Functions of long non-coding RNA in Arabidopsis thaliana

ABSTRACT

A major part of the eukaryotic genome is transcribed into non-coding RNAs (ncRNAs) having no protein coding potential. ncRNAs which are longer than 200 nucleotides are categorized as long non coding RNAs (lncRNAs). Most lncRNAs are induced as a consequence of various environmental and developmental cues. Among plants, the functions of lncRNAs are best studied in Arabidopsis thaliana. In this review, we highlight the important functional roles of various lncRNAs during different stages of Arabidopsis life cycle and their response to environmental changes. These lncRNAs primarily govern processes such as flowering, seed germination, stress response, light- and auxin-regulated development, and RNA-dependent DNA methylation (RdDM). Major challenge is to differentiate between functional and cryptic transcripts. Genome editing, large scale RNAi and computational approaches may help to identify and characterize novel functional lncRNAs in Arabidopsis.

Introduction

Advances in high throughput DNA sequencing and microarray technology during the last two decades have revealed that most of the eukaryotic genomes are transcribed; however, a large number of transcripts do not code for protein.^1–4 Non-coding transcripts are classified into distinct subgroups based on their features such as size, function and subcellular localization. These transcripts include transfer RNAs (tRNAs), ribosomal RNAs (rRNAs), long non-coding RNAs (lncRNAs), micro RNAs (miRNAs), small interfering RNAs (siRNAs), small nuclear RNAs (snRNAs), and circular RNAs. Conventionally, >200 nucleotides long ncRNAs are called as lncRNAs. They are classified as intronic, exonic, intergenic, sense and anti-sense lncRNAs with respect to location in or near the protein coding genes. They also arise from gene poor regions of chromosomes like pericentromeres and telomeres. LncRNAs regulate many molecular processes ranging from transcription, splicing, mRNA stability, chromatin architecture, chromosomal interactions, nucleosome remodeling, post-translational modifications and organization of nuclear bodies.⁴ Most lncRNAs are transcribed by RNA Pol II but a small fraction is also transcribed by alternate RNA polymerases like Pol IV and V. Most of them have 5ʹ cap, poly-A tail and fewer introns.^1,5 They are less abundant, tissue-specific and show low sequence conservation across species. They can interact with DNA, RNA and proteins to accomplish their functions. They function either in cis or in trans manner and are localized in both cytoplasm and nucleus.⁴

In this review, we discuss the current advances in the field of lncRNAs in the model organism Arabidopsis thaliana. We focus on lncRNAs that are particularly well-characterized in terms of their function, phenotype and molecular mechanism (Table 1). We realized that in Arabidopsis thaliana lncRNAs are broadly involved in regulating external and internal cue dependent developmental pathways, stress response, and genome integrity. Wherever relevant, we have discussed mechanistic similarities with other model organisms to give a holistic insight into the functions of lncRNAs.

Table 1.

LncRNAs in Arabidopsis thaliana.

	LncRNA	Length in nucleotide	Orientation	Developmental role	References
1	COOLAIR I COOLAIR II	400 ~ 750	Antisense to FLC	Flowering	22, 24, 27
2	COLDAIR	1100	Sense to FLC	Vernalization	29, 30
3	COLDWRAP	316	Sense to FLC	Vernalization	30
4	MAS	~ 2000	Antisense to MAF4	Flowering	35
5	ASLa ASLb	2236 2536	Antisense to FLC	Flowering	32
6	AG-incRNA1-4	1000–1200	Sense to AG	Flower organ specification	37
7	asDOG1	413	Anisense to DOG1	Seed Germination	38,39
8	HID1	236	-	Photomorphogenesis	6, 7
9	FLORE	1163	Antisense to CDF5	Circadian rhythm	8
10	APOLO	~ 1000	Antisense to PID	Auxin signaling	10, 11
11	ASCO	786		Root development and alternate splicing regulation	15
12	ELENA1	589	-	Interaction with MED19a and FIB2	42, 43
13	IPS1	~240	-	Phosphate signaling	46
14	T5120	~670	-	Nitrate homeostasis	50
15	DRIR	755	-	Salt and drought stress management	51
16	TE-linc11195	1247	-	Seed Germination	52
17	Proximal SVK	~600	Antisense to CBF1	Heat stress management	54
18	asHSFB2a	1458	Antisense to HSFB	Heat stress management	56

Role of lncRNAs in light and auxin signaling

Plant development is orchestrated by diverse pathways regulated by factors like hormones, nutrients, light and temperature. LncRNAs play important roles in these pathways. Light is one of the important external cues that regulate many aspects of plant development throughout its life cycle. After germination, seedlings undergo skotomorphogenic growth where hypocotyl is elongated, cotyledons are closed and there is low chlorophyll synthesis. Then they switch to photomorphogenic growth where cotyledons open and greening takes place.

