Trans-spliced long non-coding RNA: an emerging regulator of pluripotency

Abstract

With dual capacities for unlimited self-renewal and pluripotent differentiation, pluripotent stem cells (PSCs) give rise to many cell types in our body and PSC culture systems provide an unparalleled opportunity to study early human development and disease. Accumulating evidence indicates that the molecular mechanisms underlying pluripotency maintenance in PSCs involve many factors. Among these regulators, recent studies have shown that long non-coding RNAs (lncRNAs) can affect the pluripotency circuitry by cooperating with master pluripotency-associated factors. Additionally, trans-spliced RNAs, which are generated by combining two or more pre-mRNA transcripts to produce a chimeric RNA, have been identified as regulators of various biological processes, including human pluripotency. In this review, we summarize and discuss current knowledge about the roles of lncRNAs, including trans-spliced lncRNAs, in controlling pluripotency.

Introduction

Pluripotent stem cells (PSCs) categorically include both embryonic stem cells (ESCs) and induced PSCs (iPSCs). Both these subcategories of human PSCs (hPSCs) are able to give rise to many cell types from our body [1, 2] and this broad differentiation potential, known as pluripotency, makes hPSCs a valuable platform for studying early events in human embryonic development. Moreover, the capacity for unlimited self-renewal makes hPSCs a feasible cell source for use in regenerative medicine. The molecular pluripotency circuit in hPSCs is composed of and maintained by multilayered, coordinated gene expression networks. Thus, the core components of this circuitry are transcription factors, especially including NANOG, OCT4 and SOX2. Collectively, these transcription factors provide a point of integration for extracellular signals and orchestrate with epigenetic modifications on histones or chromatin to maintain hESCs and iPSCs in a long-term, proliferative state while suppressing differentiation. Recent studies have shown that, in addition to protein-coding genes, lncRNAs (> 200 nucleotides) are vital to various regulatory mechanisms of pluripotency maintenance and reprogramming. As such, the expression of lncRNAs has been shown to affect promoter-driven gene transcription [3, 4], epigenetic modification of histones and chromatin [5, 6], post-transcriptional regulation [7], miRNA availability through sponge function [8, 9], and imprinting [10, 11]. Several previous reviews have comprehensively discussed the expression and function of lncRNAs in pluripotency maintenance, reprogramming and lineage differentiation, including the mechanisms by which lncRNA regulates these biological events [12–14].

Both protein-coding mRNAs and lncRNAs are spliced and poly-adenylated to generate mature transcripts. However, splicing can occur either in cis or in trans [15, 16]. Cis-splicing joins exons from a single precursor mRNA (pre-mRNA), while trans-splicing joins exons from two or more distinct pre-mRNAs [17–20]. The best-characterized example of trans-splicing involves spliced-leader (SL) RNA in trypanosomes and nematodes. In SL trans-splicing, an identical 5′ capping exon is spliced onto various pre-mRNAs by a process that is dependent on canonical spliceosome function and the secondary structure of the SL RNA. This form of trans-splicing plays a pivotal role in processing the polycistronic transcriptional units of trypanosomes [18, 21] and is involved in the growth recovery of nematodes [22]. However, trans-splicing events in higher eukaryotes are not SL-type, and the functions and mechanisms of trans-splicing in these organisms are still unclear. In higher eukaryotes, trans-spliced RNAs have been shown to regulate apoptosis and axon guidance decisions in Drosophila [23, 24] and are associated with cancer development in humans [25]. With recent advances in next-generation sequencing technology, a continuously increasing number of trans-splicing events have been discovered in human cells or tissues. These events have many functions and recently a functional role for trans-spliced RNAs in human pluripotent stem cells has been described [5].

In this review, we summarily describe the molecular circuity underlying pluripotency maintenance in pluripotent stem cells and then highlight new discoveries that demonstrate a critical role for lncRNA in regulating pluripotency. Furthermore, we provide a summary of known trans-splicing events and describe how these events function in various biological process. Finally, we discuss how trans-spliced and non-trans-spliced lncRNAs are coordinated with transcription factors, epigenetic modification complexes and signaling molecules to constitute a molecular pluripotency network in hPSCs, focusing on pluripotency maintenance in ESCs and the establishment of pluripotency through reprogramming in iPSCs.

