Assessing the contributions of noncoding RNAs in acute myeloid leukemia

Abstract

Acute myeloid leukemia (AML) is caused by uncontrolled proliferation and impaired differentiation of hematopoietic stem and progenitor cells. Historically, research has emphasized the role of protein-coding genes in the development of AML. However, with the human genome project revealing that 98% of the transcriptome consists of non–protein-coding RNAs, recent studies have explored how the large classes of noncoding RNAs (ncRNAs) contribute to AML. Although there are many types of ncRNAs, much attention has been placed on understanding the function of long ncRNAs (lncRNAs) and small ncRNAs known as microRNAs (miRNAs). lncRNAs are >200 nucleotides, whereas mature miRNAs are typically 18 to 25 nucleotides. lncRNAs are involved in miRNA and protein sequestration and act as transcriptional and translational regulators, whereas miRNAs facilitate mRNA degradation and translational inhibition. In addition to lncRNAs and miRNAs, two additional types of ncRNAs, namely small nucleolar RNAs (snoRNAs) and circular RNAs (circRNAs), have recently garnered attention for their roles in AML. Here, we discuss how these four distinct classes of ncRNAs may aid in disease diagnosis and prognosis as well as the mechanisms by which their dysregulation contributes to AML.

Introduction

Acute myeloid leukemia (AML) is the second most common form of acute leukemia in children and the most common form in adults.^1,2 AML occurs due to genetic modifications in hematopoietic stem cells (HSCs), leading to abnormal proliferation and dysregulated differentiation of myeloid progenitor cells.² These events result in the accumulation of immature blast cells in the peripheral blood, bone marrow, and organs.^2,3 Mutations in genes encoding transcription factors, splicing factors, and epigenetic regulators, such as RUNX1, SF3B1, and TET2, are associated with AML development and resistance to therapy.4, 5, 6 Although extensive work has been done to characterize and understand the contributions of protein-coding genes to AML, only recently have studies begun to explore the potential involvement of noncoding RNA (ncRNA) in this malignancy.

The human genome project revealed that ∼80% of the human genome is transcribed.⁷ Surprisingly, only 2% of the human transcriptome is composed of protein-coding genes, whereas the remaining 98% consists of ncRNAs, including tRNAs, rRNAs, small ncRNAs, and long ncRNAs (lncRNAs).^7,8 The small ncRNAs, which are typically <200 nucleotides in length, include the well-characterized microRNAs (miRNAs) as well as small nuclear RNAs (snRNAs) and small nucleolar RNAs (snoRNAs).9, 10, 11 lncRNAs, which include long intergenic ncRNA (lncRNAs) and circular RNA (circRNA), are characterized as being >200 nucleotides in length.10, 11, 12 Both small ncRNAs and lncRNAs are involved in a variety of cellular functions, including transcriptional and posttranscriptional modifications, translational regulation, and protein sequestration and localization.11, 12, 13, 14, 15 Additionally, ncRNAs are transcribed and function in tissue-specific manners.^9,10 Increasing evidence links ncRNAs to cancer, including AML.16, 17, 18 Although several recently published reviews discuss the role of lncRNAs, circRNAs, and miRNAs in AML,^17,19, 20, 21, 22, 23, 24, 25 this review focuses on several key studies and highlight newly published reports characterizing the contribution of these major ncRNA classes as well as snoRNAs to AML and their potential usage as biomarkers.

lncRNAs

The biogenesis and multifaceted functions of lncRNAs

lncRNAs are a well-conserved, large family of ncRNAs of >200 nucleotides.^10,12 lncRNAs lack functional open reading frames and therefore do not encode for proteins.⁸ lncRNA biogenesis is similar to that of mRNAs as they are transcribed by RNA polymerase II, undergo 5′ capping and canonical splicing, as well as 3′ polyadenylation.^9,12 Interestingly, several lncRNAs, such as NEAT1 and MALAT1, are produced in a noncanonical manner.²⁶ Conventional classification of lncRNAs is based on genomic location,27, 28, 29 with the most common form being lncRNAs, which do not overlap protein-coding transcripts.²⁹ Moreover, transcription of lncRNA from intronic regions, as well as in sense, antisense, and bidirectionally with regard to protein-coding genes, contributes to the diversity of lncRNAs.27, 28, 29

Studies have shown lncRNAs interact with DNA, RNA, and proteins through various sequence domains to act as molecular scaffolds.^10,13 lncRNAs are known to function in the nucleus and/or the cytoplasm to modulate gene expression through distinct mechanisms, including regulation of transcription and translation, histone modifications, miRNA sequestration, and protein localization (Figure 1).^9,10,13 For example, the well-characterized lncRNA XIST silences one X chromosome in females through recruiting heterochromatin histone proteins to the inactive X chromosome.³⁰ Additionally, the lncRNA NEAT1 functions to modulate RNA-binding protein (RBP) organization through the formation of paraspeckles.³¹ The lncRNA DUBR, which is abundant in hematopoietic stem and progenitor cells (HSPCs), has been shown to modulate hematopoietic differentiation gene programming.³² Recent improvements in lncRNA annotation have driven efforts to identify lncRNAs that contribute to cancer, including AML progression and maintenance.

Figure 1.

The diverse functions of lncRNAs. These ncRNAs are present in both the nucleus and the cytoplasm in which they have distinct interactions and functions, including transcriptional and translational regulation, splicing regulation, histone modifications, protein organization, RNA stabilization, and miRNA sequestration. Figure was created with biorender.com. Zakutansky, P. (2026) https://biorender.com/7mmtpx1.

lncRNA dysregulation contributes to AML progression

The accurate detection of lncRNA levels is paramount to properly elucidate the roles of lncRNA in normal hematopoiesis and AML. Garzon et al used a custom microarray of cytogenetically normal AML samples from patients to identify lncRNAs associated with clinical features and mutations, including NPM1, FLT3-ITD, and CEBPA.³³ Using these patient-derived wild-type and mutant samples, this study identified 180 lncRNAs in NPM1-mutant AML, 119 lncRNAs in FLT3-ITD AML, and 51 lncRNAs in CEBPA-mutant AML, although several lncRNAs were common to both the NPM1-mutant and FLT3-ITD AML samples. The authors developed a “lncRNA score” as a prognostic indicator. Although this provides preliminary evidence for how lncRNA dysregulation is correlated with different types of AML, additional work is needed to validate the usage of lncRNAs as prognostic indicators.³³ In another study using pediatric AML bone marrow compared with healthy controls, microarray analysis revealed >2400 differential lncRNAs, with 372 of those being significantly dysregulated.³⁴ Gene ontology analysis indicated that the dysregulation of identified lncRNAs affected cell cycle progression and immune response, suggesting a potential correlation between lncRNAs and AML progression.³⁴

