Biochemical Methods To Investigate lncRNA and the Influence of lncRNA:Protein Complexes on Chromatin

Abstract

Long noncoding RNAs (lncRNAs), defined as non-translated transcripts greater than 200 nucleotides in length, are often differentially expressed throughout developmental stages, tissue types, and disease states. The identification, visualization, and suppression/overexpression of these sequences have revealed impacts on a wide range of biological processes, including epigenetic regulation. Biochemical investigations on select systems have revealed striking insight into the biological roles of lncRNAs and lncRNA:protein complexes, which in turn prompt even more unanswered questions. To begin, multiple protein- and RNA-centric technologies have been employed to isolate lncRNA:protein and lncRNA:chromatin complexes. LncRNA interactions with the multi-subunit protein complex PRC2, which acts as a transcriptional silencer, represent some of the few cases where the binding affinity, selectivity, and activity of a lncRNA:protein complex have been investigated. At the same time, recent reports of full-length lncRNA secondary structures suggest the formation of complex structures with multiple independent folding domains and pave the way for more detailed structural investigations and predictions of lncRNA three-dimensional structure. This review will provide an overview of the methods and progress made to date as well as highlight new methods that promise to further inform the molecular recognition, specificity, and function of lncRNAs.

Graphical Abstract

The “noncoding RNA revolution” (Cech, Stetiz) has revealed myriad functional RNA molecules with roles extending far beyond that of a messenger between DNA and protein.¹ The world of noncoding RNAs (ncRNAs), or RNAs that are not usually translated to proteins, came to light in large part as a result of the Encyclopedia of DNA Elements (ENCODE) project.² This consortium found that while up to 90% of the genome was transcribed only 1.2% was translated to protein. Furthermore, this large pool of untranslated transcripts demonstrated biochemical indices of function traditionally ascribed solely to proteins.³ Research exploring the biological activity of these ncRNA transcripts promptly grew. Among the many newly discovered functions of noncoding RNAs, several classes are now known to play critical roles in the regulation of gene expression¹ as well as disease progression.⁴ NcRNAs are classified based on size, with small ncRNAs less than 200 n.t. and long noncoding RNAs (lncRNAs) greater than 200 n.t. Several small ncRNA classes, including microRNAs (miRNAs) and small-interfering RNAs (siRNA), regulate gene expression by forming partially complementary duplexes with mRNAs, which in turn promote mRNA degradation or inhibit mRNA translation into peptides.^5–8 LncRNAs, on the other hand, have been found to exhibit a wide range of regulatory roles, including trafficking of proteins in the cytoplasm^7,9 and epigenetic modulation in the nucleus,¹⁰ with the latter representing the most well studied function.^11,12 Despite this rapid progress in lncRNA identification, the molecular characterization of most functional lncRNAs remains unexplored.

Perhaps the most complete lncRNA story is found in the Xist transcript. Xist is necessary for X-chromosome inactivation (Xci), which allows for even genetic dosage between female (XX) and male (XY) mammals.¹³ Xist coats the chromatin in cis and binds the polycomb repressive complex (PRC2), which imparts the repressive H3K27me3 mark.¹⁴ To date, the minimal RNA binding sequence of Xist (Repeat A, (RepA)), its binding affinity for PRC2 and PRC2 subunits, and its impact on methylase activity have been reported.¹⁵ Indeed, Xist represents one of the few lncRNAs that has been characterized from the organismal down to the biophysical level, yet precise details of the molecular interactions between Xist and PRC2 have yet to be determined.

While several excellent reviews have been published recently regarding the function of lncRNAs,^12,16–18 few have focused on the molecular-level characterization of lncRNAs and lncRNA:protein complexes.^19,20 While not exhaustive, this review will focus on select methods that have enabled recent successes in the biochemical characterization of mammalian lncRNAs, specifically those involved in human epigenetic regulation, as well as highlight new and traditional RNA methods that, when applied to lncRNAs, promise to facilitate a greater understanding of how the structure, function, and molecular interactions of these novel molecules impact their activity.

DISCOVERY OF FUNCTIONAL lncRNA TRANSCRIPTS

LncRNA sequences are defined as transcripts greater than 200 n.t., but many extend into the range of 1–5 kb. LncRNA, unlike mRNA, can vary in splicing, polyadenylation, cellular localization, and the transcribing polymerase.²¹ LncRNAs were first identified through transcriptome wide analysis using either tiled cDNA microarrays or high throughput sequencing.^12,21,22 Importantly, the de novo identification of lncRNAs generally requires methods that allow for the enrichment of low abundance RNA sequences and that do not rely on the presence of a polyA tail or specific polymerase activity. The first and strongest indication of lncRNA functional relevance stemmed from the differential but specific expression levels observed in various developmental stages, tissue types, and disease states.¹⁶ To date, nearly 16,000 lncRNA sequences have been annotated in the GENCODE v23 database,²³ while nearly 300 lncRNAs with verified function have been curated into the Long Noncoding RNA Database v2.0.²⁴ Representative examples of these lncRNAs are highlighted in Table 1.

Table 1.

Examples of Mammalian lncRNA Sequences and Their Binding Partners

shorthand	name	size (kb)a,b	binding partner(s)	known biological pathways	ref
Xist	Xi-specific transcripts	17	PRC2	inactivates one X chromosome to ensure genetic equality between sexes	25
HOTAIR	HOX transcript antisense intergenic RNA	2.2	LSD1; PRC2	suppression of transcription through modification of histone 3 on the HOXD locus	26
MEG3 (GTL2)	maternally expressed gene 3 (gene trap locus 2)	1.5–1.8	PRC2	potentially forms RNA-DNA triplex structures to regulate TGF-β genes	27,
ANRIL	antisense noncoding RNA in the INK4 locus	2.2	PRC1; PRC2	regulates histone modification in the CDKN2A/B locus	29
FENDRR	fetal-lethal noncoding developmental regulatory RNA	2.6; 3.1	PRC2; TrxG/MLL	binds Foxf1 gene promoter to regulate pathways controlling fate of ES cell differentiation	30
MALAT1 (NEAT2)	metastasis associated lung adenocarcinoma transcript 1 (nuclear-enriched abundant transcript 2)	8.8	SR proteins	recruits SR proteins, a family of splicing proteins, to nuclear speckles to control alternative splicing of pre-mRNA	31
NEAT1 (VINC)	nuclear paraspeckle assembly transcript 1 (virus inducible noncoding RNA)	3.2; 22.7	P54nrb; PSP1a	forms and maintains paraspeckle structure	32
PCAT1	prostate cancer associated transcript 1	2	PRC2	post-transcriptionally silences BRCA2	33, 34
SChLAP1	second chromosome locus associated with prostate-1	1.3–1.7	SWI/SNF	binds SWI/SNF complex, leading to aberrant gene expression	35
DANCR (ANCR)	differentiation antagonizing nonprotein coding RNA (antidifferentiation ncRNA)	0.9	PRC2	promotes expression of IL6 and TNF-α at mRNA and protein levels in post-menopausal osteoporosis	36, 37
Lnc_bc060912		1.2	PARP1; NPM	binds PARP1 and NPM1, DNA damage repair proteins, to suppress cell apoptosis	38
Kcnq1ot1	KCNQ1 overlapping transcript 1	91.5	DNMT1	imprints on multiple genes, leading to gene silencing	39
SRA	steroid receptor RNA activator	0.87–0.92	ER-α; SRC-1; Sharp; SLIRP; RNA helicases p68 and p72	breast cancer tumorigenesis	40
PARTICLE	promoter of MAT2A-antisense radiation- induced circulating lncRNA	1.4	G9a; SUZ12	implication in transcription regulation via triple-helix formation	41

^aFor lncRNA with <3 transcript isoforms, the transcripts are separated as such: ##; ##.

