Long noncoding RNAs shape transcription in plants

ABSTRACT

The advent of novel high-throughput sequencing techniques has revealed that eukaryotic genomes are massively transcribed although only a small fraction of RNAs exhibits protein-coding capacity. In the last years, long noncoding RNAs (lncRNAs) have emerged as regulators of eukaryotic gene expression in a wide range of molecular mechanisms. Plant lncRNAs can be transcribed by alternative RNA polymerases, acting directly as long transcripts or can be processed into active small RNAs. Several lncRNAs have been recently shown to interact with chromatin, DNA or nuclear proteins to condition the epigenetic environment of target genes or modulate the activity of transcriptional complexes. In this review, we will summarize the recent discoveries about the actions of plant lncRNAs in the regulation of gene expression at the transcriptional level.

Introduction

More than 50 years have passed since the C-value paradox was first presented, revealing an imbalance between organismal complexity and DNA content in a wide range of species [1,2]. More recently, next-generation sequencing (NGS) has served to uncover that up to 90% of eukaryotic genomes is transcribed into RNAs, whereas only ~2% of some genomes is translated into proteins [3–5]. This breakthrough provided exciting perspectives to understand how the so-called “dark matter” of the genome may impact gene expression. Indeed, there is a highly complex and dynamic transcriptional landscape in eukaryotes. Eukaryotic genomes are pervasively transcribed and give rise to both protein-coding mRNAs and a large variety of noncoding RNAs (ncRNAs) [6–9]. Among them, long ncRNAs (lncRNAs; longer that 200 bases and showing poor coding capacity) can act directly as long transcripts or can be processed into a rich diversity of functional small RNAs. In plants, thousands of lncRNAs have been annotated in at least 40 species [10,11]. According to their genomic position in relation to protein-coding genes, lncRNAs can be classified as antisense, intronic, intergenic, or promoter transcripts [3]. Initially, lncRNAs were believed to be nonfunctional or “junk” RNAs. However, several reports have proved that they exert an important role in a wide range of biological processes including development, hormone homeostasis, response to biotic and abiotic restraints and light perception [3].

Plant lncRNAs are generally transcribed by RNA polymerase II (Pol II) or alternatively by the plant-specific RNA polymerases Pol IV and Pol V, leading to transcriptional gene silencing [12]. Although Pol IV and Pol V transcription is generally linked to the production of 24 nucleotide small interfering RNAs (24nt siRNAs) and the RNA-directed DNA methylation (RdDM) pathway [for review, see 13], Pol V transcripts have been also related to a broader spectrum of molecular mechanisms in the last few years. For instance, Pol V lncRNAs were shown to recruit components of the silencing machinery to promoter regions of protein-coding genes [14], to delimit Pol II transcriptional read-though [15], or to determine heterochromatin boundaries [16]. Plant Pol II lncRNAs have been involved in a varied assortment of molecular regulatory mechanisms impacting on gene expression, including chromatin remodeling and three-dimensional (3D) conformation dynamics, protein trafficking, modulation of alternative splicing, fine-tuning of miRNA activity, and control of mRNA translation or accumulation [3]. In this review we integrate the current knowledge on the lncRNA-mediated regulatory mechanisms modulating gene transcriptional activity in plants.

Transcriptional traffic jams: long noncoding RNAs modulating the transcriptional machinery

Pol II transcription across eukaryotic genomes not only gives birth to pre-mRNAs, but also results in the production of countless lncRNAs [3,17]. The coding and noncoding genomes integrate internal and external signals, shaping the dynamic outcoming transcriptome. Noncoding transcripts have been linked to the modulation of transcriptional activity at different levels contributing to accurate transcription traffic (understood as the harmonic activity of all transcriptional units in a given genomic region), although sometimes RNA Pol collisions may occur, for example during simultaneous transcription of sense and antisense transcripts. Several lncRNAs were shown to be recognized by proteins modulating the activity of the transcriptional machinery [18,19]. Transcription factors (TFs) typically contain effector domains that are separated from their DNA binding domains and interact with transcriptional activators or repressors. Multiple TFs may bind to different subunits of the so-called Mediator complex, which will be in charge of communicating regulatory signals from TFs directly to the Pol II enzyme. The precise mechanisms by which the Mediator complex regulates Pol II activity remain largely unknown, although it has been linked to multiple transcriptional events including transcription initiation, transcript elongation, changes in chromatin architecture and enhancer-promoter gene looping [20].

