The Long and the Short of It: NEAT1 and Cancer Cell Metabolism

Simple Summary Altered metabolism is a hallmark of most cancers. The way that cancer cells regulate their energy production to fuel constant proliferation has been of interest with the hope that it may be exploited therapeutically. The long noncoding RNA, NEAT1, is often dysregulated in tumours. NEAT1 RNA can be transcribed as two isoforms with different lengths, with each variant responsible for different functions. This review explores how the isoforms contribute to cancer metabolism. Abstract The long noncoding RNA NEAT1 is known to be heavily dysregulated in many cancers. A single exon gene produces two isoforms, NEAT1_1 and NEAT1_2, through alternative 3′-end processing. As the longer isoform, NEAT1_2 is an essential scaffold for nuclear paraspeckle formation. It was previously thought that the short NEAT1_1 isoform only exists to keep the NEAT1 locus active for rapid paraspeckle formation. However, a recent glycolysis-enhancing function for NEAT1_1, contributing to cancer cell proliferation and the Warburg effect, has been demonstrated. Previous studies have mainly focused on quantifying total NEAT1 and NEAT1_2 expression levels. However, in light of the NEAT1_1 role in cancer cell metabolism, the contribution from specific NEAT1 isoforms is no longer clear. Here, the roles of NEAT1_1 and NEAT1_2 in metabolism and cancer progression are discussed.


Background
In healthy cells, glucose uptake stimulates cell growth through the activation of intracellular signalling pathways, including glycolysis [1]. In cancer, genetic and epigenetic alterations switch these pathways to a permanently "on" state, promoting continuous growth, which depletes key metabolites. As tumour size increases, oxygen availability decreases, limiting the oxygen-dependent final step of the electron transport chain (ETC). This shift from oxidative phosphorylation (OXPHOS) to aerobic glycolysis is termed the Warburg effect [2][3][4]. Although there is an increase in both glycolytic enzymes and activity in many cancers, the Warburg effect alone does not explain why cancer cells maintain enhanced glycolysis in normoxic cell culture nor in circulating blood [4].
In general, the functions of long noncoding RNAs (lncRNAs) are not well-understood due to their generally low expression levels and high tissue specificity [5]. An exception to this is nuclear enriched abundant transcript 1 (NEAT1) dysregulation in many diseases [6]. The highly conserved, single exon, intergenic lncRNA is transcribed near the multiple endocrine neoplasia locus on human chromosome 11q13.1 [7]. Altered 3 -end processing results in two transcripts: 3.7 kb NEAT1_1 and 22.7 kb NEAT1_2; the latter is a welldocumented and essential architectural scaffold for subnuclear paraspeckles (see review by The NEAT1 gene gives rise to two isoforms with identical 5 sequences. The paraspeckleindependent NEAT1_1 undergoes canonical 3 polyadenylation whilst the blocking of 3 polyadenylation via competitive binding of hnRNPK to the CFIm complex yields the paraspeckle-essential NEAT1_2. The recruitment of paraspeckle proteins to the hydrophilic and hydrophobic regions of NEAT1_2 leads to phase-separated paraspeckles.