Although investigators initially relied on microarrays to identify lncRNAs, more advanced methods such as RNA sequencing (RNA-seq), CRISPR activation, and CRISPR interference screens are now commonplace to explore lncRNA function.³⁵ Using RNA-seq to identify RUNX1-interacting RNAs in THP-1 cells, Trinh et al identified a novel lncRNA within the upstream regulatory element (URE) of the PU.1 myeloid gene, named LOUP.³⁶ LOUP was shown to be predominantly expressed in myeloid tissues, with its loss in U937 cells and human CD34⁺ HSPC resulting in decreased PU.1 levels. Depletion of LOUP led to increased cell proliferation and reduced myeloid differentiation. The authors also determined that LOUP promotes PU.1 expression by facilitating chromatin looping between the URE and the proximal promoter region.³⁶ However, in samples from patients with t(8;21) AML, both LOUP and PU.1 were decreased compared with normal karyotype controls. Depletion of RUNX1::ETO led to increased LOUP and PU.1, suggesting RUNX1-ETO regulates PU.1 by limiting URE accessibility, thus inhibiting LOUP transcription.³⁶ This work highlights how AML-associated mutations directly modulate both a lncRNA and mRNA, raising the broader question of how many other lncRNAs are directly regulated by fusion proteins in AML variants.

Several well-known lncRNAs, including XIST, HOTAIR, and CRNDE, have been implicated in various cancers, including AML (Table 1).^21,22,37,38 HOTTIP, a widely studied oncogenic lncRNA, was shown to be elevated in samples from patients with nucleophosmin1 (NPM1)-mutant and mixed lineage leukemia (MLL)-rearranged samples compared with wild-type controls.³⁹ Loss of HOTTIP in MOLM13 cells altered HOXA and hematopoietic gene levels, whereas in vivo ectopic expression of Hottip increased HSC self-renewal and induced an AML-like disease by perturbing the HOXA topologically associated domain.³⁹ Luo et al further reported that HOTTIP interacts with CTCF (CCCTC-binding factor)-binding sites to form R-loops to define CTCF (CCCTC-binding factor) topologically associated domain boundaries and promote β-catenin–driven leukemogenesis.⁴⁰ Another comprehensively explored lncRNA, MALAT1, has previously been shown to promote metastasis and correlate with poor overall survival in numerous cancer studies.⁴¹ In a study from 2024, MALAT1 was shown to be overexpressed in the TGCA and BEAT AML cohorts in addition to multiple myeloid cell lines.⁴² In Tet2-mutant hematopoietic cells, MALAT1 was shown to act as a downstream effector of TET2 loss of function to promote myeloid skewing, HSPC expansion, NF-κB–driven inflammation, and possibly shielding P65 from dephosphorylation.⁴²

Table 1.

lncRNAs associated with AML

lncRNA	Expression levels in AML	Function in AML	Direct targets	Indirect targets	References
CCAT1	•High levels in patients with AML compared with healthy controls •High levels in the HL-60, Kasumi-1, NB4, U937, HEL, and MEG01 AML cell lines compared with the HS-5 cell line	•Promotes cell proliferation and migration •Inhibits apoptosis •Acts as a sponge for miR-490-3p to promote the expression of MAPK1 and c-Myc	↓ miR-490-3p	↑ MAPK1 ↑ c-Myc	43
CRNDE	•High levels in APL compared with non-APL AML and healthy bone marrow	•Promotes cell proliferation •Blocks differentiation •Suppresses miR-181 to modulate NOTCH2	↓ miR-181	↑ NOTCH2	44
UCA1	•High levels in the KG-1, U937, THP-1, and HL-60 AML cell lines compared with the HS-5 cell line	•Promotes cell proliferation •Inhibits apoptosis •Suppresses miR-296-3p, resulting in increased Myc	↓ miR-296-3p	↑ Myc	45
	•Analyzed in HL-60 and U937 AML cell lines; however, expression levels were not discussed	•Promotes colony formation, migration, and invasion •Inhibits apoptosis •Overexpression leads to increased expression of METTL14 and m6A methylation •Indirectly regulates CYP1B1 and CXCR4 through modulating METTL14	↑ METTL14	↑ CYP1B1 ↑ CXCR4	46
HOTAIR	•High levels in the HL-60, NB4, U937, and THP-1 AML cell lines after all-trans retinoic acid treatment	•Promotes cell differentiation •Modulates cell cycle progression •Suppresses miR-17-5p, resulting in increased p21	↓ miR-17-5p	↑ p21	47
HOTTIP	•High levels in samples from patients with NPM1-mutant and MLL-rearranged AML	•Regulates HSC self-renewal and differentiation •Activates posterior HOXA and hematopoietic genes	↑ HOXA9 ↑ HOXA10 ↑ HOXA11 ↑ HOXA13 ↑ MEIS1 ↑ PBX3	↑ TWIST1	39
HOXBLINC	•High levels in NPM1c+ AML	•Modulates chromatin-modifying genes	↑ HOXA9 ↑ HOXA10 ↑ MEIS1 ↑ RUNX1 ↑ MLL1 ↑ SETD1A	NR	48
LOUP	•High levels in the U937 and HL-60 AML cell lines •Low levels in samples from patients with t(8;21) AML compared with samples from patients with normal karyotype	•Inhibits proliferation •Promotes myeloid lineage •Facilitates chromatin looping with RUNX1 to promote PU.1 expression •LOUP is inhibited by RUNX1-ETO in t(8;21) AML	↑ PU.1	NR	36
PVT1	•High levels in AML bone marrow compared with healthy control •High levels in the HL-60, K562, THP-1, NB4, and U937 AML cell lines compared with healthy control	•Promotes cell proliferation, migration, and invasion •Inhibits apoptosis and the miR-29 family to regulate WAVE1	↓ miR-29a ↓ miR-29b ↓ miR-29c	↑ WAVE1	49
XIST	•High levels in the THP-1, HL-60, TF-1, NB4, and U937 AML cell lines compared with the HS-5 cell line	•Promotes cell proliferation •Inhibits apoptosis •Acts as a sponge for miR-142-5p, which regulates PFKP	↓ miR-142-5p	↑ PFKP	50
SNHG5	•High levels in MSC samples from patients with AML compared with healthy donor MSC samples	•SNHG5 overexpression promotes chemotherapy resistance to daunorubicin •Induces CAF-like phenotype and autophagy by altering ATG5 •Interacts with PTBP1 to regulate ATG5 stability	↑ PTBP1	↑ ATG5	51
TUG1	•High levels in the PBMCs of patients with AML compared with healthy PBMCs •High levels in the HL-60, KG-1, MOLM-14, and MOLM-13 AML cell lines compared with healthy PBMCs	•Promotes cell proliferation and glycolysis •Acts as a competing endogenous RNA to suppress miR-185	↓ miR-185	NR	52
LINC-PINT	•Low levels in the bone marrow samples from patients with AML compared with healthy control •Low levels in the Kasumi-1, THP-1, HL-60, and U937 AML cell lines compared with the HS-5 cell line	•Overexpression inhibited cell proliferation and cell cycle progression •Overexpression promoted apoptosis •Acts as a sponge for miR-767-5p, which suppresses SUZ12	↓ miR-767-5p	↑ SUZ12	53
MEG3	•Low levels in the PBMCs from patients with M5 AML/minimal residual disease compared with samples from those with M5 AML •Low levels in the U937 and THP-1 Adriamycin-resistant AML cell lines compared with WT cell lines	•Upregulation of MEG3 inhibits cell proliferation and cell growth •Decreased expression associated with poor overall survival •Acts as a sponge for miR-155, which modulates ALG9	↓ miR-155	↑ ALG9	54
TPTEP1	•Low levels in the bone marrow samples from patients with AML compared with healthy controls •Low levels in the SHI-1, THP-1, HL-60, and U937 AML cell lines compared with CD34+ cells	•Overexpression inhibits cell proliferation •Overexpression promotes apoptosis •Inhibits miR-4295 to upregulate GADD45α	↓ miR-4295	↑ GADD45α	55