In Arabidopsis, the lncRNA ELF18-INDUCED LONG-NONCODING RNA1 (ELENA1) was shown to interact in vitro and in vivo with Mediator subunit 19a (MED19a) [21]. ELENA1 is transcriptionally accumulated in response to the pathogen-associated molecular patterns (PAMPs) flg22 and elf18, derived from bacterial flagellin and the translation elongation factor Tu (EF-Tu), respectively. ELENA1 enhances plant resistance against bacterial pathogens by activating the transcription of the PATHOGENESIS-RELATED 1 (PR1) gene. It was shown that ELENA1 can interact with the promoter region of PR1, promoting MED19a enrichment over the target region. MED19a regulation over the PR1 locus is dependent on ELENA1 accumulation (Figure 1(a); ²¹).

Figure 1.

Long noncoding RNAs modulate the plant transcriptional machinery

(a). The lncRNA ELENA1 recruits the Mediator complex to the PR1 gene promoter region. The Med19a subunit of Mediator directly interacts with ELENA1, activating PR1 transcription in response to ELF18.(b). The lncRNA SVALKA locus is subjected to transcriptional Pol II read-through in response to prolonged exposure to cold, triggering the transcription of the exosome-sensitive lncRNA asCBF1 upon poly-A site cleavage of SVALKA. asCBF1 transcription causes Pol II head-to-head collision with the CBF1 locus, giving place to a stalled CBF1 mRNA.

In contrast to ELENA1, the promoter-associated lncRNA HIDDEN TREASURE 1 (HID1) was characterized as a negative regulator of gene transcription [22]. HID1 is a highly conserved transcript of 236nt in length, regulating photomorphogenesis in Arabidopsis. HID1 interacts with chromatin in the region included in the first intron of the 5ʹ UTR of its target gene PHYTOCHROME-INTERACTING FACTOR 3 (PIF3), acting as a transcriptional repressor. Knocking down of HID1 in Arabidopsis led to increased levels of PIF3 mRNA and an elongated hypocotyl phenotype in response to far red, in agreement with PIF3 function. Based on gel filtration analyses, it was proposed that HID1 participates in big chromatin-related ribonucleoprotein complexes, although the identity of these protein partners remains uncertain [22].

The lncRNA SVALKA (meaning “cool” in Russian) was found to be induced by low temperatures [23]. Early events in the Arabidopsis cold response include rapid transcriptional induction of genes encoding the C-repeat/dehydration-responsive element Binding Factors (CBFs). CBFs are highly conserved TFs that promote cold tolerance in many plant species and their encoding genes are often arranged in a single cluster [24]. Interestingly, two lncRNA loci were identified in the Arabidopsis CBF cluster. The first one is SVALKA, which is transcribed on the antisense strand between CBF3 and CBF1, as an independent transcriptional unit. CBF1 mRNA dynamic accumulation in response to cold treatment is dependent on the negative role of SVALKA. The second one is asCBF1, a lncRNA transcribed downstream of SVALKA and overlapping the end of the CBF1 locus. This cryptic noncoding transcript depends entirely on SVALKA Pol II read-through and is only detectable in exosome mutants, hinting at a very fast asCBF1 RNA turnover. It was proposed that under cold treatment, SVALKA triggers asCBF1 transcription, which overlaps the CBF1 locus, causing Pol II transcriptional collision (Figure 1(b), ²³).

It was shown in yeast that head-to-head Pol II collision results in the enzyme stopping, and the removal of collided Pol II from the DNA template occurs via ubiquitylation-directed proteolysis [25]. Read-through transcription of the lncRNA CUT60 was also shown to modulate the transcriptional activity of a neighboring gene encoding a mitochondrial-located protein, hinting at a lncRNA-mediated mechanism regulating transcriptional termination of protein-coding genes [26].

