The Emerging Field of Noncoding RNAs and Their Importance in Pediatric Diseases

Dr Barton Childs’ landmark 1999 text, Genetic Medicine – A Logic of Disease, led to the development of a new curriculum at Johns Hopkins School of Medicine entitled “Genes to Society.”¹ A core part of contemporary medical school and pediatric residency training is individualized medicine. Trainees are now being equipped for an era of personalized, “precision” medicine. Childs said that, in the next century, medicine will be focused on the treatment of individuals rather than disease. This raises the question of how different are individuals at the genomic level? Early after the discovery of the genetic code, it was recognized that some 3 million single nucleotide polymorphisms could be used to distinguish individuals, based on their location within our unique DNA. Other sources of variation in the genome range from few-base-pair insertions and deletions and short tandem repeats to mega-base-pair variations owing to cytogenetic deletions, insertions, or aneuploidy. Once the human genome was sequenced, it became clear that more than 10% of the genome consists of copy number variations that also generate a unique signature for each individual.² The fascinating observation that progression from simple organisms to higher-order organisms and ultimately humans was not associated with a significant increase in the number of genes, but rather, an increasing amount of DNA sequences unassociated with genes, so-called junk DNA. We now know that a substantial portion (≥30%) of this junk DNA is transcribed into noncoding RNA (ncRNA)—RNA that, instead of coding for proteins, serves a direct or indirect regulatory function for those genes that do code for proteins. To date, more than 18 000 distinct ncRNAs have been identified. In many cases, these ncRNAs serve as precursors to generate small inhibitory RNAs, which regulate target gene expression. Other ncRNAs bind proteins and serve as protein translocators and/or facilitate the formation of multiprotein complexes. Interestingly, the transcripts of 88% of single nucleotide polymorphisms that have been associated with different phenotypes are actually located within ncRNAs. These single nucleotide polymorphisms could result in different secondary structures, and are thereby likely responsible for differential functions of the ncRNA.

In 2012, the ENCODE project consortium revealed that as much as 80% of the human genome is actively transcribed.³ However, only about 2% of the genome is protein coding, suggesting the rest of the transcripts are ncRNA transcripts. This discovery led to great interest in unraveling the mechanism governing the biogenesis and function of ncRNAs. Although a vast majority of ncRNAs are transcribed at low levels, and may be transcriptional noise, without any functional role, recent advances of next-generation high-throughput sequencing coupled with functional analyses have yielded numerous discoveries of functional ncRNAs across species.^4,5 The most abundant portion of ncRNAs are housekeeping ribosomal RNAs and transfer RNAs, which are well-known for their functional role in normal cellular processes for quite some time.⁴ Recent experiments have primarily focused on ncRNAs other than ribosomal RNAs and transfer RNAs. This minor fraction has been shown to play crucial roles in a myriad of physiologic processes, including genome integrity maintenance, innate immunity, neurodevelopment, and stem cell proliferation and differentiation, as well as diseases such as cancer. These new regulatory noncoding transcripts are traditionally divided into 2 major groups based on their size and an arbitrary cutoff: short noncoding RNAs (18–200 nucleotides) and long ncRNAs (lncRNAs) (>200 nucleotides). In addition, a novel class of ncRNAs was recently discovered, the circular RNAs (circRNAs), named because of the circular nature of the transcript generated from back-splicing of pre-mRNA and covalent linking of 3′ and 5′ ends.^6–9 The size of circRNAs ranges from less than 200 to several thousand nucleotides (Figure 1).

Figure 1.

A schematic illustrating various ncRNAs. The biogenesis mechanisms and functions of various ncRNAs are shown (miRNAs, left; lncRNAs, middle; circRNAs, right). pri-miRNA, primary microRNA.

