ORIGINAL RESEARCH published: 04 June 2019 doi: 10.3389/fpls.2019.00669 The Four FAD-Dependent Histone Demethylases of Arabidopsis Are Differently Involved in the Control of Flowering Time Damiano Martignago 1,2 , Benedetta Bernardini 1 , Fabio Polticelli 1,3 , Daniele Salvi 4 , Alessandra Cona 1 , Riccardo Angelini 1 and Paraskevi Tavladoraki 1* 1 Department of Science, Roma Tre University, Rome, Italy, 2 Centre for Research in Agricultural Genomics, Spanish National Research Council–Institute for Food and Agricultural Research and Technology–Autonomous University of Barcelona–University of Barcelona, Barcelona, Spain, 3 ‘Roma Tre’ Section, National Institute of Nuclear Physics, Rome, Italy, 4 Department of Life, Health and Environmental Sciences, University of L’Aquila, L’Aquila, Italy In Arabidopsis thaliana, four FAD-dependent lysine-specific histone demethylases (LDL1, LDL2, LDL3, and FLD) are present, bearing both a SWIRM and an amine oxidase domain. In this study, a comparative analysis of gene structure, evolutionary Edited by: Ana Margarida Fortes, relationships, tissue- and organ-specific expression patterns, physiological roles and University of Lisbon, Portugal target genes for the four Arabidopsis LDL/FLDs is reported. Phylogenetic analysis Reviewed by: evidences a different evolutionary history for the four LDL/FLDs, while promoter Sureshkumar Balasubramanian, activity data show that LDL/FLDs are strongly expressed during plant development Monash University, Australia Xuncheng Liu, and embryogenesis, with some gene-specific expression patterns. Furthermore, South China Botanical Garden (CAS), phenotypical analysis of loss-of-function mutants indicates a role of all four Arabidopsis China LDL/FLD genes in the control of flowering time, though for some of them with *Correspondence: Paraskevi Tavladoraki opposing effects. This study contributes toward a better understanding of the LDL/FLD paraskevi.tavladoraki@uniroma3.it physiological roles and may provide biotechnological strategies for crop improvement. Specialty section: Keywords: flowering time, histone demethylases, FLD, FLC, FWA, LDL, LSD1 This article was submitted to Plant Metabolism and Chemodiversity, INTRODUCTION a section of the journal Frontiers in Plant Science Histone methylation is involved in a wide range of biological processes (Pfluger and Wagner, 2007). Received: 28 January 2019 It decorates both transcriptionally silenced and active chromatin domains, depending on which Accepted: 02 May 2019 residues are methylated and the degree of methylation. One of the most relevant and studied histone Published: 04 June 2019 marks in plants is the methylation of lysine 3 on histone 4 (H3K4me). H3K4 can be mono-, di-, Citation: and tri-methylated (respectively me1, me2, me3) by different classes of SET domain-containing Martignago D, Bernardini B, methyltransferases and this process is reversed by histone demethylases in a dynamic fashion. Two Polticelli F, Salvi D, Cona A, Angelini R types of lysine-specific histone demethylases are present in both animals and plants, the Jumonji and Tavladoraki P (2019) The Four C (JmjC) domain-containing histone demethylases and the FAD-dependent histone demethylases FAD-Dependent Histone (JHDMs and LSDs, respectively; Shi et al., 2004; Tsukada et al., 2005; Xu et al., 2015; Gu et al., Demethylases of Arabidopsis Are Differently Involved in the Control 2016). In animals, two LSDs are found, LSD1 and LSD2 (Shi et al., 2004; Karytinos et al., 2009), of Flowering Time. which contain a FAD-dependent amine oxidase (AO; Polticelli et al., 2005) domain and a SWIRM Front. Plant Sci. 10:669. domain (Stavropoulos et al., 2006). LSD1 has also an ∼100 amino acid protruding domain, known doi: 10.3389/fpls.2019.00669 as ‘Tower’ domain, which interacts with the corepressor CoREST, among other proteins, and is Frontiers in Plant Science | www.frontiersin.org 1 June 2019 | Volume 10 | Article 669 Martignago et al. Histone Demethylases and Flowering Time required for LSD1 catalytic activity on nucleosomes (Shi et al., FLD, the latter playing a more prominent role (Jiang et al., 2007). 2005; Chen et al., 2006; Stavropoulos et al., 2006; Yang et al., Consistently, ldl1ldl2 mutants display increased H3K4me3 2006; Burg et al., 2015). Unlike LSD1, LSD2 does not have the levels at FLC as compared to wild-type plants, but to a lesser ‘Tower’ domain and does not interact with CoREST, but possesses degree than ldl1fld mutants. LDL1 and LDL2, but not FLD, both a CW-type zinc finger motif and a C4H2C2-type zinc finger are additionally involved in the control of H3K4 methylation motif joined by a linker domain composed of two α-helices. state at FWA, a homeodomain-containing transcription factor This suggests that LSD2 may interact with different targets or which interferes with floral transition (Jiang et al., 2007). co-regulatory molecules and may be involved in transcriptional Altogether, these data suggest that the Arabidopsis LDL/FLD programs distinct from those of LSD1 (Burg et al., 2015). gene family plays a critical role in the histone methylation pattern Arabidopsis thaliana has four homologs of the human LSD1 of flowering genes. A similar function was also suggested for (HsLSD1) gene: At1g62830 (LSD1-LIKE1; LDL1), At3g13682 LDL/FLD homologs in other plant species (Hu et al., 2014; Gu (LDL2), At4g16310 (LDL3), and At3g10390 (FLD), all bearing et al., 2016; Shibaya et al., 2016). both a flavin AO domain and a SWIRM domain (Shi Recent studies have evidenced the involvement of the et al., 2004; Jiang et al., 2007; Spedaletti et al., 2008). Like LDL/FLD gene family also in several developmental and stress HsLSD1, Arabidopsis LDL1 is able to specifically demethylate defense processes (Yu et al., 2016). In fact, LDL1 is involved H3K4me2 and H3K4me1 peptides and to discriminate between in root elongation and lateral root initiation (Krichevsky et al., different epigenetic marks (Forneris et al., 2005; Spedaletti 2009; Singh et al., 2012). In addition, LDL1 and LDL2 repress the et al., 2008). Furthermore, LDL1 interacts with a SET-domain expression of seed dormancy-related genes and act redundantly histone methyltransferase and a histone deubiquitinase to form in repressing seed dormancy (Zhao et al., 2015). Furthermore, co-repressor complexes (Krichevsky et al., 2007, 2011). However, FLD is required for activation of systemic acquired resistance, plant LDL/FLDs are probably directed to their substrates by through a FLC-independent pathway, and for up-regulation mechanisms different from those of their animal counterparts of important modulators of plant immune responses (Singh (Sadiq et al., 2016). Indeed, plants do not encode CoREST et al., 2013, 2014; Banday and Nandi, 2018). In wheat, a LDL1- homologs. In addition, LDL/FLDs do not interact with plant homolog is up-regulated in heat-primed plants suggesting a role homologs of SFMBT1, which functions as part of the LSD1- of this gene family in the epigenetic mechanisms regulating stress based repressor complex and is known to bind different forms memory (Wang et al., 2016). of methylated histones (Tang et al., 2013; Zhang J. et al., 2013; The increasing evidence for the involvement of the LDL/FLD Sadiq et al., 2016). gene family in different physiological processes raises the need Most of the physiological studies on the Arabidopsis for a comparative analysis of this gene family. To this end, in LDL/FLDs focus on their role in the control of flowering the present study the gene and protein structure, as well as the time. The developmental transition from the vegetative to the evolutionary history of all four LDL/FLDs have been dissected. reproductive stage is a critical event in the plant life cycle. In Furthermore, the tissue- and organ-specific expression patterns A. thaliana, a complex regulatory network controls the timing of the four LDL/FLDs were analyzed. Phenotypical analyses of floral transition, a key component of which is FLOWERING of loss-of-function mutants for all four LDL/FLD genes were LOCUS C (FLC), a MADS-box transcriptional regulator that also performed, with particular attention to the flowering time, inhibits floral transition largely by reducing the expression revealing functional differences among them. of flowering-time integrators, such as SUPPRESSOR OF OVEREXPRESSION OF CONSTANS 1 (SOC1) and FLOWERING LOCUS T (FT) (He et al., 2003; He, 2009). This regulatory MATERIALS AND METHODS network integrates the endogenous developmental state of the plant (autonomous pathway and gibberellin-dependent pathway) Protein Sequence Homology with environmental cues (Amasino and Michaels, 2010). FLD is Search and Retrieval involved in the autonomous pathway by constitutively repressing The amino acid sequence of LSD1-like proteins from various FLC. Indeed, Arabidopsis fld loss-of-function mutants are plant and animal organisms were retrieved by sequence similarity known to be late-flowering or non-flowering due to increased searches in BLASTP (Altschul et al., 1997) using the amino acid FLC expression levels (Sanda and Amasino, 1996; Chou and sequence of HsLSD1 and HsLSD2, as well as of the A. thaliana Yang, 1998; He et al., 2003; Jiang et al., 2007). FLD is also LDL1, LDL2, FLD, and LDL3 as query sequences. The amino acid required in chromatin silencing of FLC mediated by the RNA- sequence of additional LSD1-like proteins was