HID1 (HIDDEN TREASURE I) is a 236 nt lncRNA produced ubiquitously from a polycistronic cluster of four ncRNAs (nc3017, nc3018, nc3019, and nc3020/HID1) in various light conditions like continuous far red, red, blue and in dark (Figure 1(a)). It promotes photomorphogenesis by regulating PIF3 (PHYTOCHROME INTERACTING FACTOR 3) and possibly chlorophyll biosynthesis genes, PORs (PROTOCHLOROPHYLLIDE OXIDOREDUCTASES).^6,7 In red light, hid1 plants are phenocopy of skotomorphogenic growth. HID1 directly binds to PIF3 locus and negatively regulates its transcription. Additionally, it represses expression of PORs.⁷ It functions in trans. In silico analysis shows that HID1 makes a four stem-loop (SL1 to SL4) structure necessary for its function; perturbation of which leads to hid1 phenotype. HID1 is shown to interact with large protein complexes in gel filtration experiments. However, the identity of the proteins is not yet revealed. Predicted secondary structure and a stretch of nucleotide sequence of HID1 are conserved throughout land plants.⁶ It will be interesting to identify interacting proteins and more targets of HID1 to assess how widespread is the role of this lncRNA in plant life and the mechanism(s) of its action.

Figure 1.

Role of lncRNA in plant development A. HID1 regulates photomorphogenesis- HID1 is located in an ncRNA gene cluster. In continuous red light, HID1 lncRNA regulate cotyledon opening and hypocotyl elongation by binding to the PIF3 locus and inhibits its transcription. HID1 also inhibits POR A/B/C genes and regulate seedling greening B. FLOREregulates CDF5 and promotes flowering – FLORE is the antisense RNA of CDF5. FLORE is regulated in a circadian manner and shows antiphasic expression with CDF5 which is a negative regulator of flowering. FLORE and CDF5 mutually inhibit each other’s expression. C. Regulation of auxin-responsive genes by APOLO lncRNA- APOLO lncRNA is located around 5kb upstream of PID locus. In the absence of auxin, both PID and APOLO are repressed due to the formation of a repressive loop between them mediated by RdDM components and LHP1. Upon auxin induction, the loop is opened which is facilitated by the RDD pathway resulting in the activation of APOLO and PID. APOLO is a positive regulator in cis at PID locus and in trans at other auxin-responsive genes (shown as X and Y). It functions by the formation of an R loop and decoying LHP1 from the targets there by opening of the repressive loop at these loci and activates them

Apart from photomorphogenesis, light is the key regulator of circadian rhythm. Similar to many circadian/clock proteins, many lncRNAs also oscillate in a circadian fashion. In a microarray-based screen, 744 such lncRNAs are identified. Among these lncRNAs, a novel 1163 nt lncRNA, FLORE (CDF5 LONG NONCODING RNA), is transcribed in antisense orientation to a circadian protein coding gene, CDF5 (CYCLING DOF FACTOR 5)⁸ (Figure 1(b)). FLORE has 4 splice variants with varied intron sizes. FLORE and CDF5 reciprocally inhibit transcription of each other which is necessary to maintain proper oscillations of both transcripts leading to contrasting flowering time phenotypes. It can function both in cis and trans. FLORE overexpression lines, cdf5 mutant and cdf-q (quadruple mutant of CDF1, 2, 3 and 5) show early flowering by up regulating FT (FLOWERING LOCUS T) and CO (CONSTANS) expression. On the other hand, CDF1 overexpression lines show late flowering. However, the mechanism by which FLORE and CDF5 inhibit each other’s expression remains to be identified as siRNA and RdDM is unlikely to involve in the mutual antagonism. Also, the significance of different splice variants of FLORE is also not dissected yet.

Phytohormone auxin is a major regulator of plant root development. APOLO (AUXIN REGULATED PROMOTER LOOP RNA)/npc34 lncRNA, is identified by an in silico approach.⁹ It is located at ~5.1 kb upstream of the auxin signaling kinase PID (PINOID) gene, which regulates the polar localization of auxin polar transporter PIN-FORMED 2 (PIN2) in root cells (Figure 1(c)). APOLO region is a target of RdDM pathway and is heavily methylated at cytosines. Both PID and APOLO are responsive to exogenous auxin treatment. In APOLO knockdown lines, PID levels are also decreased, suggesting that APOLO positively regulates PID. Downregulation of APOLO results in reduced gravitropism and longer root phenotypes similar to pid. A repressive chromatin loop is identified between APOLO and PID locus in response to auxin. This loop is dynamically regulated by APOLO, Polycomb Responsive Complex1 (PRC1) member, LHP1 (Like Heterochromatin Protein 1), RdDM and DNA demethylase (ROS1, DML2, and DML3 (RDD)) components. Loss of LHP1 and RdDM pathway components lead to reduced chromatin loop formation whereas loss of APOLO and RDD pathway enhances the formation of the loop. Upon auxin treatment, RDD demethylates the heavily methylated APOLO locus and mediates the opening of the loop resulting in the expression of both APOLO and PID. After reaching the expression maxima, APOLO is downregulated by PolIV-PolV mediated RdDM pathway which in turn enhances the chromatin loop formation resulting in repression of both APOLO and PID. Direct interaction between APOLO and LHP1 is involved in this process.¹⁰ APOLO functions in trans and targets many auxin responsive genes by interacting with LHP1, forming an R loop through sequence complementarity to the target loci, and decoying LHP1.¹¹ In human R loop formation promotes antisense transcription and gene silencing.^12,13 Future studies will reveal whether such mechanism is also widely used in Arabidopsis. Recent finding has also shown that APOLO regulates root hair elongation, especially in cold through positively regulating ROOT HAIR DEFECTIVE 6 (RHD6). It localizes to the RHD6 locus in complex with transcription factor, WRKY42. RdDM and Polycomb complex are also likely to be involved in this process.¹⁴ Together, APOLO lncRNA has a prominent role in root development in response to auxin and cold by decoying chromatin modifiers, R loop formation and regulating chromatin architecture of the target loci (Figure 1(c)).