An upward-pointing arrow (↑) indicates that the target is upregulated by the lncRNA; a downward-pointing arrow (↓) indicates that the target is downregulated by the lncRNA.

APL, acute promyelocytic leukemia; MSC, mesenchymal stem cells; NR, not reported; PBMCs, peripheral blood mononuclear cells; WT, wild-type.

Several novel lncRNAs have recently been implicated in AML. For example, TPTEP1, which previously was reported as an oncogenic lncRNA in non–small cell lung cancer and colorectal cancer,^56,57 was found to be downregulated in the bone marrow samples from patients with AML compared with healthy controls, with reduced levels associated with poor survival.⁵⁵ Overexpression of TPTEP1 inhibited tumor cell proliferation and promoted apoptosis in AML cell lines. Mechanistically, TPTEP1 was shown to suppress miR-4295, thereby upregulating the tumor suppressor GADD45α. Conversely, TPTEP1 hypermethylation upregulated miR-4295, which decreased GADD45α. This, in turn, led to increased AML proliferation, suggesting a tumor suppressive role for TPTEP1 in AML.⁵⁵ Intriguingly, lncRNAs may also be associated with chemotherapy resistance; SNHG5 was elevated in mesenchymal stem cells of patients with AML, particularly in relapse or drug-resistant cases.⁵¹ Biomarker analysis revealed that elevated SNHG5 drives a cancer-associated fibroblast phenotype via inducing autophagy and upregulating ATG5.⁵¹ These studies highlight diverse mechanisms by which lncRNA dysregulation influences AML progression and underscore the needed to elucidate whether lncRNA may be harnessed for therapeutic development.

miRNAs

Biogenesis and functions of miRNAs in hematopoiesis

One of the most well-studied classes of ncRNAs are miRNAs. miRNAs are typically 18 to 24 nucleotides in length and function by binding to the 3′ untranslated regions of target mRNAs to facilitate their degradation or suppress their translation58, 59, 60 (Figure 2). miRNA biogenesis begins with RNA polymerase II–mediated generation of a primary miRNA transcript (pri-miRNA) with a stem-loop structure (Figure 2).58, 59, 60, 61 The microprocessor complex, consisting of DROSHA and DGCR8, cleaves the pri-miRNA to form a 60- to 80-nucleotide–long hairpin structure known as the precursor miRNA (pre-miRNA). The pre-miRNA is then exported to the cytoplasm by Exportin-5 and RAN-GTP (Ras-related nuclear protein bound to guanosine triphosphate), at which point DICER cleaves the hairpin loop of the pre-miRNA, resulting in a miRNA duplex. The guide strand of the miRNA duplex is loaded onto and stabilized by Argonaute 2 (AGO2) forming a multiprotein complex known as the RNA-induced silencing complex (RISC), whereas the passenger strand of the duplex is subsequently degraded.58, 59, 60 Importantly, nucleotides 2 through 8 of the mature miRNA, referred to as the seed sequence, are essential in guiding the miRNA to bind target mRNAs.60, 61, 62

Figure 2.

miRNA biogenesis and function. miRNAs are produced via the processing of pri-miRNA to produce pre-mRNA, which is exported to the cytoplasm and further processed to produce mature miRNA. Their functions include translation repression and mRNA degradation. pri-miRNA, primary miRNA; RISC, RNA-induced silencing complex; RNA Pol II, RNA polymerase II. Figure was created with biorender.com. Zakutansky P. (2026) https://biorender.com/g8pbz4y.

