1 INTRODUCTION

Cancer is traditionally viewed as a complex disease arising from the gradual accumulation of genetic and epigenetic alterations in protein-coding genes (Hanahan & Weinberg, 2011). Over the past decades, detailed mechanistic studies of cancer-associated proteins have linked specific protein aberrations to the acquisition of cancer hallmarks and enabled the development of targeted therapies for previously intractable cancers. The discovery that over 70% of the human genome is transcribed to yield hundreds of thousands of noncoding RNAs has transformed our view of the functional genomic space (Bertone et al., 2004; Carninci et al., 2005; Djebali et al., 2012; Iyer et al., 2015; Kapranov et al., 2007). A growing number of studies have since sought to determine the contributions of this expansive noncoding transcriptome to tumorigenesis.

Gene expression profiling by RNA sequencing (RNAseq) has revealed that a subset of noncoding RNAs, termed long noncoding RNAs (lncRNAs) on account of their length of more than 200 nucleotides, are frequently differentially expressed in tumor samples compared to normal tissues (Yan et al., 2015). Importantly, the increased or decreased expression of some lncRNAs has been found to strongly correlate with cancer progression and patient prognosis. In fact, a survey of lncRNA expression patterns across different cancer types has suggested that the deregulation of lncRNA expression may exhibit a higher specificity to cancer type and grade compared to changes in the expression of messenger RNAs (mRNAs; Yan et al., 2015). These correlative findings have provided initial indications that lncRNAs may represent an unexplored reservoir of diagnostic and prognostic markers in cancer (Arun et al., 2018).

The identification of recurrent cancer-associated genetic aberrations in several lncRNA-producing loci has raised the possibility that lncRNAs may be drivers, and not simply passengers, of cancer development (Beroukhim et al., 2010; Prensner & Chinnaiyan, 2011; Rheinbay et al., 2020). Efforts to model the recurrent genetic alterations of lncRNAs in cellular and organismal models of cancer have provided evidence that some lncRNAs are functional mediators of cancer-relevant processes (Olivero & Dimitrova, 2020). These studies have revealed that genetic aberrations in lncRNA loci can contribute to the acquisition of cancer hallmarks, such as hyperproliferation, enhanced survival, altered metabolism, and increased metastatic dissemination (Huarte, 2015). LncRNAs have also been reported to support the development of drug resistance and immune evasion (Bester et al., 2018; Joung et al., 2017). These examples have suggested that lncRNAs have the capacity to directly modulate diverse aspects of tumorigenesis.

Despite a growing appreciation of the diverse roles of cancer-associated lncRNAs, their mechanisms and functional elements remain poorly characterized. To date, only a handful of studies have examined the sequence and structural basis for lncRNA activity. The scarcity of mechanistic insights has limited our understanding of the contributions of lncRNAs to cancer development and impeded efforts to exploit their therapeutic potential. Here, we describe the current paradigms for the molecular activities of lncRNAs. By highlighting the experimental strategies used to characterize individual lncRNAs, we aim to define the relationship between lncRNA mechanisms and cellular and molecular pathologies in cancer. We propose that an integrative approach is essential for the validation of lncRNAs as drivers and therapeutic targets in cancer and other diseases.