retrieved from the binding protein FCA (Liu et al., 2007). Furthermore, the physical National Center for Biotechnology Information (NCBI) database interaction between FLD and the histone deacetylases HDA5 based on sequence annotation. Abbreviations and accession and HDA6 plays an important role in the control of both H3 numbers are listed in Supplementary Table 1. To determine acetylation and H3K4 trimethylation at FLC and its homologs SWIRM and AO domains, multiple amino acid sequence MADS AFFECTING FLOWERING 1 (MAF1), MAF4 and MAF5 alignments were performed using Clustal Omega (Sievers (Yu et al., 2011; Luo et al., 2015). Indeed, fld mutants display et al., 2011). For genomic exon–intron structure comparisons, altered H3 and H4 acetylation levels at FLC (He et al., 2003; manual alignment between genomic and cDNA sequences was Zhang Y. et al., 2013; Hu et al., 2014). FLC is down-regulated performed. Information on intron number was additionally also by LDL1 and LDL2, which act in partial redundancy with obtained from the NCBI database. Frontiers in Plant Science | www.frontiersin.org 2 June 2019 | Volume 10 | Article 669 Martignago et al. Histone Demethylases and Flowering Time Molecular Modeling the SAIL library (SAIL_640_B10.v1; Sessions et al., 2002). The Molecular models of A. thaliana LDL3, and LDL3 homologs from presence of T-DNA insertion was confirmed by PCR, and Physcomitrella patens (PpLDL3) and Selaginella moellendorffii homozygous mutant plants were selected. RT-PCR analysis using (SmLDL3) have been built using the ab initio/threading protocol primers upstream and downstream from the T-DNA insertion implemented in the I-TASSER pipeline (Yang et al., 2015). confirmed the absence of correct mRNA for the corresponding No query/template alignment has been provided in input as genes, whereas qRT-PCR analysis confirmed reduced gene- I-TASSER uses LOMETS (Local Meta-Threading Server) to specific expression levels (Supplementary Figure 2). Primer thread the query sequence through a representative library of sequences are listed in Supplementary Table 2. PDB structures and select the folds compatible with the sequence of the query protein (Wu and Zhang, 2007). Best models have Construction and Characterization of been selected on the basis of the I-TASSER quality score (C-score) Arabidopsis Transgenic Plants whose values range from −5 to 2, higher values indicating higher To construct LDL/FLD::GFP-GUS transgenic Arabidopsis plants, quality models (Yang et al., 2015). C-score values of the selected 2- to 3-kb promoter regions including the 50 UTR were models for Arabidopsis LDL3, SmLDL3 and PpLDL3 are −0.27, amplified from Arabidopsis genomic DNA by PCR and 0.86, and 0.6, respectively. cloned into the pDONR207 vector (Invitrogen) via Gateway Technology (Invitrogen). Sequences of oligonucleotides used Phylogenetic Analyses for the amplification of promoter regions are shown in Amino acid sequences were aligned with MAFFT v.7 (Katoh Supplementary Table 2. Following sequencing, promoter regions and Standley, 2013) using the E-INS-i iterative refinement were inserted into the Gateway binary vector pKGWFS7 algorithm. Two alignments were built, one with the entire vector (Karimi et al., 2002) in-frame with the downstream protein sequence, and another one including only amino acids green fluorescent protein (GFP) and β-glucuronidase (GUS) of the AO domain. For each alignment, the optimal model of reporter genes. The resulting constructs were used to transform protein evolution was selected by ModelTest-NG v0.1.51 under A. thaliana wild-type plants by the Agrobacterium tumefaciens- the corrected Akaike Information Criterion. The JTT model mediated floral dip transformation method (Bent, 2006). (Jones et al., 1992) with gamma distributed rates across site Independently transformed plant lines were tested by PCR. (+G) was selected for both alignments. Phylogenetic analyses were performed with the Maximum Likelihood method using Histochemical GUS Assay RAXML v.8.2.10 (Stamatakis, 2014) with the PROTGAMMAJTT GUS staining of Arabidopsis LDL/FLD::GFP-GUS transgenic substitution model. Node support was evaluated with 1,000 rapid plants was performed essentially as previously described (Fincato bootstrap inferences. The sequence of the polyamine oxidase 1 of et al., 2012). Briefly, samples were gently soaked in 90% (v/v) A. thaliana (AtPAO1; At5g13700; Supplementary Table 1) was cold acetone for 1 h at −20◦ C, rinsed with 50 mM sodium used as outgroup. Phylogenetic analyses were computed in the phosphate buffer pH 7.0, vacuum infiltrated in staining solution CIPRES Science Gateway V. 3.32 (Miller et al., 2010). (1 mM 5-bromo-4-chloro-3-indolyl-β-D-glucuronide, 2.5 mM potassium ferrocyanide, 2.5 mM potassium ferricyanide, 0.2% Plant Material Triton X-100, 10 mM EDTA, 50 mM sodium phosphate buffer, All experiments were performed with Arabidopsis ecotype pH 7.0) and incubated at 37◦ C for 18 h. Chlorophyll was extracted Columbia-0 plants grown under long-day (16 h day/8 h with ethanol:acetic acid (3:1). Samples were kept in 70% ethanol. night) photoperiod conditions. To determine the flowering time To improve destaining, reproductive organs were washed with (expressed as the number of rosette leaves at bolting), seeds Hoyer’s light medium (Stangeland and Salehian, 2002). were sown in a 3:1 soil:perlite mixture and plants were grown to mature stage. For RT-PCR and qRT-PCR analyses, seedlings were Quantitative RT-PCR Analysis grown for 7 days on plates containing half-strength Murashige Total RNA was isolated from whole Arabidopsis seedlings and Skoog basal medium supplemented with Gamborg’s vitamins using the RNeasy Plant Mini kit (QIAGEN) and treated with and 0.5% (w/v) sucrose (1/2MS) and solidified with 0.7% agar. RNase-free DNase during RNA purification (RNase-Free DNase Then, seedlings were transferred in 6-well plates containing 1/2MS Set; QIAGEN) according to the manufacturers’ protocol. RNA liquid medium and were left to grow for 7 more days. concentration was measured with a NanoDrop ND-1000 UV- Vis spectrophotometer (NanoDrop Technologies). Synthesis of Characterization of Loss-of-Function cDNA and PCR amplification were carried out using GoTaq R LDL/FLD Mutants 2-Step RT-qPCR System (Promega). The qPCR reactions were Arabidopsis ldl1, ldl2, and fld loss-of-function mutants were performed in a Corbett RG6000 (Corbett Life Science, QIAGEN) obtained from the SALK collection (SALK_142477.31.30.x, following the program: 95◦ C for 2 min then 40 cycles of SALK_146346.52.50.x, and SALK_015053.35.80.x, respectively; 95◦ C for 3 s and 60◦ C for 30 s. Primers were designed using Alonso et al., 2003), while ldl3 mutant was obtained from Primer3 software (Untergasser et al., 2007) and tested for specificity using Primer-BLAST. UBIQUITIN-CONJUGATING 1 https://github.com/ddarriba/modeltest ENZYME 21 (UBC21, At5g25760) was chosen as a reference 2 http://www.phylo.org/ gene (Czechowski et al., 2005). Primer sequences are listed in Frontiers in Plant Science | www.frontiersin.org 3 June 2019 | Volume 10 | Article 669 Martignago et al. Histone Demethylases and Flowering Time Supplementary Table 2. Relative expression levels are expressed domain, demethylase activity has been shown for Arabidopsis as fold-changes (2−11Ct ). Reactions were performed in triplicate LDL1 (Spedaletti et al., 2008), as shown for the mouse LSD2 and mean values ± SE were calculated. At least three independent which also lacks the ‘Tower’ domain (Karytinos et al., 2009). biological replicates were performed for each experiment, and Conversely to LDL1, LDL2, and FLD, a small insertion (about mean values of relative expression levels from the different 33 amino acids) is present inside the AO domain of LDL3 biological replicates are shown. (Figure 1), which, however, has low sequence similarity to the HsLSD1 ‘Tower’ domain. Molecular modeling of LDL3 indicates that this region is probably unstructured (Figure 2). Nonetheless, RESULTS it cannot be excluded that this region becomes structured upon interaction with, yet unknown, binding partners. Furthermore, The Arabidopsis LDL/FLD Gene Family comparative analysis of the LDL3 model with respect to the three- The Arabidopsis LDL1, LDL2, and FLD display a high amino dimensional structure of HsLDS1 in complex with a substrate- acid sequence identity with each other (48–52%; Spedaletti et al., mimic peptide (Forneris et al., 2007) and to the molecular model 2008) and a shared gene structure, although with a different of LDL1 (Spedaletti et al., 2008) indicates that almost all of the number of introns (Figure 1). In particular, LDL1 gene has no residues involved in substrate binding in HsLSD1 are conserved intron, LDL2 has one and FLD four, one of the FLD introns in both LDL1, whose demethylase activity has been demonstrated at the same position as the single intron in LDL2 (Figure 1, experimentally (Spedaletti et al., 2008), and LDL3 (Table 1). red diamonds). These similarities suggest that LDL1, LDL2, and This analysis suggests that LDL3 is a lysine demethylase with a FLD are recent derivatives of a common ancestor. In contrast, substrate specificity similar to that of LDL1. LDL3 gene structure is different from that of the other three LDL/FLD genes displaying seven introns, all of them at different Evolutionary History of Plant LDL/FLDs position with respect to the FLD