Several miRNAs have emerged as potential regulators of hematopoiesis. For example, overexpression of miR-99a was shown to increase HSPC proliferation, resulting in the blockage of myeloid and erythroid lineage development.⁶³ Importantly, miR-99a was shown to target NIPBL (Nipped B-like protein), leading to increased self-renewal ability and the inhibition of HSPC differentiation.⁶³ Alternatively, knockout of miR-144/451 in vivo led to increased Myc expression, resulting in decreased erythroid differentiation.⁶⁴ Numerous studies have shown that miRNAs can affect cancer progression by acting as either oncogenes or tumor suppressors.^65,66 For example, miR-183 was reported to directly target PTEN mRNA, suppress protein translation, and promote cell migration and invasion in breast cancer cell lines, suggesting it acts as an oncogene.⁶⁷

miRNAs function as both oncogenes and tumor suppressors in AML

One of the first studies to report the involvement of a miRNA in hematopoiesis was published in 2004 and showed that miR-181 promotes the differentiation of murine HSCs toward the B-cell lineage.⁶⁸ Alternatively, Kong et al showed in PUER cells that the miR-23a cluster of miRNAs, which consists of miR-23a, miR-24, and miR-27a, is a direct transcriptional target of PU.1, and inhibits B-cell development in vitro and in vivo to promote a myeloid lineage bias.⁶⁹ Numerous studies have since investigated the distinct expression of miRNAs in AML (Table 2).⁷⁰ For example, a report by Li et al analyzed the levels of 435 miRNAs in primary t(8;21), inv(16), and KMT2A-rearranged AML specimens, as well as multiple AML cell lines with those same alterations.⁷¹ Microarray analysis revealed 41 differential miRNA levels in the various subtypes of patients with AML compared with healthy controls, including miR-17, miR-20, miR-126, and miR-181. The authors further showed that miR-126 is upregulated in AML1-ETO, and targets PLK2, a tumor suppressor gene, suggesting miR-126 may exert an oncogenic role in leukemogenesis by regulating PLK2.⁷¹ Interestingly, miR-1 was reported to be increased in samples from patients with FLT3-ITD1 AML, in which its increase led to increased oxidative phosphorylation and decreased survival.⁷²

Table 2.

miRNAs associated with AML

miRNA	Expression levels in AML	Role in AML	Targets	References
miR-1	•High levels in patients with FLT3-ITD+ AML compared to those with FLT3-ITD− AML	•High miR-1 levels correlated with poor patient survival •Overexpression promotes oxidative phosphorylation via glutaminolysis	NR	72
miR-9	•Low levels in bone marrow samples from patients with t(8;21) AML compared with patients with CML and healthy controls •Low levels in the NB4, HL-60, Kasumi-1, SKNO-1, and KG-1a AML cell lines compared with CD34+ cells	•Overexpression inhibits cell proliferation, migration, and invasion •Overexpression promotes apoptosis •Downregulates CXCR4	↓ CXCR4	73
miR-20a-5p	•Low levels in the bone marrow samples from patients with AML compared with age-matched healthy controls •Low levels in the Kasumi-1, THP-1, U937, and HL-60 AML cell lines compared with the HS-5 cell line	•Low levels of miR-20a-5p correlated with poor survival •Overexpression inhibits cell proliferation and cell cycle progression •Overexpression promotes apoptosis •Downregulates PPP6C	↓ PPP6C	74
miR-99a	•High levels in LSCs from patients with AML compared with non-LSCs •High levels in the KG-1a and KG-1 AML cell lines compared with the K562 cell line	•Overexpression promotes colony formation, cell proliferation, and cell cycle progression •Targets CTDSPL, PP3CA, and PRDM1	↓ CTDSPL ↓ PP3CA ↓ PRDM1	75
miR-100	•High levels in bone marrow samples from pediatric patients with AML compared with healthy controls •High levels in the Kasumi-1 and MV-4-11 AML cell lines compared with healthy controls	•Knockdown inhibits cell viability •Knockdown promotes apoptosis •Suppresses ATM	↓ ATM	76
miR-125b	•Baseline expression not discussed •Used the KG-1a and HL-60 AML cell lines	•Overexpression inhibits cell proliferation and invasion •Overexpression induces cell cycle arrest •Overexpression promotes apoptosis	↓ NF-κB signaling pathway	77
miR-126	•High levels in samples from patients with t(8;21) and inv(16) AML compared with healthy controls •Used the THP-1 and ME-1 cell lines	•Overexpression inhibits apoptosis •Overexpression promotes cell viability •Predicted to target PLK2	↓ PLK2	71
miR-133a	•High levels in the HL-60 AML cell line after treatment with ATRA	•Downregulates CDX2 via translation inhibition •Overexpression inhibits cell proliferation •May serve as a prognostic biomarker	↓ CDX2	78
miR-135a	•High levels in the HL-60 AML cell line after treatment with ATRA	•Downregulates CDX2 via translation inhibition •Overexpression inhibits cell proliferation •May serve as a prognostic biomarker	↓ CDX2	78
miR-155	•High levels in FLT3-ITD+ mice compared with WT mice •High levels in samples from patients with FLT3-ITD+ compared with FLT3-ITD WT controls	•Promotes myeloid cell production, proliferation, and expansion •Knockdown inhibits colony formation in primary AML samples •Inhibits interferon signaling by suppressing Cebpb and Stat1	↓ Cebpb ↓ Stat1	79
miR-181	•High levels of miR-181 family in the PBMNCs and CD34+ HSCs from patients with AML compared with healthy controls •High levels of miR-181 family in patients with M1-M3 AML compared with patients with M4-M5 AML and healthy controls	•Negatively regulates granulocytic and macrophage-like differentiation •Targets PRKCD, affecting P38 and C/EBPα	↓ PRKCD ↓ CTDSPL ↓ CAMKK1	80
miR-193a	•High levels in Kasumi-1 and U937T cells treated with MLT1	•Inhibits AML1-ETO expression by binding to the 3′ UTR of ETO	↓ 3′ UTR of ETO	81
miR-196b	•High levels in DNMT3A/FLT3-mutant AML	•Blocks differentiation and promotes survival	↓ TLR7/8 signaling pathway	82
miR-223	•Low levels in serum samples from patients with AML compared with healthy controls	•Associated with poor prognosis	NR	83
miR-335-3p	•Low levels in samples from patients with AML compared with healthy controls •Low levels in the THP-1, U937, and Kasumi-1 AML cell lines compared with the HS-5 cell line	•Overexpression inhibits cell proliferation and cell cycle progression •Overexpression promotes apoptosis •Downregulates eIF3E	↓ eIF3E	84
miR-486	•High levels in patients with AML from the TCGA-LAML cohort compared with healthy donors	•miR-486 levels negatively correlated with patient survival •Promotes cell proliferation in TCGA-LAML cohort. •Directly targets and suppresses SOCS2 to promote JAK-STAT signaling	↓ SOCS2	85

An upward-pointing arrow (↑) indicates that the target is upregulated by the miRNA; a downward-pointing arrow (↓) indicates that the target is downregulated by the miRNA.