Further research will be needed to determine to what extent the transcriptional activity of the plant genomes is generally fine-tuned by Pol II collision of coding vs. noncoding convergent transcriptional units. Interestingly, Pol IV and Pol V activity was linked to Pol II termination. Nuclear run-on assays revealed that in pol IV or pol V mutants, Pol II occupancy downstream of poly(A) sites increased for approximately 12% of protein-coding genes. Moreover, the role of Pol IV and Pol V in limiting Pol II read-through was proposed to be independent of siRNA biogenesis or cytosine methylation for the majority of target genes [15].

Repression vs. expression: long noncoding RNAs mediating histone modifications

Polycomb Group (PcG) proteins form an epigenetic memory system that is conserved in plants and animals and controls gene expression during development. Polycomb Repressive Complex 2 (PRC2) is linked to the deposition of H3K27me3 and to transcriptional repression in animals and plants. H3K27me3 then assists to recruit PRC1 in animals and the PRC1-like components LIKE HETEROCHROMATIN PROTEIN-1 (LHP1) and AtRING1 in plants [27]. In Arabidopsis, LHP1 recognizes PRC2-mediated H3K27me3 deposition and participates in the subsequent spreading of the repressive mark [28]. LHP1 belongs to the family of chromodomain-containing proteins. Certain members of this family from the animal kingdom recognize RNA in vivo [29,30]. LHP1 was shown to bind to RNAs in vitro[31] and the lncRNA APOLO in vivo [32]. It was proposed that APOLO over-accumulation can decoy LHP1 away from multiple common targets across the Arabidopsis genome, impacting H3K27me3 deposition and chromatin conformation [33]. PRC2 components were also shown to interact with lncRNAs in animals and plants. Several examples in animals suggest that PRC2 identifies site-specific target genes via regulatory RNAs, thus pointing to a functional general interplay between lncRNAs and chromatin modifiers [34,35]. However, it was also suggested that PRC2 can promiscuously bind to RNA in vivo and in vitro [36,37]. More recently, it was concluded that promiscuous and specific RNA-binding activities of PRC2 are not mutually exclusive in vitro, whereas binding specificity in vivo remains unclear [38]. In plants, fewer examples of PRC2-interacting lncRNAs exist, although they may be considered just as the tip of the iceberg.

In Arabidopsis, AGAMOUS (AG) encodes a MADS TF involved in the specification of stamens and carpels in flowers. AG tissue-specific expression largely depends on the region encompassed in the second intron of the gene, including sequences targeted by several positive and negative regulators [39–43]. AG second intron was also found to be repressed by PRC2 in vegetative tissues [44]. Interestingly, it was recently shown that AG second intron encodes several noncoding RNAs. Among them, AGAMOUS INTRONIC RNA 4 (AG-incRNA4) is directly recognized in vitro and in vivo by the PRC2 component CURLYLEAF (CLF), thereby recruiting the complex to repress AG transcription. The absence of CLF or AG-incRNA4 results in the de-repression of AG and a reduction of H3K27me3 marks, suggesting an AG-incRNA4-dependent PRC2 repression. In addition, it was shown that AG-incRNA4 also recruits PRC2 to the promoter region of AG-incRNA4 itself, within the AG intron, down-regulating its own expression in a negative feedback loop (Figure 2(a); ⁴⁵).

Figure 2.

