Introduction

The research journey of Small nucleolar RNAs (snoRNAs) traces back to 1968 when Weinberg identified six 100–180 nucleotides single-stranded RNAs within HeLa cell nucleoplasm and nucleoli—though their nature and localization remained enigmatic.1 By the 1980s, refined techniques distinguishing small nucleolar RNAs (snoRNAs) from small nuclear RNAs (snRNAs) enabled Zieve’s seminal discovery of SNA/U3 (later designated SNORD3A) in the nuclear compartment.2 The advent of high-throughput sequencing in the early 21st century unveiled snoRNAs’ multifaceted roles, revealing their orchestration of pathogenesis across benign and malignant disorders.3–7 Cancer regulation has since emerged as a pivotal biomedical frontier, evidenced by proliferating studies on snoRNA-mediated oncogenesis.

SnoRNAs originate from introns of host genes, liberated via alternative splicing of primary transcripts to evade exonuclease degradation through snoRNP complex formation. Canonically classified by conserved motifs: C/D-box snoRNAs recruit fibrillarin for 2′-O-methylation of target RNAs, while H/ACA-box variants bind dyskerin to catalyze pseudouridylation—functional divergence dictated by distinct ribonucleoprotein topologies.8,9 Additionally, non-canonical subtypes include Cajal body-localized scaRNAs and orphan snoRNAs, the latter remaining underexplored in oncogenic contexts.10,11 Critically, recent advances illuminate non-canonical snoRNA functions—modulating acetylation, chromatin accessibility, and histone marks—thereby mechanistically bridging critical gaps in understanding snoRNA-driven carcinogenesis.12,13

Oncogenic aberrations spanning gene duplication, transcription, and translation collectively disrupt snoRNA homeostasis, establishing a feedforward loop that accelerates tumor progression.14 Guided by the classic hallmarks of cancer delineated in Hanahan and Weinberg’s seminal review, this review synthesizes how snoRNAs and their derivatives subvert malignancy through canonical and non-canonical mechanisms. It concurrently evaluates the clinical utility of intra- and extracellular snoRNA signatures for cancer diagnosis and prognostic stratification.15–19

Structural and Functional Classification of snoRNAs

SnoRNAs are single-stranded, intron-derived noncoding RNAs of 60–300 nucleotides.20 Structurally, they are classified into two canonical (shown in Figure 1) and two noncanonical categories based on conserved sequence motifs and cellular location. Understanding this structural logic is essential, because each class has distinct vulnerabilities to dysregulation in cancer.

Figure 1 The structures of two canonical snoRNAs: (a) C/D box snoRNA; (b) H/ACA box snoRNA.

Canonical ClassesC/D Box snoRNAs (SNORDs)

The first canonical class, designated as C/D box snoRNAs (SNORDs), features evolutionarily conserved sequence motifs critical for ribonucleoprotein assembly. These motifs include a 5′-terminal C-box (consensus sequence 5′-RUGAUGA-3′, where R represents purine residues adenine or guanine) and a structurally complementary 3′-terminal D-box (5′-CUGA-3′).21 Most C/D box snoRNAs additionally contain C′ and D′ boxes—sequence elements resembling the C and D boxes but with 1–2 nucleotide variations.22 These molecules are structurally characterized by terminal regions flanked by structurally defined inverted repeats, typically spanning 4–5 nucleotides in length.23 Critically, the spatial proximity between the 5′-end C box and 3′-end D-box, stabilized through terminal base-pairing interactions, enables formation of a conserved “stem-bulge-stem” secondary structure known as the kink-turn (K-turn) motif.24 This architectural element adopts a characteristic sharp-angled hairpin conformation.25 As the conformationally defined functional core of C/D box snoRNP complexes, C/D box snoRNAs serve as molecular scaffolds that govern the assembly of core proteins—including NOP56, NOP58, 15.5K (SNU13), the 2′-O-methyltransferase fibrillarin, and fibrillarin-like proteins.26–30 This scaffolding is indispensable for protein complex stability, subnuclear localization, and catalytic functionality. The assembled proteins collectively confer snoRNP resistance to exonucleolytic degradation while ensuring precise nucleolar compartmentalization essential for ribosomal RNA modification.31–33 Functionally, the antisense guide region—spanning 10–21 nucleotides upstream of the D or D′ box motifs—base-pairs complementarily with target RNAs. This precise positioning directs the methyltransferase fibrillarin (and its orthologous proteins) to catalyze site-specific 2′-O-ribose methylation precisely at the fifth nucleotide upstream of the bound D/D′ motif.34–36 Additionally, evolutionarily conserved C/D box snoRNAs choreographs 18S rRNA acetylation in eukaryotes by recruiting the acetyltransferase complex. This facilitates enzymatic deposition of N4-acetylcytidine (ac4C) modifications at helix 45 of the small ribosomal subunit, a critical structural element for translational fidelity.37–39

H/ACA Box snoRNAs (SNORAs)

The second class of snoRNAs, H/ACA box snoRNAs, as known as SNORAs, functionally rely on their characteristic “hairpin-hinge-hairpin-tail” secondary structure. This architecture incorporates a central hinge region harboring the conserved ANANNA sequence (H-box motif), with the defining ACA-box (or infrequent AUA variant) invariably positioned precisely 3 nucleotides upstream of the 3′ terminus.40,41 Each hairpin domain forms an asymmetric internal loop that collectively establishes two distinct pseudouridylation pockets.42,43 Embedded within these pockets, antisense guide sequences confer target RNA specificity through complementary base-pairing interactions, thereby directing precise positioning of the H/ACA snoRNP complex. This ribonucleoprotein complex utilizes the snoRNA as a structural scaffold to assemble the pseudouridine synthase dyskerin—the human ortholog of yeast CBF5P—along with its regulatory subunits GAR1P, NHP2P, and NOP10P.21,41,44,45 Catalytic activity follows stringent spatial constraints: upon target RNA engagement, dyskerin’s active site modifies specific uridine residues located exactly 14–15 nucleotides upstream of either the H-box or ACA-box.46 Through an RNA-dependent catalytic mechanism involving 180-degree rotation of the glycosidic bond, it mediates uridine-to-pseudouridine (Ψ) isomerization.47 This post-transcriptional modification significantly enhances RNA structural stability and plays indispensable roles in ribosome biogenesis and pre-mRNA splicing regulation.48 Both SNORDs and SNORAs are predominantly nucleolar, directing modifications essential for ribosome biogenesis and function.

