The plant noncoding transcriptome: a versatile environmental sensor

Abstract

Plant noncoding RNA transcripts have gained increasing attention in recent years due to growing evidence that they can regulate developmental plasticity. In this review article, we comprehensively analyze the relationship between noncoding RNA transcripts in plants and their response to environmental cues. We first provide an overview of the various noncoding transcript types, including long and small RNAs, and how the environment modulates their performance. We then highlight the importance of noncoding RNA secondary structure for their molecular and biological functions. Finally, we discuss recent studies that have unveiled the functional significance of specific long noncoding transcripts and their molecular partners within ribonucleoprotein complexes during development and in response to biotic and abiotic stress. Overall, this review sheds light on the fascinating and complex relationship between dynamic noncoding transcription and plant environmental responses, and highlights the need for further research to uncover the underlying molecular mechanisms and exploit the potential of noncoding transcripts for crop resilience in the context of global warming.

This review discusses how various noncoding RNAs influence plant interactions with the environment.

Variation of plant noncoding genomes

RNA molecules play vital roles in all extant forms of life. They serve as templates for protein synthesis (mRNA), but also exert regulatory effects on DNA (by transcriptional gene silencing (TGS) or mediating epigenetic modifications), other RNAs (by post‐transcriptional gene silencing (PTGS) or splicing), or proteins (by translation inhibition and ribosome assembly). Studies over the last few decades have pointed to the presence of large amounts of transcribed RNA not encoding proteins (ENCODE Project Consortium, 2012). Some noncoding transcripts were found to derive from mRNA splicing, whereas many correspond to RNA genes comprising the translation machinery (i.e., tRNA and rRNA); yet the vast majority was first undefined, and later classified as potential regulatory RNAs. Currently, it is estimated that 97% of the human genome is transcribed, and ~2% of this fraction contains protein‐coding exons (ENCODE Project Consortium, 2012). The percentage of exon protein‐coding genes relative to total genome size varies across species. For example, considering the genome composition, the plant model Arabidopsis thaliana exhibits a genome in which around 25% is dedicated to encoding proteins, specifically exons. This proportion is significantly higher not only when compared to humans but also in comparison to crops such as Triticum aestivum (wheat), Zea mays (maize), and the closely related Arabidopsis lyrata (Hu et al, 2011; Fig 1A). This observation underscores the need to expand the studies of the noncoding genome in a wider range of plant species, since conclusions drawn from studies about the relatively compact genome of Arabidopsis do not necessarily reflect the behavior of other organisms. The percentage distribution of noncoding RNAs (ncRNAs) in the Arabidopsis genome is mainly represented by long ncRNAs (lncRNAs, Budak et al, 2020; Fig 1B). The biological relevance of a growing number of characterized lncRNAs, hints at the noncoding genome as a versatile feature in gene expression regulation at different levels. In essence, this review not only explores the intricate interplay between dynamic noncoding transcription and plant responses to the environment but also underscores the imperative for continued research to unveil the underlying molecular mechanisms, offering a promising avenue for harnessing the potential of noncoding transcripts to enhance crop resilience within the context of a changing climate.

Figure 1. RNA classes and their distribution in the genome.

(A) Dot plot of the percentage of the exons of protein‐coding genes relative to total genome size in different species. (B) Dot plot graph showing the percentage of each RNA class to the total genome size of Arabidopsis thaliana. The x‐axis displays different classes of RNAs, including: Intergenic + intronic regions, TE (transposable elements), UTR's and ncRNAs (noncoding RNAs). The dot plot (upper right) illustrates the percentage distribution of ncRNAs in the A. thaliana genome. The data for the different species were obtained from Ensembl. Plant data were obtained from Ensembl Plants Release 56. Data for C. albicans and S. cerevisiae were obtained from Ensembl Fungi release 56. And the Vertebrate species data was obtained from Ensembl Release 109. Once the data from the GTF annotation files were obtained from the respective sources, it was processed to extract and analyze the relevant information for generating the plot.

Regulatory noncoding RNAs in plants

Plant regulatory ncRNA can be divided into two main groups: small noncoding RNA (sRNA; 20–24 nucleotides (nt)) and lncRNA (>200 nt). In plants, sRNAs regulate growth and reproduction, control transposon activity, and participate in adaptive responses to biotic and abiotic stresses. Plant sRNAs can be subdivided into different groups, among which the two main classes are 20–22 nt microRNAs (miRNAs) and the 21–24 nt small interfering RNAs (siRNAs). Endogenous loci transcription generates double‐strand (ds)RNA precursors which can be processed into active miRNAs, capable of regulating target mRNAs by PTGS. Alternatively, exogenous or endogenous RNA‐dependent RNA Polymerase‐produced dsRNA can generate siRNAs, which mediate PTGS (e.g., trans‐acting siRNA (tasiRNA)) or RNA directed DNA methylation (RdDM), which results in TGS (mediated by 24 nt siRNAs; Wendte & Pikaard, 2017; Zhan & Meyers, 2023). In addition to their roles in gene silencing, emerging evidence also indicates that sRNA can regulate DNA repair and pre‐mRNA splicing (Wang & Chekanova, 2016).

Transfer RNA (tRNA) undergoes extensive modifications that affect its structure and stability. These environmentally responsive modifications, mediated by specific enzymes, contribute to various physiological and pathological processes (Orellana et al, 2022). Recent studies on plant tRNA biology shed light on the regulation of tRNA biogenesis in plants and its impact on plant growth and development (Johnson et al, 2016; Soprano et al, 2018). Mature tRNAs undergo processing to produce tRNA‐derived fragments (tRFs), which play vital roles in bacteria, animals, and plants. In plants, tRFs do not adopt the tRNA structure, but instead function as sRNAs and participate in various stress responses (Ma et al, 2021). However, the mechanisms by which stress affects tRF levels remain mostly unclear (Alves & Nogueira, 2021). Remarkably, it was shown that the regulation of genes by tRFs can occur across kingdoms, including the symbiotic relationship between legumes and nitrogen‐fixative bacteria (Assmann et al, 2023).

In addition to sRNA, lncRNAs have recently emerged as essential components of the eukaryotic transcriptome (Rinn & Chang, 2020). The analysis of genomes across several plant species has revealed the presence of numerous lncRNAs (Lucero et al, 2021). In plants, lncRNAs are transcribed by RNA Polymerase II (RNAPII). Alternatively, shorter transcripts can be produced by RNAPIV and V, which are unique to plants and lead to RdDM and TGS. LncRNAs can be classified into several types based on their location and characteristics: (i) natural antisense transcripts (lncNATs) are transcribed in the opposite direction of a protein‐coding gene, and overlap with at least one coding exon; (ii) intronic lncRNAs start within an intron in either direction and end without overlapping with any exon; (iii) promoter lncRNAs are transcripts that originate from the promoter region of a protein‐coding gene; and (iv) long intergenic ncRNAs (lincRNAs) are distinct transcriptional units located at least 1 kb away from neighboring protein‐coding genes (Ariel et al, 2015). Growing evidence suggests lncRNAs can regulate gene expression as long transcripts, without undergoing processing into active sRNAs. Plant lncRNAs mediate their crucial functions by various mechanisms, including chromatin structure modification, modulation of alternative splicing, fine‐tuning miRNA activity, and controlling mRNA translation or accumulation (Lucero et al, 2021). LncRNAs can adopt complex secondary structures, which tend to be more conserved than primary sequences and are thought to contribute to their molecular functions (Novikova et al, 2013). The regulation of lncRNA function is attributed mainly to changes in their structure, which can be influenced by a range of factors such as temperature, protein interactions, DNA and RNA binding, ionic interactions, post‐transcriptional modifications, and sequence mutations (Rouskin et al, 2014; Spitale et al, 2015; Lackey et al, 2018; Aydın et al, 2023).