Human pluripotent stem cells (hPSCs): hESCs and iPSCs

During preimplantation stages, a fertilized egg undergoes a series of cleavage divisions and forms a compact embryonic sphere, known as a morula, which consists of blastomeres. The human morula further develops into a blastocyst, which contains an inner cell mass (ICM) and a fluid-filled cavity surrounded by a thin layer of trophectodermal epithelium. The ICM eventually gives rise to the embryo and the trophoblasts contribute to placenta formation. At this preimplantation stage, hESCs may be derived from the ICM and these cells will retain the pluripotency that is characteristic of the blastocyst. The successful derivation of hESCs has facilitated studies of regulatory mechanisms that operate during early embryo development and has also benefited studies on regenerative medicine. However, the generation of hESCs critically compromises the integrity of human embryos, thus raising the ethical question of whether it is acceptable to destroy human embryos for the purpose of hESC derivation. Fortunately, in 2007, Shinya Yamanaka’s group first demonstrated that iPSCs can be reprogrammed from human somatic cells by the expression of four transcription factors, OCT4, SOX2, KLF4 and c-MYC [26]. These iPSCs possess functional characteristics that are similar to hESCs, but unlike hESCs, the derivation of human iPSCs does not involve human embryos. Therefore, the use of iPSCs avoids the complex ethical issues surrounding the derivation and use of hESCs, while providing a viable source of hPSCs for experimentation and potential clinical application. Various in vitro and in vivo studies have shown that iPSCs are able to give rise to cell types representing all three embryonic germ layers [27–29], including endodermal hepatocytes [30], mesodermal cardiomyocytes [31], and various neuronal subtypes found in brain [32, 33] or spinal cord [27] (ectoderm). As such, iPSC technology provides an easy and efficient means to generate hPSCs from individuals with familial or sporadic forms of disease, as well as unlimited numbers of specific types of human cells that are not otherwise accessible for disease modeling and drug discovery. Moreover, iPSCs can be derived from any individual, providing immune-compatible, person-specific iPSCs for medical use. The iPSC reprogramming process also offers an opportunity to explore how the molecular circuity of pluripotency is established and maintained [34]. Overall, hESCs and iPSCs represent two valuable, but distinct, platforms for regenerative medicine and basic research on pluripotency circuitry.

The transcriptional regulatory network in pluripotency maintenance

OCT4/Oct4 (homeodomain transcription factor), SOX2/Sox2 (HMG-box transcription factor) and NANOG/Nanog (homeodomain transcription factor), which are all highly expressed in the ICM, the epiblast and undifferentiated PSCs, are the core molecules of a transcriptional regulatory network that controls pluripotency maintenance [35–38]. The functional roles of OCT4/Oct4, SOX2/Sox2 and NANOG/Nanog in pluripotency were first revealed in vivo by creating null mutations in mouse embryos. Knockout of Oct4 in mice prevented the formation of a pluripotent ICM population within blastocysts [39], and knockout of Sox2 expression caused a similar phenotype [38]. Ablation of Nanog is lethal in early embryonic stages, suggesting that its role is also critical in early development [36, 40]. In agreement with these findings, in vitro studies of mouse ESCs (mESCs) showed that disruption of Oct4 or Sox2 expression results in the loss of pluripotency and promotes trophoblastic differentiation [39, 41]. On the other hand, overexpression of Oct4 or Sox2 in mESCs can perturb the pluripotent state and, respectively, promote differentiation toward primitive endoderm or neuroectoderm [42, 43]. Perturbing Nanog expression in mESCs causes the loss of pluripotency and promotes in vitro differentiation toward an extraembryonic endoderm lineage [36]. Similar to the mouse studies, knockdown of OCT4 expression in hESCs induces trophectoderm differentiation [44, 45]. Additionally, overexpression of OCT4 in hESCs does not necessarily induce autonomous differentiation or an exit from pluripotency, but under certain conditions may promote in vitro differentiation toward an endodermal lineage [46]. Knockdown of NANOG expression induces neuroectodermal subsets of genes with anterior–posterior identities, while overexpression does not induce differentiation, but instead completely prevents neuroectodermal differentiation [47]. Functional studies in both mESCs and hESCs have revealed that OCT4/Oct4, SOX2/Sox2 and NANOG/Nanog can bind their own promoters/enhancers and the promoters of other genes encoding pluripotency-associated factors to activate expression [37, 41, 48–52]. Further, these three transcription factors are known to occupy the promoters of genes that specify differentiation of extra-embryonic, endodermal, mesodermal and ectodermal lineages, and inactivate expression [37].

