significant increase in YSL1 expression in clf shoots regardless of the iron conditions (Figure 4(a); Table 1), we predicted that H3K27me3 deposition might be associated with the repression of YSL1 in iron deficiency. To our knowledge, the YSL1 locus was identified as an H3K27me3 target in at least one epigenetic study⁸³ but was not among the H3K27me3 direct targets identified in multiple other studies that detected other iron homeostasis genes.^84-88 Furthermore, whether or not H3K27me3 deposition on YSL1 depends on iron condition had not been studied. Our ChIP-qPCR showed that H3K27me3 accumulated at the YSL1 locus in wild type shoots under iron deficiency (Figure 4(b)), verifying that YSL1 is a direct target of H3K27me3, and revealed a strong correlation between H3K27me3 levels and YSL1 repression in wild type and clf shoots (Figure 4). Furthermore, our data revealed about 2-fold enrichment of H3K27me3 in wild type shoots from plants treated in iron-deficient media than in iron-sufficient media, which was consistent with the significant repression of YSL1 under iron deficiency (Figure 4; Table 1). While H3K27me3 deposition was low in clf shoots under both iron conditions, there was a clear correlation between H3K27me3 levels and the extent of YSL1 repression. The H3K27me3 levels in iron-sufficient clf shoots were only about half (~48%) of that in iron-deficient clf shoots, inversely correlating with the relative level of YSL1 under the two conditions (Figure 4; Table 1). These results suggest the involvement of H3K27me3 in iron-dependent regulation of YSL1 expression.

IMA1 is a direct target of PRC2-mediated H3K27me3

IMA1 transcripts accumulated under iron deficiency in both wild type and clf shoots, consistent with previous studies,^45,46 but the induction of IMA1 in clf shoots was less than that in wild type shoots; the fold change induced by iron deficiency in clf and wild type was 402 and 1548, respectively (Figure 5(a); Table 1). We noted that the IMA1/FEP3 locus was identified as an H3K27me3 target in multiple large-scale epigenomic datasets (Table 2).^8,85-88 To examine if the iron-regulated expression of IMA1/FEP3 and its differential expression in iron-deficient wild type and clf shoots could be explained by H3K27me3 enrichment, we performed ChIP-qPCR to quantify H3K27me3 deposition in wild type and clf shoots treated under iron deficient and sufficient conditions. The amount of H3K27me3 deposition inversely correlated with IMA1 transcript levels under different iron conditions in wild type (Figure 5), which suggests that the IMA1 locus is a direct target of H3K27me3 and indicates that H3K27me3 contributes to iron-dependent regulation of IMA1 in wild type. However, IMA1 expression was significantly lower in iron-deficient clf shoots than in wild type (Figure 5(a)), despite the drastically reduced level of H3K27me3 deposition in clf regardless of the iron conditions (Figure 5(b)). Based on the role of IMA1 as a regulator of the iron-deficiency response^45,46 and the enhanced iron levels in clf shoots (Figure 2(c)), we speculate that the reduced IMA1 transcript levels in iron-deficient clf shoots are most likely the effect of enhanced iron uptake and increased level of iron in iron-deficient clf plants. Although IMA1 mRNA levels were extremely low in iron-sufficient conditions, we detected a notable difference in IMA1 transcript levels in iron-sufficient wild type and clf shoots (Figure 5(a)), which contain similar levels of iron (Figure 2c). Based on these results, we postulate that H3K27me3 deposition at the IMA1 locus may play a limited role compared to the regulation by iron levels.

Figure 5.

IMA1 transcript level and H3K27me3 deposition in wild type and clf shoots. (a) RNA-seq read count histograms of IMA1 in wild type and clf shoots under iron deficient and sufficient conditions. Gene diagrams depict introns (line) and exons (boxes) and are aligned with read counts. The read count range is denoted in the upper left corner of each diagram. Read count scale is equivalent for wild type and clf within each gene. (b) H3K27me3 deposition at IMA1 locus. ChIP-qPCR signal was normalized with input DNA, and H3K27me3 enrichment relative to that of wild type grown under iron-sufficient conditions was plotted. Error bars represent standard error of the mean of four biological replicates and significant differences compared to wild type based on t-tests are denoted (n = 4; *: p < .05).