Long noncoding RNAs recruit Polycomb repressive complex 2 to target loci

(a). The intronic lncRNA AG-incRNA4 is responsible for the recruitment of PRC2 to the intron 2 of the AGAMOUS locus by direct interaction with CLF, modulating the deposition of the repressive histone modification H3K27me3 in a tissue-specific manner. CLF-dependent H3K27me3 represses AGAMOUS in leaves and at the same time impairs AG-incRNA4 transcription in a negative feedback loop.(b). The vernalization-triggered epigenetic silencing of FLOWERING LOCUS C (FLC) is tightly regulated by 3 lncRNAs and dynamic chromatin 3D conformation. Early FLC transcriptional accumulation correlates with a gene looping encompassing the full locus, likely responsible for Pol II recycling and COOLAIR distal isoform transcription. In response to cold, the gene looping is opened, COOLAIR proximal isoform is preferred and the intronic lncRNA COLDAIR begins to be transcribed, likely recruiting PRC2 for H3K27me3 deposition. Long exposure to cold also induces the transcription of the promoter lncRNA COLDWRAP. COLDWRAP recruits PRC2 upstream FLC for histone modifications, setting the basis for the formation of an intragenic chromatin loop until the first intron of FLC. As a result, Pol II processivity is impaired and FLC is epigenetically shut off.

Another key regulatory gene encoding a MADS TF in Arabidopsis is FLOWERING LOCUS C (FLC), a developmental regulator that controls the switch from vegetative to reproductive developmental stages. FLC represses flowering and its transcriptional accumulation quantitatively correlates with flowering time [46,47]. Remarkably, FLC transcriptional regulation involves the activity of three lncRNAs. A set of antisense transcripts, collectively named as COOLAIR, fully encompasses the FLC locus [48–50]. COOLAIR transcripts are polyadenylated at multiple sites with proximal polyadenylation being promoted by alternative developmental pathways, including vernalization, e.g. the prolonged exposure to cold epigenetically repressing FLC [51]. The use of the COOLAIR proximal poly(A) site results in down-regulation of FLC expression in a process requiring FLOWRING LOCUS D (FLD), an H3K4me2 demethylase [49]. FLD activity results in H3K4me2 demethylation in the gene body of FLC and in its transcriptional repression [49,52].

An additional lncRNA coming from the FLC locus was identified [53]. COLD ASSISTED INTRONIC NONCODING RNA (COLDAIR) is fully encoded in the sense strand of the first intron of FLC. Similar to COOLAIR, its accumulation oscillates in response to low temperatures. However, unlike COOLAIR, it was proposed that COLDAIR physically associates with the PRC2 component CLF to target FLC. Thus, COLDAIR would be required for establishing stable repressive chromatin at FLC through its interaction with PRC2 [53].

More recently, a third lncRNA modulating FLC transcription was identified [54]. COLDWRAP lncRNA is derived from the FLC proximal promoter, its induction by cold occurs later than COOLAIR and COLDAIR, and it physically interacts with the PRC2 complex. Strikingly, COLDWRAP functions with the lncRNA COLDAIR to retain Polycomb at the FLC promoter through the formation of a repressive intragenic chromatin loop. The COLDWRAP locus is spatially brought together with FLC first intron blocking Pol II transcription. It was proposed that FLC vernalization-triggered Polycomb silencing is coordinated by multiple lncRNAs in a cooperative manner to form a stable repressive chromatin structure, including a 3D configuration switch of the locus (Figure 2(b), ⁵⁴).

Three dimensions expand linear two dimension regulations: long noncoding RNAs shaping 3D chromatin conformation

The nucleus of each cell is a highly organized arrangement of DNA, RNAs and proteins dynamically assembled and regulated in different cellular states [55,56]. The 3D organization of the eukaryotic nucleus has emerged as an important feature in the complex network of mechanisms behind gene activity and genome connectivity dynamics, which can be evidenced in the regionalized chromosomal spatial distribution and the clustering of diverse genomic regions with similar expression patterns [57]. The comparison of genome topology among plant model species suggests that chromatin organization in plants might be more diverse than in multicellular animals. Topologically associated domains (TADs), which appear to be a prevalent structural feature of genome packing in many animal species [58], are not prominent in A. thaliana exhibiting an extremely compact genome. Differently, it was shown that a quarter of the rice genome is covered by thousands of distinct TAD-like structures. Rice TADs boundaries are associated with euchromatic epigenetic marks and active gene expression [59]. A comparative analysis of the epigenome of crops with genome sizes ranging from 0.4 to 2.4 Gb, suggested that less compact plant genomes can be divided into mammalian-like A/B compartments. However, unlike mammalian short-range chromatin loops that are enriched at the TAD border, plant chromatin loops often link gene islands outside the repressive domains and are closely associated with active compartments [60], further hinting at commonalities and singularities across kingdoms ⁷⁹.