NEAT1_1 Enhances the Warburg Effect by Accelerating Glycolytic Metabolite Flux
Although both isoforms are abundant in the nucleus, NEAT1_1 is also exported to the cytoplasm [12]. In 2019, Adriaens et al. suggested that NEAT1_1 is potentially nonfunctional and serves to only keep the transcriptional locus active, making the switch to NEAT1_2 rapidly available during stress induction [13]. However, a recent study has discovered a novel mechanism of action for NEAT1_1 in the cytoplasm, in both in vitro and in vivo breast cancer (BC) models [12]. The authors demonstrated that the translocation of NEAT1_1 from the nucleus to the cytoplasm, through binding with the nuclear speckle-associated protein pinin (encoded by the PNN gene), occurs in a glucose-dependent manner [12,14] ( Figure 2). Interestingly, the depletion of pinin reduces cytoplasmic NEAT1_1 even after glucose stimulation, indicating the importance of pinin in the nucleocytoplasmic transport of NEAT1_1 [12]. Importantly, in the cytoplasm, NEAT1_1 interacts with the glycolytic enzymes phosphoglycerate kinase (PGK1), phosphoglycerate mutase (PGAM1), and alpha enolase (ENO1) to promote glycolysis, enhancing growth, proliferation, invasion, and metastasis ( Figure 2) [12,15]. Nucleocytoplasmic transport of NEAT1_1 enhances the Warburg effect via the binding to glycolytic enzymes. TDP-43, ARS2, and CFIm promote the canonical 3 polyadenylation of NEAT1_1, which can then bind to pinin for nuclear export to the cytoplasm. Once in the cytoplasm, NEAT1_1 can bind with the glycolytic enzymes PGK1 (A), PGAM1 (B), and ENO1 (C) to enhance glycolytic flux, and hence the Warberg effect, in transformed cells.
PGK1 is constitutively expressed in all somatic and premeiotic cells and is essential in the glycolysis pathway, but it is implicated in cancer as it encompasses many characteristics of an oncogene [16]. This bilobed enzyme has both nucleotide-binding and catalytic domains and is involved in the conversion of 1,3-bisphosphoglycerate (1,3-BPG) to 3phosphoglycerate (3-PG) in glycolysis (Figure 2A). It also catalyses the first ATP of anerobic glycolysis [16,17]. This substrate-level phosphorylation is of great significance in the continuous production of cellular energy under hypoxic conditions [16]. PGK1 also acts as a protein kinase after translocation to the mitochondria, where it directly phosphorylates pyruvate dehydrogenase kinase isozyme 1 (PDHK1) [17]. Phosphorylated PDHK enhances pyruvate dehydrogenase E1α, which inactivates pyruvate dehydrogenase, preventing the conversion of pyruvate to coenzyme A (CoA), thus suppressing mitochondrial pyruvate metabolism and increasing lactate production [18]. This rate-limiting enzyme plays a role in the promotion of tumourigenesis through the activation of oncogenic pathways, such as Akt/mTOR, Myc, and ß-catenin, and post-translational modifications, such as phosphorylation, acetylation, ubiquitination, and succinylation (as reviewed by Liu et al., 2022) [19]. In vitro research by Gou et al. [20] and Wang et al. [21] have shown that the small-molecule inhibitor of PGK1, NG52, had a dose-dependent effect on the proliferation of ovarian cancer and glioma cells, respectively. PGAM1 is a highly conserved glycolytic isomerase enzyme involved in the conversion of 3-PG to 2-phosphoglycerate (2-PG), whilst also supporting antioxidative defences by reducing mitochondrial reactive oxygen species (ROS) ( Figure 2B) [22,23]. Several studies have linked PGAM1 to cancer progression. Earlier works report PGAM1 knockdown by short hairpin RNA (shRNA) results in an increase in 3-PG and a subsequent decrease in 2-PG, whilst also decreasing glycolysis, pentose phosphate pathway flux, biosynthesis, and cell proliferation in diverse solid and leukaemia cell lines [24]. Investigation of dysregulated PGAM1 levels in human urothelial bladder cancer tissues found a positive correlation with histological-grade tumours, when compared to adjacent normal tissue [25]. Loss of functional tumour-suppressor protein p53, encoded by tumour protein 53 (TP53), is common in cancer, and it has been found to upregulate both NEAT1 and PGAM1 [23]. Similar to many metabolic enzymes, PGAM1 asserts a nonenzymatic function, as the physical interaction with checkpoint kinase 1 (Chk1) increases proliferation, specifically in RAS-driven cancer cells [26].
Alpha enolase (ENO1), encoded by ENO1, is another tumour-related, multifunctional protein, which is responsible for the conversion of 2-PG to phosphoenolpyruvate (PEP) in glycolysis ( Figure 2C). An increase in ENO1 expression has been reported in human diseases (i.e., systemic sclerosis, type II diabetes mellitus, lupus, Alzheimer disease) and many cancers, as well as being involved in cell growth and hypoxia tolerance [27][28][29][30][31][32][33][34]. Additionally, silencing ENO1 reduces the rate of glycolysis in cell lines, favouring OXPHOS even when glucose influx remains high [35]. ENO1 expression is correlated with colorectal cancer (CRC) progression, and the newly identified protein translational modification, lysine crotonylation, has been identified at lysine residue 420 in CRC cell lines [33]. Interestingly, although ENO1 is reportedly overexpressed in many human diseases and cancers, in non-small-cell lung cancer (NSCLC), ENO1 is downregulated at the protein level even though its mRNA levels remain elevated, suggesting post-transcriptional regulation [36,37].
Given the cancer-specific roles for each of these enzymes, the role that NEAT1_1 plays, either in glycolysis or in sequestering enzymes from other activities, requires further investigation.  [38][39][40]. Paraspeckle formation relies solely on the generation of NEAT1_2 in the nucleus, which sets them apart from cytoplasmic stress granules and P bodies, which require multiple proteins and RNA elements to form [8]. Paraspeckle formation occurs only following the recruitment of proteins, such as non-POU-domain-containing octamer-binding protein (NONO), splicing factor proline and glutamine-rich (SFPQ) and fused in sarcoma (FUS) proteins, among others, to the midregion of NEAT1_2 transcript. Once localised, the high concentration of molecules aggregates to form a distinct spheroid with spatial organisation [41] (Figure 1). Hydrophilic proteins bind 3 and 5 regions of the NEAT1_2 transcript to form the paraspeckle shell, whilst the middle segment of the transcript forms the hydrophobic core [8,10,38,42]. Individ-ual paraspeckles are spheroidal, but during stress conditions, they can be linked together to generate elongated paraspeckle structures [15]. In HeLa cells, there are~5-20 paraspeckles per nucleus [43]. However, in 2020, Grosch et al. reported that the size of the nuclei likely influences paraspeckle abundance in human pluripotent stem cells [43]. Regardless of basal paraspeckle abundance, their numbers seem to increase during stress, suggesting that NEAT1_2 accumulates.