introns (Figure 1). Furthermore, To investigate the evolutionary history of the plant LDL/FLDs, the amino acid sequence identity of LDL3 with the other three a phylogenetic analysis of the amino acid sequence of 159 LDL/FLDs is low (25–30%). LDL3 amino acid sequence (1,628 LDL/FLD homologs from 57 different representative animal, amino acids) is also significantly longer than that of the other plant and algal species (Supplementary Table 1) was carried LDL/FLDs (746–884 amino acids). In particular, LDL3 displays out. Maximum likelihood phylogenetic trees of both the full- longer N-terminal and C-terminal extensions, as well as a larger length (Figure 3) and AO domain (Supplementary Figure 1) region linking SWIRM and AO domains (SWIRM/AO distance), amino acid sequences show four main clades: clades AI and in respect to those of LDL1, LDL2, and FLD. Interestingly, at the AII grouping, respectively, HsLSD1 and HsLSD2 homologs from C-terminal extension of LDL3, a putative structured domain with animals, and clades PI and PII grouping LDL/FLD homologs some similarity to transcription factor IIS was identified, which from plants and algae. Relationships between these animal and may have a regulatory role (Figure 1). DNA-binding domains are plant clades are unclear since the most basal nodes of the also present in the fungal (SWIRM1 and SWIRM2) and HsLSD2 tree lack strong bootstrap support, in agreement with a recent homologs, which display a HMG box and a zinc finger domain, phylogenetic analysis of the AO domains based on a limited respectively (Nicolas et al., 2006; Zhang Q. et al., 2013). number of plant and animal LDL/FLDs (Zhou and Ma, 2008). LDL1, LDL2, and FLD do not have the HsLSD1 protruding Within both clade PI and clade PII, green algae LDL/FLDs form a ‘Tower’ domain. However, despite the absence of the ‘Tower’ subclade that is sister to the clade formed by land plant LDL/FLDs (Figure 3 and Supplementary Figure 1), in agreement with organism relationships. Phylogenetic relationships between plant LDL/FLDs indicate that plant LDL1, LDL2, and FLD homologs share a recent common ancestor (Figure 3, node c; bootstrap support, DBS = 100), whereas plant LDL3 homologs belong to a different evolutionary lineage (node b; BS = 98), consistently to what has been suggested for the Arabidopsis LDL/FLDs based on gene structure analysis. Moreover, within clade PI, plant LDL1, LDL2, and FLD homologs form three well supported clades (clades PIa, PIb1, and PIb2, respectively), with LDL2 homologs sister to FLD homologs (node e; BS = 80). The LDL/FLD homologs of PIa, PIb1, and PIb2 clades are well distributed among the various flowering plant species, being present both in dicotyledonous and monocotyledonous plants, as well as in FIGURE 1 | Domain organization of the four Arabidopsis LDL/FLD proteins. Amborella trichopoda, which represents the sister lineage to all All the four proteins contain a SWIRM domain and an amine oxidase domain. other extant flowering plants (Amborella Genome Project, 2013). LDL3 also presents an insertion in the amine oxidase domain and a putative In these clades, phylogenetic relationships between LDL/FLD structured domain with similarity to the transcription factor IIS. Intron positions amino acid sequences closely reflect evolutionary relationships are shown. Numbers indicate the length of the amino acid primary sequence. The schematic representations are in scale. between plant families to which they belong (Figure 3). This phylogenetic pattern suggests that LDL1, LDL2, and FLD genes Frontiers in Plant Science | www.frontiersin.org 4 June 2019 | Volume 10 | Article 669 Martignago et al. Histone Demethylases and Flowering Time FIGURE 2 | Schematic representation of the three-dimensional structure of human LSD1, Arabidopsis LDL3, as well as Physcomitrella patens and Selaginella moellendorffii LDL3 homologs. The three-dimensional structure of human LSD1 (HsLSD1) is that determined by Forneris et al. (2007) in complex with substrate-like peptide inhibitor. The three-dimensional structure of Arabidopsis LDL3, as well as of Physcomitrella and Selaginella LDL3 homologs (AtLDL3, PpLDL3, SmLDL3, respectively) were obtained through molecular modeling approaches based on the three-dimensional structure of HsLDS1 and the molecular model of Arabidopsis LDL1 (Spedaletti et al., 2008). The FAD cofactor is shown in ball-and-stick representation and colored in blue. Shaded ellipses highlight the ‘Tower’ domain of HsLSD1 and PpLDL3, and the corresponding regions in SmLDL3 and AtLDL3. have evolved through gene duplications. A first duplication max. Furthermore, the LDL/FLD homologs of group PII bear would have occurred in correspondence of node c and a second a higher number of introns (5 to 19 introns, Table 2 and one at node e. The fact that Amborella trichopoda shows one copy Supplementary Table 2) than those of clade PI. Whether such of each of LDL1, LDL2, and FLD genes with sister relationships to the corresponding clades formed by flowering plant LDL/FLD homologs indicates that such duplication events occurred TABLE 1 | Substrate binding residues in human LSD1, and orthologous residues before the split between eudicots and monocots. Most likely in Arabidopsis LDL1 and LDL3. duplications took place early during the diversification of land HsLSD1 LDL1 LDL3 plants, as suggested by the occurrence of a supported subclade of FLD (Figure 3, node h; BS = 98) clustering homologs found Met332 Ser∗ Ser in the moss Physcomitrella patens, the liverwort Marchantia Cys360 Cys Cys polymorpha and the seedless ancient vascular plant Selaginella Asp375 Asp Asp moellendorffii. According to this scenario, gene duplications Glu379 Glu Gln would have been followed by LDL1 and LDL2 gene loss in these Asn383 Asn Asn ancient plants. Phylogenetic relationships between plant LDL1, Phe538 Leu Tyr LDL2, and FLD homologs also account for their shared gene Ala539 Ala Gly structure. Indeed, with some exceptions, in most flowering plants Asn540 Asn Cys LDL1 homologs lack introns, LDL2 homologs display one intron, Asp553 Asp Asn whereas FLD homologs have four to five introns, all at conserved Asp556 Asp Asp positions (Table 2 and Supplementary Table 2). This suggests His564 His His that sequential insertions have occurred first in the common Trp695 Tyr Trp ancestor of LDL2 and FLD genes (node e) and later in the ancestor Tyr761 Tyr Tyr of FLD genes (node g). Binding residues have been selected on the basis of the three-dimensional The phylogenetic pattern observed in lineage PII provides no structure of human LSD1 (HsLSD1) in complex with a substrate-mimic peptide evidence for old duplication events behind the diversification (PDB code 2V1D; Forneris et al., 2007). Orthologous residues in Arabidopsis LDL1 and LDL3 have been identified on the basis of structure-based sequence alignment of LDL3 homologs, whereas species-specific duplication events using the molecular models of Arabidopsis LDL1 (Spedaletti et al., 2008) and LDL3 might account for the occurrence of multiple LDL3 genes in (present work). ∗ Conserved or conservatively substituted residues are shown in the rosids Gossypium hirsutum, Populus trichocarpa, and Glycine black, non-conserved residues in red. Frontiers in Plant Science | www.frontiersin.org 5 June 2019 | Volume 10 | Article 669 Martignago et al. Histone Demethylases and Flowering Time Pd-AtPAO1 Arabidopsis thaliana 100 f-Sw1 Schizosaccharomyces pombe f-Sw2 Schizosaccharomyces pombe 100 a-CiLSD2 Ciona intestinalis a-EpLSD2 Exaiptasia pallida a-TrLSD2 Takifugu rubripes a-XlLSD2 Xenopus laevis a-XtLSD2 Xenopus tropicalis a-TgLSD2 Taeniopygia guttata a-GgLSD2 Gallus gallus a-MdLSD2 Monodelphis domestica a-LaLSD2 Loxodonta africana a-EcLSD2 Equus caballus a-MmLSD2 Mus musculus AI a-HsLSD2 Homo sapiens a-MfLSD2 Macaca fascicularis a-AmLSD2 Ailuropoda melanoleuca a-SsLSD2 Sus scrofa a-TtLSD2 Tursiops truncatus a-DlLSD2 Delphinapterus leucas a-CiLSD1 Ciona intestinalis 100 a-TcLSD1 Tribolium castaneum a-DmLSD1 Drosophila melanogaster a-GgLSD1 Gallus gallus a-TgLSD1 Taeniopygia guttata a-MdLSD1 Monodelphis domestica a-MmLSD1 Mus musculus a-LaLSD1 Loxodonta africana a-AmLSD1 Ailuropoda melanoleuca a-HsLSD1 Homo sapiens AII a-MfLSD1 Macaca fascicularis a-EcLSD1 Equus caballus a-SsLSD1 Sus scrofa a-TtLSD1 Tursiops truncatus a-DlLSD1 Delphinapterus leucas 100 ch-OtLDL1 Ostreococcus tauris ch-OlLDL1 Ostreococcus lucimarinus p-AmtLDL1 Amborella trichopoda p-ZrLDL1 Zostera marina p-MaLDL1a Musa acuminata 89 p-MaLDL1b Musa acuminata p-BdLDL1 Brachypodium distachyon p-AgLDL1 Aegilops tauschii d p-OsLDL1 Oryza sativa p-SiLDL1 Setaria italica p-SbLDL1 Sorgum bicolor p-ZmLDL1 Zea mays p-VvLDL1 Vitis vinifera 98 p-GmLDL1a Glycine max a p-PvLDL1 Phaseolus vulgaris p-GmLDL1b Glycine max p-StLDL1 Solanum tuberosum PIa p-SlLDL1a Setaria italica p-NtLDL1a Nicotiana tabacum p-NtLDL1b Nicotiana tabacum p-PrLDL1 Prunus persica p-RcLDL1 Ricinus communis p-PtLDL1 Populus trichocarpa p-TcLDL1 Theobroma cacao p-GhLDL1 Gossypium hirsutum 100 p-CsLDL1 Citrus sinensis p-BrLDL1 Brassica rapa c p-AtLDL1 Arabidopsis thaliana p-EsLDL1 Eutrema salsugineum p-AmtLDL2 Amborella trichopoda p-MaLDL2 Musa acuminata 100 p-ZmLDL2 Zea mays p-SiLDL2 Setaria italica f p-OsLDL2 Oryza sativa p-HvLDL2 Hordeum vulgare p-BdlDL2 Brachypodium distachyon p-VvLDL2 Vitis vinifera p-PrLDL2 Prunus persica p-CusLDL2 Cucumis sativus p-MtLDL2 Medicago