2 EXPERIMENTAL STRATEGIES FOR STUDYING LNCRNA MECHANISMS

The identification of as many as 100,000 lncRNAs in mammalian genomes has opened the possibility that these newly discovered RNA molecules may harbor a wide range of novel functional and mechanistic features, akin to proteins (Derrien et al., 2012; Iyer et al., 2015). However, the observation that lncRNAs are characterized by relatively poor evolutionary conservation at the sequence level has impeded initial efforts to define conserved functional elements (Cabili et al., 2011; Ulitsky, 2016). Instead, it has been hypothesized that lncRNAs may act through the formation of intricate intramolecular secondary and tertiary structures capable of interfacing with DNA, RNA, and proteins (Guttman & Rinn, 2012). In the context of this mechanistic paradigm, trans-acting lncRNAs have been proposed to play diverse regulatory roles throughout the nucleus and in the cytoplasm (Kopp & Mendell, 2018; Rinn & Chang, 2012; Figure 1a). Examples of trans-acting lncRNAs include non-coding RNA activated by DNA damage (NORAD) and metastasis-associated lung adenocarcinoma transcript 1 (MALAT1), which are two scaffold RNAs that act in the context of cytoplasmic NP (NORAD-PUM) bodies (Elguindy & Mendell, 2021; Lee et al., 2016; Tichon et al., 2016) and nuclear speckles (Tripathi et al., 2010), respectively (Figure 1a). Cis-acting lncRNAs, on the other hand, have been defined based on their accumulation near their sites of transcription and their roles in the local regulation of gene expression (Gil & Ulitsky, 2020; Kopp & Mendell, 2018; Figure 1b). Well-established examples of repressive cis-regulatory lncRNAs include X-linked lncRNAs with essential roles in X chromosome inactivation, such as X inactive specific transcript (XIST; Sahakyan et al., 2018), and lncRNAs expressed from imprinted loci, such as antisense of IGF2R nonprotein coding RNA (AIRN; Nagano et al., 2008) and KCNQ1 overlapping transcript 1 (KCNQ1OT1; Mohammad et al., 2008; Pandey et al., 2008; Figure 1b). Beyond dosage compensation, antisense noncoding RNA in the INK4 locus (ANRIL; Yap et al., 2010), myeloid RNA regulator of Bim-induced death (Morrbid; Kotzin et al., 2016), and plasmacytoma variant translocation 1, isoform b (Pvt1b; Olivero et al., 2020) have been proposed to repress the expression of their neighboring genes. On the other hand, HOXA distal transcript antisense RNA (HOTTIP; Wang et al., 2011), LincRNA-p21 (Dimitrova et al., 2014), and LincRNA-Cox2 (Carpenter et al., 2013) have been implicated in local gene expression activation.

Categorizing lncRNAs based on subcellular localization and site of action. (a) Left—following transcription, trans-acting lncRNA localize to distant sites within the nucleus and/or are exported to the cytoplasm; right—NORAD largely localizes to the cytoplasm, where it interacts with PUM proteins to form phase-separated condensates known as NP bodies (Elguindy & Mendell, 2021; Lee et al., 2016), while MALAT1 remains within the nucleus and localizes to nuclear speckles (Hutchinson et al., 2007); (b) Left—cis-acting lncRNAs accumulate at or near their site of transcription. Right—XIST (red) is expressed from and spreads on the inactive (Xi) but not the active (Xa) X chromosome (green) in female mammalian cells. The lncRNA AIRN (red) is expressed from and represses the expression of genes from the imprinted IGF2R locus (green) on the paternal (Pa) but not maternal (Ma) allele

Notably, simply because a lncRNA is transcribed, it does not necessarily mean that it produces a functional molecule. Recent studies provide evidence that functional lncRNA loci can act through sequence- or even transcription-independent mechanisms (Espinosa, 2016) (Figure 2). On the one hand, several studies described lncRNA-producing loci that can modulate the epigenetic and transcriptional landscape of nearby genes through the act of transcription (Anderson et al., 2016; Engreitz et al., 2016; Isoda et al., 2017; Latos et al., 2012). On the other hand, a different set of studies determined that the lncRNA transcripts may in fact be nonfunctional byproducts of underlying DNA regulatory elements such as enhancers (Engreitz et al., 2016; Paralkar et al., 2016). These observations have raised questions about the functional significance of specific features of lncRNA transcripts, such as RNA sequences or structural motifs. In this section, we briefly describe the common genetic, molecular, and biochemical tools used to elucidate the mechanistic basis of lncRNA activities, with a specific focus on the unique experimental and conceptual challenges that emerge when characterizing cis-acting lncRNAs (Bassett et al., 2014; Kopp & Mendell, 2018).