ASCO (Alternative Splicing Competitor) lncRNA, a 786 bp long transcript is one of the lncRNAs which binds to nuclear speckle RNA- binding proteins (NSRs).^15,16 NSRs play a major role in regulation of alternative splicing (AS) of mRNAs in response to auxin which is important for lateral root formation. ASCO hijacks NSRs by competitive binding to mRNAs and modulates alternative splicing of RNAs thereby regulating lateral root development.¹⁵ Other than auxin responsive root development, ASCO is responsive to biotic stress, mediated via flg22 (a synthetic peptide from bacterial flagella protein flagellin). This process is likely to be mediated by interaction of ASCO with splicing factors, PRP8a and SmD1b.¹⁷ Thus, ASCO is a trans acting lncRNA regulating splicing of the targets. RNA binding motifs, specificity of RNA binding and target selection of ASCO remain to be uncovered.

LncRNAs in flowering

Flowering time determines the reproductive success of plant which is governed by at least five different pathways- vernalization, temperature, hormones (GA), photoperiod and autonomous pathways.¹⁸ The key player in vernalization and autonomous pathways is a MADS box transcription factor FLC (FLOWERING LOCUS C), which is a negative regulator of flowering. Several lncRNAs modulate FLC expression via vernalization and autonomous pathways. COOLAIR (cold induced long antisense intragenic RNA), COLDAIR (COLD ASSISTED INTRONIC NON-CODING RNA), ASL (ANTISENSE LONG) and COLDWRAP (COLD OF WINTER-INDUCED NONCODING RNA FROM THE PROMOTER) negatively regulate FLC, whereas MAS is a positive regulator of FLC expression.

Winter annuals flower after prolonged exposure to cold; this phenomenon is called vernalization.¹⁹ FLC is downregulated during cold and its downregulation is maintained after cold which allows flowering in the warmer period.

COOLAIR is a set of antisense lncRNAs transcribed from downstream of 3ʹ untranslated region of FLC, fully or partially encompassing the gene body (Figure 2). COOLAIR transcripts are polyadenylated and contain one or more introns. Depending upon their sizes and polyadenylation sites there are two major variants; classI and II. ClassI COOLAIR is ~400 nt proximally polyadenylated and does not contain the first exon of FLC in the mature form. ClassII is ~750 nt, distally polyadenylated and contains the first exon of FLC with at least two minor variants, IIa and IIb.²⁰

Figure 2.

lncRNAs regulate flowering via FLC. FLC is a negative regulator of flowering. lncRNAs COLDWRAP, COLDAIR, COOLAIR and ASL negatively regulate its expression. COLDWRAP and COLDAIR are sense lncRNAs and form repressive chromatin loop at FLC locus mediated by PRC2 via deposition of H3K27me3 marks. COOLAIR is an antisense lncRNA with two isoforms, classI and classII. COOLAIR represses FLC expression through H3K4me2 and K3K36me3 demethylation. ASL is another negative regulatory antisense transcripts from FLC which functions by deposition of H3K27me3. MAS is a natural antisense lncRNA from MAF4 locus which recruits COMPASS complex to MAF4 and modulate its expression. MAF4 positively regulates FLC during vernalization

COOLAIR splicing and 3ʹ processing is regulated by autonomous pathway genes (FCA- FLOWERING CONTROL LOCUS A, FPA and FY), spliceosome complex protein (PPR8), and the polyadenylation cleavage factor (CstF64 and CstF77). COOLAIR functions in both autonomous and vernalization pathway to negatively regulate FLC expression.²¹ It is up regulated upon cold treatment during vernalization. It directly binds to FLC locus in cis and leads to H3K4me2 demethylation via histone demethylase FLD (FLOWERING LOCUS D) causing dampening of FLC sense transcription.²² Besides, it is also involved in the loss of H3K36me3, an activating histone modification from FLC locus during vernalization.²⁰ During vernalization, COOLAIR affects the kinetics of downregulation of FLC. Earlier it was considered that COOLAIR function is independent of PRC2, but recent studies show that COOLAIR is required for CLF recruitment via FCA at FLC locus upon vernalization.^20,23 In addition to transcriptional regulation, COOLAIR co-transcriptionally influences 5ʹ capping and stability of FLC sense transcript.²⁴ COOLAIR is involved in R loop formation on its own promoter in cis. A homeodomain protein AtNDX (Arabidopsis thaliana NODLULIN HOMEOBOX) stabilizes this R loop. FCA, FPA and some of the 3ʹ end processing components also promotes R loop on COOLAIR which promotes repression of FLC.^25,26 Thus, COOLAIR employs multiple mechanisms in FLC regulation by autonomous and vernalization pathways. In several Brassicaceae species, the secondary structure of COOLAIR is conserved,²⁷ suggesting that it might be important in its function. In mammals R loop promotes antisense transcription which needs to be tested in case of R loop in FLC regulation.¹³ Interestingly, in perennial species, COOLAIR splice variants different from Arabidopsis are found. In these species, COOLAIR is also cold inducible.²⁸ It has been proposed that alternate splicing of COOLAIR plays a role in natural variation in vernalization response.²⁴ It appears that COOLAIR loss-and gain-of-function alone does not show vernalization mediated flowering time alteration making its role convoluted. Whether or not COOLAIR influence histone post translational modifications on the FLC locus by directly recruiting chromatin modifiers remains to be seen.