Collectively, the previously mentioned studies have revealed a mechanism in which OCT4/Oct4, SOX2/Sox2 and NANOG/Nanog can bind together at their own promoters and promoters of genes encoding other core pluripotency-associated transcription factors, forming an interconnected auto-regulatory loop to activate downstream pluripotency-associated effectors and repress lineage-associated gene expression.

Signaling pathways that regulate pluripotency in hPSCs

The convergence of extracellular signaling events acting on intrinsic core pluripotency-associated transcription factors, OCT4, SOX2 and NANOG, to regulate gene expression at promoters and enhancers is a common theme in the literature describing pluripotency maintenance [53–55]. Interestingly, while hPSCs and mESCs share the same core pluripotency-associated transcription factors, the supporting signaling pathways have proven to be quite different. In mESCs, it known that activation of LIF and BMP signaling pathways is essential for pluripotency maintenance [53, 54]. However, LIF and BMP do not promote human pluripotency, and, alternatively, activation of Activin/Nodal, FGF, and WNT signaling pathways is critical regulators [56–59]. These findings have led to the discovery of two distinct pluripotency states, namely naïve and primed pluripotency, which describe ICM-derived mESCs and epiblast-derived epiblast stem cells (EpiSCs), respectively [60, 61]. Since mouse EpiSCs and hPSCs have similar growth requirements, as well as similar cellular and molecular characteristics, it is believed that hPSCs are in a primed pluripotency state and, thus, are likely to originate from human epiblasts [59].

The roles of Activin/Nodal and FGF signaling pathways in maintaining pluripotency of hPSCs were first described by Vallier et al. and Ludwig et al., who demonstrated that inhibition of Activin/Nodal signaling pathway in hESCs promotes in vitro differentiation [55, 62]. Further studies showed that activation of the Activin/Nodal signaling mediator, SMAD2/3, promotes NANOG expression [56, 57] and inhibits differentiation-related BMP signaling by competing with the common mediator, SMAD4 [63] (Fig. 1a). However, neither Activin nor Nodal alone is sufficient to sustain hESCs in an undifferentiated state for an extended period. FGF2 can contribute to pluripotency maintenance in hESCs by activating the MAPK/ERK signaling cascade as a competence factor, which is necessary to cooperate with Activin/Nodal signaling [55] (Fig. 1b). FGF2 may also act to stimulate PI3K/AKT signaling and enhance SMAD2/3 occupancy on the promoters of pluripotency-associated genes [64, 65]. In addition to facilitating pluripotency-related gene transcription, both FGF-activated MAPK/ERK and PI3K/AKT signaling cascades can repress differentiation-related BMP signaling pathways to further promote pluripotency [57, 66, 67] (Fig. 1b).

Fig. 1.

Signaling pathways in human pluripotency maintenance. In hPSCs: a Activin/Nodal signaling molecules induce SMAD2/3 phosphorylation and interaction with SMAD4. The SMAD2/3–SMAD4 complex then translocates to the nucleus, where it promotes NANOG expression. NANOG cooperates with OCT4 and SOX2 to maintain pluripotency in hPSCs. b FGF2 activates phosphorylation of PI3K/AKT or MAPK/ERK signaling cascades, which cooperate with SMAD2/3–SMAD4 complexes to promote pluripotency in hPSCs. c Canonical Wnts inhibit the activity of the APC/AXIN/GSK-3β complex, by which β-catenin is protected from degradation. β-Catenin translocates to the nucleus, where it cooperates with TCF/LEF family members to regulate expression of target genes. Whether the canonical Wnt/β-catenin cascade maintains pluripotency or promotes differentiation of hESCs remains controversial