ATRA, all-trans retinoic acid; CML, chronic myeloid leukemia; LSCs, leukemia stem cells; NR, not reported; PBMNCs, peripheral blood mononuclear cells; UTR, untranslated region; WT, wild-type.

The correlation between miRNA levels and patient survival has led to the proposal that miRNAs may serve as prognostic or diagnostic biomarkers.^65,66,86,87 A report from Ellson et al proposed a new prediction model for survival based on the levels of 37 miRNAs.⁸⁸ Analysis of pediatric AML samples from the TARGET-AML cohort demonstrated that high levels of miR-155-5p were associated with poor survival, whereas the upregulation of miR-199a, miR-139-5p, miR-100, miR-335, and miR-196b was associated with increased survival.⁸⁸ In a different study, miR-133a and miR-135a were shown to inhibit cell proliferation of AML cell lines by suppression of CDX2 translation.⁷⁸ Interestingly, both miR-133a and miR-135a levels are increased in pediatric patients with AML with complete remission, suggesting these miRNAs exert a tumor suppressive function, have a potential role as prognostic biomarkers, and may serve as therapeutic targets.⁷⁸

Children with Down syndrome (DS) have a 150-fold increased risk for developing AML, referred to as myeloid leukemia of DS (ML-DS).^89,90 Interestingly, infants with DS often develop a preleukemia event known as transient abnormal myelopoiesis (TAM), which results in the proliferation of myeloid blasts containing a mutation in GATA1.⁹¹ Five miRNAs (miR-99a, miR-125b, miR-155, miR-802, and let-7c) are encoded on chromosome 21.⁹¹ Wagenblast et al reported increased levels of miR-99a, miR-125b-2, miR-155, and let-7c in long-term HSCs isolated from trisomic fetal liver tissue compared with disomic fetal livers.⁹² Moreover, deletion of these chromosome 21 miRNAs in fetal liver cells, along with the introduction of a GATA1 mutation, led to a reduction in the blast population in mice that had undergone transplantation. Interestingly, the authors concluded that although the miRNAs are involved in preleukemia initiation, they are dispensable for the further progression to ML-DS.⁹² Therefore, whether these miRNAs could be used as therapeutic agents to target the transformations at identified preleukemic stages is an interesting question worth further investigation. Of note, miRNAs encoded on other chromosomes are also upregulated in ML-DS. One example is miR-486-5p, encoded on chromosome 8, which was reported to be increased in the bone marrow isolated from pediatric patients with ML-DS.⁹³ Interestingly, the short form of GATA1 (Gata-binidng factor 1) (GATA1s) has been shown to directly regulate the transcription of miR-486. Furthermore, miR-486 was identified as a promoter of AKT activation, via enhanced phosphorylation, leading to increased survival of ML-DS cells.⁹³ Although these, and other, studies provide evidence of the association of miRNAs in the development of ML-DS, further work is needed to elucidate the potential usage of miRNAs as therapeutic targets or biomarkers for ML-DS.

circRNAs

circRNA biogenesis, regulation, and molecular functions

circRNAs are an increasingly studied form of lncRNA. They are generated through alternative splicing of linear mRNA, in which the downstream 5′ splice site is covalently joined to an upstream 3′ splice site across exons (Figure 3A).^14,94, 95, 96 In this model, known as conventional back-splicing, circRNAs are produced, and an alternatively spliced linear mRNA can be subsequently generated from the noncircularized exons.^14,94 Circular intronic RNAs are also produced from the spliced intronic regions of linear mRNA.⁹⁵ It should be noted that two alternative back-splicing models have been described, including the direct back-splicing model and the lariat intermediate model. For more details, the reader is referred to other publications.^14,96 Of note, whereas mRNA and linear lncRNAs require 5′ capping and 3′ polyadenylation, circRNAs are distinct in that they require neither modification, leading to their increased transcript stability.^94,95

Figure 3.

circRNA and snoRNA formation and function. (A) Schematic depicting circRNA back-splicing, which occurs through different mechanisms to produce multiple forms of circRNAs. circRNAs have multifaceted predicted functions in AML, including miRNA sponging, protein organization, and translation regulation. (B) Schematic depicting key differences between box C/D and box H/ACA snoRNAs, their snoRNP complexes, and respective functions in facilitating rRNA modifications. Alu, Alu repeat elements; ciRNA, circular intronic RNA; ecircRNA, exonic circRNA; EIcircRNA, exon-intron circRNA; Me, 2’O-methyl modification; RBP, RNA-binding protein; ψ, pseudouridylation modification. Figure was created with biorender.com. Zakutansky P. (2026) https://biorender.com/so2bc12.

Functionally, studies have reported that circRNAs participate in translational regulation, splicing modulation, and miRNA sequestration through scaffolding and protein recruitment.97, 98, 99, 100 For example, CDR1as, also known as CiRS-7, was the first reported circRNA to act as a sponge for multiple miRNAs, including miR-7 for which it contains >70 binding sites.¹⁰¹ Work by Dou et al in oral squamous cell carcinoma cells showed the sequestration of miR-7 led to altered regulation of MAPK/AKT signaling proteins, indicating that CDR1as/ciRS-7 can function by indirectly regulating gene expression.¹⁰² Furthermore, circRNAs have been reported to interact and modulate chromatin through R-loops.¹⁰³ For example, circSMARCA5 was shown to bind to its parent locus SMARCA5 to form an R-loop and led to the transcriptional pausing of SMARCA5 at exon 15.¹⁰⁴ These functions, among others, are important to normal cellular function and have recently begun to be explored in AML.