Layers of overlapping functional elements in cis-regulatory loci. Cis-acting lncRNA-producing loci may act through one or more tightly coupled mechanisms. DNA elements such as promoters, enhancers, and CTCF binding sites may bind transcription factors that directly regulate local gene expression (Alexanian et al., 2017; Paralkar et al., 2016). Alternatively, the transcriptional process, including associated phenomena such as splicing, may enact cis-regulation through transcriptional interference or by modulating the local transcriptional and epigenetic landscape (Allou et al., 2021; Engreitz et al., 2016; Latos et al., 2012). Finally, sequence and structural features of the lncRNA itself may enable it to interact with proteins that activate or repress local gene expression (Pandey et al., 2008; Pandya-Jones et al., 2020) 2.1 Distinguishing between trans- and cis-acting lncRNAs

A combination of loss-of-function experiments, gene expression profiling, and subcellular localization studies has been applied to differentiate between trans- and cis-acting lncRNAs. In loss-of-function studies, transient downregulation has been successfully achieved for many lncRNAs using antisense oligonucleotide (ASO) gapmer-mediated knockdown or RNA interference (RNAi), although the mechanism by which RNAi mediates the degradation of nuclear and chromatin-associated lncRNAs has remained unclear (Lennox & Behlke, 2016; Stojic et al., 2018). In parallel, studies have accomplished stable inhibition of target lncRNAs through genetic deletion approaches, such as promoter or locus deletions in cell lines and animal models (Bassett et al., 2014). While deletions effectively abolish lncRNA expression, even minimal genetic perturbations have the potential to destroy functional regulatory DNA sequences, such as transcription factor or CTCF binding sites (Bassett et al., 2014) (Figure 3). To overcome this caveat, alternative genetic approaches have aimed to insert transcriptional terminators to suppress transcription from endogenous lncRNA loci (Kopp & Mendell, 2018). This has been achieved through the knock-in of polyadenylation signals (PAS) or transcriptional STOP cassettes to induce premature transcriptional termination.

Inhibition of lncRNAs through genetic deletion approaches. Large deletions, such as ones that entail the excision of a locus or promoter, abolish lncRNA expression but can also grossly affect chromatin architecture and/or inadvertently delete functional DNA elements. Mutagenesis of DNA regulatory motifs, such as transcription factor binding sites, introduces finer genetic modifications but cannot dissociate between functional RNA and DNA elements

Following lncRNA inhibition, targeted quantitative real-time PCR (qRT-PCR) analysis or unbiased RNAseq profiling have been used to determine whether a given lncRNA acts locally to regulate the expression of its neighboring genes or has global regulatory functions. Notably, inhibition of many cis-acting lncRNAs has been found to produce relatively mild effects on the expression of neighboring genes (Gil & Ulitsky, 2018). This observation has suggested that many cis-acting lncRNAs are primarily fine-tuners of gene expression, promoting or suppressing the expression of target genes past or below a functional threshold, respectively (Dimitrova et al., 2014; Olivero et al., 2020). On the other hand, global gene expression changes following lncRNA inhibition should be interpreted cautiously as trans effects can, in some cases, be explained through cis-regulatory functions. As an example, promoter deletion of the lncRNA LincRNA-p21 leads to a global suppression of Polycomb target genes but these gene expression changes are an indirect consequence of the role of LincRNA-p21 as a local transcriptional activator of the cyclin-dependent kinase inhibitor 1a, also known as p21 (Cdkn1a/p21; Dimitrova et al., 2014).