truncatula p-GmLDL2 Glycine max p-PvLDL2 Phaseolus vulgaris p-PtLDL2 Populus trichocarpa PIb1 p-RcLDL2 Ricinus communis p-CsLDL2 Citrus sinensis p-NtLDL2b Nicotiana tabacum PI 80 p-NtLDL2a Nicotiana tabacum e p-SlLDL2a Solanum lycopersicum p-StLDL2 Solanum tuberosum p-SlLDL2b Solanum lycopersicum p-TcLDL2 Theobroma cacao p-BrLDL2 Brassica rapa p-EsLDL2 Eutrema salsugineum p-CrLDL2 Capsella rubella p-AlLDL2 Arabidopsis lyrata p-AtLDL2 Arabidopsis thaliana p-PpFLDb Physcomitrella patens 98 p-PpFLDa Physcomitrella patens p-SmFLD Selaginella moellendorffii h p-MpFLD Marchantia polymorpha p-AmtFLD Amborella trichopoda p-ZrFLD Zostera marina 70 p-MaFLD Musa acuminata g 100 p-SiFLD Setaria italica p-SbFLD Sorgum bicolor p-ZmFLD Zea mays i p-OsFLD Oryza sativa p-BdFLD Brachypodium distachyon p-HvFLD Hordeum vulgare p-VvFLD Vitis vinifera p-SlFLDa Solanum lycopersicum p-StFLD Solanum tuberosum p-SlFLDb Solanum lycopersicum PIb2 p-PrFLD Prunus persica p-GmFLD Glycine max p-RcFLD Ricinus communis p-PtFLDb Populus trichocarpa p-PtFLDa Populus trichocarpa p-GhFLDa Gossypium hirsutum p-GhFLDb Gossypium hirsutum p-GhFLDc Gossypium hirsutum p-EsFLD Eutrema salsugineum p-BnFLD Brassica napus p-BrFLD Brassica rapa p-CrFLD Capsella rubella p-AtFLD Arabidopsis thaliana p-AlFLD Arabidopsis lyrata ch-OtLDL3 Ostreococcus tauris ch-OlLDL3 Ostreococcus lucimarinus 98 p-SmLDL3 Selaginella moellendorffii p-PpLDL3 Physcomitrella patens p-AmtLDL3 Amborella trichopoda p-MaLDL3 Musa acuminata b p-SiLDL3 Setaria italica p-SbLDL3 Sorgum bicolor p-ZmLDL3 Zea mays p-OsLDL3 Oryza sativa p-BdLDL3 Brachypodium distachyon p-TuLDL3 Triticum urartu p-AgLDL3 Aegilops tauschii p-NtLDL3 Nicotiana tabacum p-SlLDL3 Solanum lycopersicum p-StLDL3 Solanum tuberosum p-VvLDL3 Vitis vinifera p-CrLDL3 Capsella rubella p-AtLDL3 Arabidopsis thaliana p-AlLDL3 Arabidopsis lyrata p-EsLDL3 Eutrema salsugineum PII p-BrLDL3 Brassica rapa p-BnLDL3 Brassica napus p-CusLDL3 Cucumis sativus p-PrLDL3 Prunus persica p-MtLDL3b Medicago truncatula p-MtLDL3a Medicago truncatula p-GmLDL3b Glycine max p-GmLDL3a Glycine max p-RcLDL3 Ricinus communis p-PtLDL3a Populus trichocarpa p-PtLDL3b Populus trichocarpa p-CsLDL3 Citrus sinensis p-TcLDL3 Theobroma cacao p-GhLDL3a Gossypium hirsutum p-GhLDL3b Gossypium hirsutum p-GhLDL3c Gossypium hirsutum 0.8 FIGURE 3 | Phylogenetic tree of full-length amino acid sequence of LDL/FLD homologs in representative plant species. Animal HsLSD1 and HsLSD2 homologs, as well as the two Schizosaccharomyces pombe homologs SWIRM1 and SWIRM2 (Nicolas et al., 2006) are also included in this analysis. Phylogenetic analyses were performed with the Maximum Likelihood method using RAXML v.8.2.10 (Stamatakis, 2014) with the PROTGAMMAJTT substitution model. Node support was evaluated with 1,000 rapid bootstrap inferences. The sequence of the polyamine oxidase 1 of A. thaliana (AtPAO1; At5g13700; Supplementary Table 1) was used as outgroup. Frontiers in Plant Science | www.frontiersin.org 6 June 2019 | Volume 10 | Article 669 Martignago et al. Histone Demethylases and Flowering Time TABLE 2 | Characteristics of LSD1-like proteins in various organisms. AO domain (Table 2), at the same position of the ‘Tower’ domain in the animal LSD1 homologs. This insertion is smaller Intron ‘Tower’ N-terminal C-terminal SWIRM/AO than the HsLSD1 ‘Tower’ domain, being of 47 amino acids in Clade Name number domain extension extension distance S. moellendorffii, 50 amino acids in A. trichopoda, 39 to 43 amino PIa pLDL1-like 0∗ Absent 136 125 22 acids in monocots, 33 to 39 amino acids in dicots, and 24 amino PIb1 pLDL2-like 1 Absent 64 153 19 acids in Ostreococcus lucimarinus. Only the LDL3 homolog of PIb2 p-FLD-like 4 Absent 109 205 19 P. patens (PpLDL3) displays an insertion of a size (94 amino PIb2 p-MpFLD NA Absent 113 161 18 acids) similar to that of the animal ‘Tower’ domain (Table 2). PIb2 p-PpFLDa 1 Absent 0 147 18 Molecular modeling analyses indicate that this insertion may PIb2 p-PpFLDb 1 Absent 0 147 18 adopt a fold similar to that of the HsLSD1 ‘Tower’ domain PIb2 p-SmFLD 2 Absent 27 157 18 (Figure 2). In contrast, the S. moellendorffii insertion appears PI ch-Ol1 1 Absent 27 56 23 unstructured (Figure 2). The functional significance of these PI ch-Ot1 0 Absent 99 57 37 structural differences among the different plant and animal PII p-LDL3-like 10 38 551 586 159 LDL/FLDs and LSDs are not clear so far. PII p-PpLDL3 19 94 451 618 264 PII p-SmLDL3 15 47 137 503 66 Expression Pattern of the Four PII ch-Ol3 0 24 44 473 117 Arabidopsis LDL/FLD Genes PII ch-Ot3 0 8 46 1506 105 AI a-LSD1-like 16 98 138 19 16 During Seedling Development AII a-LSD2-like 20 Absent 290 0 20 Since information concerning the tissue- and organ-specific gene - f-Sw1 0 21 155 243 19 expression pattern may be useful to determine physiological - f-Sw2 2 37 396 236 27 roles, promoter regions of the four Arabidopsis LDL/FLD genes were cloned upstream of a GFP-GUS fusion gene, The number of introns in the corresponding genes and the number of the amino acid residues constituting the ‘Tower’ domain, the N- and C-terminal and LDL/FLD::GFP-GUS transgenic Arabidopsis plants were extensions and the distance between the SWIRM and amine oxidase (AO) obtained. Histochemical GUS staining of developing seedlings domains are indicated. ∗ For p-LDL1-, p-LDL2-, p-FLD-, p-LDL3-, a-LSD1-, showed that LDL1 is expressed in the shoot apical meristem and a-LSD2-like proteins, mean values from all the flowering plant (p-) and (SAM), in the newly emerging leaves (Figures 4A,B), and in the animal (a-) species considered in the present study (Supplementary Figure 2) are indicated. Complete data are reported in Supplementary Figure 2. Pp, root tip (Figure 4C). Cotyledons also appeared stained mainly Physcomitrella patens; Sm, Selaginella moellendorffii; Mp, Marchantia polymorpha; at the tips and along the vascular system (Figure 4A). Strong Ol, Ostreococcus lucimarinus; Ot, Ostreococcus tauri; f-Sw1 and f-Sw2, LSD1-like LDL1-specific GUS staining was also observed in trichomes proteins from Schizosaccharomyces pombe. NA, genomic sequence not available. (Figure 4D). LDL2-specific GUS staining was observed in the root elongation and differentiation zones up to the meristematic region and in the columella of primary and secondary roots difference among LDL/FLDs of clade PI and clade PII has a (Figures 4E,G,H). LDL2 expression was also observed in the physiological significance is still unknown. Also animal LSD1 SAM and the newly emerging leaves (Figures 4E,F). FLD is and LSD2 homologs have a large number of introns (larger expressed in the root apex of primary (Figures 4I,K) and than the plant FLD and LDL3 homologs; Table 2), not at secondary roots (Figure 4L), in the SAM and in the newly conserved positions when the genes of the two animal clades are emerging leaves (Figures 4I,J). It is also expressed in the compared. These data suggest a different evolutionary history for vascular system of cotyledons, roots (Figures 4I,L) and fully the two animal clades. developed leaves. LDL3 is expressed in newly emerging leaves Phylogenetic results can also provide important insight (Figures 4M,N), in the columella and in the root vascular system into the structural evolution of LDL/FLDs and LSDs. Several (Figures 4M,O). LDL3-related GUS staining was also observed structural differences can be pointed out among the different in the vascular system of leaves (Figure 4P), in guard cells plant and animal clades. In particular, all LDL/FLD homologs (Figure 4Q), and in trichomes (Figure 4N). of group PII have long N- and C-terminal extensions, as well as long SWIRM/AO regions, as compared the LDL/FLD homologs of group PI, with the exception of the LDL3 homologs of the Expression of Arabidopsis LDL/FLD two green algal species and S. moellendorffii which display short Genes During Flower Development N-terminal extensions (Table 2). Also animal LSD homologs of and Embryogenesis both clade AI and AII have large N-terminal extensions, but LDL1-related GUS signal was observed in young, completely small or null C-terminal extensions (Table 2). In particular, closed floral buds (Figure 5A). Staining in developing anthers LSD2 homologs display longer N-extensions with respect to and in particular in both anther tapetum and filaments was the LSD1 homologs, probably through acquisition of DNA- or also observed (Figure 5B). In later steps of flower development, protein-interaction domains. Indeed, HsLSD2 possesses both strong GUS staining was evident in mature pollen grains CW-type and C4H2C2-type zinc finger motifs (Burg et al., 2015). (Figure 5C), while in non-fertilized ovules only faint staining Furthermore, similarly to the Arabidopsis LDL3, all LDL/FLD was observed. During embryo development, LDL1-related GUS homologs of group PII, except the Ostreococcus tauri one, are staining was present in the funiculus of the fertilized ovule, characterized by the presence of a small insertion inside the mainly at the ovule proximal region (Figure 5D). Furthermore, Frontiers in Plant Science | www.frontiersin.org 7 June 2019 | Volume 10 | Article 669 Martignago et al. Histone Demethylases and Flowering Time FIGURE 4 | Promoter activity of Arabidopsis LDL/FLD genes during vegetative development. Histochemical GUS staining of LDL/FLD::GFP-GUS Arabidopsis seedlings is shown. Bars indicate 1 mm in (A,E,I,M), 200 µm in (N), 100 