Discussion

Accumulating evidence suggests that regulation in response to iron availability occurs over multiple stages of gene expression, in addition to that at the transcriptional level. However, the effect of chromatin remodeling on iron homeostasis is not well understood. Compared to roots, significantly fewer studies have investigated the response to iron deficiency in shoot tissues. In this study, we extend our previous research, which focused on H3K27me3-mediated regulation of iron-acquisition genes in the roots,⁴⁷ by providing evidence that H3K27me3 mediates iron homeostasis in shoot tissues. We show that YSL1 and IMA1, which play critical roles in signaling iron status from the shoots,^42,45,46 are direct targets of H3K27me3 (Figures 4 and 5). Furthermore, we report the transcriptional changes and increased iron translocation in clf shoots that were observed under iron-deficiency treatment. To our knowledge, the effect of H3K27me3 on iron homeostasis in shoots has not been previously reported.

Under prolonged iron deficiency, the iron content of clf roots and shoots is higher compared to wild type (Figure 1), as expected from the transcriptional response in roots and growth phenotype of clf in low iron.⁴⁷ However, after a 3-d treatment under iron deficiency, clf plants enhance iron translocation to the shoots rather than accumulating iron in the roots (Figure 2(d)). Considering that YSL1 functions as a transporter to load iron-NA to the leaves,⁸² the higher induction of YSL1 under iron-deficiency in clf shoots as compared to wild type (Figure 4(a)) agrees with the evidence of increased in iron translocation to the shoots of clf (Figure 2(d)). Furthermore, the inverse correlation between H3K27me3 deposition on the YSL1 locus and iron-dependent YSL1 expression levels in wild type and clf indicates the direct involvement of H3K27me3 in iron-dependent regulation of YSL1 (Figure 4). With respect to the role of YSL1 in signaling iron status from shoots to roots,⁴² the higher induction of YSL1 in iron-deficient clf shoots might imply the generation of a stronger mobile signal from shoots that triggers iron uptake in response to iron deficiency. We speculate that stronger YSL1-mediated iron signaling in iron-deficient clf than in iron-deficient wild type may have contributed to the more robust induction of FIT-dependent iron-acquisition genes in clf, in addition to the effect of the reduced H3K27me3 deposition in the mutant, as detected in our previous study.⁴⁷

Another major regulator of iron translocation is OPT3. Iron loading to the phloem by OPT3 is necessary for iron signaling from shoots to roots to regulate iron-deficiency response, and opt3 mutants constitutively express IRT1 and FRO2 due to decreased iron in the phloem.^31,89 We observed differential expression of OPT3 in clf and wild type after iron-deficiency treatment; the fold change induction of OPT3 was lower in clf than in wild type for both roots⁴⁷ and shoots Table 1). We postulate that increased iron influx in clf under iron deficiency might have influenced the reduced induction of OPT3. Whether the observed differences in OPT3 or YSL1 transcript levels leads to differences in xylem or phloem iron levels of clf and wild type plants remains to be understood.

Based on our transcriptomic analysis, the profile of iron-regulated genes that were differentially expressed in clf shoots revealed two modes of response. First, iron-deficient clf shoots are undergoing transcriptional changes to buffer iron-overload and mitigate iron-induced oxidative stress. This idea is represented by the decreased repression, i.e. higher transcript levels, of VTLs, ROS scavenger genes, and ferritin genes (Figure 3), which are induced by iron supply or oxidative stress.^79,90Whether the higher expression of these genes is a direct effect of chromatin remodeling due to decreased H3K27me3 in iron-deficient clf shoots remains to be determined. VTL1/2/5 loci have been identified as H3K27me3 targets from epigenomic profiling studies, but FER1/4 were not detected among the target loci (Table 2).^8,85-88,91 FER1 is transcriptionally induced by the presence of iron,⁹² but its mRNA is also subjected to iron or oxidative stress-induced degradation.⁹³ Thus, post-transcriptional regulation might have contributed to the difference in FER1 transcription levels between iron-deficient clf and wild type shoots (Figure 3). We also note that in a recent study conducted with whole seedlings, it was shown that FER1/3/4 were not subjected to H3K27me3 mark deposition under control conditions.⁹⁴ In contrast, histone marks that positively regulate gene expression, H3K4me3 and H3K9ac were present on the FER1/3/4 loci.