Noncanonical ClassesSmall Cajal Body-Specific RNAs (scaRNAs)

ScaRNAs represent the third major class of small nucleolar RNAs, and their name derives from a unique UGAG localization motif (a Cajal body-targeting element). This motif drives their specific subnuclear enrichment within dynamic intranuclear compartments termed Cajal bodies.49 These RNAs exhibit remarkable structural heterogeneity: some scaRNAs contain canonical C/D box domains, others harbor H/ACA domains, while chimeric variants integrate both structural modules.50–52 This architectural diversity enables dual catalytic functions—2′-O-methylation and pseudouridylation (Ψ).53 Beyond the core UGAG motif, scaRNAs incorporate two critical localization elements: the CAB box and GU-repeat sequences. The CAB box directly binds the transport adaptor protein TCAB1 (Telomerase Cajal body protein 1), whereas GU-repeats recruit the nuclear transport receptor Transportin-1.54,55 These components cooperatively assemble into a nucleocytoplasmic shuttling complex that actively targets scaRNPs to Cajal bodies.56 Within this compartment, scaRNPs precisely modify functionally critical sites of small nuclear RNAs (snRNAs), exemplified by 2′-O-methylation in the 3′ stem-loop region of U1 and U2 snRNAs.57,58 Such modifications enhance snRNA structural stability and promote efficient assembly of Sm protein rings, ultimately ensuring the maturation and functional activation of small nuclear ribonucleoprotein (snRNP) subunits. As core components of the spliceosome, these snRNP subunits play an essential role in maintaining pre-mRNA splicing fidelity.55 Although less studied in cancer, emerging evidence links scaRNA13 mutations to dyskeratosis congenita and increased tumor susceptibility.

Orphan snoRNAs

A subset of snoRNAs lacks identifiable guide sequences or canonical targets. These orphan snoRNAs exert regulatory functions through diverse mechanisms: modulating pre-mRNA splicing, altering polyadenylation site (PAS) selection, regulating ribosome/snRNA biogenesis, acting as chromatin-associated RNAs, or processing into miRNAs or piwi-interacting RNAs (piRNAs).59–63 Such activities underpin roles in development, cardiovascular diseases, and oncogenesis.64–66 Nevertheless, the biological significance of most orphan snoRNAs in cancer remains enigmatic.

Origin and Processing of snoRNAs

The majority of human snoRNAs are embedded within introns of protein-coding genes or non-protein-coding snoRNA host genes (SNHGs).67,68 These intronic snoRNAs evade exonucleolytic degradation by associating with ribonucleoproteins (RNPs) that block trimming destruction.69 High-throughput sequencing further reveals that snoRNAs can undergo processing into shorter functional fragments, expanding their regulatory repertoire.70 The biogenesis of snoRNAs involves transcription, splicing, end trimming, and ribonucleoprotein assembly. Each step presents opportunities for dysregulation in cancer (shown in Figure 2).

Figure 2 The generation process of snoRNAs.

Small Nucleolar RNA Host Gene (SNHG)

SNHGs are genomic loci whose primary transcripts undergo processing to yield both snoRNAs and long noncoding RNAs (lncSNHGs).71,72 Full-length transcripts of SNHGs—comprising exons and introns—are exported to the cytoplasm as lncSNHGs, modulating diverse biological processes.73 In contrast, nuclear SNHG transcripts, predominantly intronic, are processed into 65–300 nt snoRNAs that localize to nucleoli.74 A single SNHG can yield multiple snoRNAs; for instance, the SNHG1 transcript serves as the precursor for eight independently processed SNORDs (SNORD22, SNORD25-SNORD31).67,68,75,76 To date, the HUGO Gene Nomenclature Committee has identified 33 SNHGs, including SNHG1–SNHG33, GAS5, MEG8, and DANCR. lncSNHGs exhibit dual roles as oncogenes or tumor suppressors, regulating processes such as miRNA sponging, protein ubiquitination, and DNA methylation. Their dysregulation impacts cancer hallmarks including epithelial-mesenchymal transition (EMT), proliferation, and apoptosis evasion.77–81 This review, however, focuses more on snoRNA-specific mechanisms.

Processing of snoRNAs

The maturation of snoRNAs involves species-specific pathways. In lower eukaryotes, snoRNA biogenesis mostly occurs predominantly through standalone transcription followed by precise 5′ and 3′ end trimming82,83 In mammals and other higher eukaryotes, however, most snoRNAs derive from intronic processing of SNHG transcripts. This mechanism requires the coordinated integration of snoRNA biosynthesis with host transcript splicing, nucleocytoplasmic transport, and translational regulation.84–86 This section focuses exclusively on the biogenesis of intron-embedded snoRNAs.