Wnt ligands are secreted glycoproteins that are involved in regulating diverse biological processes, such as cell proliferation, differentiation, migration, and asymmetric cell division. It is established that Wnt ligands exert their functions through the canonical Wnt/β-catenin cascade, or non-canonical Wnt/JNK and Wnt/Calcium-related pathways [68]. The activation of canonical Wnt/β-catenin signaling protects β-catenin from degradation by inhibiting the APC/AXIN/GSK-3β complex (Fig. 1c). Subsequently, stabilized β-catenin can translocate to the nucleus, where it interacts with TCF/LEF family transcription factors to regulate gene expression. Activation of canonical Wnt/β-catenin signaling maintains the naïve pluripotency state of mESCs [69], but the role of canonical Wnt/β-catenin signaling is less clear in hPSCs. Sato et al. reported that activation of Wnt/β-catenin signaling by either WNT3A or GSK-3β inhibitor, BIO, can maintain pluripotency in hESCs [70]. In contrast, Bone et al. and Nakanishi et al. demonstrated that WNT3A or BIO induces primitive and definitive endoderm differentiation of hESCs [71, 72]. Other studies further showed that the activation of canonical Wnt/β-catenin signaling may maintain the expression of pluripotency markers but does not support the long-term maintenance of hESCs [73–75]. Therefore, the role of canonical Wnt/β-catenin signaling in the maintenance of pluripotency or promotion of differentiation in hESCs will require further study.

Unlike canonical Wnt/β-catenin signaling, non-canonical Wnt signaling is usually recognized as a differentiation-related pathway. For example, the non-canonical Wnt/JNK cascade is known to regulate cell polarity during morphogenesis processes, such as gastrulation and neural tube closure during early embryonic development [76–78]. On the other hand, non-canonical Wnt/Calcium-related cascades control differentiation to specific cell lineages, as well as migration. This non-canonical pathway operates via activation of heterotrimeric G proteins to promote phospholipase C-mediated release of calcium from intracellular stores, which in turn activates downstream effectors such as PKC, CamKII and NFAT [79, 80].

LncRNA regulation of pluripotency circuitry

lncRNA, which refers to non-coding RNA transcripts longer than 200 nucleotides, is an emerging category of molecules that function in a variety of biological processes, including X-chromosome inactivation [81], imprinting [82], epigenetic modifications [5, 83], transcriptional regulation [84, 85], miRNA sequestration [8], nuclear trafficking [86] and nuclear body formation [2, 87]. The development of platforms for screening lncRNAs, such as targeted microarrays or next generation sequencing, has accelerated the discovery of lncRNA functions in various cell types. Accumulating evidence shows that lncRNA also functions in pluripotent stem cells. Dinger et al. were the first to show that the expression of a set of lncRNAs is controlled by NANOG, OCT4 and SOX2 during mESC differentiation and directly confirmed the regulatory role of NANOG and OCT4 on lncRNA promoters by chromatin immunoprecipitation (ChIP) [88]. Sheik Mohamed et al. demonstrated that modulation of lncRNA-AK141205 expression induced the loss of pluripotency and promoted differentiation of mESCs, suggesting a functional role for lncRNA in pluripotency maintenance [89].

The roles of lncRNAs in hPSCs have also been studied. Loewer et al. first discovered over 100 lncRNAs with significantly higher expression levels in hPSCs, as compared to fibroblasts [9]. Among these hPSC-enriched lncRNAs, lncRNA-ROR was shown to modulate NANOG, OCT4 and SOX2 expression in hESCs by competing with miR-145/181a/99b [8, 9], while lncRNA-HPAT5 counteracted let-7 activity to protect hESCs from differentiation [90]. Furthermore, Xu et al. demonstrated that the lncRNA, GAS5, can act as an miR-2467/3200/let7e sponge to regulate NODAL signaling and maintain hESC pluripotency [91]. These findings are summarized in Fig. 2a.

Fig. 2.

The roles of lncRNAs in regulating hPSCs pluripotency. a lncRNAs (lncRNA-ROR, lncRNA-HPAT5 and lncRNA-GAS5) act as miRNA sponges to inhibit miRNA activity, promoting pluripotency maintenance of hPSCs. b lncRNA-TUNA interacts with PTBP1, hnRNPK, and NCL1 to regulate NANOG and SOX2 expression through promoter occupancy. c lncRNAs (tsRMST and lncRNA-ES1/ES2/ES3) interact with the PRC2 complex and pluripotency-associated factors (NANOG or SOX2) to inhibit expression of linage-associated genes. d lncRNA-SRA interacts with MLL1 complex, PRC2 complex and the pluripotency-associated factor, NANOG, to promote bivalent domain formation