In the last few years, many lncRNAs have been implicated in shaping 3D nuclear organization [61]. Interestingly, RNase-A microinjection into the nucleus followed by microscopic observation served to show that long nuclear-retained RNAs maintained higher order chromatin in an open configuration, whereas the structure of heterochromatin domains showed a reduced dependence on RNA [62,63]. More recently, genome-wide chromatin conformation capture (Hi-C) analyses were performed in mammalian cells exposed or not to RNase, before and after crosslinking, or using exogenous transcriptional inhibitors [64]. Interestingly, this molecular approach revealed that TAD boundaries are largely unaffected by RNase treatment, although a subtle disruption of compartmental interactions was observed. In contrast, transcriptional inhibition led to weaker TAD boundaries, hinting at substantial differences in the relative contribution of RNA abundance vs. active transcription in nuclear organization [64].

In plants, few examples of lncRNAs participating in genome topology determination have appeared in the last years. As mentioned before, the Arabidopsis lncRNA COLDWRAP can modulate the chromatin conformation of its own region, forming an intragenic chromatin loop repressing the transcription of its neighboring gene FLC (Figure 2(b); ⁵⁴). To this end, COLDWRAP recruits the PRC2 machinery to establish the histone modifications needed to anchor the chromatin loop (Figure 2(b)).

Another well documented case of a lncRNA and RdDM-associated chromatin loop formation involves the action of APOLO in Arabidopsis [32]. This lncRNA is encoded 5 kb upstream of the gene PID, a key regulator of polar auxin transport, and is transcribed by RNA Pol II and Pol V. Dual transcription of APOLO regulates the formation of a chromatin loop encompassing the promoter of its neighboring gene PID. Exogenous auxin treatment results in active DNA demethylation of the APOLO locus together with a rapid opening of the loop, leaving the intergenic region available for TFs recognition. Thus, Pol II divergent transcription of the APOLO and PID loci begins. PID transcripts are shuttled to the cytoplasm for translation, whereas APOLO RNA is accumulated in the nucleus. Gradually, APOLO Pol II transcripts recruit PRC1 components for loop re-formation, as well as Pol V to the APOLO locus to trigger RdDM essential for loop stabilization. As a result, the chromatin loop starts to be steadily re-conformed and the basal chromatin conformation is gradually restored, ensuring down-regulation of PID and APOLO transcript levels. PRC2 components additionally contribute to the accurate closing of the chromatin loop. Thereby, APOLO-mediated dynamic local chromatin conformation determines PID expression pattern (Figure 3(b), ^{32, 3}).

Figure 3.

Long noncoding RNAs shape chromatin 3D conformation modulating transcriptional activity

(a). An intergenic chromatin loop includes the promoter region of the lncRNA APOLO and its neighbor gene PID. In response to auxin, the chromatin loop is opened and bidirectional transcription starts. APOLO recognizes by sequence complementarity and R-loop formation the PID locus in cis and a subset of auxin-responsive genes in trans, decoying the plant PRC1 component LHP1 away from target loci and opening local chromatin loops. As a result, APOLO co-regulates spatially independent loci across the Arabidopsis genome, modulating auxin transport and lateral root development.(b). The inverted repeat-derived lncRNA ncRNAW6 is encoded in the proximal promoter of the WRKY6 gene in Sunflower. In cotyledons, ncRNAW6 transcription generates the accumulation of 24nt siRNAs, triggering local DNA methylation and the formation of a gene looping encompassing the whole WRKY6 locus. As a result, WRKY6 transcription is enhanced thanks to Pol II recycling, whereas the bidirectionality of the promoter region is blocked and ncRNAW6 transcription is impaired. Gradually, the chromatin loop is re-opened and ncRNAW6 is increasingly accumulated, starting the cycle again. On the other hand, an alternative intragenic chromatin loop is formed in leaves, allowing bidirectionality of the promoter, maintaining ncRNAW6 steady-state and blocking Pol II transcription before the end of the WRKY6 locus.