Following co-transcriptional splicing, excised host introns release snoRNA-containing precursors lacking 5′ caps or poly-A tails.87 Nuclear processing of precursor snoRNAs initiates with recognition of their conserved structural motifs (eg, C/D boxes) and flanking sequences by dedicated chaperones, which coordinate subsequent end-trimming and RNP assembly.88,89 Studies in yeast demonstrate that snoRNA-bearing intron lariats are linearized by the debranching enzyme Dbr1 or endonuclease Rnt1.90–92 Then pre-snoRNAs undergo distinct 5′ and 3′ end processing pathways. At the 5′ terminus, Nop1p and Rnt1p recognize cap-proximal target sites to initiate decapping and endonucleolytic cleavage. Conversely, at the 3′ terminus, initial endonucleolytic cleavage generates entry points for exonucleases that progressively trim the trailer sequence.93–95 Dysregulation of trimming enzymes (eg, the exosome complex) can alter snoRNA stability and expression levels in tumors.96

Assembling of snoRNPs

To prevent post-transcriptional degradation by exonucleases, nascent snoRNAs immediately associate with specific proteins during transcription, forming stable snoRNP complexes.97 For C/D box snoRNPs, assembly initiates with Snu13 binding to both the canonical C/D motif and the distal C’/D’ motif, inducing a kink-turn structural arrangement. This facilitates sequential recruitment of Nop58, NOP56, and finally fibrillarin.98,99 In H/ACA box snoRNP assembly, NAF1P serves as a scaffold that co-transcriptionally recruits CBF5P (dyskerin), NHP2P, and NOP10P through simultaneous interactions with RNA polymerase II’s C-terminal domain (CTD) and the 3′ terminal ACA motif of snoRNAs.100 Gar1p is incorporated post-transcriptionally.101 Both pathways critically depend on chaperone systems including heat shock protein 90 (Hsp90) and the R2TP complex, which stabilize core proteins and fine-tune hierarchical assembly.102,103 Cancer-associated mutations or stress conditions that impair chaperone function can lead to incomplete assembly, exposing snoRNAs to degradation or causing their mislocalization.104

snoRNA-Derived RNAs (sdRNAs)

High-throughput sequencing has identified processed shorter snoRNA degradation fragments (sdRNAs), expanding the research scope for novel snoRNA functions. These fragments are found across all tested species, from mammals to viruses.63,105,106 The processing and excision of mature sdRNAs are largely considered to follow the classical miRNA biogenesis pathway.107 In fact, mature sdRNAs and mature miRNAs are nearly indistinguishable structurally.70 One exception is that sdRNA length varies slightly depending on whether the parental snoRNA belongs to the C/D box family (approximately 27 nucleotides) or H/ACA box family (17–19 nucleotides), whereas miRNA length is defined as 21–22 nucleotides.108 Full-length snoRNAs are cleaved into shorter transcripts by the microprocessor complex composed of Drosha and DGCR8, analogous to the conversion of primary miRNAs into pre-miRNAs.109 Subsequently, snoRNA fragments are exported from the nucleus to the cytoplasm via Exportin-5 in a RAN-GTP-dependent manner.110 In the cytoplasm, snoRNA fragments are further processed by the enzyme Dicer to generate mature sdRNAs, which then bind Argonaute-2 to form RNA-induced silencing complexes (RISC).111–114 At this stage, mature sdRNAs function as molecular guides directing RISC to target mRNAs. Additionally, alternative biogenesis pathways distinct from miRNA-type sdRNAs exist. For example, piRNA-like sdRNAs follow a unique biosynthetic pathway where snoRNA transcripts are exported to Yb bodies, undergo 3′-end cleavage by Zucchini (ZUC) enzyme, and subsequently receive PIWI-domain attachment at the 5′ end.115–117 These intermediate piRNA-like sdRNAs are further processed by Papi-dependent trimming enzymes, transported to the cytoplasm, and may re-enter the nucleus to suppress transcription.118,119

snoRNA Dysregulation in Cancer

Following their discovery via high-throughput sequencing, dysregulated expression of snoRNAs has been observed in various cancers such as colorectal cancer (CRC), lung cancer and gastric cancer (GC).120–123 And disruption of snoRNA homeostasis in cancer is not a single event but the consequence of convergent alterations acting at the DNA, RNA, and protein levels. These five layers—genetic, epigenetic, post-transcriptional, degradative, and spatial—often co-exist and reinforce each other, creating a feedforward loop that amplifies malignant phenotypes.

Genetic Alterations

Genetic alterations can include abnormal gene structure, copy number variations, and base sequence alteration.124,125

Abnormal Gene Structure

Deletions in chromosome 6q, particularly the 6q14-q16 region, have been studied in breast and prostate cancers, where snoRNA U50 (SNORD50) is considered an important tumor suppressor gene in this segment. Its deletion is associated with tumor initiation and progression. When it undergoes a homozygous 2-bp (TT) deletion mutation, it exhibits transcriptional downregulation.126,127 Zhao et al found that in multiple myeloma patients, high expression of SCARNA22 could distinguish patients with t(4;14). Compared to patients without t(4;14), certain H/ACA box snoRNAs targeting pseudouridylation (such as SNORA20, SNORA24, and SNORA14B) were highly expressed in patients carrying t(4;14). Conversely, patients carrying t(11;14) exhibited high expression of C/D box snoRNAs but lower levels of H/ACA box snoRNAs.128 Simultaneously, Chu et al found that snoRNA ACA11 was highly expressed in t(4;14)-positive multiple myeloma and other cancers.65 Interestingly, among 1545 breast cancer samples with gene fusion events, 1155 (75%) contained at least one fusion fragment related to a SNHG. This led to upregulated snoRNA expression, potentially due to a stronger promoter post-fusion.129

Copy Number Variations

In hepatitis B virus-associated hepatocellular carcinoma (HCC), amplification of the chromosomal 15q13.3 region elevates SNORA18L5 expression, which correlates with reduced overall survival in patients.130 Moreover, the snoRNA gene locus encoding SNORD50A-SNORD50B exhibits copy number loss in 10–40% of tumors across 12 common cancer types, with this deletion associated with poorer survival outcomes.131 And copy number loss of SNORD genes located on chromosome 14q32 was observed in 27.3% of infantile diffuse high-grade astrocytomas.132

Base Sequence Alteration

Biallelic mutations in the SNORD118 gene, which encodes the C/D box snoRNA U8, cause leukoencephalopathy with cerebral calcifications and cysts (LCC).133 In an obese mouse model, SNORD46, through a G11A mutant, enhanced binding affinity to IL-15, leading to suppressed lipolysis and obesity.134 These genomic events establish that snoRNA genes are not passive passengers but active drivers embedded within the chromosomal chaos of cancer.