The impact of ncRNAs can be assessed across various temporal scales. An instantaneous effect occurs in response to environmental cues, wherein the interaction of ncRNAs with partner molecules, their structure, or their half‐life is modified to integrate these signals. On this scale, transcriptional activity of the ncRNA locus may have an impact on the expression of nearby genes. Secondly, ncRNAs not only contribute to immediate environmental adaptation but also have the potential to influence pathways across generations, thereby supporting longer‐term adaptation to new habitats or climatic conditions. Finally, over evolutionary timescales, ncRNAs and other cellular pathways interact dynamically in response to environmental changes. This process may involve a gradual fine‐tuning of lncRNAs (including their sequence, structure, half‐life, and interaction partners) to enhance or modify their regulatory potential. Additionally, novel functions of ncRNAs may arise from pervasive transcripts or through mutation or neo‐functionalization of protein‐coding as well as transposon‐derived transcripts, driven by the selection pressure induced by environmental changes.

Noncoding RNA structure dynamics in response to environmental cues

Coding and noncoding RNA‐related functions (including RNA transcription, sRNA biogenesis, RNA splicing, protein translation, and RNA stability) can be affected by RNA secondary and tertiary structure produced by inter‐ and intramolecular base pairing, which can form hairpins, apical and internal loops, three‐way junctions, pseudo‐knots, riboswitches, among others (Fig 2A and B; Box 1). Formation of such structures strongly depends on the environment and consequently, these structures are highly dynamic and may refold in response to external signals such as temperature, pH, or the presence of specific molecules. For instance, an illustrative schematic representation of a structural analysis is shown in Fig 2C, demonstrating multiple individual RNA structures from the in vivo SHAPE mutational profile. Remarkably, this method was used to identify different structural conformations of the lncRNA COOLAIR in warm and cold conditions (Yang et al, 2022). It is important to note that in a living cell, the stability and conformation of RNA molecules can be influenced by a combination of endogenous and exogenous conditions, as mentioned above. Consequently, these factors can lead to a reduced number of conformations or even result in a single dominant conformation of the RNA molecule.

Figure 2. RNA structures motifs, riboswitches and structure ensembles.

(A) Scheme showing different RNA structure motifs and G‐quadruplexes. (B) A schematic representation of a riboswitch in Arabidopsis thaliana, with the TPP riboswitch aptamer domain in both the “ON” (left) and “OFF” (right) states. (C) Visual depiction of the output of a structural analysis method called DaVinci (Determination of the Variation of the RNA structure conformation through stochastic context‐free grammar) from Yang et al, 2022. A PCA plot representing the reconstruction of coexisting alternative RNA conformations. Each point corresponds to an individual RNA structure, and the three clusters correspond to alternative RNA conformations, each with a different proportion. This schematic representation was created for illustrative purposes to demonstrate the functionality of the DaVinci method. It does not correspond to any specific RNA molecule but serves as an example to showcase the ability of the method to reconstruct alternative RNA conformations.

Box 1. RNA structures motifs and riboswitches.

RNA molecules fold into complex secondary structures that play crucial roles in their functions. These structural motifs are essential for regulating gene expression and serve as versatile sensors in plants. While there are many other RNA structure motifs and riboswitches that play important roles in gene expression regulation and act as sensors in plants, some critical examples include:

‐ Hairpins are one of the most common RNA structure motifs, formed by a short stretch of nucleotides folding back on themselves to create a stem‐loop structure. In plants, functional RNA hairpins have been described as located in 5′ and 3′ UTR (Leppek et al, 2018). In addition, there are essential features on the pri‐miRNA hairpin that are essential for DICER‐LIKE1 (DCL1), recognition and consistent processing (Bologna et al, 2013a; Chorostecki et al, 2017; Axtell & Meyers, 2018).

‐ Terminal Loop are small loops at the ends of RNA hairpins. In Arabidopsis it was shown that pri‐miRNA terminal loops play also a crucial role in miRNA biogenesis, and their size, shape, and distance relative to miRNA:miRNA* duplexes can impact miRNA accumulation in vivo (Zhu et al, 2013).

‐ Bulges are formed when an unpaired nucleotide is within a double‐stranded RNA region. It was discovered that closing bulges directly beneath the loop‐distal DCL1 cleavage site has opposite effects on the efficiency of first and second DCL1 cleavages (Song et al, 2010).

‐ Multi‐branch Loops are complex RNA structures that contain three or more stems coming together at a single junction. Branched loops are important elements in miRNA biogenesis in plants. In Arabidopsis, the biogenesis of MIR157c is dependent on the recognition of this type of structure (Moro et al, 2018).

‐ Pseudoknots are complex RNA structures that form when different regions of an RNA molecule interact, resulting in overlapping base‐pairing interactions. The pseudoknot structure that induces the formation of the tRNA‐like structure (TLS) at the end of the genome of plant viruses (e.g., turnip yellow mosaic virus) is one of the best‐studied examples of RNA pseudoknots (Brierley et al, 2007).

‐ Three‐way Junctions are formed when three stems meet at a single point. In Arabidopsis, DYSKERIN was found to bind telomerase RNA (AtTR) with high affinity and specificity in vitro. This binding is facilitated by a plant‐specific three‐way junction structure (Song et al, 2021).

‐ Kissing Loops are formed when two bases between two hairpin loops pair. Recently, they have been described essential in enhancing the translation of plant RNA viruses through their interaction with 3′ CITEs and 5′ sequence elements (Urquidi Camacho et al, 2020).

‐ G‐quadruplexes are specialized RNA structures that form when guanine‐rich regions of an RNA molecule come together to create a stable four‐stranded structure. They have been identified in various plant species, and their presence and stability can depend on environmental cues such as light (Volná et al, 2021), temperature (Yang et al, 2020b), and stress (Mullen et al, 2010). Recently, it was determined through in vivo RNA chemical structure profiling that hundreds of rG4s structures exhibit significant folding in Arabidopsis and rice (Yang et al, 2020b).

‐ Riboswitches: Riboswitches are functional RNA elements usually located in the 5′ UTR of mRNA molecules. Different RNA structural switches with clear regulatory roles have been reported in different species (Spitale & Incarnato, 2023). Arabidopsis Thiamine pyrophosphate (TPP) riboswitch controls gene expression in response to intracellular levels of TPP. The structure of the TPP riboswitch changes between the ON and OFF states, which are characterized by the aptamer domain's switch helix being open or closed, respectively (Fig 3B).

Some plant lncRNAs are specifically regulated in response to abiotic stress, such as cold, heat, drought, flood, salt, and nutrient availability (Di et al, 2014; Ding et al, 2019; Pang et al, 2019; Baruah et al, 2021; Bakirbas & Walker, 2022). Interestingly, recent research has shown that some lncRNAs can adopt multiple structures at different temperatures (Ageeli et al, 2019; Chorostecki et al, 2021a), allowing them to respond to changes in the thermal environment. This suggests that lncRNA structure can be highly dynamic and responsive to environmental changes, allowing them to modulate gene expression integrating various stimuli. Moreover, given their rapid evolution and sensitivity to environmental changes, plant lncRNAs have been proposed as potential environmental sensors or effectors that can help plants adapt to changing conditions (Yu et al, 2019; Blein et al, 2020).