COLDAIR is another negative regulator of FLC expression (Figure 2). It is transcribed in sense orientation from the first intron of FLC and is ~1100 nt long lncRNA. This is also cold inducible but it is induced subsequently to COOLAIR. Its expression peaks during vernalization and then goes down after 20 days of vernalization even if the plants are continued to be kept in cold. It is 5ʹ capped but unlikely to be polyadenylated.²⁹ Like COOLAIR, COLDAIR also associates with FLC chromatin.³⁰ COLDAIR interacts with PRC2 and recruits it to the FLC locus. It is capable of functioning in trans. A hairpin loop in COLDAIR is likely to be important in its binding to PRC2, as mutations abolishing the hairpin lead to loss-of-function of this lncRNA. Without vernalization, knockdown of COLDAIR results in late flowering whereas vernalization of COLDAIR knockdown plants enhances early flowering phenotype, implying its vernalization dependent and independent function.^29,30 Unlike COOLAIR, COLDAIR does not affect the kinetics of FLC expression during vernalization.²⁰ FLC regains its expression due to loss-of-function of COLDAIR. Since the expression of COLDAIR goes down after vernalization, it is proposed that it is a part of the epigenetic memory of cold.^29,30 After vernalization, COLDAIR knockdown lines regain of FLC expression, but plants show early flowering phenotype which is exactly opposite of the well-established role of FLC as a negative regulator of flowering. Ironically, while COLDAIR expression is high during vernalization, it does not appear to influence FLC level; possibly epigenetic mechanisms could explain COLDAIR effect after vernalization. Future work will reveal the proposed epigenetic mechanisms and hopefully resolve the contradictions.

In addition to COOLAIR and COLDAIR, lncRNA COLDWRAP is induced during cold. (Figure 2). It is a 316 nt long sense transcript, originating 225 bp upstream to the FLC transcription start site. Unlike COLDAIR which transiently associates with PRC2, COLDWRAP shows stable association with PRC2 during and after vernalization. Cooperative action of lncRNAs COLDAIR and COLDWRAP maintain PRC2 mediated FLC repression. Like COLDAIR, it is also capable of function in trans. Interestingly, 3C (Chromatin Conformation Capture) studies have identified the role of COLDWRAP in the formation of a repressive chromatin loop between the FLC promoter and the first intron. The loop exists before vernalization and is enhanced during cold. The level of repressive chromatin loop formation decreases in the mutant of COLDAIR and PHD-PRC2 mutants vin3 and vil1-1. Together, these results indicate the significance of cooperation between COLDAIR and COLDWRAP lncRNAs in the formation of a repressive chromatin loop on FLC which is involved in establishing stable FLC repression mediate by PHD-PRC2 during vernalization.³¹

ASLs are lncRNAs transcribed from the same promoter as that of COOLAIR with two alternately spliced isoforms, ASLa and ASLb, of length 2236 and 2536 nt, respectively. It is a 5ʹ capped, PolII transcribed RNA without PolyA tail. It encompasses 3ʹUTR and part of first intron of FLC (Figure 2). The 5ʹ region of ASL overlaps with COOLAIR and 3ʹ region overlaps with COLDAIR but it functions distinctly than either of these two lncRNAs. Exosome complex protein, AtRRP6L regulates ASL expression by positively influencing its biogenesis by unknown mechanism. It also physically interacts with ASL. ASL physically associates with H3K27me3; loss of its biogenesis factors RRP6L1 and 2 result in loss of H3K27me3 on FLC locus. It appears to function through PRC2 but nature of relationship between the two is not clear. It has been proposed that ASL lncRNA plays role in FLC silencing via autonomous pathway.³²

Apart from FLC, its paralogs are involved in flowering time regulation. MAF4 is a paralog of FLC; it is expressed during cold and prevents precocious flowering in suboptimal vernalization condition.^33,34 MAS (MAF4 Anti Sense) is a long non-coding RNA transcribed from downstream of the transcription termination site and extends till the first intron of MAF4 (Figure 2).³⁵ MAS positively regulates MAF4 expression in cis by recruiting COMPASS like Trithorax complex to deposit H3K4me3. MAF4, in turn regulates FLC positively.³⁵

Arabidopsis Polycomb-COLDAIR/COLDWRAP and Trithorax-MAS interactions are reminiscent of mouse Polycomb-Xist and WDR5A -HOTTIP (HOXA transcript at the distal tip) interactions, respectively. Xist lncRNA transcribed from one of the two X chromosomes of a female mouse, functions in cis to tether PRC2 to the X chromosome to inactivate it and accomplish dosage compensation. Similarly, lncRNA HOTTIP interacts with trithorax core protein, WDR5 to activate HOXA loci in mice.³⁶ In a nutshell, lncRNAs are players in regulating expression of FLC and flowering time via autonomous and vernalization pathways.