Epigenetic Abnormalities

Epigenetic abnormalities (eg, histone modifications, DNA methylation, and post-transcriptional modifications) lead snoRNA abnormal expression, with post-transcriptional modifications addressed in a dedicated section. For example, Ho et al found in a bisphenol A (BPA) exposure-induced prostate cancer model that BPA exposure altered the recruitment of H3K9me3, H3K4me3, and H3K27me3 in the 5′ regulatory/exon sequences of five SNORDs, thereby reducing the expression of these genes.135 Hypermethylation of promoter-proximal CpG islands at host gene loci for SNORD123, U70C, and ACA59B in neoplastic cells, inducing transcriptional silencing through methyl-CpG binding domain protein recruitment.136 Thus, epigenetic therapy (eg, DNA methyltransferase inhibitors) might reactivate tumor-suppressive snoRNAs, an underexplored avenue.

Post-Transcriptional Processing Defects

As mentioned earlier, post-transcriptional pre-snoRNAs require processing steps including splicing, 5′ and 3′ end modifications, and protein assembly to become functional. Abnormalities in these steps have been observed during tumor initiation and development.

Splicing Factor Alterations

Serine-arginine protein kinase 1 (SRPK1), a key regulator of alternative splicing, was found to be elevated in GC. Co-immunoprecipitation analysis suggested it may bind to nucleolin (NCL) in the nucleolus to exert splicing effects on SNORA42, SNORA74A, and SNORD10.137 The splicing factor polyadenylate-binding nuclear protein 1 (PABPN1), in hepatoblastoma (HB), binds to Splicing Factor Proline- and Glutamine-Rich (SFPQ) and induces intron retention after alternative splicing of SNHG RNAs (SNHG1, SNHG8, SNHG10, SNHG12, SNHG19, SNHG29), downregulating intronic snoRNAs. For instance, overexpression of intron-retained SNHG19 promoted HB cell growth, while overexpression of spliced SNHG19 suppressed HB cell growth both in vitro and in vivo. Furthermore, SNORD60 overexpression was confirmed to significantly inhibit HB cell proliferation.138

3′ End Processing Defects

Dyskeratosis congenita is characterized by increased tumor susceptibility. In these patients, mutations in the N-terminal extension domain (NTE) of dyskerin were accompanied by downregulation of scaRNA13. CRISPR/Cas9 experiments revealed that NTE mutations led to abnormal 3′ oligoadenylation extension of the scaRNA13 transcript via the PARN (encoding poly(A) specific ribonuclease)-PAPD5 (non-canonical poly(A) polymerase) axis.113,139

Protein Assembly Faults

Peroxiredoxin 1 (Prx1) has snoRNA-binding functions; however, under oxidative stress, conformational changes in Prx1’s cysteine residues lead to dissociation. When Prx1 is overexpressed, it stabilizes binding snoRNAs, thereby increasing snoRNA expression levels—a mechanism worthy of attention in subsequent cancer research.140 These examples highlight that post-transcriptional control is a major nexus for snoRNA dysregulation.

Degradation Abnormalities

SnoRNAs are metabolically stable but not immortal. In rapidly proliferating tumors, increased rRNA processing can consume snoRNAs faster than they are synthesized, leading to relative deficiency.141 Furthermore, SNORD50A/B competitively bind K-Ras, blocking farnesyltransferase access; when these snoRNAs are deleted, K-Ras hyperprenylation and ERK signaling are potentiated—culminating in tumorigenesis.131 This competitive binding paradigm—snoRNAs acting as decoys for oncogenic proteins—represents a non-canonical degradation-related mechanism with therapeutic implications.

Mislocalization

Proper subcellular localization is essential for snoRNA function. For example, under palmitate-induced lipotoxic conditions, RPL13a-encoded snoRNAs (U32a, U33, and U35a) accumulate in the cytoplasm of C2C12 myoblasts while nuclear levels remain unchanged.142 SNORA18L5 overexpression in HBV-related HCC can cause excessive retention of RPL5 and RPL11 in the nucleolus. This retention prevents them from binding to MDM2. This culminates in augmented MDM2-mediated ubiquitination and proteasomal degradation of p53.130 Notably, snoRNAs are also released into body fluids within extracellular vesicles. Urinary exosomal SNORD99, SNORD22, SNORD26, and SNORA50C distinguish clear cell renal cell carcinoma from urolithiasis with high accuracy (AUC = 0.811, p = 0.0091), underscoring the diagnostic potential of mislocalized snoRNAs.143

snoRNAs in Cancer Hallmarks

In several cancers, snoRNAs contribute to the following tumor hallmarks through their canonical functions in 2′-O-methylation and pseudouridylation modifications, as well as non-canonical roles in histone modification and competitive binding inhibition. Some examples are briefly illustrated in Figure 3.

Figure 3 The examples of snoRNAs and SNHGs related to cancer hallmarks.