Temperature is one of the most direct physical parameters that influence RNA structure dynamics, and in turn, thermosensitive RNA structures can be used by a cell to perceive and respond to its temperature environment. Indeed, many thermosensitive RNA structures (RNA thermoswitches) with biological function have been identified in bacteria and viruses (Chowdhury et al, 2006; Thomas et al, 2022), suggesting the temperature responsiveness of RNA secondary structures has been harnessed multiple times during evolution. It was shown that temperature sensitivity of RNA structures is exploited by bacteria to fabricate RNA thermosensing systems that allow a rapid adaptive response to temperature change (Mandin & Johansson, 2020). In contrast to mammals, plants cannot regulate their body temperature, suggesting that warmth‐sensitive RNA structure may mediate rapid responses to the environment in these sessile organisms. Interestingly, recent discoveries revealed that RNA thermoswitches are also present in plants, where they influence RNA stability and pre‐mRNA splicing and translation efficiency in a temperature‐dependent manner (Chung et al, 2020). Moreover, several RNA chaperones have been shown to be involved in the temperature‐response in plants (Zhu et al, 2007). Similar effects were found in plants defective in RNA nucleo‐cytoplasmic transport (Gong et al, 2005). RNA processing has also been shown to be temperature dependent. While there is increasing evidence of temperature dependence in RNA splicing (Neumann et al, 2020), a temperature change that induced a minimal variation on the miRNA precursor structure was shown to be detrimental enough to completely impair the full function of the resulting miRNA (Kim et al, 2016). In contrast, the potential role of lncRNAs as thermosensors in higher organisms remains unexplored. Therefore, it is possible that lncRNAs may have a function as thermosensors in higher organisms, given their sensitivity to temperature changes and the potential impact on their stability and folding.

In addition to temperature, other environmental cues such as salt, magnesium, and osmotic stress have the potential to modulate base pairing, leading to RNA structural changes. For example, an osmotic stress‐responsive promoter upstream of transcripts in poplar revealed their increased stability and ability to induce target gene expression was mediated through secondary structure‐dependent interactions (Song et al, 2019b). Moreover, an analysis of RNA‐protein interactions demonstrated a substantial alteration in RNA secondary structure under salt stress conditions—suggesting a notable impact on RNA folding independent of overall changes in RNA‐protein interactions (Kramer et al, 2020). Further, the Arabidopsis riboswitch k‐junctions was shown to fold in response to the addition of magnesium or sodium ions (Song et al, 2019b; Li et al, 2023). Altogether, a high correlation exists between temperature/environment and RNA‐related processes.

Resolving the structure of RNA molecules is crucial for determining their function and responsiveness to environmental cues (Box 2; Table 1). Nevertheless, within a cell, multiple copies of the same RNA molecule can assume various conformations, each of them with distinct structural and functional properties (Fig 2C), boosting the versatility of a particular transcript. The regulatory functions of RNA structures rely on their ability to adjust the relative abundance of specific conformations within the ensemble in response to environmental cues (Ganser et al, 2019). Recent attempts have been made to develop computational methods to reconstruct alternative RNA conformations from data obtained through structure probing techniques. These methods discussed in a recent review (Aviran & Incarnato, 2022) are clearing the path towards investigating RNA structure ensembles, even within the living cell setting.

Box 2. RNA secondary structures prediction and experimental approaches.

RNA secondary structure prediction is a crucial tool for understanding the functional roles of noncoding RNAs in plants. Several computational algorithms exist for RNA secondary structure prediction, including RNAStructure (Reuter & Mathews, 2010), MFold (Zuker, 2003) and RNAfold (Lorenz et al, 2011). However, evaluations of these in silico methods demonstrate an average accuracy of less than 40% in secondary structure topology (Zhao et al, 2018b), indicating a significant need for improvement to achieve reliable results. This is especially problematic for long RNA molecules, as highlighted by a recent comparison of computational methods (Bugnon et al, 2022).

Due to the inherent limitations of being in silico, computational methods cannot fully capture RNA folding and dynamics. Hence, experimental approaches are necessary to directly measure RNA structure. Several experimental approaches have been developed in the last few years to investigate specific structural features of RNA molecules. These approaches offer insights into the structural state at single‐nucleotide resolution using different methods. The methods can be classified into two main categories: biochemical techniques and proximity ligation‐based methods. (i) Biochemical techniques involve enzymatic or chemical probes to probe RNA structure. Enzymatic‐based methods are effective for transcriptome‐wide RNA structure analysis but are typically used in vitro. Chemical‐based methods are widely used for in vivo RNA structure studies in plants, enabling gene‐specific and genome‐wide investigations (Table 1). (ii) Proximity ligation‐based methods involve crosslinking and detecting intra‐ and intermolecular interactions between RNA molecules (Table 1). Recently, several methods have been developed for deconvoluting RNA structure ensembles, which can provide insights into dynamic conformations specific to certain conditions (Table 1). Experimental approaches offer valuable insights into RNA structure and dynamics at single‐nucleotide resolution, complementing computational methods and advancing our understanding of the plant noncoding transcriptome as a versatile environmental sensor.

Table 1.

RNA structure probing methods for individual RNAs, genome‐wide scale and structure ensemble.

Method	Description	Application	Reference
PARS	Enzymatic‐based method for transcriptome‐wide RNA structure analysis	In vitro	Kertesz et al (2010)
nextPARS			Chorostecki et al (2021b)
Frag‐Seq			Underwood et al (2010)
Structure‐Seq	Chemical‐based method for examining in vivo RNA structures study in both gene‐specific and genome‐wide studies	In vivo	Ding et al (2014)
DMS‐Seq			Rouskin et al (2014)
SHAPE‐Seq			Lucks et al (2011)
icSHAPE			Spitale et al (2015)
DMS‐MaPseq			Tomezsko et al (2021)
CLASH	Crosslinking and proximity ligation‐based method for the direct detection of both intra‐ and intermolecular interactions between RNA molecules	In vivo	Kudla et al (2011)
COMRADES			Ziv et al (2018)
PARIS			Lu et al (2016)
RING‐MaP	Methods for RNA structure ensemble deconvolution.	In vitro	Homan et al (2014)
Rsample		In vivo	Spasic et al (2018)
DREEM			Tomezsko et al (2020)
DRACO			Morandi et al (2021)

miRNA biogenesis dynamics in response to environmental cues

In addition to their function in developmental processes and genome integrity, sRNAs play key roles during the adaptation of plants to environmental stimuli. Several studies have shown that miRNAs and siRNA are essential to balance plant metabolism during drought, salinity, high or low temperatures, light intensity, and nutrient stresses. Heat is one of the most serious stresses affecting plant growth, development, and crops yield. In Arabidopsis, heat stress induces miR160 transcription and represses expression of ARF transcription factor, which regulates expression of heat shock proteins (HSP) genes, resulting in plant thermotolerance (Lin et al, 2018). Several miRNAs, such as miR393, miR396, miR398, miR166, and miR169, are also involved in temperature responses (Song et al, 2019a; Ma et al, 2022). Further, tasiRNA accumulation is found in cold stress conditions and controls thermosensitive genic male sterility in wheat (Tang et al, 2012). In addition, heat‐responsive biogenesis of tasiRNAs was also linked to HEAT‐INDUCED TAS1 TARGET 5 (HTT5) lncRNA activation, which drives early flowering and attenuates immunity. This regulatory hub was shown to induce transmitted phenotypes related to heat, ensuring reproductive success and transgenerational stress adaptation (Liu et al, 2019).