Implications of circRNAs in hematopoiesis and AML

Although the first circRNA was discovered >40 years ago, it was not until 2012 when a large number of circRNAs were identified by RNA-seq that we began to truly understand their biology.¹⁰⁵ A 2018 study by Nicolet et al identified >14 000 distinct circRNAs using previously published data of primary human hematopoietic cells, many of which were expressed in cell-type–specific manners.¹⁰⁶ Interestingly, the authors discovered that transcript levels of circRNAs were altered during differentiation. Furthermore, analysis of the circular-over-linear ratio of each transcript was reported to be dependent on the differentiation stage during hematopoiesis. Moreover, it was reported that the highest number of circRNA transcripts were detected in platelets, followed by red blood cells and granulocytes.¹⁰⁶

Microarray analysis of bone marrow cells from 5 pooled pediatric AML cases and healthy controls revealed ∼2000 differentially expressed circRNAs.¹⁰⁷ CircRNF220, a previously uncharacterized circRNA, was found to be upregulated in patients with AML compared with controls, except for those patients with M6/M7 AML. Overexpression of circRNF220 in HL-60 and THP-1 AML cells increased proliferation and inhibited apoptosis. Moreover, as with many circRNAs (Table 3), circRNF220 was reported to act as a miRNA sponge, in this case for miR-30a.¹⁰⁷ A second study also reported increased levels of circRNF220 in AML cell lines compared with HS-5 control cells.¹⁰⁸ Mechanistically, this study determined that circRNF220 suppressed miR-330-5p by acting as a miRNA sponge, leading to the induction of SOX4, which was shown to drive proliferation and decreased apoptosis.¹⁰⁸ Together, these reports suggest that circRNF220 may influence AML progression by interacting with miRNAs and relieving their targets from negative regulation.

Table 3.

circRNAs dysregulated in AML

circRNA	Alternative name	Expression levels in AML	Function in AML	Target miRNAs	Target genes and protein	References
hsa_circ_0004291	circPOLA2	•High levels in the bone marrow samples from patients with AML compared with healthy controls •High levels in the Kasumi-3 and Kasumi-6 AML cell lines compared with the HS-5 cell line	•Promotes cell proliferation •Suppresses miR-34a	↓ miR-34a	NR	109
hsa_circ_0012152	circRNF220	•High levels in the bone marrow and peripheral blood of patients with AML compared with healthy controls	•Promotes cell proliferation •Inhibits apoptosis •Suppresses miR-30a, which inhibits MYSM1 and IER2	↓ miR-30a	↑ MYSM1 ↑ IER2	107
hsa_circ_0012152	circRNF220	•High levels in the peripheral blood of patients with AML compared with healthy controls •High levels in the TBP-1, MV4-11, KG-1, HL-60, and AML2 AML cell lines compared with the HS-5 cell line	•Knockdown of circRNF220 inhibited cell proliferation, cell cycle progression, and invasion •Knockdown of circRNF220 promoted apoptosis •Inhibits miR-330-5p, which suppresses SOX4	↓ miR-330-5p	↑ SOX4	108
hsa_circ_0000488	circDLEU2	•High levels in bone marrow samples from patients with M4/M5 AML compared with healthy controls •High levels in the MOLM-13, HL-60, and MV-4-11 AML cell lines compared with healthy human BMSC	•Promotes cell proliferation in vitro •Inhibits apoptosis in vitro •Promotes in vivo tumor formation •Inhibits miR-496 to promote PRKACB	↓ miR-496	↑ PRKACB	110
hsa_circ_0075001	circNPM1	•High levels in peripheral blood samples from patients with AML compared with healthy controls •High levels in the THP-1 and HL-60 AML cell lines compared with the HS-5 cell line	•Knockdown of circNPM1 inhibited colony formation, cell cycle progression, migration, and invasion •Knockdown of circNPM1 promoted apoptosis •Inhibited miR-345-5p, which regulates FZD5	↓ miR-345-5p	↑ FZD5	111
hsa_circ_0033144	circBCL11B	•High levels in samples from patients with AML compared with healthy HSPCs	•Knockdown of circBCL11B inhibited proliferation in the absence of FCS •Knockdown of circBCL11B in the absence of FCS promoted apoptosis	NR	NR	112
hsa_circ_0006332	circMYBL2	•High levels in patients with FLT3-ITD+ AML compared with healthy controls	•Knockdown of circMYBL2 inhibits proliferation of FLT3-ITD+ leukemic cells and colony formation •Knockdown of circMYBL2 promotes apoptosis •Enhances PTBP1 binding to FLT3 mRNA	NR	↑ PTBP1	113
hsa_circ_0001821	circPVT1	•High levels in bone marrow samples from patients with AML compared with healthy controls •High levels in the KG-1a, THP-1, and HL-60 AML cell lines compared with healthy control	•Knockdown of circPVT1 inhibits proliferation and migration •Knockdown of circPVT1 promotes apoptosis •Stabilizes c-Myc protein expression	NR	↑ CXCR4, ↑ c-Myc	114
hsa_circRNA_103104	circZBTB46	•High levels in the bone marrow and peripheral blood samples from patients with AML compared with healthy controls •High levels in the HL-60, K562, and THP-1 AML cell lines compared with the HS-5 cell line	•Inhibits ferroptosis •Promotes cell proliferation and cell cycle progression	↓ miR-326 ↓ miR-339-3p ↓ miR-671-5p	↑ SCD	115
hsa_circ_100290	NR	•High levels in the bone marrow samples from patients with M2-M6 AML compared with healthy controls •High levels in the MV-4-11, Kasumi-1, HL-60, and AML2 AML cell lines compared with the HS-5 cell line	•Knockdown inhibits cell proliferation and cell cycle progression •Knockdown promotes apoptosis •Suppresses miR-203 to promote RAB10	↓ miR-203	↑ RAB10	116
hsa_circ_0006404	circFOXO3	•Low levels in the bone marrow samples from patients with AML compared with healthy controls	•Diagnostic biomarker	NR	NR	117
hsa_circ_0001206	circCRKL	•Low levels in the bone marrow samples from patients with AML compared with healthy controls •Low levels in the KG-1a, KG-1, Kasumi-1, THP-1, and MOLM-13 AML cell lines compared with the HS-5 cell line	•Overexpression of circCRKL inhibits cell proliferation and cell cycle progression •Suppresses miR-196a-5p and miR-196b-5p to promote p27	↓ miR-196a-5p ↓ miR-196b-5p	↑ p27	118

An upward-pointing arrow (↑) indicates that the target is upregulated by the circRNA; a downward-pointing arrow (↓) indicates that the target is downregulated by the circRNA.

BMSC, bone marrow stem cells; FCS, fetal calf serum; NR, not reported.