Short ncRNAs are further categorized into several classes, including small nuclear RNAs, which are key components of the spliceosome and play an important role in pre-mRNA splicing, and small nucleolar RNAs, which regulate ribosomal RNA modification. Additional classes of short (21–30 nt) ncRNAs include small interfering RNAs (siRNAs), Piwi-interacting RNAs, and microRNAs (miRNAs or miRs), which are core components of RNA interference (RNAi), an evolutionarily conserved process that regulates gene expression in a sequence-specific manner. The siRNAs are derived from long double-stranded RNA precursors. They join and guide Argonaute protein-containing complexes to target mRNAs by complementary base pairing, leading to gene silencing via RNA degradation. Viral RNA replication intermediates (in the form of double-stranded RNAs) can be converted into siRNAs, which in turn can target viral RNA transcripts for degradation via RNAi.^10–17 Thus, RNAi is a key constituent of antiviral defense mechanism in diverse host organisms. The miRNAs are transcribed as long stem-loop primary transcripts, which are processed into 22- to 24-nt segments.^18–29 Similar to siRNAs, miRNAs are loaded into and guide Argonaute complexes to target mRNAs by imperfect base pairing between miRNAs and target mRNAs, and decrease protein output by mRNA destabilization and translation inhibition.^30–32 Piwi-interacting RNAs are primarily produced in germ cells and control the expression of transposable elements, thereby contributing to the maintenance of germ-line genome integrity.³³

The lncRNAs are characterized based on location and orientation of resulting transcript into long intergenic ncRNAs, natural antisense transcripts, enhancer RNAs, and bidirectional transcripts.^5,34 LncRNAs are involved in functionally diverse mechanisms. They can serve as transcriptional repressors (eg, XIST), enhancers by promoting activating interactions between promoters and distal regulatory elements (eg, LUNAR1), miRNAs sponges (eg, TUG1), and hubs for protein-protein and protein-nucleic acid interactions (eg, SPRIGHTLY).^5,34–39

The circRNAs are generated by back-splicing reactions during pre-mRNA processing. Emerging evidence has delineated diverse modes of action of circRNAs. For example, the mouse circRNA CDR1as/CiRS-7 shelters miR-7 and impacts brain development.^9,40,41 In addition, the circRNA SRY plays a prominent role in male sex determination.^9,40 Furthermore, select intron-containing circRNAs can interact with U1 small nuclear ribonucleoprotein particle and promote host gene transcription in the nucleus.⁴² Moreover, circRNAs can modulate gene expression by competing with linear splicing.⁴³ Lastly, selective circRNAs can give rise to functional polypeptides, thereby expanding the complexity of the proteome.^44,45

Cross-regulation among various classes of RNAs has been well-documented (Figure 2). For example, the lncRNA TUG1 can sequester and functionally inhibit the activity of a number of miRNAs.^38,46–49 In addition, the lncRNA H19 serves as a source of precursors for miR-675 biogenesis.^50–53 An elegant example of ncRNA cross-regulation is the miR-671-CDR1as-miR-7 axis: the circRNA CDR1as carries 1 near perfect binding site for miR-671 and dozens of imperfect binding sites for miR-7. Engagement of miR-671 with CDR1as results in CDR1as degradation, which leads to downregulation of miR-7 owing to loss of protection by CDR1as.^9,40,41 Lastly, ncRNAs (lncRNAs and circRNAs) and protein-coding mRNAs can engage with and therefore compete for the same pool of miRNAs, resulting in cross-regulation. In fact, cross-regulation among various mRNAs and between mRNAs and lncRNAs via miRNA engagement provides support for this model.⁵⁴

Figure 2.

A schematic depicting cross-regulation among various classes of ncRNAs. On the one hand, miRNAs can bind and downregulate the expression of target lncRNAs; on the other hand, select lncRNAs can serve as miRNA precursors or sequestrate and functionally inhibit miRNAs. In addition, select circRNAs can inhibit or protect miRNAs by physically associating with miRNAs. Conversely, miRNAs that carry perfect complementarity to circRNAs can lead to target circRNA degradation. Last, both lncRNAs and circRNAs can serve as competing endogenous RNAs by sequestrating miRNAs, thereby modulating the availability of miRNAs to engage with and functionally inhibit target mRNAs.