While siRNAs originate from dsRNA recognized by DICER‐LIKE proteins (DCL2, DCL3 or DCL4) to produce abundant phased and/or overlapping siRNAs species, miRNAs are produced by DCL1 from imperfect stem‐loop structures of self‐complementary transcripts (pri‐miRNA) generated from endogenous loci (Bologna & Voinnet, 2014). The folded pri‐miRNAs are recognized and cleaved by DCL1 assisted by the dsRNA‐binding (DRB) proteins HYPONASTIC LEAVES 1 (HYL1) and SERRATE (SE) to release the mature miRNA duplexes. Each duplex consists of a miRNA strand (guide) and a miRNA* strand (passenger). In contrast to siRNA biogenesis, DCL1 precisely processes the pri‐miRNA to release a main and unique miRNA duplex. The miRNA* strand is usually degraded while the miRNA strand is retained on an AGO protein to form an RNA‐induced silencing complex (RISC; Yu et al, 2017; Zhan & Meyers, 2023).

The secondary structure of the miRNA precursor has important consequences for miRNA processing (Bologna et al, 2013a). Depending on the processing mechanism, the lower stem, upper stem and/or internal bulges are essential for miRNA biogenesis. For example, it was described that while a ~ 15 nt segment in the lower stem plays a crucial role for the correct processing of several miRNA precursors (such as pri‐miR164, pri‐miR167, pri‐miR168, pri‐miR171, pri‐miR398) (Mateos et al, 2010; Song et al, 2010); the structure of the upper stem of pri‐miR156, pri‐miR159, and pri‐miR319 is essential for their correct processing (Fig 3A; Bologna et al, 2009, 2013b). Just one nucleotide change converting a match to mismatch (or the opposite) could be enough to improve efficiency or abolish miRNA biogenesis (Rojas et al, 2020). In addition, the identity of the nucleotides (C, G, A or U) at unpaired positions in the precursor can also confer structural constraints resulting in altered miRNA biogenesis (Rojas et al, 2020).

Figure 3. miRNA biogenesis in response to temperature.

(A) Plant miRNA processing mechanisms and their respective RNA structure determinants. (B, C) Temperature‐dependent miRNA biogenesis determined by the RNA secondary structure (B) or by DCL1 cofactors, SE and HYL1 (C).

Although siRNA biogenesis in plants was reported to be temperature‐sensitive (higher siRNA abundance correlating with higher temperature) (Szittya et al, 2003), temperature‐dependent miRNA accumulation relies on the specific miRNA precursor secondary structure. Few studies highlighted a connection between temperature and miRNA biogenesis. For example, analyzing the effects of mutations that can perturb the RNA secondary structure demonstrated that the upper stem of pri‐miR156a plays a role in ambient temperature‐responsive processing to regulate flowering. Mutations in the upper stem near the first cleavage site of pri‐miR156a affect miR156 processing at 23°C, but not at 16°C. This was the first study linking temperature, RNA secondary structure, and miRNA processing in plants (Fig 3B; Kim et al, 2016). Aside from secondary structure of specific miRNAs, a link was found later between temperature and sRNA biogenesis through HYL1 and SE. Two different works suggested the absence of HYL1 or SE might be bypassed by the lower temperature. Growing hyl1 or serrate mutants at 16°C partially rescued both morphological and reproductive defects and restored the production of some miRNAs to wild type levels (Fig 3C; von Born et al, 2018; Ré et al, 2019). In addition to biogenesis, other steps of the miRNA pathway rely on RNA structure, such as ARGONAUTE sRNA sorting or miRNA/mRNA regulation (Zhang et al, 2014; Yang et al, 2020a; Martín‐Merchán et al, 2023). For example, a conserved motif, which enables flexible internal base‐pairing underlying at least three metastable structural configurations, was found in the miR168 precursors. These configurations promote the production of alternative isomiR168 variants with distinctive AGO sorting properties and modes of action (Iki et al, 2018). These results suggest that RNA structure flexibility could adjust miRNA functions under specific situations. Moreover, it has been proposed that a single‐strandedness of the two nucleotides immediately downstream of the miRNA target site, named Target Adjacent nucleotide Motif, can promote mRNA cleavage (Yang et al, 2020a). This single‐strandedness together with the miRNA/mRNA duplex interaction could be also a potential step regulated by external factors such as temperature. Further studies should be performed to explore how the miRNA duplex stability, miRNA sorting, miRNA/mRNA recognition, or mRNA cleavage response confer resistance to environmental changes.

LncRNA interactome dynamics in response to environmental cues

Functional and mechanistic paradigms of plant lncRNAs depend on their association with alternative molecular partners, including other RNAs, proteins, and/or DNA (Lucero et al, 2021). RNA interacting partners will ultimately determine the transcript's molecular and cellular function. In plants, lncRNAs have been implicated in alternative splicing, nucleo‐cytoplasmic protein traffic, translation regulation, epigenetic reprogramming, and miRNA titration (Ariel et al, 2015; Romero‐Barrios et al, 2018). LncRNA activity largely relies on RNA structure, which is sensitive to environmental cues. Therefore, lncRNA interactome dynamics provide a major hub for signal integration controlling adaptation and development (Manavella et al, 2022). Although a growing number of plant lncRNAs have been implicated in a wide range of developmental events (Lucero et al, 2021), only a limited portion have been recognized within ribonucleoprotein complexes, revealing their molecular functions and extensive impacts on development (Table 2, Fig 4).

Table 2.

Molecular and physiological functions of known lncRNA‐integrating ribonucleoprotein complexes.