Sustained Proliferation

The acquisition of limitless replicative potential represents a pathognomonic characteristic of malignant progression. Tumor cells subvert physiological constraints on mitotic entry and DNA replication fidelity.15 Within this framework, snoRNAs demonstrate context-dependent duality in modulating tumor proliferation: functioning as oncogenic drivers in specific malignancies while acting as tumor suppressors in others, as evidenced across HCC, lung cancer, ovarian cancer, CRC, breast cancer, and bladder cancer.144–149

Transcription Regulation

Certain snoRNAs potentiate proliferation through direct interactions with sequence-specific transcription factors. For instance, SNORA74A expression positively correlates with disease severity and poor prognosis in HCC. Mechanistically, SNORA74A is highly expressed in HCC stem cells and binds DDB1- and Cul4-associated factor 13 (DCAF13). This binding prevents E2F2 ubiquitination and degradation, thereby stabilizing E2F2. Stabilized E2F2 increases NOTCH3 transcription, activating the Notch3 growth pathway to drive self-renewal of HCC stem cells and hepatocarcinogenesis.150 Similarly, SNORA38B overexpression in non-small cell lung cancer (NSCLC) promotes proliferation by directly binding E2F1 to regulate the GAB2-mediated AKT/mTOR pathway.151 Furthermore, snoRNAs directly drive ribosomal translocation into the nucleolus, as exemplified by SNORA68, which binds U2AF2 to drive RPL23 relocation from the nucleoplasm to the nucleolus. This relocation amplifies c‑Myc translation and enhances stemness properties in triple‑negative breast cancer (TNBC).152

Epigenetic Regulation

SnoRNAs engage in the modification of nucleic acids and histones. SNORA37 overexpression in gastric cancer—clinically linked to poor prognosis—drives tumor growth in vitro and in vivo by hijacking alternative splicing machinery. Specifically at the mechanism level, this snoRNA facilitates cap methyltransferase 1 (CMTR1) binding, stabilizes CMTR1- ELAVL1 (Embryonic Lethal Abnormal Visual 1) complexes to nuclear-retain ELAVL1, and ultimately reprograms CD44 splicing.153 Another example is SNORD9, whose overexpression drives ovarian cancer proliferation through dual mechanisms. It binds m6A methyltransferase METTL3 to elevate m6A modification of NFYA, and mediates 2′-O-methylation of the m6A reader IGF2BP2, thereby stabilizing NFYA mRNA and upregulating downstream proteins CCND1, CDK4, and VEGFA.154 SNORA28 further exemplifies snoRNA-mediated proliferation in CRC by recruiting Bromodomain-containing protein 4 (BRD4) to increase H3K9 acetylation at the Leukaemia Inhibitory Factor Receptor (LIFR) promoter, thereby enhancing LIFR transcription and activating the JAK1/STAT3 pathway. This activation accelerates tumor growth and concurrently induces radioresistance.155

Tumor-Suppressive Function

Conversely, Ke et al revealed tumor-suppressive functions of snoRNA-related elements, extending mechanistic insight into the SNHG18/c-Myc/p21 axis. Targeting this axis inhibits bladder cancer progression by triggering ubiquitin-dependent c-Myc degradation and subsequent p21-induced G0-G1 cell cycle arrest.156

Inhibition of Cell Death

Cell death is categorized as programmed (eg, apoptosis, autophagy, pyroptosis, necroptosis, ferroptosis, cuproptosis, panoptosis, disulfidptosis) or non-programmed.157–159 SnoRNAs block multiple cell death pathways.

Pyroptosis

SNORD99 upregulation in endometrial cancer catalyzes 2′-O-methylation of GSDMD via forming the SNORD99-FBL RNP complex. This modification downregulates pyroptosis executors (GSDMD, caspase-1, NLRP3), directly inhibiting pyroptotic cell death and ultimately driving malignant progression.160

Autophagy

Through mediating 2′-O-methylation at 28S rRNA-C3680, SNORD88C enhances the translation of stearoyl-CoA desaturase-1 (SCD1), the key lipogenic enzyme for monounsaturated fatty acid (MUFA) synthesis. This leads to elevated lipid peroxidation and activated mTOR signaling. These effects ultimately suppress autophagy, thereby promoting tumor progression in non-small cell lung cancer (NSCLC).161

Ferroptosis

Lipid peroxidation suppression underlies SNORA56-mediated ferroptosis inhibition in CRC. This suppression is achieved through pseudouridylation of 28S rRNA-U1664, which enhances translation of the glutamate-cysteine ligase (GCLC) catalytic subunit.162

p53-Related Death

SnoRNAs dictate divergent mechanisms of p53 degradation to modulate cell death. Specifically, SNORD50A/B deletion enhances oncogenicity in KRAS-mutant cell line.131 Paradoxically, as demonstrated by Su et al, such deletion confers survival advantage in p53 wild-type breast cancer by disrupting the tripartite motif-containing 21 (TRIM21)-guanosine 5′-monophosphate synthase (GMPS) complex. This enables nuclear translocation of GMPS, where it stabilizes p53 via USP7-mediated deubiquitination, thereby suppressing malignant phenotypes.163 Conversely, SNORD6 overexpression in cervical cancer potentiates E6-mediated ubiquitin-proteasomal degradation of p53, effectively blocking apoptosis.164 Moreover, the p53-p21 pathway is downregulated by SNORA24 through ubiquitin-proteasomal co-degradation in CRC.165

Invasion and Metastasis

Invasion and metastasis are critical drivers of tumor progression, and snoRNAs contribute to these processes through multiple mechanistic layers.