^bFor lncRNA with ≥3 transcript isoforms, a range of transcript lengths are shown: ##-##.

CELLULAR LOCALIZATION AND FUNCTION

Cellular Localization

The localization of lncRNA transcripts further supports specific cellular functions and has been examined with traditional RNA detection methods including real-time quantitative polymerase chain reactions (RT-qPCR), Northern blotting, and RNA-fluorescent in situ hybridization (RNA-FISH).

RT-qPCR

The standard determination of lncRNA cellular localization relies upon cell fractionation and RT-qPCR. The first step, cell lysis, uses detergents that maintain segregation between the nuclear and cytoplasmic transcripts.⁴² Following separation, the transcripts are reverse transcribed and the cDNA submitted to RT-qPCR to quantitatively compare RNA levels in the separated cellular fractions. In a recent example, Kanduri and co-workers used these methods to determine the distribution of MEG3 lncRNA and found it to be almost exclusively expressed in the nuclear fraction.²⁸ Compared to sequencing methods, RT-qPCR is limited to discrete sequences but can be highly quantitative even when performed on small amounts of sample.

RNA-FISH

RNA-FISH uses nucleic acid probes to fluorescently track the localization of a target transcript inside a cell through sequence-specific hybridization. In general, a fluorescently tagged short RNA construct (10–50 n.t.) with unique complementarity to the target sequence is added to chemically fixed cells.⁴³ Ideally, the probe sequence selectively binds to the transcript of interest, allowing for direct imaging of the target. Improved sensitivity now enables the visualization of a target transcript at the single-molecule level while limiting the off-target effects. Furthermore, with the use of dozens of fluorescently labeled probes that effectively tile the target sequence, one can achieve significant enrichment over background/off-target complementation.^44,45 Both cytoplasmic and nuclear-localizing lncRNAs, including Xist, NEAT1, and MALAT1, have been studied using RNA-FISH.⁴⁶

Recently, Lee and co-workers used three-dimensional (3D) stochastic optical reconstruction microscopy (STORM), a high-resolution microscopy method, in conjunction with RNA-FISH to study the colocalization of PRC2 to Xist (Figure 1).⁴⁷ In this two-color 3D STORM experiment, PRC2 was probed using an enhancer of zeste homologue 2 (EZH2)-antibody, and Xist was probed using RNA-FISH. Spectral overlap confirmed the expected colocalization, and over 75% of the end-to-end distances between the PRC2 and Xist fluorescent signals were within 100 nm. The resolution of STORM combined with the sequence-specific probing enabled by RNA-FISH will continue to generate crucial insight into the function of lncRNA and lncRNA:protein complexes.

Figure 1.

Nebulous microscopy image of Xist RNA using traditional technology in MEF cells by RNA FISH (A) as compared to 3D-STORM resolution of the same cells (B). Reprinted (adapted or reprinted in part) from ref 47 with permission. Copyright 2015 National Academy of Sciences.

Encoded Fluorescent Tagging

Similar to fluorescent protein tagging methods, RNA transcripts can be encoded with non-native sequences for fluorescence detection, though generation of the fluorescent signal usually requires a secondary step.^48,49 Perhaps the most common method, MS2 tagging, takes advantage of the high affinity of the MS2 viral coat protein to the stem loop repeat MS2 aptamer sequence.⁵⁰ When the MS2 protein is conjugated to a fluorescent protein, selective binding of the conjugate to the MS2 aptamer sequence results in fluorescent labeling. Visualization of NEAT1 lncRNA, for example, was achieved through incorporation of the MS2 aptamer downstream of the lncRNA.⁵¹ Cotransfection of the NEAT1-MS2 aptamer construct and an enhanced yellow fluorescent protein (EYFP)-MS2 protein construct effectively allowed in situ fluorescent tagging of the NEAT1 lncRNA.

Jaffrey and co-workers developed an alternative small molecule based method that relies on interactions between an RNA aptamer sequence (Spinach)⁵² and the cognate small molecule, 3,5-difluoro-4-hydroxybenzylidene imidazolinone (DFHBI), a derivative of the native green fluorescent protein (GFP) chromophore. In this method, the Spinach aptamer is encoded downstream of the target RNA sequence. Rather than requiring coexpression of a reporter protein, as in the case of MS2, the cell-permeable DFHBI molecule fluoresces upon binding the Spinach aptamer. The Jaffrey lab has continued to improve the stability and quantum yield of these RNA reporter complexes,⁵³ and in the future these constructs may offer advantages in lncRNA detection over the bulky protein-based labeling approaches. Although powerful, encoded tags can interfere with native RNA folding, which is an important consideration when using these methods.

Cellular Function

Functional studies of lncRNAs generally begin with knock-down and overexpression methods, both in cell culture and animal models, though knockout/knockin methods using CRISPR-Cas9 technology have also been employed.

RNA Knockdown and Rescue

RNA interference (RNAi) methods such as small interfering RNA (siRNA) and short hairpin RNA (shRNA) allow for loss-of-function studies in cell culture or mouse models and are often the first experiment performed to investigate lncRNA function. RNAi protocols typically use transcripts of ~20–40 n.t. that are complementary to the target transcript. Upon target binding, the newly formed duplexes are degraded through existing cellular machinery.⁵⁴ For instance, Chinnaiyan and co-workers developed siRNA and shRNA sequences toward a novel prostate-cancer associated lncRNA, second chromosome locus associated with prostate-1 (SChLAP1; also called LINC00913).³⁵ Addition of siRNA or shRNA reduced invasion in cell culture and metastasis in mouse models, respectively. As a complement to knock-down studies, “rescue” experiments are some of the most compelling confirmations of function. After successful inhibition with siRNA, reintroduction of an siRNA-resistant target transcript should lead to recovery of lncRNA function and restore the native phenotype. Because the siRNA developed was specific for one SChLAP1 isoform, the same culture could be complemented with a second native but siRNA resistant SChLAP1 isoform. As expected, the invasive phenotype was restored, confirming the impact of SChLAP1 on this pathway.