The key to differentiating cis versus trans activities has been determining whether exogenously introduced transcripts can rescue lncRNA function in loss-of-function experiments. Ectopic constructs are expected to recapitulate the function of trans-acting but not cis-acting lncRNAs, since the latter would not localize to the endogenous lncRNA locus. In the case of the cis-regulatory LincRNA-p21, exogenous overexpression of LincRNA-p21 in LincRNA-p21-deficient cells does not rescue Cdkn1a/p21 or Polycomb target expression levels (Dimitrova et al., 2014). Similarly, the introduction of Pvt1b from an ectopic construct does not recapitulate transcriptional repression of Myc (Myelocytomasis), while overexpression of Pvt1b from the endogenous locus following CRISPR activation (CRISPRa) is able to suppress Myc (Olivero et al., 2020). An exception to this rule is the X-inactivation lncRNA Jpx, which can activate Xist expression through both cis and trans mechanisms in a dose-dependent manner, likely through its ability to act as a decoy and sequester negative regulators of Xist both locally and globally (Carmona et al., 2018; Sun et al., 2013; Tian et al., 2010).

Determining the localization of a lncRNA by subcellular fractionation or fluorescence in situ hybridization has also provided direct clues to whether a lncRNA may function in cis or in trans. For example, subcellular fractionation and RNA FISH visualization of cis-regulatory lncRNAs, such as XIST and lncRNAs from imprinted loci, have revealed their close association with chromatin, while trans-acting lncRNA, such as MALAT1 and NORAD1, have been detected in the nucleoplasmic and cytoplasmic fractions, respectively (Elguindy et al., 2019; Tripathi et al., 2010). The recent development of single-molecule RNA fluorescence in situ hybridization (smRNA-FISH) has transformed the field by allowing the visualization of individual lncRNA molecules, including lowly expressed lncRNAs (Cabili et al., 2015; Raj et al., 2008; Raj & Rinn, 2019). smRNA-FISH has been successfully used to establish the accumulation of functional intergenic RNA repeat element (FIRRE), LincRNA-p21, and Pvt1b at their sites of transcription by co-localizing exon- and intron-specific probes (Dimitrova et al., 2014; Hacisuleyman et al., 2014; Olivero et al., 2020). In a different instance, immunofluorescence (IF)/smRNA-FISH combined with high-resolution microscopy has revealed detailed mechanistic insights, such as the key role of the central region of the nuclear paraspeckle assembly transcript 1 (NEAT1) RNA in organizing structural proteins within the core of nuclear paraspeckles (Yamazaki et al., 2018).

2.2 Identification of binding partners of lncRNAs

Many lncRNAs have been proposed to operate through their interactions with DNA, RNA, and proteins. Thus, experimental identification of lncRNA interaction partners has been key to the mechanistic characterization of both trans- and cis-acting lncRNAs. Recently, a number of techniques have been developed for the unbiased identification of lncRNA-interaction partners. Approaches such as chromatin isolation by RNA purification (ChIRP) and capture hybridization analysis of RNA targets (CHART) have been used to map the genomic regions occupied by a lncRNA of interest (Chu et al., 2011; Simon et al., 2011). These techniques have provided valuable information about the genomic targets of the structural trans-acting lncRNAs NEAT1, MALAT1, and FIRRE (Hacisuleyman et al., 2014; West et al., 2014). On the other hand, RNA and DNA split-pool recognition of interactions by tag extension (RD-SPRITE), developed to comprehensively examine RNA–DNA and RNA–RNA interactions, has revealed the widespread roles of cis-acting lncRNAs in forming stable nuclear compartments in spatial proximity to their transcriptional loci. An additional suite of techniques has focused on the detection of lncRNA–RNA interactions (Engreitz et al., 2014; Lu et al., 2016). For example, RNA antisense purification (RAP) has been used to systematically map RNA–RNA interactions (RAP-RNA) and to facilitate the identification of RNAs that interact with a lncRNA of interest (Engreitz et al., 2014). Alternatively, techniques such as psoralen analysis of RNA interactions and structures (PARIS) can be used to study entire RNA interactomes (Lu et al., 2016). Importantly, these and similar approaches also detect intramolecular base-pairing interactions, and therefore have the potential to yield further insights into functional secondary structures within a lncRNA of interest. Finally, biochemical tools have been successfully used to identify proteins that bind lncRNAs directly or as part of a multi-factor complex. A series of approaches based on cross-linking, followed by pull-down of RNA species with labeled probes, and mass spectrometric identification of associated proteins have been developed to identify lncRNA-binding proteins. In the case of XIST, three independent strategies led to the identification of over 80 XIST binding proteins (Chu et al., 2015; McHugh et al., 2015; Minajigi et al., 2015). Subsequent studies validated many of these interactions and elucidated how they contribute to the process of X chromosome inactivation (Brockdorff et al., 2020). On the other hand, NORAD was found to specifically associate with the Pumilio (PUM) proteins PUM1 and PUM2 through a series of 15 repetitive Pumilio response elements (PREs; Lee et al., 2016; Tichon et al., 2018). An emerging theme from these studies has been the potential role of lncRNAs–protein interactions in nucleating phase-separated compartments with unique biophysical and molecular characteristics, such as the XIST cloud (Pandya-Jones et al., 2020), NEAT1-containing paraspeckles (Yamazaki et al., 2018), and NORAD-dependent NP bodies (Elguindy & Mendell, 2021).