Alternative Splicing

By guiding 2′-O-methylation of small nuclear RNAs (snRNAs), certain snoRNAs induce spliceosome dysfunction and global alternative splicing changes that favor metastasis. For example, SNORD67 promotes lymph node metastasis and distant spread in breast cancer through upregulating 2′-O-methylation on U6 snRNA, which alters mRNA splicing profiles.166 Similarly, SNORD89 is elevated in endometrial cancer—particularly in lymph node metastatic specimens—where it drives proliferation and migration by catalytically mediating 2′-O-methylation of the pro-apoptotic factor Bim, as validated by antisense oligonucleotide (ASO)-directed silencing.167

Stabilized Metastasis-Related mRNAs

Other snoRNAs directly bind RNA stability regulators to prolong the half-life of mRNAs encoding pro-metastatic proteins. SNORA71A is markedly upregulated in metastatic breast cancer specimens. It enhances metastatic competence by upregulating ROCK2 (a TGF-β signaling suppressor) through two mechanisms: elevating ROCK2 mRNA and protein abundance, and directly binding the mRNA stability regulator G3BP1 to prolong ROCK2 transcript half-life.168 In normal pregnancy—a physiologically invasive process—SNORD88B enhances trophoblast invasion and migration by recruiting splicing factors SRSF1 and U2AF1 to regulate G3BP1 pre-mRNA splicing; additionally, it stabilizes G3BP1 mRNA through METTL14-mediated m6A modification and IGF2BP2 binding, further reinforcing pro-invasive function.169

ceRNA Networks and Signaling Pathway

A third group of snoRNAs functions as competitive endogenous RNAs (ceRNAs) or directly modulates signaling cascades. SNHG12, delivered via gastric cancer-derived extracellular vesicles, induces peritoneal metastasis through the miR-129-5p/E2F7/MAPK/ERK axis, triggering mesothelial-mesenchymal transition (MMT) in peritoneal cells.170 SNHG6 promotes ovarian cancer migration and epithelial-mesenchymal transition (EMT) via the miR-543/YAP1 pathway, while SNORA50C, acting downstream of its host gene SNHG25, facilitates neuroblastoma metastasis through HDAC1-mediated signaling.171,172

Dual-Pathway

Some snoRNAs exert metastatic effects through simultaneous actions on both rRNA maturation and non-canonical RNA decay. SNORD11B functions as a dual-pathway oncogenic driver in colorectal cancer. It catalyzes site-specific 2′-O-methylation at G509 of 18S rRNA to potentiate ribosome maturation, while concurrently inducing Nm-mediated degradation of pri-let-7a at the G225 site, thereby suppressing tumor-suppressive let-7a-5p expression. Overexpression of SNORD11B shows strong clinical association with lymph node metastasis progression (p < 0.001).173

Collectively, these examples illustrate that snoRNAs drive metastatic dissemination through diverse, often non-canonical mechanisms—ranging from spliceosome remodeling and mRNA stabilization to ceRNA networks and dual-pathway regulation. Many of these snoRNAs are enriched in metastatic specimens, underscoring their clinical utility as prognostic markers or therapeutic targets.

Angiogenesis

Tumors hijack angiogenesis to fuel growth, with emerging roles for snoRNAs. Exosomal SNHG12 transfers from breast cancer cells to endothelial cells, where it binds PBRM1, liberating MMP10 to potentiate angiogenesis.174 In GC, SNHG22 stabilizes HMGA1 through miR-361-3p sponging, activating Wnt/β-catenin signaling to potentiate tumor vascularization.175 And SNHG14 drives HCC angiogenesis by inducing PABPC1 via H3K27 acetylation and upregulating PTEN signaling.176 Hypoxic glioblastoma (GBM)-derived exosomes are significantly enriched for SNORD116-21 alongside pro-angiogenic regulators (LOX, TSP1, VEGF, ADAMTS1) and enhance endothelial progenitor cell tube formation, suggesting a role for exosomal snoRNAs in tumor angiogenesis.177 These findings position snoRNAs as non-cell-autonomous regulators of the tumor microenvironment.

Metabolic Reprogramming

SnoRNAs regiment oncogenic metabolic reprogramming by mobilizing key metabolic effectors, thereby fueling accelerated proliferative demands of malignant phenotypes. In breast cancer, sdRNA-93 enhances invasiveness by regulating sarcosine metabolism via Pipox.178 In response to UV-C stress, specific snoRNAs—including SNORD113, SNORA70, SNORA46, SNORA3, SNORA26, and SNORA20—can maintain cellular homeostasis and enhance stress resilience in murine fibroblasts.179 Meanwhile, SNORD44/49A promote stress granule assembly via UBAP2L/G3BP1 interactions, enhancing cell survival under metabolic stress.180 Emerging evidence indicates snoRNAs modulate lipid metabolism, exemplified by U60 snoRNA’s regulation of cellular cholesterol homeostasis.181 Furthermore, Michel et al identified three snoRNAs (U32a, U33, U35a) encoded within the RPL13a locus that mediate palmitate-induced lipotoxic stress and oxidative damage.142 In GBM, Cui et al delineated an SNORD113-3/ADAR2/PHKA2/EBF1 regulatory axis governing glycolysis and lipid metabolism. SNORD113-3 overexpression attenuated glycolysis, lactate production, glucose consumption, lipid anabolism. Crucially, SNORD113-3 overexpression potently inhibited tumorigenesis.182 Parallelly in CRC, SNHG16 upregulation engages in ceRNA-mediated metabolic reprogramming by derepressing key lipogenic genes—including SCD, PCSK9, SQLE, ACLY, and HSD17B7—to drive lipid metabolic flux.183 But, the metabolic functions of snoRNAs remain underexplored relative to proliferation and metastasis.

Immune Evasion

Tumors evade immune cytotoxic through immunoediting and suppression of immune cells, with emerging roles of snoRNAs in modulating these processes.