Functionally, these ncRNAs species regulate cellular identity and function primarily via modulating transcriptional and post-transcriptional gene expression. Given the diverse physiological processes regulated by ncRNAs, it is not surprising that dysregulation of ncRNA expression can contribute to pathological conditions.^55–57 There is now a growing body of evidence implicating ncRNAs in normal development, health, and dysfunctions. We summarize current knowledge on the role of ncRNAs in several diseases and the implications for therapy (Table). We note that extensive literature covering the topic of ncRNAs in pediatric diseases is still developing because the field of is at early stages. In particular, the pediatric translational research applications to date in the rapidly emerging basic science field of ncRNA remain fewer in number than among studies focused on diseases that primarily affect adults. The examples discussed herein are primarily based on studies involving cultured cells or animal models and focus heavily on basic science, and thus may not be specific to children. However, we believe that knowledge gained from these preclinical/early stage clinical studies discussed here will help us to better understand the molecular mechanisms underlying these diseases in the pediatric patient population and facilitate the development of novel diagnostic and therapeutic tools.

Table.

ncRNAs implicated in various diseases

Diseases	ncRNA	Mechanism	References
β-thalassemia	miR-486-3p and miR-210	Repressing BCL11A	58
	miR-23a	Repressing KLF-2	59
	miR-15a and miR-16-1	Repressing MYB	60
	miR-27a	Repressing Sp1	59
	HMI-LNCRNA	Repressing γ-globin	61
DMD	miR-486	Repressing PTEN and Foxo1a	62
	miR-21	Repressing PTEN and SPRY-1	63
	miR-29a and miR-29c	Repressing COL3A1, FBN1 and YY1	63
	linc-MD1	Sequestering miR-133 and miR-135	64
	circ-ZNF609	Encoding a new protein	44,45
Rett syndrome	miR-199a	Derepressing mTOR signaling	65
	AK087060	Dysregulation upon MeCP2 loss contributes to Rett syndrome	66-68
	AK081227	Elevated expression correlates with downregulation of Gabrr2	68-70
Glioma	miR-9/9*	Repressing SOX2, PTCH1, FOXP1 and CAMTA1	71-73
	miR-17-92 cluster	repressing CFTG	74,75
	miR-17, miR-19a/b, miR-26a and miR-221/222	Inhibiting PTEN signaling	76-79
	let-7a, miR-101, miR-124, miR-138, miR-214 and miR-708	targeting EZH2-dependent epigenetic mechanisms	80-85
	miR-7	Repressing EGFR	86-88
	miR-128	Repressing EGFR, WEE1, MSI1, and RPS6KB1	89-91
	miR-34a	Repressing CDK6 and CCND1	92
	miRNA-100	Repressing PLK1	93
	H19	Sequestering miR-140 and miR-29a	94,95
	HOTAIR	Sequestering miR-326 and interacting with EZH2	96,97
	MEG3	Activating p53	98
	TUG1	Interacting with EZH2, SUZ12 and YY1	99
	GAS5	Binding to miR-196a-5p and upregulating FOXO1	100
Medulloblastoma	miR-17-92 cluster	Positive effector of Shh-mediated proliferation	101,102
	miR-10b	Expression positively correlated with BCL2 expression	101,103
	miR-21	Repressing PDCD4	101,104
	miR-124	Repressing CDK6 and SCL16A1	105,106
	miR-218	Repressing NANOG RICTOR, CTSB and CDK6	107,108
	miR-125b	Repressing SMO and LIFRα	109,110
	miR-326	Repressing SMO	109
	MIR100HG	Sequestering miR-19a-3p, miR-19b-3p and miR-106a-5p and derepressing CDK6, MYCN, SNCAIP and KDM6A	111
	PVT1	Stabilizing MYC	112

mTOR, mammalian target of rapamycin; PTEN, phosphatase and tensin homolog.