µm in (B–D,F–H,J–L,O,P), and 20 µm in (Q). developing and mature embryos presented staining at the mature embryos (Figures 6E–G). This staining was extended central part of cotyledons (Figures 6A–C) and this pattern in the entire embryo, excluding only the embryonal root was maintained in fully developed embryos, as observed by tip. In embryos within the imbibed seeds, the expression GUS histochemical analysis of imbibed seeds (Figure 6D). pattern resembled the one of the young seedlings, with strong The LDL1 expression in imbibed seeds is in agreement promoter activity in SAM, cotyledons, and in the root elongation with the public Arabidopsis microarray database and the zone (Figure 6H). reported essential role of LDL1 and LDL2 in seed dormancy FLD-related GUS staining was observed in the anther– (Zhao et al., 2015). filament junction and in the tapetum (Figures 5I,J), but LDL2-specific GUS staining was observed in developing not in mature pollen grains. Ovules also appeared stained pistils and anthers of floral buds (Figure 5E). Later during (Figures 5K,L). Following fertilization, strong FLD-specific development, mature pollen grains (Figure 5F) and embryo staining was observed in developing embryos at the heart and sacs (Figures 5G,H) were stained too, embryo sacs presenting torpedo stages and in mature embryos (Figures 6I–L). Mature a strong signal at the micropylar end (Figure 5H). Following embryos in imbibed seeds also presented staining (Figure 6M). fertilization, strong LDL2-specific staining was observed in Interestingly, FLD expression was evident in the provascular developing embryos at the heart and torpedo stages and in tissues of embryonic roots and cotyledons (Figure 6L, arrows), Frontiers in Plant Science | www.frontiersin.org 8 June 2019 | Volume 10 | Article 669 Martignago et al. Histone Demethylases and Flowering Time FIGURE 5 | Promoter activity of Arabidopsis LDL/FLD genes during reproductive development. Histochemical GUS staining of LDL/FLD::GFP-GUS Arabidopsis transgenic plants in inflorescences is shown. Bars indicate 1 mm in (A,E,G,I,K,M,O), 100 µm in (B–D,F,J,N), and 20 µm in (H,L,P). similarly to FLD expression in root, cotyledon and leaf vascular staining of the SAM and root tip was evident (Figure 6Q). The system of young seedlings (Figures 4I,L). same staining pattern was observed in mature embryos inside LDL3-specific staining was observed in peduncles, sepals imbibed seeds (Figure 6R). (Figure 5M), stamen filaments (Figure 5O) and mature pollen grains (Figure 5N). Pistils were also stained (Figures 5M,O), in particular ovules (Figures 5O,P). Following fertilization, LDL3 Mutant Plants Show LDL3-specific staining was observed both in developing and in Early-Flowering Phenotype mature embryos (Figures 6N–Q). Funiculus of fertilized ovules To elucidate the physiological roles of the four Arabidopsis presented staining as well (Figure 6O). In mature embryos, LDL/FLD genes, loss-of-function mutants for each of the four Frontiers in Plant Science | www.frontiersin.org 9 June 2019 | Volume 10 | Article 669 Martignago et al. Histone Demethylases and Flowering Time FIGURE 6 | Promoter activity of Arabidopsis LDL/FLD genes during embryogenesis and in imbibed seeds. Histochemical GUS staining of LDL/FLD::GFP-GUS transgenic plants in siliques (A–C,E–G,I–L,N–Q) and imbibed seeds (D,H,M,R). Black arrows indicate LDL1-specific staining in embryo (A), FLD-specific staining in provascular tissues of roots and cotyledons in embryo (L) and LDL3-specific staining in globular stage of embryo (N). Bars indicate 50 µm. Arabidopsis LDL/FLD genes were identified and characterized experimental conditions unless the fld mutant plants were treated to confirm disruption of gene expression (Supplementary with gibberellins or grown under low-temperature conditions for Figure 2). The ldl/fld mutants were initially examined for prolonged periods, further confirming that FLD is involved in flowering time by measuring the number of rosette leaves upon the autonomous pathway controlling flowering time. Under the bolting. In agreement with previously published data (He et al., same conditions, the ldl1 and ldl2 mutants displayed only a very 2003), the fld mutant presented an extremely late-flowering short, not statistically significant, delay in flowering (Figure 7) phenotype. Indeed, floral transition was not obtained under our as previously reported (Jiang et al., 2007). Conversely, data Frontiers in Plant Science | www.frontiersin.org 10 June 2019 | Volume 10 | Article 669 Martignago et al. Histone Demethylases and Flowering Time FIGURE 7 | Flowering time of ldl/fld loss-of-function Arabidopsis mutants for LDL/FLDs. Flowering time of ldl1, ldl2, ldl3 mutants and wild-type plants (WT) is expressed as the number of rosette leaves at bolting. The fld mutant plants presented non-flowering phenotype under our growth conditions. Mean values (n > 8) of a representative analysis out of at least three repetitions are shown and bars indicate standard error. Asterisk indicates statistically significant differences from WT plants (one-way ANOVA test, p < 0.05). presented here evidence that ldl3 mutant display early-flowering phenotype (Figure 7). FLC Is Up-Regulated in ldl1, ldl2, and fld Mutants, but Down-Regulated in ldl3 Mutants To verify whether the early-flowering phenotype of the ldl3 mutant depends on FLC, as the late-flowering phenotype of ldl1, ldl2, and fld mutants does (He et al., 2003; Jiang et al., 2007), a comparative analysis of the FLC expression levels in the four lsd/fld mutants was performed. The qRT-PCR analysis evidenced twofold and fourfold increase in FLC expression FIGURE 8 | Relative expression levels of FLC and FWA in Arabidopsis levels in the ldl1 and ldl2 mutants, respectively, comparing mutants for LDL/FLDs. Two-week-old seedlings of ldl1, ldl2, fld, and ldl3 to the wild-type plants, as opposed to a 100-fold increase mutants, and wild-type plants (WT) were analyzed for FLC and FWA in the fld mutant (Figure 8), consistently with the flowering expression levels by qRT-PCR. Data for FLC are shown in logarithmic scale. phenotypes. Conversely, a twofold decrease in FLC expression Numbers are mean values ± SE of three independent replicates. Asterisks indicate statistically significant differences from WT plants (one-way ANOVA levels was observed in the ldl3 mutant as compared to the test, p < 0.05). wild-type plants (Figure 8), which is also consistent with the early-lowering phenotype of this mutant. These results suggest that the various members of LDL/FLD gene family contribute in a different way to the control of FLC expression and thus that fld mutants were previously shown to display altered H3K4 to the flowering time, LDL3 having an opposing effect with trimethylation levels at MAF4 and MAF5 (Yu et al., 2011). respect to the others. FWA is a transcription factor which participates in the control MAF1 to MAF5, MADS-containing transcription factors of floral transition and which has been demonstrated to be under homologs to FLC, also contribute to the control of the flowering epigenetic control. In particular, FWA is silenced during plant time (Ratcliffe et al., 2003; Rosloski et al., 2013). Data from vegetative development and in the sporophytes by repressive qRT-PCR analysis evidenced no statistically significant difference DNA methylation in its 50 region, its expression being confined to in the expression levels of MAF1 to MAF4 in all four ldl/fld the central cell of the female gametophytes and to the endosperm. mutants as compared to the wild-type plants (Supplementary Moreover, fwa epi-alleles cause a late-flowering phenotype due Figure 3). These data indicate that Arabidopsis LDL/FLDs are not to ectopic FWA expression in sporophytic tissues (Soppe et al., involved in the regulation of the MAF gene family despite the fact 2000; Kinoshita et al., 2004). In addition, FWA was shown Frontiers in Plant Science | www.frontiersin.org 11 June 2019 | Volume 10 | Article 669 Martignago et al. Histone Demethylases and Flowering Time to be ectopically activated in rosette leaves of ldl1 and ldl2 In the present study, an analysis of the main LDL/FLD target mutants, but not of fld mutants, suggesting that LDL1 and genes showed that all four LDL/FLDs are involved in the control LDL2 contribute to the repression of FWA expression during of FLC expression. In particular, in agreement with previously vegetative development (Jiang et al., 2007). Here, to determine the published data (Jiang et al., 2007), LDL1, LDL2, and FLD were specific contribution of the four LDL/FLDs to FWA regulation, shown to have a repressive effect on FLC expression levels, with qRT-PCR analysis was performed. Results showed a strong the effect of FLD being much more pronounced than that of increase in FWA expression levels in the ldl2 mutant and a LDL1 and LDL2. Instead, LDL3 has an enhancing effect on FLC smaller one in the ldl1 and fld mutants (Figure 8). Instead, expression. Indeed, ldl3 mutant plants display decreased FLC no difference in FWA expression levels was observed in the transcript levels, as compared to the wild-type plants, while ldl1, ldl3 mutant as compared to the wild type plants. These data ldl2, and fld mutants display increased FLC levels (Figure 8). suggest that it is mainly LDL2, among the four Arabidopsis These differences in FLC expression levels reflect the