Table 2.

Shoot iron homeostasis genes identified as H3K27me3 targets from existing epigenomic profiling datasets.

Gene ID	Gene Name	Annotated Function
At1g21140	Vacuolar Iron Transporter-Like 1 (VTL1)	iron transport into the vacuole
At1g47395	Fe-Uptake Inducing Peptide 2/Ironman 2 (FEP2/IMA2)	regulator of the iron deficiency response
At1g47400	Fe-Uptake Inducing Peptide 3/Ironman 1 (FEP3/IMA1)	regulator of the iron deficiency response
At1g65730	Yellow Stripe Like 7 (YSL7)	iron-nicotianamine transmembrane transporter
At1g76800	Vacuolar Iron Transporter-Like 2 (VTL2)	Iron transport into the vacuole
At2g30766	Fe-Uptake Inducing Peptide 1/Ironman 3 (FEP1/IMA3)	regulator of the iron deficiency response
At3g25190	Vacuolar Iron Transporter-Like 5 (VTL5)	iron transport into the vacuole
At4g24120	Yellow Stripe Like 1 (YSL1)	iron-nicotianamine transmembrane transporter
At4g25100	Fe-Superoxide Dismutase 1 (FSD1)	dismutates superoxide during oxidative stress
At5g04950	Nicotianamine Synthase 1 (NAS1)	synthesizes nicotianamine
At5g51720	NEET	2Fe-2 S cluster binding

Second, the transcriptional profile indicated that clf shoots were sensing less iron deficiency than wild type after iron-deficiency treatment (Figure 3). In clf shoots, multiple known iron-deficiency response genes, including bHLH38/39/100/101, BTS, PYE, and PYE-target genes, are not as strongly induced as in wild type (Figure 3). However, we note that the induction level of these genes was similar between clf and wild type in roots.⁴⁷ This discrepancy between the relative fold changes in the roots and shoots of wild type and clf indicates that the decreased induction of iron-deficiency response genes in the shoots of iron-deficient clf is most likely a response to a higher level of iron in their shoots, rather than a direct effect of H3K27me3. Decreased content of chlorophyll and other light-harvesting pigments is one of the main consequences of iron deficiency on photosynthesis.^95-97 Thus, the reduced repression of genes involved in photosynthesis and tetrapyrrole biosynthesis is likely due to the less severe iron deficiency perceived in clf leaves, as a result of more iron translocation to the shoots compared to wild type. In addition, the lower level of IMA1 transcripts in clf compared to wild type implies that clf shoots sense less iron deficiency than wild type (Figure 5). We detected iron-dependent H3K27me3 deposition on IMA1 locus in wild type shoots (Figure 5), which strongly suggests that PRC2-mediated H3K27me3 controls the expression of IMA1 under iron deficiency. However, the undetectable level of IMA1 transcripts in iron-sufficient clf shoots indicates that the extremely low level of H3K27me3 is not sufficient to induce IMA1 expression (Figure 5). We postulate that iron-responsive transcriptional regulator(s) are necessary to fully induce IMA1 under iron-deficient conditions, and it is most likely that the higher level of iron in iron-deficient clf shoots limits IMA1 transcript levels. We further speculate that the bHLH121/URI and ILR3 complex may be responsible for the difference in IMA1 levels in iron-deficient wild type and clf, because bHLH121/URI and ILR3 are ubiquitously expressed^22,23,98 and bHLH121/ILR3 directly activates IMA1 under iron deficiency.²³ Meanwhile, neither bHLH121 nor ILR3 have been identified as H3K27me3 targets from epigenomic profiling datasets (Table 2),^8,85-88,91 which indicates that these transcription factors are largely H3K27me3 independent.