What Is Driving the NEAT1 Isoform Switch?
Since the formation of paraspeckles is dependent on the transcriptional read-through of NEAT1_1 to NEAT1_2, understanding the mechanistic control of this isoform switch is crucial. Polyadenylation (polyA) signals terminate the primary NEAT1 transcript, resulting in canonical processing of the 3 polyA tail and consequently a short NEAT1_1 lncRNA [44]. The long NEAT1_2 isoform is generated when heterogeneous nuclear ribonucleoprotein K (hnRNPK) competes with cleavage-and-polyadenylation-specific factor 6 (CPSF6) for nudix hydrolase 21 (NUDT21) binding, inhibiting the CPSF6-NUDT21 (CFIm) complex from forming, and facilitating the 3 -end polyA ( Figure 2) [44,45]. In vitro binding assays have demonstrated that inhibiting the formation of the CFIm complex prevents the polyadenylation of NEAT1_1 and increases the nuclear levels of NEAT1_2 and, thereby, paraspeckles [44,46]. Additionally, recent work has reported that arsenic resistance protein 2 (ARS2) acts as a chaperone, guiding CFIm to the NEAT1 transcript to facilitate the polyadenylation of NEAT1_1 in osteosarcoma cell lines [47]. Furthermore, the knockdown of ARS2 leads to an increase in, and the preferential stabilisation of, NEAT1_2 transcripts [47]. Moreover, RNA binding protein (RBP) transactive response (TAR) DNA binding protein 43 kDa (TDP-43) directly represses the formation of paraspeckles, but it increases NEAT1_1 transcription by binding the NEAT1_1 GU-rich motifs upstream of the polyA site [48,49]. Although TDP-43 has been thoroughly investigated in amyotrophic lateral sclerosis and has been linked to altered miRNA expression, the understanding of its involvement in a cancer context remains limited [49][50][51][52][53][54][55]. In summary, the isoform switch from NEAT1_1 to NEAT1_2 may involve several factors with context-specific roles; their oncogenic relevance requires further clarification.

NEAT1 and Paraspeckles Alter Metabolism via Mitochondria
Mitochondrial function reaches beyond the established role in energy generation. In addition to ATP production, mitochondria generate macromolecules which alleviate mitochondrial stress [56]. Interestingly, the mitochondrial stressors, FCCP, rotenone, and doxycycline, all increase NEAT1 levels, in part through ATF2-induced NEAT1 expression [10]. Mito-nuclear communication is crucial for ensuring cellular homeostasis during mitochondrial stress, and recently it has been hypothesised that NEAT1 and paraspeckles may play a role in mitochondrial homeostasis [57,58]. Mitochondrial fusion and fission are controlled by dynamin-related GTPases MFN1/MFN2 and DRP1, respectively [58]. NEAT1depletion in HeLa and HEK293 cells resulted in mitochondrial elongation and the enhanced retention of mito-mRNAs encoding functional mitochondrial proteins, such as cytochrome c, subunits of NADH dehydrogenases, and carnitine o-palmitoyl transferase [10]. This was further supported by reduced DRP1 but stable MFN1 and MFN2 expression, increased mitochondrial mRNAs (mito-mRNAs) exported from the nucleus, reduced respiration capacity, ATP generation, extracellular acidification rate (ECAR), and proliferation [10]. On the contrary, NEAT1 overexpression showed an increase in DRP1 expression and DRP1 phosphorylation, corresponding to fragmented mitochondria and increased mitochondrial DNA (mtDNA), and this was phenocopied by deleting the NEAT1_1 polyadenylation signal [10].