differences LDL/FLDs, that is involved in FWA repression during vegetative in flowering time, fld displaying a non-flowering phenotype, growth of Arabidopsis plants. This well correlates with the while ldl3 an early-flowering phenotype (Figure 7). LDL1, LDL2, LDL2 expression in the female gametophyte presenting a pattern and FLD repress also FWA expression LDL2 having a more similar to that of FWA (Kinoshita et al., 2004). Indeed, both pronounced effect than LDL1 and FLD (Figure 8). The lack of LDL2 and FWA are expressed in embryo sacs (Figure 5N) differences in flowering time among ldl1 and ldl2 mutants and (Kinoshita et al., 2004). wild-type plants, despite the altered FLC and FWA expression levels, suggests a quantitative effect of FLC and FWA on floral transition. Differences in tissue- and temporal-specific expression DISCUSSION pattern, as well as in protein/protein interactions among the four LDL/FLDs may also explain the different flowering phenotypes of In the present work, a comparative study on gene structure, the four ldl/fld mutants. phylogenetic relationships, spatio-temporal expression patterns, Previous studies have shown that LDL1, LDL2, and FLD physiological roles and target genes for the four LDL/FLD genes repress FLC transcription by reducing H3K4 methylation levels was performed, which evidenced several similarities, but also at specific regions of the FLC chromatin (Jiang et al., 2007). important differences among them. This raises the question of which are the underlying mechanisms Data from exon/intron structure analyses and phylogenetic determining the opposing effects of the different LDL/FLDs on studies suggest that the LDL1, LDL2, and FLD homologs of FLC expression levels considering the similarity of the catalytic the various plant species derive from a single copy of the sites (Table 1) (Spedaletti et al., 2008). To get through these LDL/FLD gene present in the early ancestor of land plants mechanisms, the LDL3 substrate specificity, the FLC chromatin through two duplication events. Furthermore, the different regions with which LDL3 specifically interacts and the LDL/FLD number of introns observed in LDL/FLD genes is likely the specific partners have to be determined. On the other hand, result of two sequential insertion events occurred first in the physiological significance of the different/opposing effects of the common ancestor of LDL2 and FLD genes and later in the four LDL/FLDs on floral transition is not clear so far. They the ancestor of FLD genes. This process might have brought may contribute to a fine-tune regulation and optimization of about differences in expression levels and function among the flowering time. LDL/FLD homologs of clade PI. Indeed, it has been shown FLC expression is promoted by FRIGIDA (FRI) and is that although all three LDL1, LDL2, and FLD act redundantly repressed by sets of genes in the autonomous and vernalization in the control of the flowering time, FLD plays a major role pathways (Amasino and Michaels, 2010; Yang et al., 2017). FLC in this process (Jiang et al., 2007). Intron acquisition occurred is expressed in shoot and root apical regions, as well as in leaf also during evolution of the LDL3 homologs which might vasculature, in pollen mother cells, in the tapetum surrounding have contributed to increase gene expression levels. Differently these cells, and in the anther connective tissue, but not in mature from LDL1, LDL2, and FLD, LDL3 homologs are characterized pollen grains (Bastow et al., 2004; Michaels et al., 2005; Sheldon by the presence of a small insertion in the same position et al., 2008; Choi et al., 2009). It is also expressed in the ovule as that of the ‘Tower’ domain in HsLSD1. This insertion, integuments before and after pollination, but not in the female which in P. patens is long enough to allow the presence of gametophytes. FLC is additionally expressed in the developing a structural domain similar to the HsLSD1 ‘Tower’ domain, embryo during all stages of embryogenesis reaching a maximum became shorter during the transition from the moss P. patens (94 when the seed has been fully formed (Sheldon et al., 2008; Choi amino acids), to the ancient vascular plant S. moellendorffii (47 et al., 2009; Berry and Dean, 2015). In old embryos, FLC is amino acids), to the basal angiosperm A. trichopoda (50 amino expressed in the provascular tissue of both the embryonic roots acids), to dicots and monocots (33–47 amino acids). The and cotyledons. Thus, FLC expression is repressed in mature evolutionary pressure leading to such changes is not known male and female gametophytes to be reactivated after fertilization, yet. It is possible that these changes have been accompanied by in reprogramming processes that are considered important for evolution of new protein/protein interaction motifs. Altogether, plant reproduction, mainly ensuring a vernalization requirement these data indicate a different evolutionary history of the in each generation (Berry and Dean, 2015). Little is known so far two main plant LDL/FLD clades, similarly to the two animal about how the several FLC regulators control FLC transcription LSD clades. in the various developmental stages (Choi et al., 2009). Frontiers in Plant Science | www.frontiersin.org 12 June 2019 | Volume 10 | Article 669 Martignago et al. Histone Demethylases and Flowering Time The promoter activity studies presented here evidence that all repression of FWA expression in ovules 48 h after pollination four Arabidopsis LDL/FLD genes are expressed in SAM and/or (Kinoshita et al., 2004). newly emerging leaves. The Arabidopsis LDL/FLD genes are Altogether, data presented here suggest functional differences also expressed in roots, though with a gene-specific pattern. among the four Arabidopsis LDL/FLD genes, even among Furthermore, all LDL/FLD genes, except LDL2, are expressed the LDL1, LDL2, and FLD, which are recent derivatives of in the vascular system of the roots and/or leaves. LDL3, a common ancestor gene. It is possible that following gene differently from the other LDL/FLD genes, is also expressed in duplication, LDL1, LDL2, and FLD genes have undergone sub- guard cells. LDL/FLDs are also expressed during reproductive functionalization or neo-functionalization which might have development, though with some differences from each other. helped in the optimization of the regulatory network controlling In particular, prior to fertilization all four LDL/FLDs are floral transition and defense responses (Zhou and Ma, 2008). expressed in ovules. However, while LDL1, FLD and LDL3 are Several studies have evidenced a relevant role of the different expressed in the entire ovule, probably mainly involving the epigenetic mechanisms in the control of plant developmental and ovule integuments, LDL2 is specifically expressed in the embryo defense/adaptation processes and their impact on agronomical sacs. Furthermore, LDL1, LDL2, and LDL3 are expressed in traits other than flowering time, such as yield and fruit ripening mature pollen grains, as opposed to FLD that is not expressed (Gallusci et al., 2017; Giovannoni et al., 2017; Annacondia et al., in male gametophytes. Following fertilization, all four LDL/FLDs 2018). In this context, the contribution of the LDL/FLD gene are expressed in developing embryos. LDL1 and LDL3 are family in these processes should be analyzed and involved target additionally expressed in the funiculus of developing embryos. genes should be identified. These pieces of information may Altogether, the promoter activity studies presented here show provide novel biotechnological strategies for crop improvement. that the four LDL/FLDs, both the FLC repressors and the FLC activator, display overlapping and complementary expression patterns with respect to each other and to FLC, thus not AUTHOR CONTRIBUTIONS allowing to assign a specific role to each of them in FLC regulation at certain developmental stages. What appears to PT conceived the research plan. DM and PT designed and be an important difference among the different LDL/FLDs is performed the experiments. BB, DS, and PT contributed to the the lack of FLD-specific expression in pollen grains. These phylogenetic analyses. FP performed the molecular modeling data exclude the possibility that FLD, which among the four analyses. DM, DS, FP, and PT wrote the manuscript. AC and RA LDL/FLDs is the best FLC repressor, is responsible for FLC provided advice and comments for the manuscript. repression in pollen grains, in agreement with previous data showing that FLC expression pattern during gametogenesis and embryogenesis is not altered in an fld genotype (Choi FUNDING et al., 2009). LDL1 and LDL2 may have a role in this process, although it is again difficult to explain the presence of LDL3, This work was supported by the Italian Ministry of Education, which acts as an FLC activator, in pollen. Further detailed University and Research (Grant to Department of Science, analyses of FLC expression pattern in single and multiple University ‘Roma Tre’-‘Dipartimenti di Eccellenza’, ARTICOLO ldl/fld mutants may give useful information on the specific 1, COMMI 314–337, LEGGE 423 232/2016), and University contribution of the different LDL/FLDs to FLC regulation in ‘Roma Tre’. DS is currently supported by the program ‘Rita Levi a tissue- and organ-specific way and in the reprogramming Montalcini’ (MIUR, Ministero