Though RNAi has been crucial to the study of many lncRNA functions, it is important to note that lncRNA-targeted RNAi methods can be hindered by several factors, including the often low levels of native lncRNA expression,⁵⁵ the nuclear localization of many lncRNAs,⁴⁶ and the more structured nature of lncRNAs compared to mRNAs.⁵⁶ As a result, several siRNA sequences are often screened before identifying a unique transcript for effective lncRNA knockdown. The use of chemically modified oligonucleotides has addressed some of these limitations. Corey and co-workers, for example, recently developed ssRNA sequences with chemical modifications to the backbone that facilitate nuclear localization and cotranscriptional silencing of ncRNA.⁵⁷ Antisense oligonucleotides (ASOs) are synthetic nucleic acid derivatives that, while often less effective than siRNA, are more stable to degradation and more easily access nuclear RNA sequences.⁵⁸ For example, concerns over the efficiency of nuclear siRNA targeting led Jeang and co-workers to use ASOs complementary to NEAT1 to confirm the decrease in nuclear paraspeckle formation previously observed with siRNA.⁵⁹ The impact of cellular environments on siRNA must also be taken into account. For example, several authors have performed knockdown on MALAT1 using siRNA and shRNA in different cell lines and identified distinct functional pathways of MALAT1 activity.^60–62 These results indicate that either siRNA and shRNA knockdown are not analogous methods or that the function of MALAT1, and potentially other lncRNAs, varies greatly between cell lines.

CRISPR-Cas9

The CRISPR-Cas9 system improves upon the shortcomings inherent to RNAi.⁶³ Bacteria naturally harbor the CRISPR-Cas9 system as a means of adaptive immunity, and scientists have engineered this variable immune system to cleave directed DNA sites in systems that range from isolated plasmid⁶⁴ to human cells.⁶⁵ Whereas RNAi uses transcript complementarity to target transcribed RNA, leaving the potential for incomplete targeting and missed transcripts, the CRISPR-Cas9 system uses gene complementarity to bind and excise the target gene.^64–66 One potential drawback to this technique, however, is that the removed genomic DNA may also play a regulatory or structural role. In CRISPR-Cas9 studies on Haunt lncRNA, Shen and co-workers utilized a combined knockout/knockin strategy to systematically delete regions of the Haunt gene without changing the amount of bulk genomic DNA.⁶⁷ These studies revealed opposing functions of the Haunt DNA sequence and the lncRNA transcript, clearly demonstrating the importance of adequate controls and the power of the CRISPR-Cas9 system to study lncRNA function.

IDENTIFICATION OF lncRNA BINDING PARTNERS

As previously discussed, lncRNA can perform explicit biological functions, e.g., epigenetic regulation, when bound to a protein or protein complex (Table 1). In many cases, knockdown and related methods give insight into pathways that might be influenced by lncRNA transcripts, allowing researchers to infer which proteins might be involved. On the basis of these insights, protein-specific immunoprecipitation (IP) methods and RNA pull-down assays can be pursued. Methods that specifically focus on chromatin-bound lncRNAs are discussed in Methods for Identifying and Characterizing lncRNAs and lncRNA:Protein Complexes and Their Influence on Chromatin.

RIP(-Seq)

Antibody-based RNA immunoprecipitation (RIP) methods isolate RNA:protein complexes from cell culture.^68,69 This well-established method isolates the protein of interest by exposing freshly harvested cell lysates to immobilized antibodies (Figure 2). While not always necessary, cross-linking is often employed prior to lysis to ensure coprecipitation of the RNA of interest. Following isolation, the cross-links can be reversed to allow for RNA isolation and downstream analyses such as RT-qPCR. In addition, other protein binding partners can be confirmed through Western blot analysis or identified de novo through proteomics. In early studies of HOTAIR, for example, antibodies to both EZH2, a member of the PRC2 complex, and lysine-specific demethylase 1A (LSD1) were used to IP HOTAIR.¹⁰ In the event that other RNA sequences may be bound to the protein target, the isolated RNA can be submitted to microarray or sequencing analysis for binding partner identification on a transcriptome wide scale.⁶⁸ Using this method, Cech and co-workers were able to evaluate the promiscuity of the PRC2 complex for RNA binding via RIP-seq.^70,71 Numerous combinations of these capture-and-characterize methods and the utility of Next Generation Sequencing (NGS) can be found in recent reviews.^72–76

Figure 2.

RNA- and protein-centric immunoprecipitation assays. (A) Protein-centric pull-down to identify RNA–protein interactions, using either native or denaturing methods. After immunoprecipitation, transcripts are separated and submitted for sequencing. (B) A schematic of PAR-CLIP, where the red X indicated the point mutation inserted during reverse transcription (RT). (C) In vitro methods of RNA-centric pull-downs to identify RNA-binding proteins (D) cell culture methods of RNA-centric pull-downs (E) RNA:protein complexes from (C) and (D) are separated, and proteins are submitted for proteomics. Reprinted (adapted or reprinted in part) with permission from ref 77. Copyright 2015 Nature.

PAR-CLIP

Photoactivatable-ribonucleoside-enhanced cross-linking and immunoprecipitation (PAR-CLIP) can reduce the number of false positives in traditional RIP experiments by photo-cross-linking newly synthesized RNA directly to protein binding partners that are in close proximity.^78,79 First, cells are treated with photoreactive thioribonucleoside analogues. These analogues are incorporated into newly transcribed RNA, and exposure to UV light induces cross-linking of the nucleosides to nearby cysteine residues on the protein surfaces. Following IP, the RNA and protein can be separated through proteinase digestion, and the RNA fraction is reverse-transcribed into cDNA before being submitted to sequencing. This method not only allows the discovery of novel protein-bound RNA transcripts but also identifies the protein-binding domain of the RNA through the detection of distinct RT-induced mutations in the cDNA caused by the unnatural ribonucleosides. As with all cross-linking and IP techniques, robust background controls and statistical analyses must be performed.⁷⁹ PAR-CLIP has been employed by Reinberg and co-workers to identify lncRNA binding partners of the jumonji family ARID-containing protein 2 (JARID2),⁸⁰ which was discovered to be an adjunct component in PRC2 pathways.^81,82 Through PAR-CLIP with a JARID2-specific antibody, Reinberg and co-workers identified 53 lncRNAs that were bound to both JARID2 and EZH2.