In addition to functioning as a miRNA sponge, lncRNAs can also form complexes with proteins to regulate pluripotency maintenance. For example, Rana’s group demonstrated that lncRNA-TUNA forms a complex with RNA-binding proteins, PTBP1, hnRNPK and NCL, to regulate NANOG, SOX2 and FGF4 expression through promoter occupancy [92] (Fig. 2b). Ng et al. identified 36 lncRNAs with expression profiles that mirrored expression of OCT4 and NANOG during hESC differentiation. The authors further demonstrated that impaired expression of three of the identified hESC-enriched lncRNAs (lncRNA-ES1, lncRNA-ES2 and lncRNA-ES3) in hESCs resulted in upregulation of early lineage-associated genes and downregulation of pluripotency-associated genes by a process dependent on interaction with SUZ12, a PRC2 repressive complex component, and SOX2 [83]. Yu et al. reported that a trans-spliced lncRNA, lncRNA-tsRMST interacts with NANOG and PRC2 complex to repress differentiation-associated genes and signaling pathways to promote pluripotency maintenance [5, 93] (Fig. 2c). Further, a recent report showed that lncRNA-SRA can interact with both the MLL1/SET1 activating complex and the PRC2 repressive complex, suggesting that lncRNA may also be involved in bivalent domain formation, which is functionally important in the balance of pluripotency and differentiation [94] (Fig. 2d). Finally, lncRNA has also been shown to contribute to pluripotency reprogramming. For instance, disruption of lncRNA-SRA was found to decrease the number of reprogrammed iPSC colonies [95], and lncRNA-ROR and lncRNA-HPAT5 act as miRNA sponge to promote pluripotency reprogramming and iPSC generation [9, 90]. In contrast, lncRNA-p21 hampers pluripotency reprogramming of iPSCs by interacting with SETDB1 and DNMT1 to sustain H3K9me3 modifications and CpG methylation [96]. Based on this mounting evidence, lncRNA can be considered as an emerging player in pluripotency maintenance and reprogramming. For further detailed examples regarding the expression, function roles and the mechanism by which lncRNA regulates pluripotency and reprogramming, readers are referred to the review by Ghosal et al. [14] and Chakrabarti et al. [97].

Trans-spliced RNA (tsRNA)

RNA splicing removes introns from pre-mRNA and joins exons to generate mature and functional mRNA. Cis-splicing combines exons from the same pre-mRNA transcript, whereas trans-splicing uses two or more pre-mRNA transcripts to form chimeric, non-colinear transcripts, which may either encode new protein products or serve as regulatory non-coding RNAs. Trans-splicing can occur within a single gene or between different genes [15, 16] and was first discovered in Trypanosoma brucei and other trypanosomes, where the process entails a short-SL RNA being spliced onto the 5′ termini of all mRNAs [98–100]. SL-type trans-splicing was later found to occur in nematodes, such as Caenorhabditis elegans and Panagrellus redivivus [20, 101]. Intriguingly, SL-type trans-splicing can be processed by HeLa cell extracts, suggesting that the machinery is broadly conserved in eukaryotes [20, 102–104]. However, no observations of SL-type splicing events have been reported in higher eukaryotes. In higher eukaryotes, the most well-known trans-spliced RNAs are Mod4, which regulates apoptosis, and Lola, which regulates axon guidance, in Drosophila [23, 24, 105–107]. In a global exploration of trans-splicing events in insect linages by Kong et al., Mod4 trans-splicing was found to be conserved in two Diptera and two Lepidoptera species [108]. Interestingly, Kong et al. also showed that 146 trans-spliced RNAs resemble cognate genes in different species, suggesting that trans-splicing may function as a buffer system to preserve the function of genes that have undergone “breakup” during the evolution of insect linages [108].

In contrast with the numerous trans-splicing events that have been identified in insects or lower eukaryotes, only around 20 have been identified in humans to date. Moreover, more than half of the known human trans-splicing events were identified in cancer cell lines or tissues, and the functions of most human trans-splicing events are still not well characterized (Table 1). The most prominent examples of known functional trans-splicing in normal human cells are JAZF1-JJAZ1 and tsRMST. JAZF1-JJAZ1 is translated into a chimeric protein with anti-apoptotic function and is believed to be a prerequisite for chromosomal exchange [16, 109, 110], while tsRMST recruits the PRC2 complex to repress differentiation-associated genes in hESCs [5].

Table 1.