dell’Istruzione dell’Università e processes. It is also likely that a balanced activity of different FLC della Ricerca) for the recruitment of young researchers at the regulators is necessary for proper FLC levels to be established University of L’Aquila. at the various developmental stages. On the other hand, the high expression levels of LDL/FLDs during gametogenesis and embryogenesis suggests a function for this gene family in the ACKNOWLEDGMENTS transgenerational reset of epigenetic memory, known to affect not only DNA methylation level, but also histone methylation We are grateful to Plant Systems Biology (University of Gent) (Zheng et al., 2016). for the kind gift of the pKGWFS7 and pK2GW7 binary FWA is specifically expressed in the female gametophytes, vectors, the Nottingham Arabidopsis Stock Center for the mainly in the central cell and in the developing endosperm, SAIL_640_B10.v1 mutant, and Arabidopsis Biological Resource for 48 h after pollination (Kinoshita et al., 2004). However, Center for the SALK lines. in the present study, a similar expression pattern has been evidenced for LDL2 (Figure 5H), which, among the three LDL/FLDs, appears to have a major role in the control FWA SUPPLEMENTARY MATERIAL expression during vegetative growth of Arabidopsis plants, repressing it. LDL2 may be necessary together with FWA The Supplementary Material for this article can be found online activators in multi-protein complexes for optimal FWA levels at: https://www.frontiersin.org/articles/10.3389/fpls.2019.00669/ in embryo sacs. It is also possible that LDL2 is responsible for full#supplementary-material Frontiers in Plant Science | www.frontiersin.org 13 June 2019 | Volume 10 | Article 669 Martignago et al. Histone Demethylases and Flowering Time REFERENCES He, Y., Michaels, S. D., and Amasino, R. M. (2003). Regulation of flowering time by histone acetylation in Arabidopsis. Science 302, 1751–1754. doi: 10.1126/ Alonso, J. M., Stepanova, A. N., Leisse, T. J., Kim, C. J., Chen, H., Shinn, P., et al. science.1091109 (2003). Genome-wide insertional mutagenesis of Arabidopsis thaliana. Science Hu, Q., Jin, Y., Shi, H., and Yang, W. (2014). GmFLD, a soybean homolog of the 301, 653–657. doi: 10.1126/science.1086391 autonomous pathway gene FLOWERING LOCUS D, promotes flowering in Altschul, S. F., Madden, T. L., Schäffer, A. A., Zhang, J., Zhang, Z., Miller, W., et al. Arabidopsis thaliana. BMC Plant Biol. 14:263. doi: 10.1186/s12870-014-0263-x (1997). Gapped BLAST and PSI-BLAST: a new generation of protein database Jiang, D., Yang, W., He, Y., and Amasino, R. M. (2007). Arabidopsis relatives of search programs. Nucleic Acids Res. 25, 3389–3402. the human lysine-specific demethylase1 repress the expression of FWA and Amasino, R. M., and Michaels, S. D. (2010). The timing of flowering. Plant Physiol. FLOWERING LOCUS C and thus promote the floral transition. Plant Cell 19, 154, 516–520. doi: 10.1104/pp.110.161653 2975–2987. doi: 10.1105/tpc.107.052373 Amborella Genome Project (2013). The amborella genome and the evolution of Jones, D. T., Taylor, W. R., and Thornton, J. M. (1992). The rapid generation flowering plants. Science 342:1241089. doi: 10.1126/science.1241089 of mutation data matrices from protein sequences. Comput. Appl. Biosci. 8, Annacondia, M. L., Mageroy, M. H., and Martinez, G. (2018). Stress response 275–282. regulation by epigenetic mechanisms: changing of the guards. Physiol. Plant. Karimi, M., Inzé, D., and Depicker, A. (2002). GATEWAYTM vectors for 162, 239–250. doi: 10.1111/ppl.12662 Agrobacterium-mediated plant transformation. Trends Plant Sci. 5, 193–195. Banday, Z. Z., and Nandi, A. K. (2018). Arabidopsis thaliana GLUTATHIONE- doi: 10.1016/S1360-1385(02)02251-3 S-TRANSFERASE THETA 2 interacts with RSI1/FLD to activate systemic Karytinos, A., Forneris, F., Profumo, A., Ciossani, G., Battaglioli, E., Binda, C., acquired resistance. Mol. Plant Pathol. 19, 464–475. doi: 10.1111/mpp. et al. (2009). A novel mammalian flavin-dependent histone demethylase. J. Biol. 12538 Chem. 284, 17775–17782. doi: 10.1074/jbc.M109.003087 Bastow, R., Mylne, J. S., Lister, C., Lippman, Z., Martienssen, R. A., and Dean, Katoh, K., and Standley, D. M. (2013). MAFFT multiple sequence alignment C. (2004). Vernalization requires epigenetic silencing of FLC by histone software version 7: improvements in performance and usability. Mol. Biol. Evol. methylation. Nature 427, 164–167. doi: 10.1038/nature02269 30, 772–780. doi: 10.1093/molbev/mst010 Bent, A. (2006). Arabidopsis thaliana floral dip transformation method. Methods Kinoshita, T., Miura, A., Choi, Y., Kinoshita, Y., Cao, X., Jacobsen, S. E., et al. Mol. Biol. 343, 87–103. doi: 10.1385/1-59745-130-4:87 (2004). One-way control of FWA imprinting in Arabidopsis endosperm by Berry, S., and Dean, C. (2015). Environmental perception and epigenetic memory: DNA methylation. Science 303, 521–523. doi: 10.1126/science.1089835 mechanistic insight through FLC. Plant J. 83, 133–148. doi: 10.1111/tpj. Krichevsky, A., Gutgarts, H., Kozlovsky, S. V., Tzfira, T., Sutton, A., Sternglanz, R., 12869 et al. (2007). C2H2 zinc finger-SET histone methyltransferase is a plant-specific Burg, J. M., Link, J. E., Morgan, B. S., Heller, F. J., Hargrove, A. E., and McCafferty, chromatin modifier. Dev. Biol. 303, 259–269. doi: 10.1016/j.ydbio.2006.11.012 D. G. (2015). KDM1 class flavin-dependent protein lysine demethylases. Pept. Krichevsky, A., Zaltsman, A., Kozlovsky, S. V., Tian, G. W., and Citovsky, V. Sci. 104, 213–246. doi: 10.1002/bip.22643 (2009). Regulation of root elongation by histone acetylation in Arabidopsis. Chen, Y., Yang, Y., Wang, F., Wan, K., Yamane, K., Zhang, Y., et al. (2006). J. Mol. Biol. 385, 45–50. doi: 10.1016/j.jmb.2008.09.040 Crystal structure of human histone lysine-specific demethylase 1 (LSD1). Krichevsky, A., Zaltsman, A., Lacroix, B., and Citovsky, V. (2011). Involvement Proc. Natl. Acad. Sci. U.S.A. 103, 13956–13961. doi: 10.1073/pnas.060638 of KDM1C histone demethylase-OTLD1 otubain-like histone deubiquitinase 1103 complexes in plant gene repression. Proc. Natl. Acad. Sci. U.S.A. 108, 11157– Choi, J., Hyun, Y., Kang, M. J., In Yun, H., Yun, J. Y., Lister, C., et al. (2009). 11162. doi: 10.1073/pnas.1014030108 Resetting and regulation of flowering locus C expression during arabidopsis Liu, F., Quesada, V., Crevillen, P., Baurle, I., Swiezewski, S., and Dean, C. reproductive development. Plant J. 57, 918–931. doi: 10.1111/j.1365-313X.2008. (2007). The Arabidopsis RNA-binding protein FCA requires a lysine-specific 03776.x demethylase 1 homolog to downregulate FLC. Mol. Cell 28, 398–407. doi: 10. Chou, M. L., and Yang, C. H. (1998). FLD interacts with genes that affect different 1016/j.molcel.2007.10.018 developmental phase transitions to regulate Arabidopsis shoot development. Luo, M., Tai, R., Yu, C. W., Yang, S., Chen, C. Y., Lin, W. D., et al. (2015). Regulation Plant J. 15, 231–242. of flowering time by the histone deacetylase HDA5 in Arabidopsis. Plant J. 82, Czechowski, T., Stitt, M., Altmann, T., Udvardi, M. K., and Scheible, W.-R. 925–936. doi: 10.1111/tpj.12868 (2005). Genome-wide identification and testing of superior reference genes for Michaels, S. D., Himelblau, E., Kim, S. Y., Schomburg, F. M., and Amasino, R. M. transcript normalization in arabidopsis. Plant Physiol. 139, 5–17. doi: 10.1104/ (2005). Integration of flowering signals in winter-annual Arabidopsis. Plant pp.105.063743 Physiol. 137, 149–156. doi: 10.1104/pp.104.052811 Fincato, P., Moschou, P. N., Ahou, A., Angelini, R., Roubelakis-Angelakis, K. A., Miller, M. A., Pfeiffer, W., and Schwartz, T. (2010). “Creating the CIPRES Science Federico, R., et al. (2012). The members of Arabidopsis thaliana PAO gene Gateway for inference of large phylogenetic trees,” in Workshop on Gateway family exhibit distinct tissue- and organ-specific expression pattern during Computing Environments (GCE), New Orleans, LA. seedling growth and flower development. Amino Acids 42, 831–841. doi: 10. Nicolas, E., Lee, M. G., Hakimi, M. A., Cam, H. P., Grewal, S. I., and Shiekhattar, 1007/s00726-011-0999-7 R. (2006). Fission yeast homologs of human histone H3 lysine 4 demethylase Forneris, F., Binda, C., Adamo, A., Battaglioli, E., and Mattevi, A. (2007). Structural regulate a common set of genes with diverse functions. J. Biol. Chem. 281, basis of LSD1-CoREST selectivity in histone H3 recognition. J. Biol. Chem. 282, 35983–35988. doi: 10.1074/jbc.M606349200 20070–20074. doi: 10.1074/jbc.C700100200 Pfluger, J., and Wagner, D. (2007). Histone modifications and dynamic regulation Forneris, F., Binda, C., Vanoni, M. A., Battaglioli, E., and Mattevi, A. (2005). of genome accessibility in plants. Curr. Opin. Plant Biol. 10, 645–652. doi: Human histone demethylase LSD1 reads the histone code. J. Biol. Chem. 280, 10.1016/j.pbi.2007.07.013 41360–41365. doi: 10.1074/jbc.M509549200 Polticelli, F., Basran, J., Faso, C., Cona, A., Minervini, G., Angelini, R., et al. (2005). Gallusci, P., Dai, Z., Genard, M., Gauffretau, A., Leblanc-Fournier, N., Richard- Lys300 plays a major role in the catalytic mechanism of maize polyamine Molard, C., et al. (2017). Epigenetics for plant improvement: current nnowledge oxidase. Biochemistry 44, 16108–16120. doi: 10.1021/bi050983i and modeling avenues. Trends Plant Sci. 22, 610–623. doi: 10.1016/j.tplants. Ratcliffe, O. J., Kumimoto, R. W., Wong, B. J., and Riechmann, J. L. (2003). 