Several factors must be considered when performing RIP, and numerous variations of RIPs and CLIPs have been developed within the past five years and are well-described in recent reviews.^83,84 For example, while cross-linking can help to capture low affinity or transient interactions, it is also prone to false positives.⁸³ Non-native RNA:protein interactions can be formed in vitro after the cell lysate has been collected.⁸⁵ The generally low expression levels of lncRNA may lead to masking of these sequences by more abundant RNAs, particularly given the logarithmic amplification of PCR.⁸⁶ Single-molecule sequencing (SMS) can be used to avoid interference from abundant RNAs of all sizes as well as other amplification-induced biases.⁸⁷ SMS covers a broad range of transcript concentrations without the need for PCR. Recently, Williams and co-workers used this technique to identify novel estrogen-regulated lncRNAs in breast cancer.⁸⁸

RNA-Centric Methods

RNA pull downs can be performed both in vitro and in cell culture and can isolate a wide array of binding partners that may or may not appear in a protein-specific IP.⁶⁹ For example, purified and biotinylated HOTAIR was immobilized on resin and exposed to HeLa cell lysates. HOTAIR was found to bind EZH2 and SUZ12, two subunits of the PRC2 complex, as well as the LSD1 complex.¹⁰ By immobilizing a series of HOTAIR deletion mutants, the fragments of HOTAIR critical to protein binding were identified. Cell culture pull downs often employ an encoded MS2-aptamer sequence that can be immobilized on an MS2-coat protein surface.⁸⁶

BIOPHYSICAL CHARACTERIZATION OF lncRNA:PROTEIN COMPLEXES

Following identification of lncRNAs and lncRNA:protein complexes, a series of in vitro techniques can be utilized for further biophysical characterization. Many studies begin with the determination of binding affinity and selectivity of lncRNA and cognate protein binding partners. The determination of the structural components of these interactions, however, has been investigated in far fewer cases, though novel techniques used in the characterization of other long RNA sequences such as viral RNAs should facilitate these studies.

EMSA

Electrophoretic mobility shift assay (EMSA), also known as a gel shift assay or band shift assay, is an affinity-based technique that can be used to quantitatively or qualitatively study RNA-protein interactions based on the size difference between apo and protein-bound states of RNA.⁹⁷ One can quantify the bound versus unbound RNA by comparing the RNA-protein band versus the apo RNA band, respectively. Specificity can be evaluated through the addition of unlabeled control RNA, which would not be expected to impact complex formation, as well as through the addition of unlabeled RNA target, which should displace the labeled RNA in a dose-dependent manner. Lee and co-workers, for example, used this method to test specificity of PRC2 for RepA, the minimal binding domain of Xist.¹⁴ While unlabeled RepA led to displacement of the observed complex, competition assays with the antisense sequence and scrambled sequences showed little to no disruption in RepA:PRC2 binding, consistent with a specific complex.

Both the Cech⁷⁰ and the Lee⁸⁹ laboratories pursued quantitative analysis of HOTAIR:PRC2 interactions and found dissociation constants ranging from 75 to 100 nM between PRC2 complexes and the 5′-end domain of HOTAIR. At the same time, significant differences in the selectivity of PRC2 for HOTAIR and other lncRNAs were observed, with the Cech lab concluding that PRC2 binding affinity was strongly dependent on the length rather than the sequence of the RNA. These discrepancies led the two laboratories to collaborate on a careful comparison between the two studies,⁷¹ and the researchers found that the RNA:PRC2 EMSA assay was highly sensitive to changes in experimental conditions. In addition, PRC2 was found to have multiple lncRNA binding partners, each with similar affinity; however, PRC2 did not interact with all lncRNAs or mRNAs of similar sized transcripts, leading both groups to define PRC2:lncRNA interactions as promiscuous but specific.

Filter-Binding Assay

In addition to EMSAs, more traditional filter-binding assays can be employed to determine binding affinities of RNA:protein complexes. Radiolabeled RNA is incubated with the protein of interest and is washed over a nitrocellulose filter paper. RNA that complexes with protein remains on the nitrocellulose, and the bound versus unbound fraction can be measured via scintillation counter. Lee and co-workers utilized this method not only to confirm binding constants of lncRNA:PRC2 complexes but also to study enzymatic activity.⁸⁹ It was expected that RNA sequences with lower binding affinity to PRC2 would lower the histone methyltransferase (HMTase) activity. Methylation rates were measured by tracking the transfer of tritiated methyl groups from S-adenosyl methionine (SAM) onto recombinant histone H3 via a filter-binding assay. The rate of catalysis was thus directly proportional to the radioactivity of H3. The authors discovered that regulation by an additional PRC2-association protein, JARID2, was necessary to allow histone methylation, which is consistent with the reduction of the binding affinity of PRC2 for lncRNA in the presence of JARID2.

Surface Plasmon Resonance

Surface plasmon resonance (SPR) allows for the study of RNA-protein interactions in a label-free, dynamic environment by measuring changes in the refractive index (RI) on the surface.^90,98 Though SPR requires higher molecular weight particles to measure the RI, this may not be an issue for the study of large transcripts such as lncRNA and their oft multiprotein binding partners.⁹⁸ For example, PARTICLE, or promoter of MAT2A-antisense radiation-induced circulating lncRNA, is located within the MAT2A gene.⁴¹ Knockdown of PARTICLE in sham-irradiated cells caused overexpression of MAT2A, indicating an epigenetic regulatory role for PARTICLE. Anastasov and co-workers examined the PARTICLE:MAT2A complex using SPR. Their results show an inhibitory DNA:RNA triple helix formation between the MAT2A gene and the PARTICLE lncRNA. These effects were only seen at low doses.

The use of additional methods may be required to accurately characterize lncRNA:protein complexes (Table 2), including fluorescence anisotropy techniques⁹⁶ and ITC,⁹⁹ which would provide quantitative evaluation of binding affinities. The large size of lncRNA and associated protein complexes may serve as an advantage for fluorescence anisotropy, which depends upon changes in tumbling rate.⁹⁶ Novel methodologies to determine the specificity of RNA binding are under active investigation.¹⁰⁰ For example, specificity for another promiscuous RNA-binding protein, RNase P protein subunit C5, has been studied using high-throughput sequencing kinetics approach (HITS-KIN), which allows simultaneous monitoring of thousands of RNA sequences. Methods such as these along with studies of the impact of lncRNA on enzymatic activity will be crucial to future characterizations of the lncRNA:protein interaction.

Table 2.

Common Biophysical Techniques That Have Been Used for lncRNA, Longer RNA Transcripts or Could Be Applied in Future Studies^a

technique	shorthand	values solved	pros	cons	RNA examples	ref
electrophoretic mobility shift assay	EMSA	Kd	low-cost equipment	often requires P-32 radiolabel	HOTAIRb; Xistb	71
filter-binding assay		Kd	straight-forward method; efficient	requires radiolabeling (P-32); nonspecific binding to filter; aggregation of materials not related to native binding event	HOTAIRb	89
surface plasmon resonance	SPR	Kd	label-free	target immobilization; cannot report on 2+ state binding model	PARTICLEb	90
isothermal titration calorimetry	ITC	n, Kd, ΔH, ΔG, ΔS	label-free; kinetic and thermodynamic information; can report on 2+ state systems	high concentrations of materials	HIV-1 TARc; purine riboswitchc; 16S rRNAc	91–93
Förster resonance energy transfer	FRET	n, Kd	relatively cheap; can measure distance relationship; sensitive	requires FRET pairs that do not interfere with RNA structure; depends on conformational change and may miss static binding events.	rRNAc, tRNAc	94
fluorescence polarization/anisotropy	FP/FA	n, Kd, aggregation, folding dynamics	no energy transfer requirement; can be HTP	requires fluorescent labeling; binding must significantly change the size of the fluorophore-labeled molecule	Tetrahymena Group I Intron Ribozymec	95, 96

^aHTP = high-throughput; n = Hill’s coefficient; K_d = dissociation constant; ΔH = change in enthalpy; ΔS = change in entropy; ΔG = change in Gibb’s free energy.