Immune Phenotype Regulation

SnoRNAs remodel surface expression of immune checkpoint molecules and modulate secretion of inflammatory factors by tumor cells. In diffuse large B-cell lymphoma, SNHG14 binds miR-5590-3p to upregulate ZEB1, which transcriptionally activates PD-L1 to drive immune escape.184 In NSCLC, the SNHG12/HuR complex elevates PD-L1 and USP8 expression; critically, USP8-mediated deubiquitination stabilizes PD-L1 protein to potentiate immune evasion.185 Concurrently, SNORA38B in NSCLC induces IL-10 secretion, recruiting CD4+FOXP3+ regulatory T cells. These regulatory T cells reduce CD3+CD8+ T cell infiltration within the tumor microenvironment, thereby establishing immunosuppressive niches.151

Immune Cell Regulation

SnoRNAs regulate immune cell functionality. During immunoglobulin class-switch recombination (CSR), aSNORD1C directs fibrillarin-dependent 2′-O-methylation of R-loops on immunoglobulin heavy chain (IgH) loci, ensuring efficient CSR.186 In B16 melanoma, the Dicer-independent sdnRNA-3 polarizes macrophages toward the pro-tumor M2 phenotype by repressing Nos2 transcription via promoter H3K27me3 deposition.114 Furthermore, SNORD63-derived piR30840 binds pre-mRNA introns in CD4+ naïve T cells, promoting its degradation to suppress IL-4 expression and then inhibit Th2 lymphocyte differentiation.187

Senescence

Aberrant oncogene activation can trigger irreversible cell-cycle arrest, propelling cancer cells into oncogene-induced senescence (OIS). Emerging evidence indicates snoRNAs contribute to the sustenance of the OIS phenotype. SNORA24 is enriched in HRASG12V-induced senescent human fibroblasts, a phenomenon conserved in tumor cells from NRASG12V-driven murine HCC models. This snoRNA directs pseudouridylation at two 18S rRNA residues, enhancing translational fidelity by 10–20% to reinforce senescence-associated tumor suppression.188 Complementarily, Cheng et al demonstrated that SNORA13 complexes with RPL23 to impede its incorporation into mature ribosomes, thereby amplifying nucleolar stress responses and activating p53-mediated senescence programs.189 Thus, some snoRNAs act as tumor suppressors by enforcing senescence, a property that could be therapeutically exploited.

Epigenetics Alteration

SnoRNAs orchestrate tumor epigenetics through multifaceted modifications—including 2′-O-methylation, pseudouridylation, alternative splicing, and histone methylation—that reshape nucleic acid and protein landscapes. Illustratively, Liu et al showed that INHEG accelerates SNORD assembly to promote rRNA 2′-O-methylation, thereby enhancing translational efficiency of oncoproteins (EGFR, IGF1R, CDK6, PDGFRB) in glioma stem cells.190 In endometrial carcinoma, SNORA73B pseudouridylates MIB1, thereby stabilizing its transcript and amplifying protein synthesis. This consequently elevates Jagged-1 ubiquitination and hyperactivates the Notch pathway. Concurrently, SNORA73B redirects alternative splicing of RCC1, upregulating RCC1-T2/T3 splice variants to potentiate cellular proliferation, migration, and invasion.191 NSCLC models revealed SNHG20 recruited EZH2 (enhancer of zeste homolog 2) to deposit H3K27me3 at the P21 locus, epigenetically silencing this tumor suppressor.192 Similarly in HCC, SNHG1 cooperates with EZH2 to enforce H3K27me3-mediated transcriptional repression of P21/P15 promoters while competitively binding miR-140-5p to derepress CDK4 expression, dually activating cell-cycle signaling pathway.193 Furthermore, sdnRNA-3 imposes deterministic control over macrophage M2 polarization in melanoma by compacting chromatin architecture at the Nos2 promoter. It achieves this by recruiting Mi-2β chromatin remodelers and catalyzing H3K27me3 deposition to occlude transcriptional accessibility, consequently ablating Nos2 expression.114 This broad impact suggests snoRNAs are integral to the epigenetic machinery.

Phenotypic Plasticity

Tumor phenotypic plasticity refers to the adaptive capacity of cancer cells to remodel their identity through transdifferentiation or dedifferentiation during progression. This plasticity is critically modulated by snoRNAs. In gliomagenesis, the epigenetic regulator TRIM24 upregulates U3 snoRNA to drive anaplastic astrocytoma transformation into epithelioid glioblastoma.194 In acute myeloid leukemia, DNA damage downregulates chromatin-associated orphan snoRNA SNORA73, disrupting its complex with PARP1. This unleashes PARP1 auto-PARylation, which stabilizes the genome but also establishes a negative feedback loop that blocks differentiation and sustains a progenitor-like state.195 By depositing H3K4me3/H3K27ac/H3K9ac via trans-chromatin interactions, SNORD118 and SNORD3A block myeloid differentiation and sustain leukemia progression, illustrating how snoRNAs suppress phenotypic conversion.196

Microbiome-Host Interaction

While microbiota reign over tumor evolution, the interplay between microbial signaling and snoRNAs remains unexplored yet mechanistically enticing. Gut microbiota suppresses SNHG9 expression in small intestinal epithelial cells via myeloid cells and group 3 innate lymphoid cells (ILC3s) signaling cascades. And SNHG9 downregulation stabilizes the SIRT1-CCAR2 complex, which represses PPARγ, reducing lipid absorption.197 Notably, in colitis, this axis protects against inflammation-driven tumorigenesis by enhancing p53 acetylation.198 Complementarily, Zhang et al demonstrated that microbiota-derived valerate exerts radioprotective effects in hematopoietic systems through disrupting the LRPPRC-SNHG15 regulatory axis during γ-irradiation injury.199