Summary of trans-splicing events identified in human

Trans-splicing	Biological function	Cell types	Validation				Read through	References
			RT-PCR	Sequencing	Northern, RPA Western or FISH	Genome rearrangement
TMEM79-SMG5	Cancer marker	LNCaP cells Cancer tissue	Yes	Yes	No	No	No	[111]
TSNAX-DISC1	Cell growth	Endometrial carcinomas	Yes	Yes	Western	No	Unknown	[112]
hER-α	May modulate hER-αbinding	MCF7, T47D, ZR75, cells, mammary gland, ovary, liver, endometrium	Yes	Yes	RPA	No	–	[113]
SP1	May modulate the activity of SP1	Hep2 cells	Yes	Yes	RPA	No	–	[114]
RGS12TS	Induction of nuclear abnormalities	COS-7, HEK293T cells	Yes	Yes	Western	No	–	[115]
PAX3-FOXO1	Activates MYOD and MYOG	Rhabdomyosarcoma	Yes	Yes	FISH, Western	No	No	[116, 117]
PJA2-FER	Cancer maker	Non-small cell lung cancer	Yes	Yes	No	No	No	[118]
JAZF1-JJAZ1	Anti-apoptotic	Endometrial stromal cells	Yes	Yes	FISH	No	No	[25]
ZC3HAV1l-CHMP1A	Onset of chromosomal translocation	Mammary epithelial cells	Yes	Yes	No	No	No	[119]
AF4, AF9, ELL, ENL, MLL	Onset of chromosomal translocation	PBMCs	Yes	Yes	No	No	No	[120]
CYP3A43-3A4 CYP3A43-3A5	May alter cellular location	HepG2 cells and normal Liver tissue.	Yes	Yes	RPA	No	No	[121]
CYCLIND1-TROP2	Cell growth	OVCA-432, MCF7 cells	Yes	Yes	Northern, RPA	No	No	[122]
CoAA-RBM4	Regulates CoAA activity	Wild range of cell lines and normal tissues	Yes	Yes	Western	Unknown	Unknown	[123]
CDC2L2	Unknown	Testis	Yes	No	FISH	Unknown	–	[124]
CAMK2G-SRP72	Unknown	Islet cells	Yes	Yes	Western	Unknown	No	[125]
ACAT1-Amp	Unknown	THP-1, HEK293 cells	Yes	Yes	Northern	Unknown	No	[126–128]
SLC45A3-ELK4	Cancer maker	(Benign) prostate cancer tissues and LNCaP cells	Yes	Yes	FISH	No	No	[124]
tsRMST	Pluripotency maintenance	hESCs	Yes	Yes	RPA	No	No	[5]

RPA RNase protection assay

The function of tsRMST in human pluripotency

RMST (ENSG00000255794.6) is located on the q arm of chromosome 12 (chr12:97431653-97565015). The entire linear RMST transcript contains 13 exons and is ~ 2.6 kb in length. RMST was first identified as a cancer marker in alveolar-subtype rhabdomyosarcoma [129], and a subsequent study in mice showed that Rmst is also expressed in the ventral mesencephalic floor plate and anterior dorsal midline cells [130]. The transcript was then identified as a neurogenic lncRNA that is important for neural differentiation. This discovery was based on genome-wide profiling and functional screening of lncRNAs with differential expression during in vitro differentiation of hESCs [83]. Importantly, a further mechanismic study showed that RMST expression is important for the binding of SOX2 to neurogenic genes, such as ASCL1, NEUROG2, HEY2, and DLX1 [131].

tsRMST is the trans-spliced isoform of RMST, in which the 3′ terminus of RMST exon 11 is joined to the 5′ terminus of exon 3, forming a linear RNA with scrambled exon order (Fig. 3a). The enrichment of tsRMST in the oligo-dT purified mRNA fraction indicated that tsRMST is polyadenylated. Furthermore, tsRMST was found to be degraded by RNaseR treatment, which confirmed that tsRMST is not a circular RNA (circRNA) with scrambled exon order [5]. When the tsRMST sequence was examined by a coding potential calculator [132], it showed low protein coding potential, suggesting that tsRMST is likely to be a trans-spliced lncRNA. Furthermore, the conserved exon–intron boundaries in both RMST and tsRMST suggest that conventional splicing machinery is involved in the trans-splicing process, and the downregulation of tsRMST in differentiated hESCs suggests that trans-splicing is tightly regulated. It has been demonstrated that splicing factors and complementary sequences in flanking introns regulate the biogenesis of circRNAs, another type of alternatively spliced RNA with scrambled exon order. Therefore, it will be interesting to explore whether trans-splicing events are regulated through similar mechanisms [133–135].