Alternative Processing of lncRNAs in Cancer
Many lncRNAs are implicated in cancer, either through direct or indirect processing, as previously reviewed [59,60]. Similarly to NEAT1, the noncoding product of a neighbouring locus metastasis-associated lung adenocarcinoma transcript 1 (MALAT1) is retained in the nucleus and is involved in the nuclear architecture as well as in RNA splicing and gene-regulation [61]. Whilst MALAT1 has been found to be overexpressed in 14% of breast tumour samples, an alternatively spliced variant of MALAT1 (∆sv-MALAT1) showed decreased expression in a subset of tumours and shows potential as an individual prognostic factor for BC [62]. Additionally, similar to NEAT1, MALAT1 contains a tRNA-like cloverleaf structure near the 3 terminus; however, this structure is cleaved [63] and accumulates as MALAT1-associated small cytoplasmic RNA (mascRNA) in the cytoplasm, where it promotes global translation [64] and hepatocellular cancer cell proliferation [65]. Additionally, the lncRNA ANRIL recruits polycomb proteins to regulate target-gene expression, and the overexpression of truncated isoforms has been reported in bladder cancers [66]. Clearly, the function of many lncRNAs is nuanced and not limited to the full-length transcript. Understanding the activities of truncated forms and processed derivatives of lncRNAs will enlighten our understanding of these complex regulatory molecules.

Dysregulation of Both Short and Long NEAT1 in Cancer
NEAT1 (without distinguishing between the long and short isoform) is expressed at similar levels in most healthy tissues, but it is reported to have a relatively low expression in the brain, heart, and whole blood [6]. On the contrary, NEAT1 is upregulated in many solid cancers (reviewed in [6,[67][68][69][70]), and in most cases, it is associated with aggressive disease and poor outcomes [67][68][69]. However, only a handful of recent studies have established the relative abundance of the two isoforms, despite many reports suggesting that NEAT1 acts as either a tumour suppressor or oncogene, depending on this isoform ratio (summarised in Table 1) [70][71][72][73][74]. In some haematological malignancies, NEAT1 functions as a tumour suppressor, enhancing the expression of NEAT1_2, and thus paraspeckle formation, which may counteract oncogene-induced stressors in cancer [75,76]. NEAT1 levels are upregulated in multiple myeloma (MM) when compared to healthy controls; however, no correlation with patient prognosis has been found [77]. In chronic lymphocytic leukaemia (CLL), the total NEAT1 expression levels remained similar to those of healthy controls, but the expression of NEAT1_2 was found to be significantly higher [77]. On the other hand, in acute and chronic myeloid leukaemia (AML and CML, respectively) and acute lymphoblastic leukaemia (ALL), total NEAT1 levels decrease in patients' peripheral blood and bone marrow, and this was found to be an essential mediator of apoptosis induced by imatinib in BCR-ABL-expressing cells [75,[77][78][79][80]. NEAT1_1 transcription was analysed by using a polyA primer for cDNA generation before RT-qPCR, using primers targeting total NEAT1. NEAT1_2 transcription was analysed by using random primers for cDNA generation before primers specifically targeting the NEAT1_2 region of the transcript. ↑ NEAT1 associated with poor patient prognosis.