^bLong ncRNA example.

^cOther ncRNA example.

Investigations of lncRNA Secondary Structure

The secondary structure of RNA refers to the arrangement of nucleotides in terms of loops and base-pairings. The differences in flexibility and dynamics of nucleotides that are base-paired compared to unpaired nucleotides render the latter more reactive to chemical or enzymatic probing methods. These relative reactivities lend valuable insight into the secondary structure of the RNA, providing parameters for the prediction of secondary structure and laying the groundwork for the prediction of tertiary interactions and protein binding domains.

Chemical and Enzymatic Probing

Chemical probing is the addition of reactive small molecules that target the base, sugar, or backbone of RNA molecules. Because of the high rigidity and structure of the A-form helix, Watson-Crick base-pairs are “protected” from reacting with the chemical probe, whereas highly mobile, single-stranded regions will react more often (Figure 3). These reactions can be analyzed though primer extension, reverse transcription, and cDNA sequencing via gel electrophoresis, respectively. Enzymatic probing utilizes RNases rather than small molecules, and important structural insights can be gained from the differential reactivity of RNases¹⁰¹ (Figure 3).

Figure 3.

Schematic mapping where various chemical probing materials target on nucleobases and backbone. Reprinted (adapted or reprinted in part) with permission from ref 102. Copyright 2015 Elsevier.

The size of lncRNA sequences renders structural determination particularly difficult; to date, only two secondary structures have been characterized. In the first example, Sanbonmatsu and co-workers reported the secondary structure of the steroid receptor RNA activator (SRA).¹⁰³ The secondary structure of this 0.87 kb lncRNA was determined using selective 2′-hydroxyl acylation analyzed by primer extension (SHAPE), in-line probing, dimethyl sulfate (DMS), and RNase V1 (enzymatic) probing. More recently, the Pyle lab published the secondary structure analysis of the lncRNA HOTAIR using SHAPE, DMS, and terbium (Tb).¹⁰⁴ This work determined four independent folding domains within the 2.2 kb lncRNA by comparing the secondary structure data of overlapping fragments to the analysis of the full-length HOTAIR transcript using the shotgun secondary structure method (discussed below) (Figure 4).

Figure 4.

Secondary structure of human HOTAIR Domain 1 (1–530 n.t.; GI: 383286742). Pyle and co-workers performed chemical probing on the full-length HOTAIR transcript. Domain 1 contains the binding-domain of PRC2. Secondary structure rendered in VARNA.¹⁰⁵

Improvements in secondary structure probing are ongoing. Weeks and co-workers recently reported a new SHAPE method, SHAPE-MaP (mutational profile), that can be used to predict the secondary structure of longer and more complex transcripts by detecting reduced enzyme processivity.¹⁰⁶ This method takes advantage of the inherent ability of reverse transcriptase enzymes to either incorporate noncomplementary nucleotides or create deletions at the site of a SHAPE adduct. Additionally, Weeks and co-workers developed RNA interaction groups by mutational profiling (RING-MaP), a technique to explore correlations between clusters of chemically modified nucleobases, enabling the identification of through-space global interactions.¹⁰⁷ Das and colleagues developed a technique named Mutate-and-Map (M2), to further explore the secondary structure of RNA.¹⁰⁸ Within M2, each nucleotide is systematically mutated, and the resulting change in susceptibility of its base-pair to chemical adduction is measured and incorporated into the calculation. Chang and Kool have collaborated on the design and application of SHAPE probes that function in cell culture.¹⁰⁹ Comparison of the SHAPE profiles for the 5S rRNA in vitro and in cell culture revealed that residues of the 5S that normally bind with ribosomal proteins were less accessible in cell culture, as expected. This method should allow for probing of lncRNA secondary structure in a biological context and reveal pivotal bases involved in RNA–protein interactions.

Shotgun Secondary Structure

To identify possible independently folding domains of HOTAIR, Pyle and coworkers used shotgun secondary structure (3S) fragment analysis.¹¹⁰ Following fragmentation, independent folding domains were determined by comparing the full-length probing map with the fragment probing maps. The analysis showed that HOTAIR is divided into four independent domains, of which the 5′-end domain and the 3′-end domain are known to interact with the PRC2 and the LSD1/REST/co-REST complex, respectively.^10,70,71,89

Finally, both chemical and enzymatic probing can be used in the presence and absence of the RNA protein binding partner(s). This would not only determine the protein-binding region but could also give insight into the RNA structure rearrangement from apo to bound forms. It is also important to note that chemical and enzymatic probing methods result in structural predictions rather than absolute characterization. The probing experiments are often performed in parallel to provide a stronger model of the true base-pairings for a given structure. The solved secondary structures are thus a probability of base-pairs, a further reflection of the dynamic structure of RNA.^111,112

Computational Prediction

To produce a secondary structure, data from chemical probing and enzymatic cleavage experiments are run through a secondary structure prediction program based on free energy minimization and the RNA base-pairing energetic parameters determined by Turner and coworkers^113,114 These programs, such as Mfold¹¹⁵ or RNAs-tructure,¹¹⁶ attempt to reconcile the experimental data with free energy calculations to produce the most accurate secondary structure possible. While different programs specialize in particular chemical probing methods, all of them focus on increased sensitivity and positive predictive values (PPV) and are continually improving.

Investigations of lncRNA Three-Dimensional Structure and Tertiary Interactions

RNA three-dimensional structure is highly dependent upon backbone flexibility and/or rigidity, interhelical contacts, and coaxial stacking as well as less common motifs such as triple-helix formation or G-quadruplexes. While little is currently known about the three-dimensional structure of lncRNAs, the secondary structure probing of HOTAIR revealed significant and diverse structure, with more than half of the nucleotides found in base-pairs along with the presence of several higher order junctions.¹⁰⁴ These results, along with the known and specific protein interactions, are consistent with HOTAIR having a complex 3D structure, though such complexity may not be true of all lncRNAs.

To facilitate studies of native lncRNA 3D structure and tertiary contacts, the Pyle lab recently reported a clear protocol for the in vitro transcription and native purification of lncRNAs.¹¹⁷ Nondenaturing purification avoids non-native buffer conditions and high temperatures that would cause the lncRNA to melt and reanneal, which can result in non-native folding. Sedimentation velocity analytical ultracentrifugation (SV-AUC) analyses support the importance of native purification as the formation of multiple lncRNA conformations were observed upon exposure to denaturing conditions.