2017.04.009 Analysis of the Arabidopsis MADS AFFECTING FLOWERING gene family: Giovannoni, J., Nguyen, C., Ampofo, B., Zhong, S., and Fei, Z. (2017). The MAF2 prevents vernalization by short periods of cold. Plant Cell 15, 1159–1169. epigenome and transcriptional dynamics of fruit ripening. Annu. Rev. Plant doi: 10.1105/tpc.009506 Biol. 68, 61–84. doi: 10.1146/annurev-arplant-042916-040906 Rosloski, S. M., Singh, A., Jali, S. S., Balasubramanian, S., Weigel, D., and Grbic, V. Gu, T., Han, Y., Huang, R., McAvoy, R. J., and Li, Y. (2016). Identification and (2013). Functional analysis of splice variant expression of MADS AFFECTING characterization of histone lysine methylation modifiers in Fragaria vesca. Sci. FLOWERING 2 of Arabidopsis thaliana. Plant Mol. Biol. 81, 57–69. doi: 10. Rep. 6:23581. doi: 10.1038/srep23581 1007/s11103-012-9982-2 He, Y. (2009). Control of the transition to flowering by chromatin modifications. Sadiq, I., Keren, I., and Citovsky, V. (2016). Plant homologs of mammalian Mol. Plant 2, 554–564. doi: 10.1093/mp/ssp005 MBT-domain protein-regulated KDM1 histone lysine demethylases do not Frontiers in Plant Science | www.frontiersin.org 14 June 2019 | Volume 10 | Article 669 Martignago et al. Histone Demethylases and Flowering Time interact with plant Tudor/PWWP/MBT-domain proteins. Biochem. Biophys. Wang, X., Xin, C., Cai, J., Zhou, Q., Dai, T., Cao, W., et al. (2016). Heat priming Res. Commun. 470, 913–916. doi: 10.1016/j.bbrc.2016.01.151 induces trans-generational tolerance to high temperature stress in wheat. Front. Sanda, S. L., and Amasino, R. M. (1996). Ecotype-specific expression of a Plant Sci. 7:501. doi: 10.3389/fpls.2016.00501 flowering mutant phenotype in Arabidopsis thaliana. Plant Physiol. 111, Wu, S., and Zhang, Y. (2007). LOMETS: a local meta-threading-server for protein 641–644. structure prediction. Nucleic Acids Res. 35, 3375–3382. doi: 10.1093/nar/ Sessions, A., Burke, E., Presting, G., Aux, G., McElver, J., Patton, D., et al. (2002). A gkm251 high-throughput arabidopsis reverse genetics system. Plant Cell 14, 2985–2994. Xu, J., Xu, H., Liu, Y., Wang, X., Xu, Q., and Deng, X. (2015). Genome- doi: 10.1105/tpc.004630 wide identification of sweet orange (Citrus sinensis) histone modification gene Sheldon, C. C., Hills, M. J., Lister, C., Dean, C., Dennis, E. S., and Peacock, families and their expression analysis during the fruit development and fruit- W. J. (2008). Resetting of FLOWERING LOCUS C expression after epigenetic blue mold infection process. Front. Plant Sci. 6:607. doi: 10.3389/fpls.2015. repression by vernalization. Proc. Natl. Acad. Sci. U.S.A. 105, 2214–2219. doi: 00607 10.1073/pnas.0711453105 Yang, H., Berry, S., Olsson, T. S. G., Hartley, M., Howard, M., and Dean, C. (2017). Shi, Y., Lan, F., Matson, C., Mulligan, P., Whetstine, J. R., Cole, P. A., et al. (2004). Distinct phases of polycomb silencing to hold epigenetic memory of cold in Histone demethylation mediated by the nuclear amine oxidase homolog LSD1. Arabidopsis. Science 357, 1142–1145. doi: 10.1126/science.aan1121 Cell 119, 941–953. doi: 10.1016/j.cell.2004.12.012 Yang, J., Yan, R., Roy, A., Xu, D., Poisson, J., and Zhang, Y. (2015). The I-TASSER Shi, Y. J., Matson, C., Lan, F., Iwase, S., Baba, T., and Shi, Y. (2005). Regulation Suite: protein structure and function prediction. Nat. Methods 12, 7–8. doi: of LSD1 histone demethylase activity by its associated factors. Mol. Cell 19, 10.1038/nmeth.3213 857–864. doi: 10.1016/j.molcel.2005.08.027 Yang, M., Gocke, C. B., Luo, X., Borek, D., Tomchick, D. R., Machius, M., et al. Shibaya, T., Hori, K., Ogiso-Tanaka, E., Yamanouchi, U., Shu, K., Kitazawa, N., et al. (2006). Structural basis for CoREST-dependent demethylation of nucleosomes (2016). Hd18, encoding histone acetylase related to Arabidopsis FLOWERING by the human LSD1 histone demethylase. Mol. Cell 23, 377–387. doi: 10.1016/j. LOCUS D, is involved in the control of flowering time in rice. Plant Cell Physiol. molcel.2006.07.012 57, 1828–1838. doi: 10.1093/pcp/pcw105 Yu, C. W., Chang, K. Y., and Wu, K. (2016). Genome-wide analysis of gene Sievers, F., Wilm, A., Dineen, D., Gibson, T. J., Karplus, K., Li, W., et al. (2011). regulatory networks of the FVE-HDA6-FLD complex in Arabidopsis. Front. Fast, scalable generation of high-quality protein multiple sequence alignments Plant Sci. 7:555. doi: 10.3389/fpls.2016.00555 using Clustal Omega. Mol. Syst. Biol. 7:539. doi: 10.1038/msb.2011.75 Yu, C. W., Liu, X., Luo, M., Chen, C., Lin, X., Tian, G., et al. (2011). HISTONE Singh, S., Singh, A., Roy, S., and Sarkar, A. K. (2012). SWP1 negatively regulates DEACETYLASE6 interacts with FLOWERING LOCUS D and regulates lateral root initiation and elongation in Arabidopsis. Plant Signal. Behav. 7, flowering in Arabidopsis. Plant Physiol. 156, 173–184. doi: 10.1104/pp.111. 1522–1525. doi: 10.4161/psb.22099 174417 Singh, V., Roy, S., Giri, M. K., Chaturvedi, R., Chowdhury, Z., Shah, J., et al. (2013). Zhang, J., Bonasio, R., Strino, F., Kluger, Y., Holloway, J. K., Modzelewski, A. J., Arabidopsis thaliana FLOWERING LOCUS D is required for systemic acquired et al. (2013). SFMBT1 functions with LSD1 to regulate expression of canonical resistance. Mol. Plant Microbe Interact. 26, 1079–1088. doi: 10.1094/mpmi-04- histone genes and chromatin-related factors. Genes Dev. 27, 749–766. doi: 10. 13-0096-r 1101/gad.210963.112 Singh, V., Roy, S., Singh, D., and Nandi, A. K. (2014). Arabidopsis flowering locus Zhang, Q., Qi, S., Xu, M., Yu, L., Tao, Y., Deng, Z., et al. (2013). Structure-function D influences systemic-acquired-resistance-induced expression and histone analysis reveals a novel mechanism for regulation of histone demethylase modifications of WRKY genes. J. Biosci. 39, 119–126. LSD2/AOF1/KDM1b. Cell Res. 23, 225–241. doi: 10.1038/cr.2012.177 Soppe, W. J., Jacobsen, S. E., Alonso-Blanco, C., Jackson, J. P., Kakutani, T., Zhang, Y., Li, B., Xu, Y., Li, H., Li, S., Zhang, D., et al. (2013). The cyclophilin Koornneef, M., et al. (2000). The late flowering phenotype of fwa mutants is CYP20-2 modulates the conformation of BRASSINAZOLE-RESISTANT1, caused by gain-of-function epigenetic alleles of a homeodomain gene. Mol. Cell which binds the promoter of FLOWERING LOCUS D to regulate 6, 791–802. flowering in Arabidopsis. Plant Cell 25, 2504–2521. doi: 10.1105/tpc.113. Spedaletti, V., Polticelli, F., Capodaglio, V., Schinina, M. E., Stano, P., Federico, 110296 R., et al. (2008). Characterization of a lysine-specific histone demethylase Zhao, M., Yang, S., Liu, X., and Wu, K. (2015). Arabidopsis histone demethylases from Arabidopsis thaliana. Biochemistry 47, 4936–4947. doi: 10.1021/bi701 LDL1 and LDL2 control primary seed dormancy by regulating DELAY OF 969k GERMINATION 1 and ABA signaling-related genes. Front. Plant Sci. 6:159. Stamatakis, A. (2014). RAxML version 8: a tool for phylogenetic analysis and doi: 10.3389/fpls.2015.00159 post-analysis of large phylogenies. Bioinformatics 30, 1312–1313. doi: 10.1093/ Zheng, H., Huang, B., Zhang, B., Xiang, Y., Du, Z., Xu, Q., et al. (2016). bioinformatics/btu033 Resetting epigenetic memory by reprogramming of histone modifica- Stangeland, B., and Salehian, Z. (2002). An improved clearing method for GUS tions in mammals. Mol. Cell 63, 1066–1079. doi: 10.1016/j.molcel.2016. assay in Arabidopsis endosperm and seeds. Plant Mol. Biol. Rep. 20, 107–114. 08.032 doi: 10.1007/bf02799426 Zhou, X., and Ma, H. (2008). Evolutionary history of histone demethylase families: Stavropoulos, P., Blobel, G., and Hoelz, A. (2006). Crystal structure and mechanism distinct evolutionary patterns suggest functional divergence. BMC Evol. Biol. of human lysine-specific demethylase-1. Nat. Struct. Mol. Biol. 13, 626–632. 8:294. doi: 10.1186/1471-2148-8-294 doi: 10.1038/nsmb1113 Tang, M., Shen, H., Jin, Y., Lin, T., Cai, Q., Pinard, M. A., et al. (2013). The Conflict of Interest Statement: The authors declare that the research was malignant brain tumor (MBT) domain protein SFMBT1 is an integral histone conducted in the absence of any commercial or financial relationships that could reader subunit of the LSD1 demethylase complex for chromatin association be construed as a potential conflict of interest. and epithelial-to-mesenchymal transition. J. Biol. Chem. 288, 27680–27691. doi: 10.1074/jbc.M113.482349 Copyright © 2019 Martignago, Bernardini, Polticelli, Salvi, Cona, Angelini and Tsukada, Y., Fang, J., Erdjument-Bromage, H., Warren, M. E., Borchers, C. H., Tavladoraki. This is an open-access article distributed under the terms of the Creative Tempst, P., et al. (2005). Histone demethylation by a family of JmjC domain- Commons Attribution License (CC BY). The use, distribution or reproduction in containing proteins. Nature 439, 811–816. doi: 10.1038/nature04433 other forums is permitted, provided the original author(s) and the copyright owner(s) Untergasser, A., Nijveen, H., Rao, X., Bisseling, T., Geurts, R., and Leunissen, J. A. are credited and that the original publication in this journal is cited, in accordance (2007). Primer3Plus, an enhanced web interface to Primer3. Nucleic Acids Res. with accepted academic practice. No use, distribution or reproduction is permitted 35, W71–W74. doi: 10.1093/nar/gkm306 which does not comply with these terms. Frontiers in Plant Science | www.frontiersin.org 15 June 2019 | Volume 10 | Article 669