Besides flowering time, lncRNAs are also involved in floral organ specification. Four lncRNAs are detected from the second intron of MADS box floral homeotic gene AGAMOUS (AG). They are named AG-incRNA1to4. AG-incRNA4 is studied in greater detail. It negatively regulates AG expression by facilitating PCR2 recruitment to AG locus. RNAi lines of AG-incRNA4 phenocopy AG overexpression phenotype which includes floral defects and curly leaf phenotype. AG-incRNA4 binds with CLF and this binding is necessary for CLF recruitment and CLF mediated H3K27me3 deposition to suppress AG in vegetative tissue. AG-incRNA4 also auto-regulates its expression by recruiting CLF to its promoter.³⁷

LncRNAs in seed dormancy

Transition from seed dormancy to seed germination is an important developmental transition event in plant’s life cycle. asDOG1 (antisense Delay of Germination 1) or 1GOD a cis acting, 413 nt lncRNA transcribes from DOG1 locus. DOG1 expresses only in seeds. DOG1 transcript has 4 isoforms, among them proximally polyadenylated short DOG1 transcript establishes seed dormancy. Its expression peaks at the time of seed maturation and then declines. This event marks release of seed dormancy.

DOG1-asDOG1 has reciprocal expression profile and they negatively regulate each other’s expression. Once dormancy is established by DOG1, asDOG1 strongly suppress DOG1 expression and hence regulate the timing of seed dormancy to germination transition.³⁸ On the other hand DOG1 alternative polyadenylation site selection controls asDOG1 expression. Proximal site selection enhances and distal site selection reduces asDOG1 expression. FY (flowering time regulator protein, homologue of WDR33) promote selection of proximal polyadenylation site and CPL1 (C-TERMINAL PHOSPHATASE-LIKE 1) promotes selection of distal polyadenylation site. CPL1 indirectly regulates asDOG1 expression.³⁹ There is a striking similarity between CDF5-FLORE and DOG1-asDOG1 regulations but mechanism of sense-antisense regulation is not clear. In baker’s yeast antisense transcription influences sense transcription by influencing nucleosome occupancy and histone post-translational modifications.⁴⁰ Whether such mechanism operates in plant remains to be seen.

lncRNAs in biotic and abiotic stress response

Plants, being sessile organisms, cannot physically escape stress. They have evolved various defense mechanisms to combat stresses. Stresses are broadly classified into two categories- biotic and abiotic stress. Plants perceive stress either through receptors on the cell membrane or within the cell, and in response activate cell signaling and gene regulation machinery which optimize growth and development to overcome its adverse effects. lncRNAs modulate both biotic and abiotic stress response.

Biotic stress caused by pathogen infections hinders or adversely affect plant growth and development. Plants counteract pathogens with the innate immune system as they do not have any adaptive immunity. Microbial or pathogen-associated molecular patterns (MAMPS and PAMPS) are recognized by cell surface receptors to elicit the first line of defense called pattern triggered immunity (PTI). The second line of defense is called effector-triggered immunity where plants counteract pathogen secreted effector molecules by intracellular receptors.⁴¹ LncRNAs are known to modulate PTI in Arabidopsis.

ELENA1 (ELF18-INDUCED LONG-NONCODING RNA1), an intergenic trans acting 589 nt long lncRNA, positively regulates expression of defense response-related genes upon elf18 induction. PR1&2 (PATHOGENESIS RELATED1& 2), CRK7 (CYSTINE RICH RECEPTOR LIKE PROTEIN KINASE7), CYP82C2 (CYTOCHROME P450, FAMILY 82, SUBFAMILY C, POLYPEPTIDE 2) and ARD3 (ACIREDUCTONE DIOXYGENASE 3) are some of the identified targets of ELENA1. It physically interacts with the mediator subunit MED19a to regulate transcription of its target genes.⁴² It also interacts with another mediator subunit MED36a/FIBRILLIN2 (FIB2) and this association increases upon elf18 treatment. FIB2 is a negative regulator of innate immune response genes and its association with ELENA1 inhibits FIB2 recruitment to the target genes (Figure 3(a)).⁴³ It is likely that ELENA1 allosterically inhibits FIB2 targeting, similar to lncRNA BORDERLINE inhibiting the binding of HP1 to chromatin in yeast.⁴⁴ In spite of the fact that ELENA1 can disrupt MED19a-FIB2 interaction, it fails to induce PR1 expression in the absence of elf18 signaling, indicating that there are unidentified players which bridge ELENA1 and elf18 signaling.