Lux et al recently characterized a new circRNA, B-cell CLL/lymphoma 11B (circBCL11B), which was abundant in a subset of AML samples but absent in healthy controls.¹¹² Knockdown of circBCL11B in AML cell lines without serum was shown to decrease proliferation of AML cells, while increasing cell death.¹¹² An additional circRNA dysregulated in AML, circ-DLEU2, was shown to bind miR-496, an inhibitor of Protein kinase CAMP-activated catalytic subunit beta (PRKACB).¹¹⁰ Moreover, analysis of patient-derived AML samples revealed that high levels of circ-DLEU2 led to increased cell proliferation and in vivo tumor growth.¹¹⁰ These studies offer exciting insights into the functions of circRNAs in AML, and further research will be needed to determine their efficacy as therapeutic vulnerabilities.

snoRNAs

The biogenesis and molecular functions of snoRNAs

snoRNAs are a class of ncRNAs predominantly localized within the nucleolus and are typically 60 to 300 nucleotides in length.^119,120 snoRNAs can be transcribed from independent promoters within snoRNA host genes or from intronic regions of protein-coding and noncoding genes by RNA polymerase II or polymerase III.^119,120 snoRNAs are classified into two classes based on conserved structural motifs.^15,119, 120, 121, 122 The box C/D snoRNAs (SNORDs) contain box C and box D motifs, which interact to form a K-turn,^121,123 whereas box H/ACA snoRNAs form a hairpin-hinge–hairpin-tail structure, with the hinge region containing the box H motif and the tail region containing the box ACA motif^122,124,125 (Figure 3B). The motifs of the box C/D and H/ACA snoRNAs are essential for the formation of snoRNA ribonucleoprotein (snoRNP) complexes.¹²⁶ The formation of box C/D snoRNP complexes requires the interaction and binding of SNORDs with the essential factors FBL, NOP56, NOP58, and SNU13, whereas DKC1, GAR1, NOP10, and NHP2 are required for box H/ACA snoRNP complex formation and function^15,124,125 (Figure 3B).

In general, snoRNAs have been reported to modify rRNA and spliceosomal RNAs to regulate ribosome biogenesis and RNA splicing, respectively. They can also modulate modifications to tRNAs and snRNAs.^15,122, 123, 124, 125 The primary mechanism of action for SNORDs is the modulation of 2’-O-methylation of RNA^124,125 (Figure 3B). For example, it was shown, using in vivo mouse models, that U32A and U51 SNORDs modulated the 2′-O-methylation of Pxdn mRNA, which resulted in increased mRNA levels but decreased protein translation.¹²⁷ Unlike SNORDs, snoRNAs of the box H/ACA class are known to regulate pseudouridylation of RNA¹²² (Figure 3B). In non–small cell lung cancer, it was reported that depletion of NOP10 results in decreased rRNA pseudouridylation, and the additional knockout of SNORA65, SNORA7A, and SNORA7B resulted in decreased cell growth and colony formation.¹²⁸ Intriguingly, NPM1 was reported to interact with SNORDs and FBL (fibrillarin) in mouse embryonic fibroblasts to modulate 2’O-methylation of rRNA.¹²⁹ Moreover, in K562 cells, loss of NPM1 as well as SNORD15, SNORD47, and SNORD104 separately led to decreased colony formation. Alternatively, the inactivation of SNORD15, SNORD47, SNORD52, and SNORD58 increased erythroid differentiation, suggesting snoRNAs have specific biological functions.¹²⁹ Several recent reviews have discussed the dysregulation of snoRNAs in hematopoiesis and leukemias^130,131 (Table 4). Herein, we further discuss key reports and new studies establishing a role for snoRNAs in AML.

Table 4.

snoRNAs implicated in AML

snoRNA	Class of snoRNA	Expression levels in AML	Function in AML	References
SNORD115	Box C/D	•Low levels in patients with AML compared with CD34+ controls	•Alternative splicing of serotonin receptor 2C	132
SNORD116	Box C/D	•Low levels in patients with AML compared with CD34+ controls	•Involved in the pathogenesis of Prader-Willi syndrome
SNORD36C, SNORD88B, and SNORD88C	Box C/D	•Low levels in patients with AML compared with CD34+ controls	•Modification of the intersubunit bridge of the ribosome, leading to altered ribosome biogenesis
SNORD1A	Box C/D	•Low levels in patients with AML compared with CD34+ controls	•Modification of the peptidyl transferase center of the ribosome, leading to altered ribosome biogenesis
SNORA21	Box H/ACA	•Low levels in patients with AML compared with CD34+ controls	•Modification of the peptidyl transferase center of the ribosome, leading to altered ribosome biogenesis
SNORD14D	Box C/D	•High levels in AML1-ETO compared with control	•Deletion of SNORD14D reduces colony formation and impairs protein synthesis •Loss of SNORD14D inhibits leukemogenesis in vivo	133
SNORD35A	Box C/D	•High levels in AML1-ETO compared with control	•Loss of SNORD35A impairs rRNA methylation •Depletion of SNORD35A inhibits protein synthesis and colony formation •Loss of SNORD35A inhibits leukemogenesis in vivo
SNORD42A	Box C/D	•High levels in patients with AML compared with CD34+ cells	•Knockdown inhibits cell proliferation and colony formation •Modulates 2’-O-methylation at uridine 116 of 18S rRNA	134
SNORD114-1	Box C/D	•High levels in samples from patients with APL compared with controls	•Overexpression promoted cell proliferation •Loss of SNORD114-1 inhibited cell growth and cell cycle progression •Depletion of SNORD114-1 upregulates Rb pathway	135
SNORD3A	Box C/D	•Levels not significantly changed in patients with AML with low and high LSC frequency	•Knockdown inhibits cell proliferation and colony formation •Loss in primary AML samples inhibits proliferation	136
SNORD118	Box C/D	•High levels in patients with AML with high LSC frequency compared with patients with low LSC frequency	•Knockdown inhibits cell proliferation and colony formation •Loss in primary AML samples inhibits proliferation

APL, acute promyelocytic leukemia; LSC, leukemia stem cells; Rb, Retinoblastoma.