X-ray Diffraction and Cryo-Electron Microscopy

X-ray crystallography has been a staple of structural biology for decades. For large, dynamic, and flexible structures such as RNA, however, obtaining an atomic-resolution structure is difficult, as evidenced by the 20:1 ratio in protein to RNA structures deposited in the Protein Data Bank.¹¹⁸ While the largest ribosome structures in the PDB contain as many as 5070 nucleotides¹¹⁹ (PDB: 4UG0), the highly structured RNA–protein complex samples only a limited range of conformations after its initial folding, and lncRNAs are expected to be more conformationally dynamic in comparison. Furthermore, the well-ordered structure of the ribosome is an exception: generally, as the unit cell size increases and becomes more complex, the resolution decreases substantially. The only non-ribosomal RNA that approaches the length of a lncRNA is the Bluetongue viral RNA (BTV) at 1906 nucleotides.¹²⁰ At 10 Å resolution, the overall topology of a standard A-form helix is the only discernible structural feature (PDB: 1H1K). An additional concern with X-ray diffraction is that of structural artifacts: as the crystal itself forms, it is possible to induce structures that favor the formation of a stable crystal lattice and/or the optimization of crystal contacts but are not representative of the most biologically relevant structure. This is less of a concern for solution-phase techniques such as NMR and SAXS/SANS, which allow RNA to sample all possible conformations of a given structure and will be discussed later.

One potential solution to structural studies of lncRNAs using X-ray crystallography is to obtain X-ray diffraction structures of the independent folding domains, such as those in HOTAIR mentioned above, and use these structural fragments to reconstruct the full length structure by cryo-EM. Cryo-EM is capable of capturing conformational changes and dynamic interactions for structures that are too large for NMR yet too dynamic for crystallography.^121,122 While this would not provide a perfect rendering of the lncRNA structure, significant insight into global interactions would be possible, and it follows the trends in methodology used to elucidate the structures of other large biomolecules.^119,123,124

At the same time, X-ray crystallography has not been without success in lncRNA. Steitz and co-workers successfully crystallized the 3′-end of MALAT1 along with a portion of its A-rich 3′-UTR (PDB: 4PLX).¹²⁵ These results yield valuable insight into the local lncRNA structure and are consistent with the stabilization of MALAT1 against degradation as a potential mechanism for nuclear accumulation.

NMR Spectroscopy

Solution state nuclear magnetic resonance (NMR) spectroscopy is limited by the size of lncRNA due to both severe spectral overlap and poor resolution from increased tumbling times of large systems.¹²⁶ At the same time, the advantage of NMR lies in the potential to obtain atomic resolution ensembles in solution state conditions that are more biologically relevant than the crystal-lattice conditions needed for X-ray diffraction. Additionally, recent advances in selective labeling techniques promise to help circumvent spectral overlap in long RNA samples.¹²⁷

While a lncRNA structure via NMR has not yet been reported, Summers, Telesnitsky, and co-workers recently reported the longest RNA transcript structure (155 n.t.) solved by 2D-NMR methods, namely, the HIV-1 5′-UTR.¹²⁸ The universal issue of spectral overlap was avoided by separately deuterium-doping segments of the HIV-1 core encapsidation signal (r^CES), which encompasses the shorter fragments of the HIV-1 5′-UTR, and then annealing the 3′-end with a nonlabeled 5′-end (and vice versa) using rapid denaturation and snap cooling. This allowed for unambiguous assignment by utilizing the nuclear Overhauser effects (NOEs) to obtain sequence dependent correlations and ultimately the identification of long-range interhelical relationships. This work was able to resolve discrepancies between nonstructural studies (phylogenetic, biochemical, nucleotide reactivity, and mutagenesis) and the many NMR structure studies performed on shorter segments of the HIV-1 5′-UTR.¹²⁹

Small Angle Scattering

Small-angle scattering (SAS) provides the user with direct structural information about the size, shape, organization, and flexibility of a biopolymer and can utilize both X-ray and neutron scattering (SAXS and SANS, respectively).¹³⁰ This method also uses solution phase sample data collection, similar to conventional biomolecular NMR. Contrary to NMR, however, SAS does not have a size limit nor does it require large concentrations of sample. It is important to note that the resolution of SAS is significantly lower than that of NMR or X-ray diffraction, and the method does not provide atomic-level detail (Figure 5).

Figure 5.

A scale showing the different ranges of electromagnetic dispersion for various structural methods.

For large structures such as lncRNAs, however, SAS might be the only reasonable method for directly acquiring structural data, as evidenced by the characterization of large viral RNA structures. For example, Rein, Wand and co-workers were able to resolve the topological map of the HIV-1 Rev response element (RRE; 233 n.t.) using SAXS, revealing an asymmetric A-shaped structure.¹³¹ Musier-Forsyth and co-workers were able to characterize the 3D structure of the 5′-UTR by combining SAXS topological maps of three ~100 nucleotide fragments with previously reported NMR data. This analysis revealed tertiary interactions consistent with several proposed functions of the 5′-UTR, including tRNA mimicry.¹³²

Small-angle neutron scattering (SANS) can be used to resolve distinct classes of biomacromolecules, including nucleic acids and proteins, based on differences in the bound-atom neutron scattering lengths within each species. More specifically, proteins scatter in the range of (1–5) × 10⁻¹² cm, while nucleic acids scatter in the range of (8–12) × 10⁻¹² cm.¹³³ Despite the inherent low resolution of SAS, this fundamental characteristic of neutron scattering may make it possible to discretely study the conformation of a lncRNA molecule in the context of any number of protein binding partners.

While no studies of SAS on lncRNAs have been reported to date, it is expected that the methods described here will be directly applicable and will provide numerous insights into RNA dynamics, folding, binding, and activity. SAS may further serve as a suitable “first pass” structural assessment of a de novo complex that can be refined by obtaining high-resolution structures of individual fragments.

Computational Modeling

As with secondary structure, computational modeling can provide insight into the 3D structure of RNA.^113,114,134 Indeed, ab initio methods based only on primary sequence have improved to take into account not only canonical and noncanonical base-pairing but also some pseudoknots and unpaired regions. In 2007, Das and Baker presented the Rosetta-based fragment-based structure prediction algorithm for automatic de novo prediction of RNA 3D structures,¹³⁵ which evolved into the Fragment Assembly of RNA with Full Atom Refinement (FARFAR) algorithm now available on the Rosetta Online Server that Includes Everyone (ROSIE).^136,137 Predicted RNA structures can be inserted into a molecular dynamics (MD) simulation using force fields such as AMBER¹³⁸ or CHARMM¹³⁹ in order to further sample their conformational space.¹⁴⁰ MD simulations predict multiple thermodynamically allowed structural conformations for a given sequence, which can lend insight into the dynamic flexibility, and potentially the mode of action, of an RNA structure.