lncRNA name	Plant species	Molecular partner	Molecular mechanism	Physiological context
COOLAIR	Arabidopsis thaliana	FRI, FCA, FLD, DNA, FY/WDR33	H3K27me3 and H3K4me2 deposition and chromatin looping, protein decoy from chromatin
COLDAIR	Arabidopsis thaliana	CLF	H3K27me3 deposition and chromatin looping
AG‐intron‐4	Arabidopsis thaliana	CLF	H3K27me3 deposition
COLDRWAP	Arabidopsis thaliana	CLF	H3K27me3 deposition and chromatin looping
APOLO	Arabidopsis thaliana	LHP1, WRKY42, VIM1, DNA	R‐loop formation, H3K27me3 and meDNA deposition, protein recruitment to chromatin, and chromatin looping
DAN1	Gossypium hirsutum	DNA	R‐loop formation blocking cis ‐acting elements in gene promoters
MARS	Arabidopsis thaliana	LHP1	H3K27me3 deposition and chromatin looping
MAS	Arabidopsis thaliana	OsMOF, OsWDR5	H3K4me3 and H4K16ac deposition
LAIR	Oryza sativa	OsWDR5	H3K4me3 and H4K16ac deposition
circSEP3	Arabidopsis thaliana	DNA	Alternative splicing
SEAIRa	Arabidopsis thaliana	PUB25/26, RUB1	H2Aub deposition
ASCO	Arabidopsis thaliana	NSRs, PRP8, SmD1	Alternative splicing
HID1	Arabidopsis thaliana	ATXR7	H3K4me3 deposition
SHO cis‐NAT	Petunia hybrida	SHO mRNA	mRNA degradation
lncRNA354	Gossypium hirsutum	miR160b	miRNA sponge
FRILAIR	Fragaria ananassa	miR397	miRNA sponge
TE‐derived lncRNAs	Oryza sativa	miR171	miRNA sponge
ELENA1	Arabidopsis thaliana	FIB2, MED19a	Complex dissociation, protein recruitment to chromatin
MISSEN	Oryza sativa	HeFP	Complex dissociation
SABC1	Arabidopsis thaliana	CLF	H3K27me3 deposition
ENOD40	Medicago truncatula	RBP1	Nucleo‐cytoplasmic trafficking
AhDONE40	Arachis hypogaea	AhASHR3	H3K36me3 and H3K4me3 deposition
lncRNA23468	Solanum lycopersicum	miR482b	miRNA sponge
MSTRG18363	Solanum lycopersicum	miR‐1918	miRNA sponge
PHO1;2 cis‐NAT	Oryza sativa	PHO1;2 mRNA	Translational promotion
IPS1	Arabidopsis thaliana	miR399	miRNA sponge
PDIL1	Medicago truncatula	miR399	miRNA sponge
PILNCR1	Zea mays	miR399	miRNA sponge
VAS	Triticum aestivum	TaRF2b, PDS5A	Protein recruitment to chromatin
SVALKA, asCBF1	Arabidopsis thaliana	Pol II	Pol II collision
CRIR1	Manihot esculenta	MeCSP5	Translational promotion

, Warmth; , Development; , Hormones; , Pathogens; , Vernalization & cold; , Nutrients; , Abiotic stress; , Symbionts.

Figure 4. The lncRNA interactome in alternative molecular mechanisms.

Long noncoding RNAs (lncRNAs) interact with alternative molecular partners which will determine their molecular function. LncRNAs can interact with different classes of proteins, including chromatin modifiers, transcription factors, splicing factors, among others. As a result, specific lncRNAs were shown to modulate the chromatin state or the transcriptional regulation of target genes, as well as the alternative splicing of pre‐mRNAs typically processed by lncRNA‐associated splicing factors. In addition, some lncRNAs can interact directly with DNA forming R‐loops and regulating the activity of target loci; or with RNAs forming RNA‐pairs. RNA–RNA interactions include the role of certain lncRNAs acting as target mimicry of miRNAs, or may affect the stability or translation of natural antisense transcripts (NATs).

Long noncoding RNA interactome integrating developmental signals

Growing evidence supports the role of lncRNAs in epigenetics underlying various developmental processes in plants and animals (Rinn & Chang, 2020; Lucero et al, 2021). In Arabidopsis, two intronic lncRNAs were shown to recruit CURLY LEAF (CLF), the histone methyltransferase subunit of the Polycomb repressive complex‐2 (PRC2), which is responsible for the deposition of the repressive epigenetic mark H3K27me3 (Schuettengruber et al, 2007). These two lncRNAs, named COLDAIR and AG‐intron‐4, participate respectively in the transcriptional silencing of FLOWERING LOCUS C (FLC) and AGAMOUS (AG), encoding key flowering transcription factors from the MADS family (Heo & Sung, 2011; Wu et al, 2018). Expression of the homeotic gene AG ceases all meristem activity and facilitates the development of stamens and carpels (Gómez‐Mena et al, 2005). Upon accumulative exposure to cold, FLC is gradually repressed by histone modification and chromatin compaction, allowing the transition to flowering (Whittaker & Dean, 2017). COLDWRAP is another vernalization‐induced lncRNA, which is transcribed from the proximal promoter of Arabidopsis gene FLC. During vernalization, COLDWRAP, together with COLDAIR, directly interacts with CLF. It was suggested that COLDWRAP and COLDAIR participate in the formation of an intragenic chromatin loop between the promoter and 3′ end of the first intron of FLC, which is involved in establishing stable Polycomb‐mediated silencing of FLC (Kim & Sung, 2017). The structure dynamics of PRC2‐associated lncRNAs in response to ambient temperature remains unexplored. In addition to COLDAIR and COLDWRAP, pioneer plant lncRNA discoveries were focused on COOLAIR, an lncNAT implicated in the epigenetic control of FLC (Ietswaart et al, 2012). The interactome and the structure dynamics of COOLAIR will be discussed below in the context of the plant response to temperature. In Arabidopsis, alternative splicing of the MADS homeotic gene SEPALLATA 3 involved in flowering also depends on the activity of an lncRNA. It was demonstrated that a circular RNA (circRNA) derived from the sixth exon of the SEP3 locus forms a DNA–RNA duplex (known as R‐loop) with its own parent gene. The resulting R‐loop promotes the processing of the nascent transcript into the naturally occurring exon 6‐skipped SEP3.3 isoform, driving floral homeotic phenotypes (Conn et al, 2017).

Another lncRNA participating in the transition to flowering is MAF4 ANTISENSE (MAS), an lncNAT‐modulating transcriptional activity of the MADS AFFECTING FLOWERING4 (MAF4) locus. MAS is also induced by cold and recruits WDR5a, a core component of COMPASS‐LIKE complex, to the MAF4 locus, resulting in enhanced H3K4me3 to promote MAF4 transcription and repress Arabidopsis precocious flowering (Zhao et al, 2018a). Similarly in rice, the lncRNA LAIR is transcribed from the antisense strand of its neighboring gene LRK1 (encoding a leucine‐rich repeat receptor kinase), and it regulates rice plant growth and increases grain yield. RNA‐binding proteins OsMOF and OsWDR5, which participate in H3K4me3 and H4K16ac histone modification complexes, were found to recognize LAIR, and both the lncRNA and the two epigenetic modification proteins could target the LRK1 locus. Overexpressing LAIR increased H3K4me3 and H4K16ac levels on LRK1 chromatin and activate LRK1 transcription (Wang et al, 2018b). More recently, it was reported that the lncNAT SEAIRa is transcribed from the SE locus (Chen et al, 2023), which encodes a core protein involved in miRNA biogenesis and mRNA alternative splicing (Jozwiak et al, 2023). Interestingly, it was shown that SEAIRa recruits plant U‐box proteins PUB25/26 (not previously reported to bind RNA) and a ubiquitin‐like protein related to ubiquitin 1 (RUB1) for H2Aub deposition at specific SE exons. In addition, PUB25/26 participate in SEAIRa cleavage. Shorter SEAIRa is able to recruit PRC2 for H3K27me3 deposition at a different exon of SE. Cooperatively, SEAIRa‐mediated H2Aub and H3K27me3 deposition triggers SE repression (Chen et al, 2023).