Identification and involvement of snoRNAs in AML

Among the first studies of snoRNAs in AML, Valleron et al used microarrays in conjunction with quantitative reverse transcription polymerase chain reaction to analyze changes in snoRNA expression in AML and acute lymphoblastic leukemia compared with healthy controls.¹³⁵ Their results revealed a significant downregulation of multiple snoRNAs in AML. Interestingly, in a subtype of acute promyelocytic leukemia (APL), SNORD112-114 was ectopically increased from the DLK1-DIO3 locus. Further analysis demonstrated that SNORD114-1 promoted AML cell growth by negatively regulating cell cycle progression and the Rb/p16 pathway.¹³⁵ This study provided preliminary evidence suggesting snoRNAs may be dysregulated and affect molecular mechanisms contributing to AML. Another study used a modified sequencing approach to identify novel small ncRNA, including snoRNAs, in samples from patients with AML.¹³² This study reported that snoRNA steady-state levels are lineage restricted, with most snoRNAs being downregulated during myeloid or lymphoid differentiation.¹³² This suggests any increased levels or retainment of the snoRNAs may indicate a malfunction that could promote AML. Overall, this study determined that 102 snoRNAs were significantly reduced in AML, including 37 snoRNAs located within the DLK-DIO3 or SNURF-SNRP loci. Although some studies speculate snoRNAs may be suitable diagnostic and prognostic biomarkers,^131,137,138 additional evidence is needed to support these claims.

Although studies have screened for snoRNA expression in AML, there is still a large gap in our understanding of which snoRNAs contribute to malignancy. Yun et al conducted in situ mapping of RNA-genome interactome to identify several snoRNAs in normal and malignant hematopoiesis associated with chromatin in MV4-11 cells, including SNORD118 and SNORD3A.¹³⁶ Using patient-derived AML samples that were categorized as having high or low stem cell frequency, the authors determined that SNORD118 was significantly upregulated in the high LSC (leukemia stem cell) frequency, whereas there was no correlation between SNORD3A steady-state levels and LSC (leukemia stem cell) frequency. SNORD118, as well as SNORD3A, were shown to promote cell proliferation both in cell lines and primary AML bone marrow.¹³⁶ Interestingly, SNORD42A was shown to regulate the 2′-O-methylation of U116 on 18S rRNA.¹³⁴ Loss of SNORD42A resulted in decreased cell proliferation and cell size, and reduced translation of ribosomal proteins, suggesting SNORD42A could be a tumor-promoting snoRNA.¹³⁴

Relatively few studies in the past decade have examined the role of snoRNAs in AML subtypes. However, 1 study found that SNORDs and RNPs contribute to leukemogenesis in AML1-ETO–positive AML.¹³³ Leveraging the AML1-ETO9a (AE9a) murine transplant model, the Groucho-related amino-terminal enhancer of split (Aes) was identified as being essential for AE9a leukemogenesis. Transduction of MLL/AF-9 into mouse HSPC-enriched lineage-negative bone marrow cells led to increased snoRNA levels, irrespective of Aes.¹³³ Coimmunoprecipitation and RNA immunoprecipitation analysis of Kasumi-1 cells revealed AES protein interacted with DDX21 and snoRNAs. Microarray analysis revealed that 79 snoRNAs were downregulated after Aes deletion, resulting in rRNA methylation inhibition. Knockdown of NOP58, a core box C/D snoRNP component, reduced snoRNA levels and 2’-O-methylation.¹³³ Six SNORDs, including SNORD14D and SNORD35, were reduced by both Aes deletion and NOP58 knockdown and promoted colony formation and leukemogenesis in MV4-11 cells. This study also demonstrated that the RNA helicase DDX21 interacted with Aes and loss of DDX21 led to decreased snoRNA levels.¹³³ This novel study provided key evidence that an RNA helicase regulates snoRNAs in AML and potentially contributes to hematologic malignancies.

Interestingly, additional studies reported that the dysregulation of RNA helicases perturbs snoRNAs in hematopoiesis and AML. Chlon et al demonstrated in Ddx41-knockout mice that Ddx41 was required for the regeneration and proliferation of HSPCs and observed increased snoRNA levels.¹³⁹ Furthermore, DDX41 dysregulation disrupted snoRNA processing and expression, further leading to altered rRNA processing.¹³⁹ In a separate study, loss of Ddx10 in mouse embryonic stem cells resulted in cell cycle progression inhibition and induce apoptosis, resulting in a blockage of mouse embryonic stem cell growth.¹⁴⁰ DDX10 was determined to be essential for proper 18S rRNA maturation, with loss of DDX10 inhibiting U3 snoRNA release from pre-rRNA, leading to the dysregulation of 40S small ribosomal subunit biogenesis. Notably, DDX10 forms the NUP98-DDX10 fusion protein observed in AML, leading to a reduction of DDX10 protein.¹⁴⁰ Whether this fusion protein, or others, modulates snoRNA function and contributes to the development of AML, remains unknown. Collectively, these studies highlight the involvement of snoRNAs in AML development, although their mechanistic roles and potential as prognostic indicators or therapeutic targets will require further investigation.

Conclusions

Numerous studies have provided enticing data suggesting the dysregulation of ncRNAs, including lncRNAs, miRNAs, circRNAs, and snoRNAs, contributes to the pathogenesis of cancers, including AML.^{17,21,24,25,38,98,132} In vitro models has demonstrated that lncRNAs, miRNAs, circRNAs, and snoRNAs contribute to AML by altering cell proliferation, apoptosis, differentiation, and cell cycle progression.^{46,47,51,54,75,77,79,114,115,134,141} Additional reports have suggested that the abundance of ncRNAs may serve as diagnostic and prognostic biomarkers.^{19,20,100,137,138,142} However, variation in ncRNA levels due to alterations in biogenesis and regulatory mechanisms will be a constant challenge faced by the field. Therefore, the pathways influencing ncRNA biogenesis will need to be elucidated to confidently consider these species as biomarkers.

Many reports have shown that lncRNAs and circRNAs can directly regulate miRNAs in AML by acting as molecular sponges.^{49,54,56,102,109,111,118,143} Additionally, lncRNAs can indirectly regulate miRNAs through competitively binding to target mRNAs.¹⁴³ Separately, miRNAs affect AML cells and disease progression by regulating critical signaling cascades and transcriptional regulators.^{68,72,73,75,83,85,87} Furthermore, recent studies have implicated snoRNAs in AML by modulating 2’-O-methylation of rRNA, often resulting in altered ribosome biogenesis.^131,133,134 Together, these studies have shed light onto the multitude of mechanisms by which ncRNAs contribute to AML. Advancing our understanding of these functional mechanisms will be essential for the potential development and usage of ncRNAs as therapeutic agents.

Conflict-of-interest disclosure: J.D.C. receives research support from Syndax. The remaining authors declare no competing financial interests.