Role of ncRNAs in Pediatric Diseases

β-Thalassemia is a recessive inherited disease affecting hundreds of thousands of individuals worldwide, with symptomatic onset in childhood. Before birth, the predominant hemoglobin is α2γ2. In the last trimester, fetal γ globin synthesis decreases and adult β globin synthesis increase. A hallmark of β-thalassemia is a decrease in or absence of β globin synthesis, necessary for the predominant adult hemoglobin (α2β2), resulting in anemia and ineffective erythropoiesis. Current treatments include chelation therapy, blood transfusion, and bone marrow transplantation. Recently, a novel strategy has been explored, which involves reactivation of gene encoding fetal γ globin, to compensate for the shortage of β globin. A suite of miRNAs that functionally inhibit the γ globin gene transcriptional repressors have been identified and validated: miR-486–3p and miR-210 (which target BCL11A mRNA), miR-23a (against KLF-2), miR-15a and miR-16–1 (against MYB), and miR-27a (against Sp1).^58–60 Experimental approaches that enhance the activities of these miRNAs are expected to induce the γ globin gene, thereby harboring potential as a new approach to treating β-thalassemia. The lncRNAs have also been implicated in β-thalassemia.^61,113,114 For example, the nuclear lncRNA HMI-LNCRNA generated from the HBS1L-MYB enhancer region displays significantly higher levels of expression in erythroblasts derived from cultured adult peripheral blood cells, which express more β globin, compared with erythroblasts from cultured cord blood cells, which express more γ globin. Notably, downregulation of HMI-LNCRNA in HUDEP-2 cells, which express mostly β globin, significantly reactivates γ globin expression and promotes erythroid maturation. Thus, HMI-LNCRNA might be a potential therapeutic target for γ globin induction treatment in β-thalassemia.

Duchenne muscular dystrophy (DMD) is a lethal neuromuscular disease and is the most common muscular dystrophy affecting children. DMD is characterized by a rapid progression of muscle degeneration caused by mutations in the dystrophin gene. Several miRNAs have been implicated in DMD. For example, the muscle-enriched miRNA miR-486 is markedly decreased in the muscles of dystrophin-deficient mice and in DMD patient samples. Mechanistically, miR-486 suppresses the expression of phosphatase and tensin homolog (PTEN) and Foxo1a, negative regulators of phosphoinositide-3-kinase (PI3K)/Akt signaling, which regulates muscle hypertrophy and growth, as well as dedicator of cytokinesis 3 (DOCK3), thereby playing a key role in myotube survival.^62,115 In addition, miR-21 expression is significantly increased in DMD samples, and correlates with a significant reduction in the expression of miR-21 target transcripts including PTEN and SPRY-1 (Sprouty homolog 1), whereas miR-29a and miR-29c are significantly decreased in Duchenne muscle and myoblasts, accompanied by a concordant increase in miR-29 target transcripts, including COL3A1, FBN1, and YY1.⁶³ Several additional muscle-enriched miRNAs, such as miR-1, miR-133, and miR-206 display an increase in the serum of DMD patients and/or in muscle tissues of mouse DMD models, suggesting that they may serve as biomarkers for DMD. When it comes to lncRNAs, it has been shown that the lncRNA linc-MD1 can operate as a miRNA sponge by sequestering miR-133 and miR-135, thereby modulating the expression of Maml1, Mef2c, Myog, and Mhc, which regulate muscle-specific gene expression. In the muscle of DMD patients the level of linc-MD1 is greatly reduced, whereas linc-MD1 overexpression can rescue the defective myogenic differentiation and restore the normal expression of the aforementioned linc-MD1-regulated genes.⁶⁴ Last, a recent study provided evidence that the circRNA circ-ZNF609 can be translated into a functional protein that modulates myogenesis, thereby adding circRNAs to the list of regulatory RNAs in muscle development.^44,45