The interactome of several lncRNAs has been revealed in the response to phytohormone exposure. In Arabidopsis, the lncRNA ALTERNATIVE SPLICING COMPETITOR (ASCO) can hijack nuclear speckle RNA‐binding proteins (NSRs) in the nucleus to modulate alternative splicing of NSR pre‐mRNA targets in response to auxin during lateral root development (Bardou et al, 2014). More recently, it was shown that NSR proteins recognize multiple lncRNAs, suggesting that an interplay between lncRNAs and mRNAs in NSR‐containing complexes may integrate developmental and external signals (Bazin et al, 2018). Predictions in silico indicated that the region of ASCO best conserved across species is highly structured (Rigo et al, 2020). However, the structure–function bases of ASCO remain unexplored. In Arabidopsis, the PRC1 subunit LIKE HETEROCHROMATIN PROTEIN 1 (LHP1), which regulates H3K27me3 spreading (Veluchamy et al, 2016; Wang et al, 2016) and the chromatin 3D conformation of PRC1 target genes (Veluchamy et al, 2016), can interact with the intergenic lncRNA AUXIN REGULATED PROMOTER LOOP (APOLO). LHP1 participates in the formation of a local chromatin loop encompassing the intergenic region between the APOLO locus and its neighbor gene PINOID (Ariel et al, 2014). APOLO and PINOID are transcriptionally activated by auxin, and PINOID encodes a kinase involved in auxin efflux. More recently, it was shown that APOLO recognizes multiple spatially unrelated loci in the Arabidopsis genome via sequence complementarity and R‐loop formation. Most of these targets are auxin‐responsive genes involved in lateral root development (Ariel et al, 2020). In cotton, the process of whole‐genome duplication and hybridization results in significant variation in transcriptomes, affecting coding and noncoding genes. A specific lncRNA, known as DROUGHT‐ASSOCIATED NON‐CODING 1 (DAN1), originates from a particular cotton progenitor genome after hybridization during cotton evolution. DAN1 was proposed to use RNA–DNA interactions to regulate a subset of target genes involved in auxin‐response pathways, which are critical for drought stress regulation (Tao et al, 2021).

In Arabidopsis, the lncRNA MARNERAL SILENCING (MARS) was linked to the response to abscisic acid (ABA). MARS is actively transcribed in response to ABA from the marneral cluster, which includes the marneral synthase MRN1 gene and the two P450 cytochrome‐encoding genes CYP705A12 and CYP71A16 (Roulé et al, 2022). MARS over‐accumulation is associated with the decoy of LHP1 away from chromatin and a decrease in H3K27me3 distribution throughout the marneral cluster. Loss of H3K27me3 likely allows the formation of a chromatin loop bringing together an enhancer element enriched in ABA‐related transcription factors binding sites and the MRN1 proximal promoter, resulting in transcriptional activation of MRN1 and a delay in seed germination (Roulé et al, 2022). Although it remains uncertain what structure and/or sequence is recognized by LHP1 in APOLO and MARS, it has been determined in vitro that LHP1 can interact with RNAs through its intrinsically disordered hinge region. Interestingly, disruption of the region prevented formation of subnuclear foci in plant cells (Berry et al, 2017), hinting at a potential RNA‐mediated regulation of LHP1 localization.

In the model species A. thaliana, the promoter‐associated lncRNA HIDDEN TREASURE 1 (HID1) was found to be part of a large nuclear ribonucleoprotein complex and associates with the first intron of PHYTOCHROME‐INTERACTING FACTOR 3 (PIF3) to repress its transcription, promoting photomorphogenesis in continuous red light (Wang et al, 2014). In addition, it was found that HID1 also acts downstream of PIF3 during the dark‐to‐light transition (Wang et al, 2018a). More recently, it was shown that HID1 promotes phytochrome B‐dependent seed germination by repressing ABA biosynthesis. Interestingly, it was demonstrated that HID1 interacts with ARABIDOPSIS TRITHORAX‐RELATED7 (ATXR7), an H3K4me3 methyltransferase, inhibiting its occupancy and H3K4me3 deposition at specific ABA‐related loci (Wang et al, 2023).

RNA–RNA interactions were also observed in the context of hormone homeostasis. The SHO gene from Petunia hybrida encodes an enzyme responsible for the synthesis of plant cytokinins. It was shown that the SHO RNA is post‐transcriptionally modulated by a cis‐lncNAT triggering the production of dsRNA‐derived siRNAs and leading to SHO RNA degradation in a tissue‐specific manner (Zubko & Meyer, 2007). Moreover, an in silico analysis of siRNA accumulation over antisense loci in Arabidopsis suggested that RNA interference may constitute a gene regulatory mechanism for at least a subset of cis‐NATs (Jin et al, 2008).

RNA–RNA interactions may also occur between an lncRNA and a miRNA. If the pairing is interrupted by a mismatched loop at the expected miRNA cleavage site, the miRNA–lncRNA interaction persists, tittering miRNA abundance in the plant cell (Paschoal et al, 2018). LncRNA‐mediated miRNA target mimicry in plants has been generalized as a complex regulatory mechanism existing in all eukaryotes, including some lncRNAs collectively known as competing endogenous RNAs (ceRNAs; Kartha & Subramanian, 2014). In cotton, lncRNA354 works as an endogenous target mimic for miR160b and inhibits miR160b‐mediated degradation of the AUXIN RESPONSE FACTOR genes GhARF17 and 18 and modulates plant adaptive response to salt stress. It was proposed that the lncRNA354–miR160b module controls GhARF17/18 expression to regulate auxin homeostasis and plant growth (Zhang et al, 2021). In strawberry, the lncRNA FRILAIR functions as a sponge of miR397, thereby modulating the expression of miR397 target LAC11a involved in strawberry fruit ripening (Tang et al, 2021). Genome‐wide analyses identified of further candidate mimicry lncRNAs of 20 miRNAs conserved in Arabidopsis and rice (Wu et al, 2013). Additionally in rice, a transcript derived from a transposable element was reported to function as a miRNA sponge. Specifically, retrotransposon‐derived transcripts act as decoys for miR171, which destabilizes SCARECROW‐like transcription factor mRNA involved in root development (Cho & Paszkowski, 2017).

Long noncoding RNAs interactome in the plant–microbe interaction

In Arabidopsis, the flg22‐ and elf18‐induced lncRNA ELF18‐INDUCED LONG‐NONCODING RNA1 (ELENA1) was shown to recruit the transcriptional coactivator mediator subunit 19a (MED19a) to the PATHOGENESIS‐RELATED 1 (PR1) locus, thereby positively regulating Arabidopsis resistance against Pseudomonas syringae by modulating PR1 expression (Seo et al, 2017). In addition to MED19a, ELENA1 directly interacts with FIBRILLARIN 2 (FIB2) in the nucleoplasm and nucleolus. FIB2 possesses two RNA‐binding sites through which it can interact with RNAs of different lengths and types, including long and small RNAs (Rakitina et al, 2011). ELENA1 was proposed to dissociate the FIB2/MED19a complex, releasing FIB2 from the PR1 promoter to activate PR1 expression (Seo et al, 2019). In rice, another lncRNA was linked to the dissociation of a protein complex. The maternally expressed lncRNA MISSEN can competitively inhibit the interaction between tubulin and a helicase family protein HeFP, which negatively regulates endosperm development (Zhou et al, 2021).

In Arabidopsis, the lncRNA SALICYLIC ACID (SA) BIOGENESIS CONTROLLER 1 (SABC1) fine‐tunes SA biosynthesis by recruiting the PRC2 component CLF to its neighboring gene encoding the NAC transcription factor NAC3 to reduce its transcription via H3K27me3 deposition. NAC3 activates SA biosynthesis through direct regulation of the isochorismate synthase 1 (ICS1) gene. Upon pathogen infection, SABC1 is downregulated to trigger plant resistance to bacteria and viruses. Thus, it was proposed that SABC1 acts as a molecular switch to balance plant immunity and growth (Liu et al, 2022).