2.3 Unique challenges with the functional dissection of lncRNA-producing cis-regulatory loci

While initial studies proposed that many cis-acting lncRNAs may act as scaffolds that guide the recruitment of protein-binding partners to specific genomic locations (Rinn & Chang, 2012), subsequent models have questioned the role of the mature RNA molecule in cis-regulation. A major challenge to the functional characterization of cis-acting lncRNAs has been the difficulty of experimentally dissociating the mature transcript from the transcriptional process or the underlying DNA regulatory elements within the locus (Bassett et al., 2014; Kopp & Mendell, 2018; Figure 4). While commonly used for mRNAs and trans-acting lncRNAs, RNAi knockdown of chromatin-bound lncRNAs is expected to be ineffective. As an alternative, ASO-mediated degradation by RNAse H acts co-transcriptionally in the nucleus to efficiently deplete cis-acting lncRNAs (Figure 4a). However, recent studies have suggested that ASOs might also disrupt the act of transcription (Lai et al., 2020; Lee & Mendell, 2020).

Experimental approaches for dissecting the activity of cis-regulatory lncRNAs. Left—(a) ASO gapmers act co- or post-transcriptionally to trigger RNAse H-dependent degradation of target RNAs (Lai et al., 2020; Lee & Mendell, 2020); (b,c) Insertion of short genetic elements such as a PAS or self-cleaving ribozyme permits the stable, specific knockdown of host lncRNAs (Engreitz et al., 2016; Latos et al., 2012; Sleutels et al., 2002; Tuck et al., 2018); (d,e) CRISPRi, based on recruitment of catalytically inactive dead Cas9 (dCas9), may disrupt transcription by sterically blocking transcriptional elongation by Pol II (Dahlman et al., 2015). This inhibitory effect can be augmented by fusing Cas9 to repressive chromatin-modifying proteins such as the KRAB domain (Gilbert et al., 2013). Analogously, CRISPRa, exploits the ability of dCas9 to target fused activating domains to promoters, thereby boosting transcription (Gilbert et al., 2014); (f) Various genetic approaches, such as sequence inversion (Mohammad et al., 2008), 5′ splice site (SS) or downstream exon deletions (Allou et al., 2021; Engreitz et al., 2016) have been used to investigate the importance of specific sequence elements. Right—Table indicates whether the approach is expected to affect DNA elements (DNA), act of transcription (TXN), or the RNA molecule (RNA); “Yes”, the element is affected, “No”, the element is not affected, and “?” the effects are unknown or context-specific