Though powerful tools, both computational modeling and MD are still limited to shorter RNA sequences based on current computing power. Coarse-grain modeling overcomes the time needed to run full-length, all-atom MD simulations while still allowing the exploration of topologically allowed space. Restrictions are placed to limit the degrees of freedom within the structure, and pseudoatoms are used in lieu of full structural elements to simplify the calculation. Caution must be taken when choosing a program, however, as most coarse-grain modeling programs lack parameters unique to RNA structure. The topological mapping of RNA (TOPRNA) program,¹⁴¹ on the other hand, was parametrized using RNA structure databases such as the Nucleic Acid Data Base (NBD).¹⁴² TOPRNA reduces the secondary structure of a given transcript to four pseudoatoms corresponding to the phosphate, ribose, nucleobases, and hydrogen bonds between paired nucleobases (omitted in unpaired bases; Figure 6) and uses the CHARMM force field to explore topological 3D space within the coarse grain parameters. This program focuses on connectivity and sterics for a given transcript, ignoring noncovalent interactions. TOPRNA successfully reproduced NMR residual dipolar couplings (RDCs) of tRNA-PHE, confirming that topological restrictions are a major factor in predicting RNA 3D structure. High-resolution RNA (HiRE-RNA) coarse-grain modeling incorporates seven pseudoatoms per nucleotide rather than four.^143,144 This extra information allows for stacking and base-pairing terms, which are ignored in TOPRNA, but has a slower processing time.

Figure 6.

Conversion of known secondary structure (A) into one of four possible identities: phosphate, ribose, nucleobase, hydrogen bond (B). Three-dimensional assignment in the coarse-grain modeling program TOPRNA: gold, phosphate; yellow, ribose; green, nucleobase; white, hydrogen bond (C). Reprinted (adapted or reprinted in part) with permission from ref 141. Copyright 2014 Oxford University Press.

The ability to predict topologically allowed 3D structures from a secondary structure might lend valuable insight into the stability and accessibility of lncRNA domains. As parametrizations and computer processing improve, we can expect to see more detailed representations of RNA and access to longer constructs.

METHODS FOR IDENTIFYING AND CHARACTERIZING lncRNAs AND lncRNA:PROTEIN COMPLEXES AND THEIR INFLUENCE ON CHROMATIN

LncRNA sequences that interact with chromatin, chromatin-binding proteins, and/or genomic DNA are increasingly appreciated as regulators of gene expression.^145–147 Several recent methods have led to insight into the specific interactions of lncRNA with chromatin and the influence of lncRNA on chromatin structure.

ChIRP

A method developed by Chang and co-workers, chromatin isolation by RNA purification (ChIRP), can directly map lncRNA on the genomic scale.^148,149 ChIRP enables pull-down of lncRNA:chromatin complexes by tiling the target lncRNA sequence with short biotinylated oligonucleotides. This method removes the need for known protein partners, pathways, or structure but also permits false positives, as colocalization does not directly imply cofunctionality. Recently, Kanduri and co-workers investigated how the maternally expressed gene 3 (MEG3) lncRNA associates with its target chromatin while bound to PRC2.²⁸ The authors mapped the location of MEG3 binding using ChIRP-seq in conjunction with 4-SU photoactivatable in vivo cross-linking and discovered an overenriched GA-rich sequence motif in genomic regions associated with gene regulation. The authors used the Triplexator software^150,151 to computationally predict a higher potential for triplex formation at these GA-rich regions when compared to nonenriched regions. Follow up EMSA assays confirmed the formation of a complex between the 20 nucleotide GA-rich ssRNA sequence within the 5′-end of MEG3 and the GA-rich dsDNA sequences found at the chromatin targets. While these results show the potential for triple-helix formation, it is not possible to know from these experiments if the same segment of RNA would be spatially available for triple-helix formation within the full-length MEG3 transcript.

Chromosome Conformation Capture

The study of large-scale chromosome structure has advanced significantly within the past decade.^152–154 The development of chromosome confirmation capture (3C) and its subsequent advances (i.e., 4C,¹⁵⁵ 5C,¹⁵⁶ Hi-C,¹⁵⁷ ChIP-loop,¹⁵⁵ ChIA-PET¹⁵⁸) enables the studying of chromosome structure in both the presence and absence of other regulatory factors (Figure 7). Originally, 3C was developed to show cis-regulation of the chromosome between genes separated by many kilobases. In short, the chromatin is cross-linked and then exposed to digestion enzymes that remove any non-cross-linked DNA. Exposed DNA ends in close proximity are then ligated, and the newly generated sequences can be identified through PCR or sequencing following cross-link reversal. The successors of 3C allow for necessary prior knowledge of only one target sequence as opposed to both (4C); analysis of multiple junctions and contact points (5C); biotin-labeled ligation junction and specific purification (Hi-C); the combination of 3C and ChIP (ChIP-loop); and the removal of nonspecific interactions while also assessing de novo chromatin interactions (ChIA-PET). The method 3C has been adapted by Hoffman and Hu to study lncRNA:DNA complexation (R3C).¹⁵⁹ In the R3C method, genome-bound ssRNA, in this case lncRNA Kcnq1ot1, is cross-linked to genomic DNA and then reverse transcribed with biotin-labeling. This cDNA is subjected to digestion via EcoRI and then ligated to the original target dsDNA. The biotin-labeled cDNA:dsDNA complex is pulled down using streptavidin beads and subjected to PCR to map the chimeric structure. This promising assay is able to map binding of lncRNA to specific loci, in this case the Kcnq1 promoter region.

Figure 7.

(A) The general steps of chromosome conformation capture (3C), from left to right: covalent cross-linking of overlapping chromatin regions; chemical or auditory fragmentation; ligation of cross-linked segments; cross-link reversal and DNA purification. (B) Expansive methods of 3C that allow for higher processivity and more spatial information. Reprinted (adapted or reprinted in part) with permission from ref 152. Copyright 2014 BioMed Central Ltd.

Future Directions

lncRNA transcripts and lncRNA:protein complexes play an undeniably essential role in epigenetic regulation, impacting a wide range of biological pathways from embryonic development to the progression of metastatic cancer, yet many details of lncRNA molecular function remain to be elucidated. While robust methods exist to identify the ever-increasing number of novel lncRNA transcripts and their specific protein partners, the order and regulation of these binding events in biological systems remain unknown. Biophysical characterization of these relationships will rely on continued efforts to standardize existing protocols and the application of innovative characterization methods. The impact of lncRNA:protein interactions on the localization and chemical activity of chromatin-modifying enzymes also merits continued investigation and may benefit from the use of synthetic nucleosomes as model systems.^160,161 Arguably one of the most pressing and exciting areas of lncRNA research will involve structural elucidation, which has progressed impressively at the secondary structure level and will likely benefit from the application of coarse-grained predictions and the fitting of structural fragments to cryo-EM or small angle scattering envelopes. Furthermore, the impact of N⁶-methyladenosine (m⁶A) base modifications, discovered in lncRNA MALAT1 by Pan, He, and co-workers,^162,163 may unlock yet another level of complexity as work in the m⁶ A field progresses. The application of these and other methods to lncRNA is expected to reveal striking insights into the molecular recognition, specificity, and function of these enigmatic molecules.