In Arabidopsis, in addition to the interaction between ASCO and NSRs (Bardou et al, 2014), it was shown that ASCO is recognized in vivo by the spliceosome‐core components PRP8a and SmD1b, fine‐tuning the transcriptional response to flg22. It was thus proposed that lncRNAs may integrate a dynamic network including spliceosome proteins to modulate transcriptome reprogramming in response to pathogens (Rigo et al, 2020). In the model legume Medicago truncatula, the lncRNA EARLY NODULIN 40 (ENOD40) interacts with and regulates the nucleocytoplasmic trafficking of the RNA‐binding protein 1 (RBP1), also belonging to the NSR family (Campalans et al, 2004; Lucero et al, 2020). ENOD40 is specifically expressed during symbiotic interaction of the root with nitrogen‐fixative bacteria, suggesting that it exerts a role in alternative splicing during nodulation (Lucero et al, 2020). In the legume Arachis hypogaea (peanut), the ENOD40 locus bears an lncNAT named AhDONE40. Upon infection by the symbiotic bacteria, AhDONE40 recruits the SET domain‐containing protein AhASHR3 to activate ENOD40 through H3K36me3 and H3K4me3 deposition. Considering that ASHR3 and ENOD40 from legumes cluster away from those in actinorhizal plants and other nonlegumes in phylogenetic distance trees, it was proposed that the interaction of DONE40 with ASHR3 could have evolved for adapting the nodule organogenesis program in legumes (Ganguly et al, 2021).

In tomato, lncRNA23468, lncRNA01308, and lncRNA13262 may act as sponges of the miR482b, a miRNA involved in the response to pathogenic fungi. Upon overexpression of lncRNA23468, the activity of miR482b was blocked and the expression of the target genes NBS‐LRRs was significantly increased, boosting plant resistance to the Phytophtora infestans (Jiang et al, 2019). Similarly, the lncRNA MSTRG18363 is induced in tomato by inoculation with the rhizobacterial strain Bacillus subtilis SL18r, capable of triggering induced systemic resistance. MSTRG18363 can sequester miR1918 and increase the transcriptional accumulation of miR1918 target mRNAs, ultimately boosting plant resistance against the foliar pathogen fungus Botrytis cinerea (Zhou et al, 2020 ).

Long noncoding RNA interactome in plant nutrient signaling

Although a growing number of lncRNAs have been implicated in nutrient perception and uptake, the molecular bases behind their action remain unknown (Bakirbas & Walker, 2022; Fan et al, 2022). In rice, an lncRNA‐mediated mechanism regulating phosphate homeostasis was uncovered. The transcription of a cis‐lncNAT of the gene PHOSPHATE1;2 (PHO1;2) was found to activate the formation of a sense–antisense RNA pair which is shuttled towards polysomes. Thus, it was proposed that the cis‐lncNAT promotes translation of PHO1;2 (Jabnoune et al, 2013). Interestingly, structural analyses showed that a high GC content region in PHO1;2 includes a structure inhibiting the association of the 60S subunit to the 40S. It was proposed that upon interaction with its cis‐lncNAT, a localized sense–antisense inter‐molecular pair results in rearrangement of this inhibitory structure, leading to increased formation of the 80S complex in PHO1;2, thus enhancing its translation (Reis et al, 2021). Similarly, cis‐lncNATs have been found to regulate cognate sense mRNA translation in Arabidopsis (Bazin et al, 2017; Deforges et al, 2019). A systematic search for cis‐lncNATs influencing cognate sense mRNA translation in Arabidopsis identified novel transcripts which can repress and enhance translation of their overlapped mRNAs, respectively (Deforges et al, 2019).

In Arabidopsis, the lncRNA INDUCED BY PHOSPHATE STARVATION 1 (IPS1) was the first transcript reported to sequester a miRNA by forming an lncRNA–miRNA duplex. The IPS1 sequence is complementary to the phosphate starvation‐induced miR399, but cannot be cleaved, thus sequestering the miRNA. Notably, IPS1 overexpression results in increased transcriptional accumulation of the miR399 target PHO2 which encodes a ubiquitin‐conjugating E2 enzyme participating in phosphate homeostasis. As a result, increased levels of PHO2 cause a reduced phosphate content in shoots (Franco‐Zorrilla et al, 2007). More recently, a subset of phosphate deficiency‐induced lncRNAs (PDILs) was identified in Medicago truncatula. Interestingly, it was shown that PDIL1 suppresses degradation of MtPHO2, by sequestering miR399 (Wang et al, 2017). Similarly, the PHOSPHATE‐DEFICIENCY‐INDUCED LNCRNA1 (PILNCR1) inhibits miR399‐guided cleavage of ZmPHO2 in maize (Du et al, 2018), suggesting that target mimicry by sequence‐unrelated lncRNAs in the control of miR399 activity is a conserved mechanism across plant species.

Long noncoding RNA interactome in response to temperature changes

As plants lack the ability to regulate their own body temperature, they are highly influenced by daily and yearly temperature fluctuations, which play a significant role in their development and life cycle. Research on lncNAT COOLAIR has provided pioneering examples of plant lncRNA biology. Alternative splicing of COOLAIR is associated with the downregulation of its overlapped gene FLC during vernalization, in a process requiring the H3K4me2 demethylase FLOWERING LOCUS D (FLD; Liu et al, 2010). Moreover, COOLAIR interacts with chromatin in a local manner, leading to a switch in the FLC chromatin state, further indicating that COOLAIR plays a role in epigenetic mechanisms (Csorba et al, 2014). As described above, it was proposed that an intragenic chromatin loop mediated by the PRC2‐associated lncRNAs COLDAIR and COLDWRAP participates in FLC transcriptional repression (Kim & Sung, 2017). An additional chromatin loop was identified between the 5′ and 3′ flanking regions of the FLC locus, which is disrupted during vernalization (Crevillén et al, 2013) and depends on the SWI/SNF chromatin‐remodeling complex (Jégu et al, 2014). It was hypothesized that the perturbation of this gene looping structure could enhance the expression of COOLAIR by exposing cis‐elements that are involved in cold‐induced antisense transcription (Crevillén et al, 2013; Zhu et al, 2015). It was also shown that COOLAIR interacts with the plant‐specific transcription factor FRIGIDA (FRI). FRI acts as a scaffold protein to form a transcription activator complex, resulting in the activation of FLC transcription (Choi et al, 2011). Interestingly, a specific isoform of COOLAIR could sequester FRI away from the FLC promoter by forming nuclear condensates, and repressed its transcription (Zhu et al, 2021). In addition, the COOLAIR transcript was found to generate an R‐loop at the 3′‐end of the FLC gene (Sun et al, 2013). The stabilization of this R‐loop by COOLAIR enabled the RNA binding protein FLOWERING CONTROL LOCUS A (FCA), along with its direct partner FY/WDR33 and other 3′‐end processing factors, to polyadenylate the nascent COOLAIR transcript. This led to the clearance of the R‐loop and the recruitment of chromatin modifiers for the silencing of the FLC locus (Xu et al, 2021a). Furthermore, FCA also interacts with the PRC2 component CLF, suggesting that this allows recruitment of CLF at FLC for H3K27me3 deposition and gene silencing (Tian et al, 2019). To gain deeper insights into the influence of environmental factors on the structure‐dependent function of COOLAIR, a novel method for profiling RNA structures was developed. This groundbreaking approach enabled the determination of the precise in vivo structures of individual RNA molecules, rather than relying on average characteristics of the RNA population (Yang et al, 2022). The findings unveiled that different isoforms of the COOLAIR transcript adopt multiple structures with diverse conformational dynamics. Specifically, the primary COOLAIR isoform with distal polyadenylation, observed under warm conditions, exhibited three predominant structural conformations that underwent changes in both proportions and conformation after exposure to cold temperatures. Conversely, an alternatively spliced distal COOLAIR isoform that is significantly upregulated in cold conditions displayed a higher degree of structural diversity. Remarkably, modifications to the structure of COOLAIR through mutations resulted in alterations in both FLC expression and flowering time, indicating a crucial regulatory role of COOLAIR's structure in FLC transcription. Collectively, this analysis of alternative structures within COOLAIR provides compelling evidence that lncRNAs isoforms can assume distinct and functionally relevant structural conformations, which exhibit variations in abundance and shape in response to environmental cues (Yang et al, 2022).