Figure 3.

lncRNAs in biotic and abiotic stress A. ELENA1 promotes PR1 expression: MED19 is a component of the mediator complex. ELENA1 binds with MED19a and recruits it to pathogenesis-related genes like PR1. FIB2 binds with MED19a and inhibits its activity. ELENA1 also binds to FIB2 and inhibits MED19a-FIB2 interaction leading to PR1 gene activation. B. Read-through transcription from SVK regulates CBF1 expression: CBF1 and (SVALKA) SVK are cold-induced lncRNAs. SVALKA is transcribed from the antisense strand of CBF1. Read-through transcription of SVK results in Pol II stalling on both strands and leads to the downregulation of CBF1

Abiotic stresses like drought, salt, suboptimal nutrients, temperature fluctuations and hypoxia are known to be combated by lncRNAs. Phosphate and nitrate are among the most important nutrients for plant growth and development. PHO2 (PHOSPHATE2), an E2 ubiquitin conjugase–related protein, targets Pi transporters for degradation and negatively affects Pi accumulation in the shoot. During Pi starvation, miR399 expression is induced which targets PHO2 mRNA for cleavage and results in Pi accumulation in the shoot.⁴⁵ In addition to PHO2, TPSI (tomato phosphate starvation induced) family lncRNAs have complementary sequences to miR399 and are induced in Pi starvation conditions.⁴⁶ IPS1 (INDUCED BY PHOSPHATE STARVATION 1), a TPSI lncRNA, fine-tunes regulation of Pi homeostasis. IPS1 has sequence complementarity to miR399 with mismatches at critical positions important for miR399 mediated degradation of target mRNA. Due to these mismatches, IPS RNA is not cleaved by miR399, instead it sequesters miR399 and functions through target mimic mediated mechanism. This is similar to lncRNA MD1 sequestering miR133 to regulate muscle differentiation in animals.⁴⁷ At4, another TPSI family member, functions redundantly with IPS1. Since miR399 and TPSI family members are conserved among many plant species, similar mechanisms may operate in them as well.⁴⁶ Given that ~1000 lncRNAs are induced upon Pi starvation, it will be interesting to find whether more lncRNAs function in a similar way to understand their role in the regulation of nutrient homeostasis processes.⁴⁸ Recent work has also suggested that lncRNAs may be involved in natural variation of phosphate sensitivity.⁴⁹

To cope up with the fluctuating availability of nitrate in soil and ensure proper growth and development, plants have evolved sophisticated pathways for nitrate homeostasis. T5120 is one of the nitrate starvation-induced lncRNAs. Promoter of this lncRNA contains a cis element NRE (Nitrate Responsive cis element) which is recognized by NLP (NIN LIKE PROTEIN) transcription factors. NLP7 binds to this NRE on T5120 promoter and induces it during nitrate starvation. T5120 does not affect nitrate transport but it positively regulates nitrate assimilation genes like NIA1 (NITRATE REDUCTASE 1), NIA2 (NITRATE REDUCTASE 2), and GLN1 (GLUTAMINE SYNTHASE 1) by an unknown mechanism.⁵⁰

Salt and drought stress cause cellular dehydration and osmotic stress via overlapping mechanisms. Abscisic acid is a phytohormone involved in drought and salt stress signaling. DRIR (Drought Induced lncRNA), a 755 nt long lncRNA, is upregulated upon salt, dehydration stress and ABA treatment, and functions as a positive regulator of stress tolerance. Gain-of-function plants, show more tolerance to salt, drought stress and increased responsiveness to ABA. Transcriptomics studies of mutant and overexpression lines after drought treatment revealed that DRIR modulates the expression of large number of genes. Direct targets and mechanisms of action of DRIR are yet to be identified.⁵¹

TE-lincRNA11195 is identified as an abiotic stress induced TE-lincRNA. Its expression level changes upon salt, ABA and cold treatment. Null mutants of TE-lincRNA11195 show ABA insensitive phenotype for seed germination and post germination seedling development. RNA-seq on wild-type and TE-lincRNA11195 mutants under normal and ABA treated condition shows that the salicylic acid stimulus response genes can be the potential targets of TE-lincRNA11195.⁵²

Plants perceive temperature above and below ambient as stress.⁵³ LncRNAs play roles in combating both low and high temperatures. COR (COLD REGULATED) and CBF (CORE BINDING FACTOR) are the major cold-responsive transcription factors. After cold exposure, a cold-responsive lncRNA, SVK (SVALKA), is transcribed from the antisense strand between CBF3 and CBF1 genes (Figure 3(b)). asCBF1 (anti-sense CBF1), a read-through transcript from SVK into CBF1, is lost in uns-1 mutant (a T-DNA insertion mutant with increased distance between SVK and CBF1). Loss of asCBF1 leads to uncontrolled CBF1 expression; consequently increasing freezing tolerance associated with fitness penalties. asCBF1 negatively regulates CBF1 expression by transcriptional interference due to collision of polymerase complexes of sense and antisense CBF1 (Figure 3(b)). asCBF1 maintains controlled CBF1 expression in order to maximize freezing tolerance with minimum fitness costs.^54,