Overall, the results from this study and our prior report show that PRC2-mediated H3K27me3 plays a critical role in iron homeostasis (Figure 6). The significantly reduced level of H3K27me3 deposition in clf plants results in hyper-induction of iron-acquisition genes in the roots and more translocation of iron to the shoots. YSL1 and IMA1, which are both implicated in signaling iron status from the shoots, are direct targets of H3K27me3. However, while iron-dependent regulation of YSL1 can be explained by H3K27me3 levels, H3K27me3 appears to play a partial or indirect role in IMA1 induction under iron deficiency. Whether increased iron translocation in clf compared to wild type is an effect of H3K27me3 activity or a response to higher iron influx in clf roots remains to be elucidated. Further studies are also needed to better understand the molecular mechanisms of iron regulation mediated by chromatin remodeling in plants.

Figure 6.

Proposed model for H3K27me3-mediated regulation under iron deficiency.

Upper left: Wild type under iron-sufficient conditions (WT +Fe). H3K27me3 accumulates on FIT-dependent iron-acquisition gene loci and prevents their expression in the roots. In the shoots, moderate H3K27me3 deposition on YSL1 locus in wild type allows for moderate YSL1 expression, and the high H3K27me3 deposition on IMA1 locus mediates repression. Upper right: Wild type under iron-deficient conditions (WT – Fe). Partial removal of H3K27me3 from FIT-dependent acquisition loci allows the induction of these genes. In the shoots, the high H3K27me3 deposition on YSL1 locus prevents its expression. The decreased H3K27me3 deposition on the IMA1 locus facilitates an increase in expression, but iron-responsive transcriptional regulator(s) are necessary to fully induce IMA1. Compared to iron-deficient clf plants, less high-affinity iron uptake from the soil (orange arrow) and iron translocation from roots to shoots (green arrow) occur in iron-deficient wild type. Lower left: Iron-sufficient clf mutant (clf +Fe). The lack of H3K27me3 in clf is not sufficient to induce FIT-dependent genes due to the presence of iron that negatively regulates their expression in the roots. In the shoots, the low H3K27me3 deposition on YSL1 locus in clf mutants allows for higher expression of YSL1 than in wild type from the same iron conditions. Lower right: Iron-deficient clf mutant (clf -Fe). Induction of iron-acquisition genes is more robust in clf roots, which exhibits significantly less H3K27me3 deposition on these loci. In the shoots, the low H3K27me3 deposition on YSL1 locus in clf mutants allows for higher expression of YSL1 than in wild type from the same iron conditions. Decreased H3K27me3 deposition on IMA1 leads to an increase in expression that is dampened by higher shoot iron content compared to wild type. Compared to iron-deficient wild type, more high-affinity iron uptake from the soil (orange arrow) and iron translocation from roots to shoots (green arrow) occur in iron-deficient clf. Blue circles with black lines represent nucleosomes, with core histones and chromatin DNA, red circles represent H3K27me3 deposition, and the distance between blue circles depicts euchromatin or heterochromatin structure due to H3K27me3. Green circles with the question mark represent iron-responsive transcriptional regulator(s) necessary to fully induce IMA1, possibly bHLH121/URI and ILR3. Black arrows and their thickness represent the strength of transcriptional activity. High-affinity iron uptake under iron-deficient conditions is shown as orange arrows, and iron translocation from roots to shoots under iron-deficient conditions is indicated in green arrows. Thicker arrows represent enhanced high-affinity iron uptake from the soil or increased translocation of iron.

Funding Statement

This work was supported by the Gregory S. Call Undergraduate Research Program to EYP and EMP, Doelling Undergraduate Research Fund to EMP, and the H. Axel Schupf ’57 Fund for Intellectual Life and the National Science Foundation grant [IOS-1754969] to JJ.

Author contributions

JJ conceived the idea and supervised the project. EYP, KT, and EMP conducted ChIP-qPCR and prepared samples for ICP-MS. EYP prepared samples for RNA-seq and analyzed data. JJ primarily wrote the manuscript with contributions of EYP, KT, EMP. All authors have reviewed and approved the manuscript.

Disclosure statement

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.

Data availability statement

The RNA-seq data discussed in this publication have been deposited in NCBI’s Gene Expression 783 Omnibus,⁹⁹ shoot, homeostasis and are accessible through GEO Series accession number GSE149816. https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE149816.