Another unexplored area of lncRNA research is the development of selective small molecule probes for lncRNA. While guiding principles for small molecule:RNA recognition are still being developed,^164,165 protein-targeted small molecule ligands have been indispensible for both functional probing and therapeutic development, suggesting an enormous potential impact of lncRNA-targeted small molecules in the lncRNA field. The selective inhibition of discrete lncRNA:protein complexes, for example, would not only reveal the functional relevance of these interactions but may also allow the therapeutic potential of targeting lncRNA:protein interactions to be explored.

ABBREVIATIONS

3C

chromosome confirmation capture

4C

circularized chromosome confirmation capture

4-SU

4-thiouridine

5C

carbon-copy chromosome confirmation capture

6-SG

6-thioguanosine

AMBER

Assisted Model Building with Energy Refinement

ANRIL

antisense noncoding RNA in the INK4 locus

ARID

AT-rich interaction domain

ASO

antisense oligonucleotide

BTV

Bluetongue viral RNA

Cas9

CRISPR associated protein 9

cDNA

complementary DNA

CHARMM

Chemistry at Harvard Macromolecular Mechanics

ChIA-PET

chromatin interaction analysis by paired-end tag sequencing

ChIP

chromatin immunoprecipitation

ChIP-loop

ChIP-3C

ChIRP

Chromatin isolation by RNA purification

ChIRP-seq

ChIRP followed by high-throughput sequencing

Co-REST

corepressor for REST

CRISPR

clustered regularly interspaced short palindromic repeats

crRNA

CRISPR RNA

Cryo-EM

cryo-electron microscopy

Cy5

cyanine dye 5

DANCR

differentiation antagonizing nonprotein coding RNA

DDX3

DEAD (Asp-Glu-Ala-Asp) box helicase 3

DFHBI

3,5-difluoro-4-hydroxybenzylidene imidazolinone

DIS

dimer initiation site

DLS

dynamic light scattering

DMS

dimethyl sulfate

DNA

deoxyribonucleic Acid

DNMT1

DNA (cytosine-5)-methyl-transferase 1

dsDNA

double-stranded DNA

EED

embryonic ectoderm development protein

EMSA

electrophoretic mmobility shift assay

ENCODE

Encyclopedia of DNA elements

ER-α

estrogen receptor alpha

EYFP

enhanced yellow fluorescent protein

EZH2

enhancer of zeste homologue 2

FARFAR

fragment assembly of RNA with Full Atom Refinement

FARNA

fragment assembly of RNA

FENDRR

fetal-lethal noncoding developmental regulatory RNA

FP/FA

fluorescence polarization/anisotropy

FRABASE

RNA fragments search engine and database

FRET

Förster resonance energy transfer

GFP

green fluorescent protein

H3K27me3

trimethylation of histone 3 lysine 27

H3K9me2

demethylation of histone 3 lysine 9

hHotair

human HOTAIR

HiRE-RNA

high resolution RNA

HITS-KIN

high-throughput sequencing kinetics approach

HIV 5′-UTR

human immunodeficiency virus 5′ untranslated region

HMTase

histone methyltransferase

HOTAIR

HOX transcript antisense intergenic RNA

HTS

high-throughput screen

IP

immunoprecipitation

ITC

isothermal titration calorimetry

JARID

Jumonji, AT-rich interaction domain

JARID2

Jumonji, AT-rich interaction domain 2

Kcnq1ot1

KCNQ1 overlapping transcript 1

lncRNA

long noncoding RNA

LSD1

lysine-specific demethylase 1A

MALAT1

metastasis associated lung adenocarcinoma transcript 1

MaP

mutational profile

MBP

maltose binding protein

MD

molecular dynamics

MEG3

maternally expressed gene 3

mHotair

mouse HOTAIR

MLL

myeloid/lymphoid or mixed-lineage leukemia

mRNA

messenger RNA

MS2

male-specific bacteriophage 2

NBD

nucleic data base

ncRNA

noncoding ribonucleic acid

NEAT1

nuclear enriched abundant transcript 1

NMR

nuclear magnetic resonance

NOESY

nuclear Overhauser effect spectroscopy

NPM

nucleophosmin

Nt

nucleotides

P54nrb

54 kDa nuclear RNA- and DNA-binding protein

PAR-CLIP

photoactivatable-ribonucleoside-enhanced cross-linking and immunoprecipitation

PARP1

poly [APD-ribose] polymerase 1

PBS

primer binding site

PCAT1

prostate cancer associated transcript 1

PCR

polymerase chain reaction

PDB

Protein Data Bank

PRC1

polycomb repressive complex 1

PRC2

polycomb repressive complex 2

PSP1a

Paraspeckle protein 1a

R3C

RNA 3C (chromosome confirmation capture)

RDC

residual dipolar couplings

RepA

repeat A transcript

REST

RE1 silencing transcription factor

RIP

RNA immunoprecipitation

RNA

ribonucleic Acid

RNA-FISH

RNA-fluorescent in situ hybridization

RNAi

RNA interference

RNase

ribonuclease

ROSIE

Rosetta Online Server that Includes Everyone

roX

RNA on X chromosome

RRE

Rev response element

rRNA

ribosomal RNA

RT-qPCR

real-time quantitative polymerase chain reaction

SAM

S-adenosyl methionine

SANS

small-angle neutron scattering

SAS

small-angle scattering

SAXS

small-angle X-ray scattering

SChLAP1

second chromosome locus associated with prostate 1

SD

splice donor

SEC

size exclusion chromatography

SELEX

systematic evolution for ligands by exponential enrichment

SHAPE

selective 2′-hydroxyl acylation analyzed by primer extension

SHARP

SMRT/HDAC1-associated repressor protein

shRNA

short hairpin RNA

siRNA

small interfering RNA

SLIRP

SRA stem-loop interacting RNA binding protein

SMS

single-molecule sequencing

SPR

surface plasmon resonance

SR proteins

proteins containing long repeats of serine and arginine amino acid residues

SRA

steroid receptor RNA activator

SRC-1

steroic receptor coactivator 1

ssRNA

single-stranded RNA

STORM

stochastic optical reconstruction microscopy

SUZ12

suppressor of zeste 12

SV-AUC

sedimentation velocity analytical ultracentrifugation

SWI/SNF

switch/sucrose non-fermentable complex

TAR

trans-activation response element

Tat

trans-activator of transcription

Tb

terbium

TLE

tRNA3Lys-like element

TOPRNA

topological mapping of RNA program

tracrRNA

transactivation crRNA

tRNA

transfer RNA

TrxG

trithorax group proteins

Xci

X-chromosome inactivation

Xi

inactivated X chromosome

Xist

X-inactive specific transcript