Heat stress transcription factors (HSFs) are vital players in the regulation of stress responses and developmental pathways like root cell identity, cell fate and pathogen resistance. Upon heat stress, both sense and antisense transcript of HSFB2a (asHSFB2a) are induced. HSFB2a and asHSFB2a act in trans and negatively regulate each other. Overexpression of asHSFB2a results in defects in ovule development and increased biomass of vegetative plant compared to wild type suggesting that the optimal HSFB2a expression is important for Arabidopsis growth and development. Though their expression is required for proper female gametophyte development, overexpression leads to growth penalties.⁵⁶ The mechanism of sense-antisense inhibition is unexplored.

LncRNAs in RNA dependent DNA methylation pathway

Arabidopsis genome sequencing revealed two plant-specific DNA dependent RNA polymerases, Pol IV and Pol V. They are diligent in the suppression of transcription at heterochromatic regions and selected euchromatic loci by RNA dependent DNA methylation (RdDM) pathway (Figure 4). Pol IV produced transcripts are processed into 24 nucleotide double stranded siRNAs⁵⁷ and one strand of these 24 nt RNAs is loaded onto AGO4 (ARGONOUTE 4) and targeted to Pol V transcripts.⁵⁸ DDR complex DEFECTIVE IN RNA-DIRECTED DNA METHYLATION 1 (DRD1), DEFECTIVE IN MERISTEM SILENCING 3 (DMS3), RNA-DIRECTED DNA METHYLATION 1 (RDM1) appears to be involved in Pol V transcription and its association with chromatin. While majority of the Pol V target loci are pericentromeric and transposon families (except LTRs-long terminal repeats), few intergenic regions in euchromatic arms of chromosomes are also targeted. Pol V transcripts are approximately 200 nt in length, transcribed from both strands of DNA, and typically have a single exon. These transcripts are 5ʹ capped or tri-phosphorylated, they lack a poly-A tail and their 5ʹ and 3ʹ ends are not accurately defined and are variable.⁵⁹

Figure 4.

LncRNAs act as platforms to assemble chromatin modifiers to silence transposons. DDR complex, SUVH2-9 and DRM3 facilitate Pol V association with chromatin. Pol V transcribes around 200 nt lncRNAs. AGO4 binds to 24 nt siRNA and the resultant AGO4-siRNA complex binds to Pol V transcripts. IDN2 binds to PolV transcripts in AGO4 dependent manner. IDN2 recruits SWI/SNF complex to PolV target loci by interacting with SBI3B. SPT5L, AGO4 and RDM1 assemble a platform to recruit DRM2 to Pol V target loci. SWI/SNF complex mediates nucleosome repositioning and DRM2 mediate DNA methylation in CHH context leading to heterochromatin silencing

Pol V lncRNAs act as platforms for the assembly of proteins/protein complexes AGO4, IDN2 (INVOLVED IN DE NOVO 2), SPT5L (Suppressor of Ty insertion 5 – like), RDM1 and DRM2 (DOMAIN REARRANGED METHYLTRANSFERASE 2)^60–63 (Figure 4). This is reminiscent of animal non-coding RNA serving as a platform to recruit proteins to the target genes like lncRNAs ANRIL and HOTAIR recruit PRC2 to the target loci.^64–66

Pol V transcripts assemble IDN2, an RNA binding scaffolding protein. IDN2 physically interacts with SWI/SNF subunit SWI3B and recruits it (likely SWI/SNF complex) to the Pol V loci.⁶³ Other ATP depend chromatin remodelers may be similarly recruited and remodel nucleosomes to assist deposition of DNA methylation. In addition, SPT5L also plays a part in the RdDM pathway. With the help of Pol V transcript and PolV-AGO4 protein complexes SPT5L stably localizes to Pol V loci.⁶⁷ SPTL5 and AGO4 then make the platform ready for DRM2 localization to Pol V loci. DRM2 is the major de novo DNA methyltransferase in the RdDM pathway. Apart from the DDR complex, RDM1 is also proposed to bridge AGO4 and DRM2 to their targets.^61,62

Once assembled on Pol V loci, DRM2 deposits cytosine methylation, majorly in the CHH context. AGO4, Pol V and DRM2, as expected, affect most of the loci in a coordinated manner but other components influence distinct subsets among Pol V targets. DMS3 shows a widespread effect on methylation of the target loci, implying its role as a downstream player in the RdDM pathway.⁶⁸ IDN2 affects about one-third of the loci.⁶⁷ Similarly, SWI3B recruited by IDN2 affects CHH methylation on a subset of Pol V targets. Simultaneous targeting by Pol IV and V is suggested to be more efficient in targeting for silencing⁶⁸ (Figure 4).

Though Pol V transcripts are widespread and they form a platform for assembly of different components of RdDM, it is not clear how some of the RdDM components influence only a subset of Pol V target loci. A major area remains to be explored whether RNA sequence determines targeting mechanism.