As previously discussed, genetic deletion of a lncRNA-associated regulatory element, promoter, or locus effectively abolishes the production of the lncRNA molecule but also affects the act of transcription and perturbs the DNA sequence (Bassett et al., 2014; Figure 3). Vice versa, amplification of entire loci leads to increased lncRNA expression but also increases the copy number of cis-acting DNA regulatory sequences, such as enhancers (Tseng et al., 2014). To overcome some of the limitations, studies have employed PAS or STOP cassette insertions in endogenous lncRNA loci to disrupt transcription without removing underlying DNA elements (Allou et al., 2021; Anderson et al., 2016; Engreitz et al., 2016; Latos et al., 2012; Sleutels et al., 2002; Figure 4b). These approaches have been used to determine that the length of transcription is key to the function of some cis-regulatory lncRNAs, such as lncRNAs from imprinted loci. Weaknesses of these approaches include the observation that the insertion of a single PAS, while convenient from a genome editing perspective, can be inefficient in the context of longer or highly transcribed lncRNAs (Engreitz et al., 2016). Moreover, PAS-mediated termination still allows the production of nascent transcripts, which may include key functional elements. While more efficient in terminating transcription, larger STOP cassettes, which usually contain multiple polyadenylation sequences and a selection marker, may have additional consequences such as altering the chromatin organization of the locus.

As an emerging alternative, the insertion of self-cleaving ribozymes in lncRNA transcripts has been used to achieve transcript-specific knockdown (Tuck & Bühler, 2021). Self-cleaving ribozymes are naturally occurring bacterial and viral RNA elements, such as the Hammerhead ribozyme and the Hepatitis Delta virus ribozyme, with short sequences capable of forming tertiary structures that undergo a self-cleavage reaction (Tang & Breaker, 2000). When inserted into a lncRNA, self-cleaving ribozymes have been shown to induce the cleavage and subsequent degradation of the host transcript (Camblong et al., 2009; Figure 4c). Further high-throughput efforts have been dedicated to engineering these ribozymes into drug-responsive devices (Xiang et al., 2019). While this line of approaches has been hailed as the first genetic tool to effectively dissociate the mature RNA molecule from the act of transcription, a number of additional questions remain to be addressed. For example, it is not clear to what extent the sequence and local structural context at the insertion site may impact ribozyme activity (Tuck et al., 2018). Moreover, it is not clear how the insertion of various self-cleaving ribozymes affects the production and processing of nascent and mature lncRNA transcript (Fong et al., 2009).

Clustered regularly interspaced short palindromic repeats (CRISPR)-based tools for epigenetic control have recently gained popularity as they have the unique capability to modulate the expression of lncRNAs from their endogenous loci and are therefore particularly useful for the dissection of the functions of cis-regulatory lncRNAs (Gilbert et al., 2014; Figure 4d,e). CRISPR inhibition (CRISPRi) can disrupt transcription by sterically blocking elongating RNA Polymerase II (Pol II) or by recruiting repressive chromatin-modifying enzymes, such as Kruppel associated box (KRAB), to the lncRNA promoter (Gilbert et al., 2013; Liu et al., 2020). CRISPR activation (CRISPRa), on the other hand, can activate the expression of lncRNAs through the local recruitment of activating factors, such as p65 and heat shock transcription factor 1 (HSF1) in CRISPR-synergistic activation mediator (CRISPR-SAM; Bester et al., 2018; Dahlman et al., 2015). One CRISPRi screen highlighted the cis-repressive activities of the promoter of the lncRNA Pvt1 on the expression of the Myc oncogene (Cho et al., 2018). Combined CRISPRi and CRISPRa approaches were analogously used to determine that production of the p53-regulated lncRNA LincRNA-Gadd45γ is both necessary and sufficient for the activation of its neighbor, Gadd45γ, during the cellular response to stress (Tesfaye et al., 2021). On the other hand, CRISPR-based technologies have opened some unique challenges, including the potential of CRISPRi and CRISPRa factors to affect locations that are genomically distant but physically proximal to the promoter of the target lncRNA due to the three-dimensional chromatin organization (Thakore et al., 2015).