In winter wheat, the lncNAT VRN1 ANTISENSE (VAS) derived from TaVRN1 gene is induced specifically during the early period of vernalization. TaVRN1 encodes a MADS‐box transcription factor, lowly expressed in the vegetative stage of winter wheat (Yan et al, 2004; Trevaskis, 2010). VAS physically interacts with a bZIP transcription factor TaRF2b and recruits it to activate TaVRN1, ultimately activating the transition to flowering (Xu et al, 2021b). Interestingly, a chromatin loop structure including VAS and the TaRF2b binding site was found in the TaVRN1 locus, and the cohesion factor PDS5A was identified among VAS‐associated proteins, suggesting that VAS might also participate in the regulation of local chromatin conformation dynamics in response to cold.

In addition to vernalization, lncRNAs have also been implicated in the response to cold stress. In cassava, the expression of the lncRNA called COLD‐RESPONSIVE INTERGENIC RNA 1 (CRIR1) was found to be significantly upregulated in response to cold. CRIR1 directly interacts with a cold shock domain‐containing protein COLD SHOCK PROTEIN 5 (MeCSP5), and enhances its translational yield to help the plant cope with cold stress (Li et al, 2022). In Arabidopsis, SVALKA is a lncRNA transcribed from the antisense strand of the CBF1 gene, which encodes C‐repeat/dehydration‐responsive element Binding Factor 1. Upon exposure to cold stress, SVALKA Pol II read‐through initiates the transcription of the lncNAT asCBF1, which is located downstream of SVALKA and overlaps with the end of the CBF1 locus. It has been suggested that the SVALKA‐mediated transcription of asCBF1 leads to Pol II transcriptional collision (Kindgren et al, 2018), causing the enzyme to halt (Hobson et al, 2012), a mechanism leading to the transcriptional termination of protein‐coding genes (du Mee et al, 2018). Although the act of transcription seems per se responsible for the effect, it remains uncertain if specific SVALKA protein partners are required for this mechanism.

In Arabidopsis, the lncRNA APOLO has been shown to play a role in regulating plant responses to cold stress through its interaction with a transcription factor. Specifically, APOLO was found to directly interact with WRKY42, forming a novel ribonucleoprotein complex that shapes the epigenetic environment of the master regulator gene ROOT HAIR DEFECTIVE 6 (RHD6) and activates its transcription. This in turn modulates the transcriptional program for root hair growth, inducing cell expansion in response to cold by stimulating the expression of ROOT HAIR DEFECTIVE SIX‐LIKE2 (RSL2) and RSL4 (Moison et al, 2021). Moreover, the APOLO/WRKY42 complex was found to bind and positively regulate the expression of several genes encoding cell wall EXTENSIN (EXT) proteins, including the key regulator of root hair growth, EXT3, thus triggering elongation of root hair cells (Pacheco et al, 2021). Remarkably, it was recently shown that APOLO also plays an important role in the response to warmth. It was found that APOLO coordinates the activity of the methylcytosine‐binding protein VARIANT IN METHYLATION 1 (VIM1) and the PRC1‐component LHP1 to regulate the transcription of YUCCA2, an auxin biosynthesis gene. In response to warmth, this complex dynamically determines DNA methylation and H3K27me3 deposition over the YUCCA2 promoter, activating the plant thermomorphogenic response (Fonouni‐Farde et al, 2022). Strikingly, the human lncRNA UHRF1 PROTEIN ASSOCIATED TRANSCRIPT (UPAT), a direct interactor of the mammalian DNA methylation regulator UHRF1, can also be recognized by VIM1 and LHP1 in plant cells despite the lack of sequence homology between UPAT and APOLO. Increased levels of APOLO or UPAT interfere with VIM1 and LHP1 binding to the YUCCA2 promoter and globally alter the Arabidopsis transcriptome in a similar manner. Altogether, these findings suggest that a common specific structure between UPAT and APOLO may be responsible for their interaction with conserved protein partners. Further research will be required to uncover how their structure may vary with increasing temperature, affecting the lncRNA interactome to ultimately modulate the transcriptome in response to the environment.

Perspectives

As sessile organisms, plants are highly influenced by their environment, including light, temperature, water, nutrients, and biotic factors, which can affect their growth, development, and survival. Plants have evolved various mechanisms to adapt and respond to these environmental cues in order to optimize their development and reproductive success. The environment can impact plant RNA structure, which, in turn, can influence RNA function, plant growth and development. Additionally, environment‐responsive RNA‐binding proteins may interact with specific RNA molecules, leading to changes in RNA structure and function. Notably, protein‐RNA interactions may also be mediated by the recognition of short sequences within the transcript, which was proven for spliceosome components (Meyer et al, 2017). Thus, RNA folding may influence the availability of these boxes for protein interaction. Among other features, temperature emerges as a significant modulator of RNA structure and function in plants. Existing research has primarily concentrated on temperature‐mediated processes during seasonal changes involving prolonged exposure to high or low temperatures over an extended period. Major advances have also been made using A. thaliana, a model organism with a particularly compact genome, to investigate the impact of temperature changes on the coding and noncoding transcriptomes. However, the identification and characterization of critical RNAs potentially acting as thermosensors across species, pivotal for responding to acute temperature changes in a short time period, have yet to be elucidated. To this end, further investigations integrating genetics with RNA structure‐related approaches, together with the emerging field of epitranscriptomics (Shen et al, 2023), and the elucidation of specific lncRNA‐interactomes will be key to further unraveling the role of the noncoding genome in plant development and adaptation.

Understanding the molecular mechanisms by which plants sense and respond to rapid environmental changes at the RNA level is an important research area that remains relatively unexplored. Further studies identifying and characterizing these key RNA sensors are needed to elucidate the intricate molecular mechanisms underlying plant responses to acute temperature changes, among other influencing cues. Such research could provide valuable insights into the regulatory networks and signaling pathways involved in plant adaptation to dynamic conditions, and may have potential applications in crop improvement strategies to enhance plant tolerance to environmental stress in the face of climate change challenges.

Author contributions

Uciel Chorostecki: Conceptualization. Nicolás G. Bologna: Conceptualization. Federico Ariel: Conceptualization.

Disclosure and competing interests statement

The authors declare that they have no conflict of interest.

The EMBO Journal (2023) 42: e114400