More recently, APOLO Chromatin Isolation by RNA Purification followed by DNA sequencing (APOLO-ChIRP-Seq) served to identify a subset of APOLO targets in trans across the Arabidopsis genome [33]. Strikingly, it was shown that APOLO can coordinate the expression of multiple genes through sequence complementarity and DNA-RNA hybrid formation, referred to as R-loops. Upon target recognition, APOLO can decoy LHP1 away from target loci and modulate local 3D chromatin conformation to fine-tune gene transcription. Therefore, APOLO exerts a regulatory role in cis and in trans, controlling the expression of a subset of auxin-related genes during lateral root development (Figure 3(a), ³³).

In Sunflower, a dual gene-looping event was found to modulate the transcriptional activity of the HaWRKY6 locus encoding a TF of the WRKY family, in a tissue-specific manner [65]. In cotyledons, a chromatin loop encompassing the entire HaWRKY6 locus from the promoter to the transcription termination site, allows a more efficient usage of Pol II increasing HaWRKY6 levels. On the other hand, the formation of a second loop in leaves blocks the progression of transcription. Indeed, the leaf-specific loop shares the same 5ʹ anchor point in the promoter region but the interacting point is located in the fourth intron of the gene, blocking Pol II elongation. Interestingly, the region located in the HaWRKY6 promoter and serving as the 5ʹ anchor of the two alternative chromatin loops encodes a lncRNA. This lncRNA corresponds to a miniature inverted-repeat transposable element (MITE) transcribed and processed into 24nt small RNAs by a non-canonical autonomous RdDM pathway. Remarkably, the modulation of each chromatin loop formation depends also on 24nt accumulation and DNA methylation of the 3ʹ anchor points. The formation of the HaWRKY6 gene loop forces promoter unidirectionality, enhancing the transcription of the coding sequence but blocking the divergent transcription of the lncRNA. The eventual impairment of the divergent transcription governs the locus transcriptional dynamics, as the repressed lncRNA is required for establishment of the loop [65]. Thus, the locus reaches a self-buffered equilibrium where the formation of the loop enhances gene transcription but represses the lncRNA, progressively promoting the loop disruption (Figure 3(b)).

Squaring the circle: circRNAs meet chromatin

Circular RNAs (circRNAs) are abundant and evolutionarily conserved RNAs of largely unknown function [66]. They are covalently closed circular molecules of single-stranded RNA, resulting from a noncanonical splicing event, the so-called back-splicing. This event consists in the ligation of a downstream splice donor site reversely with an upstream splice acceptor site from the pre-mRNA, generating a circular and highly stable RNA molecule [67]. In animals, a subset of circRNAs were shown to be associated to polysomes and translated into proteins [66]. Circular lncRNAs may efficiently compete with the linear pre-mRNA for the recognition of related splicing protein complexes, or can act as miRNA sponges [68].

In plants, the accumulation of circRNAs has been linked to the response to stress [69]. Transcriptomic approaches covering a wide range of environmental constraints including oxidative stress, drought and nutrient deficiency have revealed differential expression of circRNAs. Furthermore, it was observed that in most cases circRNA abundance does not correlate with expression of their associated mRNA, suggesting a functional role of circRNAs in environmental and stress response [70]. For example, over 60 circRNAs were shown to exhibit differential expression under water deficit, which were linked to photosynthesis and hormone signaling in wheat [71]. Moreover, nutrient depletion such as phosphate, iron, and zinc were shown to modulate the accumulation of specific circRNAs in rice and barley [69,72]. Although the transcriptional behavior of circRNAs in plants hints at a role in development and the response to environmental stimuli, the underlying molecular mechanisms remain largely unknown.