Rett syndrome is a neurodevelopmental disorder associated with mutations in the MeCP2 gene encoding methyl-CpG binding protein 2. It has been reported that the MeCP2 protein associates with the miRNA biogenesis machinery and is required for appropriate post-translational processing of a suite of miRNAs.⁶⁵ Among these MeCP2-regulated miRNAs is miR-199a, which suppresses the expression of inhibitory factors of mammalian target of rapamycin (mTOR) signaling that has been implicated in Rett syndrome. Besides miRNA regulation, MeCP2 can also modulate lncRNA expression. MeCP2 loss in the mouse brain is associated with upregulation of 2 lncRNAs, AK081227 and AK087060.¹¹⁶ In particular, elevated expression of AK087060 in MeCP2 knockout mouse brain correlates with an increase in the expression of its host gene Arhgef26 encoding a Rho guanine nucleotide exchange factor that contributes to axon patterning.^66,67 Thus, it is possible that dysregulation of AK087060 and Arhgef26 upon MeCP2 loss in mouse brain contributes to Rett syndrome phenotypes. In contrast, elevated expression of AK081227 is associated with downregulation of gamma-aminobutyric acid (GABA) receptor subunit rho 2 (Gabrr2). Because dysfunction in GABAergic inhibitory neurotransmission is associated with many neurodevelopmental disorders, including Rett syndrome, and that the expression of another GABA receptor subunit member (GABRB3) is reduced in Rett syndrome, it is likely that AK081227 and Gabrr2 are candidates to be altered in Rett syndrome.^{69,70,117,118} Lastly, the observation that most circRNAs are enriched in the brain suggests a functional relevance in neurodevelopment. In fact, the circRNA CDR1as/CiRS-7 plays a key role in brain development, at least in part, by associating with and stabilizing miR-7.^9,40,41 We envision that advances in circRNA study will continue to provide insights regarding the role of circRNAs in normal neurodevelopment and neuro-logic diseases, such as Rett syndrome.

Glioma is a cancer originating in glial cells that are primarily involved in nourishment and upkeep of neighboring neurons.¹¹⁹ Gliomas are the most common brain tumor in children. The most frequent form, low-grade gliomas, are generally not associated with poor prognosis whereas high-grade glioma are often fatal.¹²⁰ Various differentially regulated miRNAs and lncRNAs have been identified in gliomas.

Oncogenic miRNAs that promote glioma pathogenesis include miR-9/9* that regulates SOX2, PTCH1, FOXP1, and CAMTA1, the miR-17–92 cluster targeting CFTG, and miR-17, miR-19a/b, miR-26a, and miR-221/222 that inhibit tumor suppressive PTEN signaling.^71–79 Tumor suppressor miRNAs, such as let-7a, miR-101, miR-124, miR-138, miR-214, and miR-708 are involved in inhibiting glioma/glioblastoma growth, particularly by targeting EZH2-dependent epigenetic mechanisms.^80–85 Other tumor suppressor miRNAs in high-grade glioma include miR-7 (targeting EGFR), miR-128 (targeting EGFR, WEE1, MSI1, and RPS6KB1), miR-34a (targeting CDK6 and CCND1), and miRNA-100 (targeting PLK1).^86–93

LncRNAs are also found to be associated with IDH1/2 mutation status and glioma grade, thereby highlighting their potential diagnostic and prognostic significance.¹²¹ Importantly, select lncRNAs have been analyzed for potential functional role in gliomas. For example, the lncRNA H19 is upregulated in gliomas where it acts as oncogene by sequestering miR-140 and miR-29a, thereby relieving their inhibitory effect on key oncogenes such as CDK6, CASH2, MDR, and so on.^94,95 Another lncRNA, HOTAIR, can activate the fibroblast growth factor (FGF), PI3K/AKT, and MEK pathways and promote glioma proliferation and metastasis by serving as a miR-326 sponge, and regulate cell cycle progression in glioma via interaction with EZH2.^96,97 Of note, HOTAIR has also been shown to be activated by epigenetic regulator BRD4 in glioma genesis, a well-known oncogene across cancer types.¹²² In addition, the lncRNA MEG3 inhibits cell proliferation via p53 activation.⁹⁸ Furthermore, the lncRNA TUG1 maintains glioma stem cells through interactions with PRC2 components (EZH2 and SUZ12) and transcription factor YY1, thereby epigenetically suppressing multiple neuronal differentiation-associated genes.⁹⁹ Moreover, the lncRNA GAS5 suppresses glioma stem cell proliferation, migration, and invasion by binding to the oncogenic miR-196a-5p and upregulating the downstream FOXO1.¹⁰⁰ Lastly, XIST is another oncogenic lncRNA that modulates epigenetic pathways and promotes cell proliferation and migration in gliomas.¹²³