Fig. 3.

tsRMST promotes pluripotency maintenance in hESCs. a The exon order of tsRMST is shown. EX exon. b In hPSCs, interaction of tsRMST with NANOG and PRC2 complex promotes occupancy on inactive genes, such as GATA4, GATA6, PAX6 and WNT5A. c The expression of GATA4 and GATA6 promote endoderm differentiation of hPSCs, while PAX6 promotes ectoderm differentiation. The expression WNT5A further activates epithelial–mesenchymal transition (EMT) to promote endoderm differentiation

tsRMST is highly expressed in hPSCs, including hESCs and human iPSCs, as compared to differentiated somatic cells. The disruption of tsRMST expression in hESCs hampers pluripotency-associated gene expression, suggesting a role in pluripotency maintenance. Mechanistically, tsRMST interacts with pluripotency factor, NANOG, and PRC2 complex component, SUZ12, which acts to suppress lineage differentiation and promote pluripotency. These interactions suggest that tsRMST may act as a co-repressor of NANOG and the PRC2 complex to maintain pluripotency. Interestingly, a study by our group found that tsRMST does not interact with SOX2; however, an interaction between SOX2 and the colinear RMST transcript was shown by Ng et al. [83]. Therefore, trans-splicing may modify the protein interactome of RMST or other lncRNAs. Considering that tsRMST and RMST contain highly similar nucleotide sequences with the only difference being in the trans-splice junction, it is reasonable to suspect that trans-splicing may alter the RNA structure to modulate the protein interactome of lncRNAs in humans. ChIP-Seq-based global analysis of gene occupancy by SUZ12 and NANOG indicated that NANOG and the PRC2 complex co-occupy inactive genes, such as early lineage-associated transcription factors PAX6, GATA4 and GATA6, as well as the signaling ligand, WNT5A [5, 93]. In hESCs with impaired tsRMST expression, NANOG and the PRC2 complex are not associated with PAX6, GATA4, GATA6 and WNT5A promoters, suggesting that tsRMST promotes NANOG and PRC2 complex occupancy on inactive genes in hESCs (Fig. 3b). With these results in mind, we proposed a model for regulation of hESCs, wherein tsRMST interacts with NANOG and the PRC2 complex to repress expression of signaling ligands and early lineage-associated transcription factors to promote pluripotency. Along these lines, we observed that downregulation of tsRMST disrupted the repressive complex and activated GATA4, GATA6, PAX6 and WNT5A expression during in vitro differentiation of hESCs. Moreover, the expression of WNT5A in differentiated hESCs further activated epithelial–mesenchymal transition to promote endoderm differentiation (Fig. 3c). Thus, tsRMST blocks differentiation by affecting multiple layers of the pluripotency maintenance machinery, specifically repressing both core transcription factors and signaling ligands.

Conclusions and perspectives

In hPSCs, the pluripotency circuitry is tightly regulated at multiple layers to orchestrate self-renewal, cell differentiation and pluripotency. Master pluripotency-associated transcription factors (i.e., NANOG, OCT4, SOX2) form the core network that promotes the pluripotency program and suppresses differentiation programs. Meanwhile, an additional layer of regulation centers around signaling molecules, such as FGF2, NODAL/ACTIVIN and WNTs, which support the pluripotency core network by activating kinase cascades. The role of lncRNA as an additional regulatory layer in pluripotency circuitry is an emerging concept and is supported by recent findings that individual lncRNAs can inhibit differentiation-associated miRNAs, such as miR-145 and miR-34a, to promote the expression of core transcription factors.

Our exploration of the functions of trans-splicing in humans has only just begun and many unanswered questions remain. How many trans-splicing products can be found in human transcriptome? Are there unknown functions for tsRMST and other trans-splicing events? How are trans-splicing events regulated? What splicing machinery is utilized in human trans-splicing and does it resemble the canonical splicing machinery? With so many central questions remaining, it will be interesting and fruitful to continue exploring the unknown field of trans-splicing in humans.