Although major function of RdDM is to suppress transposons and maintain genome integrity, this pathway is also involved in regulating small number of protein coding genes in Arabidopsis thaliana.⁶⁹ In most cases DNA methylation in/near gene represses transcription. FLOWERING WAGENINGEN (FWA), MEA, SPOROCYTLESS/NOZZLE, PHERES1 are few of the genes which are repressed by DNA methylation.^70–73,55 Selective silencing of rRNA loci is also regulated by RdDM pathway.⁷⁴

Conclusion and perspectives

LncRNAs regulate developmental processes, stress response and genome integrity in Arabidopsis thaliana. In developmental processes so far lncRNAs are majorly known to play roles in photomorphogensis, flowering, auxin mediated root development and, seed germination. Role of lncRNAs in abiotic stresses like phosphorus and nitrogen stress, salt, drought, heat and cold is reported whereas it is also shown to play role in PTI arm of immunity in Arabidopsis. lncRNAs are produced from heterochromatic region and are involved in silencing transposons through DNA methylation. A small number of protein coding genes are also regulated by DNA methylation in Arabidopsis thaliana.

Transcriptomics studies have revealed that a large number of lncRNAs are stress-inducible in nature. Probably, a fraction of induced lncRNAs are cryptic transcripts resulting from over expression of transcription factors and accessible chromatin. The major challenge in the field is to identify lncRNAs that are truly regulatory in nature. Intensive insertion mutagenesis program to find Arabidopsis lines with T-DNA insertions in the non-coding parts of the genome and large scale RNAi mediated knock down of lncRNAs coupled with phenotypic characterization will be helpful in identifying novel lncRNAs important in life cycle of plants. Utility of T-DNA insertion mutants in studying SVK (uns-1), HID1 and of RNAi in case of APOLO is well demonstrated. RNA binding/cleaving genome editing tools can also help in this aspect of lncRNA studies. Genome editing tool mediated modulation of lncRNAs from the endogenous loci can be another important way to study functions of lncRNAs.

In several cases molecular mechanisms of lncRNA function are shown. APOLO, COLDAIR and COLDWRAP are involved in chromatin loop formation. In these cases chromatin loops are repressive but it is not clear what the exact role of chromatin loop is. There are several possibilities like loop itself is sufficient to repress gene expression or loop brings repressive proteins to the locus. R loop formation is another mechanism of function in case of APOLO and COOLAIR. Genome wide R loop formation study is carried out in Arabidopsis and identifying lncRNAs associated with them is the next challenge which will help in the progress of the field. Sense-antisense mediated mutual repression is established in cases of DOG1-asDOG1, HSF2a-asHSFB2a and FLORE-CDF5 and it appears that RdDM is not involved in this repression. It is an interesting open question how this repression is achieved. RNA polymerase collision, sequestering and chromatin modifications are the possibilities worth pursuing. Like other model organisms, several Arabidopsis lncRNAs bind and recruit various proteins like chromatin modifying factors to the target loci or sequester proteins to inhibit their function. LHP1 binds to APOLO, CLF binds to COOLAIR and COLDAIR, and DNA methylation and histone modifying enzymes are recruited to target loci by Pol V transcribed lncRNAs in RdDM pathway. Although LHP1 and CLF both bind RNAs but they lack RNA recognition motifs. RNA binding amino acid residues are dispersed throughout the proteins. This raises the question on binding of specific lncRNAs to regulate specific loci through lncRNAs. It is challenging task to experimentally find generalized RNA binding mechanism in the proteins without RNA recognition motifs. In general, lncRNA-protein interactions are sparsely discovered in Arabidopsis. Computation and experimental tools to identify secondary structures in lncRNA and RNA binding capacity of chromatin modifiers added by more intensive use and improvement of the tools to study RNA-protein interactions may shed more light on molecular mechanism of lncRNA functions.

In summary, use of new tools and more intensive use of established tools will lead to discoveries of more functional lncRNA and their molecular mechanism of action in Arabidopsis thaliana.

Author contributions statement

ML conceptualized the idea of the manuscript. PJ and AG wrote the initial draft and collected the references. ML, PJ and AG revised the draft manuscript. All authors read and approved the manuscript.

Conflict of interest

All authors declare no conflict of interest.

Abbreviations

ABA	Abscisic acid
ATP	Adenosine triphosphate
bp	Basepair
HSF	Heat shock factor
H3	Histone 3
H3K4me2	Histone H3 lysine 4 dimethylation
H3K4me3	Histone H3 lysine 4 trimethylation
H3K27me3	Histone H3 lysine 27 trimethylation
H3K36me3	Histone H3 lysine 36 trimethylation
lncRNA	Long non coding RNA
MAMPS/PAMPS	Microbial or pathogen-associated molecular patterns
npc	non protein coding
nt	nucleotide
PTI	Pattern triggered immunity
Pi	Phosphate
PHD	Plant Homeodomain
Pol	Polymerase
PRC2	Polycomb Repressive Complex 2
RNAi	RNA interreference
RdDM	RNA directed DNA methylation
siRNA	small interfering RNA
3C	Chromatin Conformation Capture