In Arabidopsis, it was demonstrated that a circRNA modulates the alternative splicing of its own parent gene by directly interacting with DNA to form an R-loop [73]. The accumulation of the circRNA from exon 6 of the SEPALLATA 3 (SEP3) gene promotes the processing of the nascent transcript into the naturally-occurring SEP3.3 isoform, which consists of the exon 6-skipped transcript (Figure 4). SEP3 belongs to the MADS-box TF family and is involved in flower development in Arabidopsis. Altered SEP3 splicing results in homeotic phenotypes in the flower. Accordingly, the lines with higher levels of this circRNA produced flowers with altered floral organ number, e.g. fewer stamen and additional petals. The first genome-wide characterization of R-loops performed in Arabidopsis revealed that DNA-RNA duplexes are a common feature in this model genome [74]. More recently, a comprehensive analysis of R-loop distribution across the Arabidopsis genome during development and in response to a wide variety of stimuli indicated that DNA-RNA duplexes are fairly stable and do not depend on RNA abundance [75]. However, an exhaustive analysis linking R-loop formation, circRNAs and alternative splicing in plants is still missing to determine to what extent the observations of the SEP3 locus constitute a general regulatory mechanism [76].

Figure 4.

Circular RNAs directly interacting with DNA and modulating alternative splicing

The exon 6 of the SEP3 locus is transcribed and back-spliced into a circular RNA. The circRNA can directly interact with its parent gene DNA, forming a DNA–RNA duplex known as R-loop. As a result, skipping of SEP3 exon 6 is promoted, favoring the accumulation of the SEP3.3 mRNA isoform.

Perspectives

Growing evidence supports the role of plant lncRNAs in virtually every step of gene expression regulation, including transcriptional and post-transcriptional regulations in the nucleus, discussed here. The access to new plant genomes and to transcriptomic datasets coming from tissue-specific samples and under different conditions has revealed the high transcriptional complexity of the genome. Furthermore, the noncoding genome is emerging as a major factor in natural variation [77]. A deeper knowledge about lncRNA activity over protein-coding genes will likely shed light on the adaptation of plants to their environment, allowing new strategies for agriculture.

Functional characterization of plant lncRNAs is still in its infancy. Genome-wide identification of Mediator and Polycomb associated lncRNAs, together with a better understanding of the chromatin-related ribonucleoprotein complexes integrated by noncoding transcripts will help us to classify them and decipher particular profiles linked to their function. As shown for the LHP1-interacting lncRNA APOLO, noncoding transcripts may modulate the activity of their neighbor genes in cis by dynamically shaping 3D chromatin conformation [32]. However, other auxin-responsive genes are regulated in trans across the Arabidopsis genome through the formation of lncRNA-DNA R-loops [33]. It is reasonable to speculate that others lncRNAs also exhibit the ability to coordinate the expression of multiple genes at long distance. Further work will be needed to reveal how extensive the role of characterized lncRNAs such as ELENA1 [21] and HID1 [22] may act over other genes in the response to pathogens or light, respectively. Additionally, the identification of further lncRNA nuclear protein partners will certainly contribute to give mechanistic insights in chromatin lncRNA regulations. Recent findings in animals revealed that chromatin is densely occupied by RNA binding proteins (RBPs), especially at promoter regions [78]. The characterization of novel RBP-lncRNA ribonucleoprotein complexes controlling transcription will help to uncover potential pervasive roles for lncRNAs in gene expression.

Similarly, the integration of genetic deletions through e.g. CRISPR/Cas9 techniques [78] and chromatin genome-wide approaches will likely help to determine how general are the mechanisms described for lncRNA-mediated Pol II collision [23], alternative splicing [73], R-loop formation [33], or other reported post-transcriptional events regulating gene expression. One certitude about the plant “dark matter” of the genome, is that plenty of hidden secrets will come to light in the upcoming years.

Funding Statement

This work was supported by the Agence Nationale de la Recherche [ANR-17-EUR-0007]; Agencia Nacional de Promoción Científica y Tecnológica [PICT]; CNRS (Laboratoire International Associé NOCOSYM).

Disclosure statement

No potential conflict of interest was reported by the authors.