Additional targeted genetic perturbations of various RNA features have also informed on the diverse functional features of cis-acting lncRNAs (Figure 4f). For example, deletion of the first 5′ splice site in the lncRNA Blustr was found to recapitulate the effects of Blustr promoter deletion and PAS insertion on the nearby Scm like with four MBT domains 2 (Sfmbt2) gene, suggesting that promoter-proximal splicing is critical for cis-regulation by the Blustr locus (Engreitz et al., 2016). In contrast, deletion of three downstream exons had no effect on Sfmbt2 levels, indicating that the full-length lncRNA sequence is dispensable (Engreitz et al., 2016). In contrast, both deletion and inversion of a region of KCNQ1OT1 relieved the repression of nearby imprinted genes, suggesting a role for production of the mature lncRNA (Mohammad et al., 2008, 2010). Taken together, these examples illustrate the power of complementary genetic approaches to provide important insights into the mechanisms of cis-regulation.

Crucially, for all of these approaches, the inability to experimentally rescue the deficiency of cis-acting lncRNAs with exogenously expressed constructs has made it difficult to control for potential off-targets.

4 CONCLUSION

A growing body of work has linked alterations in lncRNAs to cancer initiation and progression. Additionally, an array of complementary genetic, molecular, and biochemical approaches has been employed in an effort to elucidate the functions and mechanisms by which these lncRNAs contribute to tumorigenesis. These studies have provided intriguing examples of how alterations in lncRNAs result in the sequestration or redistribution of lncRNA-binding partners and lead to perturbations of the epigenetic state, gene expression balance, or other processes in cancer cells. These types of gain-of-function activities have emerged as the predominant mechanism by which overexpressed trans-acting lncRNAs perturb cellular homeostasis and enable the acquisition of cancer hallmarks. A similar pattern has also been observed for cis-acting lncRNAs. On the one hand, low abundance, cis-acting lncRNAs have been found to mislocalize to distant sites in the context of oncogenic overexpression, leading to gain-of-function trans activities. As an example, LincRNA-p21, which is chromatin-associated and expressed at just a few copies per cell, appears to be highly overexpressed and to localize and function in the cytoplasm of several human cancer cell lines (Yang et al., 2014; Yoon et al., 2012). On the other hand, the oncogenic upregulation or downregulation of high abundance, cis-acting lncRNAs, such as XIST, may also result in the sequestration or release, respectively, of a large pool of protein-binding factors, including chromatin modifiers and transcriptional regulators, which may in turn lead to a global epigenetic unbalance. There have also been examples of how perturbations in cis-regulation can contribute to cellular transformation through local disruption of gene expression patterns. Taking all of these examples into account, the emerging theme is that deregulated lncRNAs frequently acquire gain-of-function activities that may or may not be related to their physiological functions. Thus, lncRNA-targeting therapies focused on lncRNA degradation or on the disruption of functional interactions may be effective in reversing cancer phenotypes.

Finally, it is important to note that, in addition to RNA-based models, the functional impact of unconventional mechanisms should also be considered. The growing number of short functional peptides encoded by putative trans-acting lncRNAs has opened new avenues for exploration. Similarly, elucidating the local interplay between regulatory DNA elements, the act of transcription, and the functional elements in the mature lncRNA molecule will be essential to dissecting the mechanisms of cis-regulatory loci.

ACKNOWLEDGMENTS

The authors are grateful to Elena Martinez for her insightful comments. Figures were created with BioRender.com.

CONFLICT OF INTEREST

The authors declare no conflicts of interest.

AUTHOR CONTRIBUTIONS

Lauren Winkler: Conceptualization (equal); writing – original draft (lead); writing – review and editing (equal). Nadya Dimitrova: Conceptualization (equal); funding acquisition (lead); writing – original draft (supporting); writing – review and editing (equal).