GAPDH
2017 [111] Tumour suppressor p53 is regarded as the guardian of the genome, but it is mutated in over 50% of malignancies, enabling cells to escape apoptotic signalling, bypass cellcycle arrest, and inhibit senescence [112,113]. It is well-established that NEAT1 is induced by p53 binding to the NEAT1 promotor to activate expression [76,112,114]. Interestingly, NEAT1 also promotes p53 and Chk1 through ATR signalling in response to replication stress [76]. Furthermore, in CRC cells, the induction of both NEAT1 isoforms were p53dependent when exposed to a chemotherapeutic agent and the topoisomerase 2 inhibitor, doxorubicin [112]. In CML, p53 mutations are uncommon [78]; instead, MYC binds to the NEAT1 promotor to enhance expression [78]. Accordingly, Ronchetti et al. reported an increase in total NEAT1 and NEAT1_2 expression in CML patients when compared to normal B-cells [77].
A positive correlation was identified for increased NEAT1 expression and higher histological grades of BC, and NEAT1 levels were elevated in patient plasma and peripheral blood although no relative abundance of NEAT1 isoforms was reported [11]. The same study reported higher expression of NEAT1 in estrogen receptor positive (ER+) breast cancers, when compared to estrogen-receptor-negative-(ER-) BC [11]. Similarly, correlation to lymph node positivity was seen with ER+, but not with ER- [11]. Evidence suggests that NEAT1 point mutations are drivers for breast and prostate cancers, regardless of little change in NEAT1 transcription levels [115][116][117]. However, a more recent study suggested these mutations were likely caused by transcription errors instead of cancerspecific selection pressures [118]. Regardless, NEAT1 downregulation has been reported in invasive breast carcinoma, oesophageal cancer, pheochromocytoma, and paraganglioma, suggesting NEAT1 plays a tumour-suppressor role in these malignancies [6]. In summary, it is apparent that NEAT1 expression varies greatly between malignancies and that relative isoform abundance may play a key role in disease progression.

Considerations for Isoform Detection of NEAT1
Most recent NEAT1 studies have concentrated on the paraspeckle-associated long isoform NEAT1_2, at the risk of overlooking essential roles for the shorter NEAT1_1, paraspeckle-independent isoform. In addition, an analysis of the literature (Table 1) shows the reliance placed upon glyceraldehyde-3-phosphate dehydrogenase (GAPDH) transcripts levels for normalizing NEAT1 expression in RT-PCR studies. As new light has been shed on the involvement of NEAT1_1 in glycolysis, a glycolytic housekeeping transcript, such as GAPDH, may not be ideal for normalizing NEAT1 expression in studies related to metabolism. Inconsistent polyadenylation and contextual processing can make lncRNAs difficult to quantify using standard RT-PCR protocols. Quantifying the relative abundance of NEAT1 isoforms is not straightforward. Kolenda et al. [119] compared cDNA synthesis protocols for various cancer-associated lncRNAs; however, isoform differentiation was not a primary outcome. Similarly, the RNA purification method used can significantly influence the relative abundance of isoforms, with a heating step liberating NEAT1_2 from paraspeckle complexes and increasing yields [120]. While oligo-dT primed cDNA might be used to specifically amplify NEAT1_1 sequences, the presence of poly-A stretches downstream in the NEAT1_2 sequence necessitates the careful calibration of reverse transcription conditions, which is rarely reported. Validated oligo-dT clamp cDNA protocols would be advantageous.
It may be expected that RNA-seq should enable accurate comparison of isoforms; however, many studies report oligo-dT primed libraries, such that the long isoform is under-represented or ignored, and even total RNA-seq data may be influenced by the aforementioned bias in isoform ratios as a result of RNA isolation methods. Newer, longread direct RNA sequencing methods promise improved qualitative and quantitative data [121].
Visualisation of NEAT1_2, employing RNA-fluorescence in situ hybridization, is often used to quantify paraspeckle abundance and can also be directed to detect NEAT1_1 [13]. Similarly, dCas13 tagging has recently been used to detect both isoforms of NEAT1 in living cells [122].

Conclusions and Future Directions
In the context of cancer, NEAT1 may have either a protective, tumour-suppressive role, or a tumour-promoting oncogenic role, depending upon the type of cancer and, most likely, also upon the specific NEAT1 isoform expressed. Many previous studies have concentrated on the total NEAT1 transcription level, rather than on the isoform ratio; hence, isoform-specific contributions are unclear. Improving detection of specific NEAT1 isoforms is crucial in understanding the tumour-promoting vs. the tumour-protective roles of NEAT1 in cancer. Future directions will compare the relevant contributions of paraspecklemediated sequestration and epigenetic regulation against the glycolytic influence of the shorter isoform on tumour progression. Whether one or both isoforms are found to be necessary for specifically maintaining neoplasia, that will impact their relative value as therapeutic targets.

Conflicts of Interest:
The authors declare no conflict of interest. The funding agencies had no role in the planning or writing of the manuscript, nor in the decision to publish.