Medulloblastoma represents the most common malignant pediatric brain tumor, localized in cerebellum. Recent genetic and epigenetic studies have characterized medulloblastoma into four clinical and molecular subgroups, namely, wingless (WNT), sonic hedgehog (SHH), group 3, and group 4 (reviewed in¹²⁴). As of now, miRNAs and lncRNAs are the predominant ncRNA species that have been investigated in medulloblastoma pathogenesis (reviewed in¹⁰¹). Both miRNAs and lncRNAs show subgroup specific expression pattern, thereby highlighting subgroup-specific molecular mechanism regulated by these ncRNA species.

Several miRNA such as miR-17–92 cluster, miR-10b and miR-21, have been shown to promote medulloblastoma proliferation and/or metastasis in vitro and in vivo.^101–104 Conversely, miR-124, miR-218, miR-125b, and miR-326 are examples of tumor suppressor miRNAs found downregulated in medulloblastomas.^{105–110,125,126} Some of these candidates also represent potential therapeutic targets. For example, the miR-17–92 cluster, which was found to be associated with SHH medulloblastoma, promotes tumor development in vivo.¹²⁷ Consequently, complete knockout or locked nucleic acid (LNA) based inhibition of the miR-17–92 cluster reduced tumor growth and improved survival in SHH medulloblastoma mice.

The lncRNAs in medulloblastoma have received comparatively little attention, as of yet. A recent genome-wide lncRNA analysis highlighted subgroup-specific lncRNA expression in medulloblastoma patients and proposed a diagnostic and prognostic model based on lncRNAs.¹²⁸ Several other in vitro studies also highlight functional role of lncRNAs in medulloblastoma. The lncRNA MIR100HG was found to act as oncogene in group 4 tumors where it sponged miR-19a-3p, miR-19b-3p, and miR-106a-5p, thereby derepressing their targets, including CDK6, MYCN, SNCAIP, and KDM6A, and promoting proliferation.¹¹¹ However, surprisingly, overexpression of MIR100HG in the group 3 cell line downregulated their proliferation. PVT1 is another oncogenic lncRNA in medulloblastoma, particularly group 3 patients, where it is frequently found fused to oncogene MYC.¹²⁹ One consequence of the resulting fusion transcript is stabilization of MYC mRNA.¹¹²

CircRNAs are also gaining interest in medulloblastoma research. It has been shown that various cancer types display distinct circRNA signatures and that circRNAs may serve as tumor biomarkers.^130,131 In addition, Lv et al recently identified and validated 33 circRNAs that are dysregulated in medulloblastoma.¹³² Interestingly, 2 circRNAs (circ-SKA3 and circ-DTL) seem to modulate expression of corresponding host genes and impact the proliferation, migration, and invasion of tumor cells. Considering that circRNA research is still at infancy, we expect that rapid progress in this field will solidify the notion that circRNAs can modulate and perhaps drive the development and progression of various cancer types, such as medulloblastoma.

Conclusions and Future Directions

We have summarized several physiologic processes regulated by ncRNAs, and provide examples of a wide variety of diseases resulting from ncRNA dysregulation. Although the examples discussed herein may not be strictly specific to children, similar, if not identical, underlying molecular mechanisms operate in both adults and children. We envision that the exciting field of transcriptomics and ncRNA research, which encompasses both basic science and translational studies, will continue to benefit from rapid advances in the development of next generation sequencing technology and bioinformatics tools. This will facilitate the elucidation of the molecular mechanism underlying the function and regulation of ncRNAs in physiological and pathological settings, and provide insights into the development of ncRNA-based diagnostic and therapeutic strategies against pediatric diseases.

Glossary

circRNA

Circular RNA

DMD

Duchenne muscular dystrophy

GABA

Gamma-aminobutyric acid

lncRNA

Long ncRNA

miRNA

MicroRNAs

ncRNA

Noncoding RNA

SHH

Sonic hedgehog

siRNA

Small interfering RNA