Clinical ApplicationsDiagnosis

The presence of tumor-derived snoRNAs in body fluids has spurred interest in liquid biopsy diagnostics. Several panels show encouraging performance. For clear cell renal cell carcinoma, urinary SNORD63, SNORD15A, SNORD35B, SNORD60, and plasma SNORD96A were validated as a composite diagnostic signature.200,201 In HBV-related HCC, a panel combining platelet SNORD12B, SNORA63, and SNORD14E achieved a diagnose AUC of 0.9047.202 For breast cancer, plasma SNORD16, SNORA73B, SCARNA4, and SNORD49B combined with CEA improve AUC to 0.832.203

Prognosis Prediction

Beyond diagnosis, snoRNAs can predict tumoral prognosis. In NSCLC, serum exosomal SNORD116 and SNORA21 decrease significantly (AUC = 0.738 and 0.761, P < 0.0001). Integration with CYFRA21-1 and CEA forms a Composite Clinical Model (CCM), boosting diagnostic AUC to 0.917. The snoRNAs further identify metastatic traits (AUC = 0.743/0.694), which increase to 0.760/0.756 when assessed by CCM.204 Han et al demonstrated that SNORA47-mediated hijacking of the EBF3-RPL11-c-Myc circuit promoted cancer stem cell enrichment while blunting paclitaxel sensitivity, ultimately compromising patient prognosis.205 High plasma SNORD33 in cisplatin-treated metastatic TNBC patients maintains drug sensitivity, extending median progression-free survival from 6.6 to 10.1 months (P = 0.005).206 Notably, a TIIsno score based on seven immune-associated snoRNAs (eg, SNORD59A, SNORD63B) predicts favorable prognosis, enhanced immune infiltration, and better response to immune checkpoint inhibitors in colorectal cancer.207

Treatment

The therapeutic targeting of snoRNAs is an emerging frontier in precision oncology. The most direct strategy involves using ASOs to specifically knock down oncogenic snoRNAs. For instance, ASO-mediated silencing of SNORA13 synergizes with 5-fluorouracil to suppress colorectal cancer growth in xenograft models, while SNORD9 ASO inhibits tumorigenicity in ovarian cancer patient-derived organoids.154,208 Given that snoRNAs frequently mediate treatment resistance through epigenetic regulation, metabolic reprogramming, and tumor microenvironment remodeling, combining snoRNA-directed therapies with conventional modalities holds particular promise. Beyond ASOs, emerging strategies include rRNA modification-targeted small molecules, epigenetic drugs to reactivate tumor-suppressive snoRNAs, and advanced delivery platforms such as exosomes and engineered probiotics for cell-type specific targeting.190

Challenges

Despite growing promise, several challenges impede clinical applications of snoRNA-based therapies. Functional redundancy represents a critical concern. Multiple snoRNAs often target identical modification sites within rRNA or snRNA, and distinct snoRNA families may converge on the same physiological pathways.209 Therefore, strategies targeting a single snoRNA may yield limited therapeutic effects, necessitating combinatorial approaches. Methodological standardization poses another major hurdle. Current qRT‑PCR protocols for snoRNA quantification vary substantially across studies, including differences in primer design, choice of endogenous reference genes (eg, U6 vs. RNU48), and normalization strategies.210,211 This variability complicates cross‑study comparisons and clinical validation. Tumor heterogeneity adds further complexity. SnoRNA expression profiles differ not only between cancer types but also among tumor subclones within the same patient. This demands personalized selection of biomarker panels or therapeutic targets.212,213 Delivery systems for snoRNA‑based therapies remain underdeveloped. While antisense oligonucleotides can effectively knock down oncogenic snoRNAs in vitro, efficient and tumor‑specific in vivo delivery platforms—such as engineered exosomes or lipid nanoparticles—with minimal off‑target effects and low immunogenicity are still needed.214 Finally, the lack of well‑characterized animal models recapitulating snoRNA dysregulation in a tissue‑specific manner limits preclinical validation of candidate targets.215 Overcoming these barriers will require concerted efforts in method harmonization, advanced delivery engineering, and development of physiologically relevant preclinical models.

Conclusion

The functional delineation of small nucleolar RNAs (snoRNAs) in cancer progression has markedly advanced, with their intricate molecular mechanisms garnering significant interest as promising therapeutic targets. This review synthesizes current knowledge on aberrant snoRNA biogenesis in malignancies, regulatory networks driving tumorigenesis, and clinical translation potential. Genetic lesions—including chromosomal rearrangements, copy number alterations, and point mutations—converge with epigenetic dysregulation (histone modification, CpG island hypermethylation) and post-transcriptional failures (aberrant degradation/localization) to disrupt snoRNA homeostasis. Dysregulated snoRNAs coordinate oncogenesis through canonical functions involving C/D-box-directed 2′-O-methylation and H/ACA-box-mediated pseudouridylation of target rRNAs/mRNAs, which activate key pathways like mTOR and MAPK signaling. Their non-canonical actions encompass chromatin compaction, histone modification, and competitive protein complex binding, thereby inhibiting tumor suppressors. These mechanisms allow the cancer to exhibit classical hallmarks, including sustained proliferation with metastatic dissemination, apoptosis evasion via p53 destabilization, and angiogenesis induction through MMP10/VEGF upregulation. Additionally, they drive metabolic adaptation via lipid enzyme reprogramming, immune escape by PD-L1 induction and immunosuppressive microenvironment remodeling, and senescence sustained through p53 pathway activation. Interestingly, gut microbiota emerges as a pivotal regulator of tumor progression, with microbial snoRNA host genes modulating intestinal epithelial p53 acetylation to influence inflammation-driven carcinogenesis. Meanwhile, activation of the microbiota-blood axis protects against radiation-induced hematopoietic damage via the LRPPRC-SNHG15 pathway. Extracellularly, tumor-specific snoRNA signatures in body fluids enable non-invasive cancer detection and prognostic stratification, demonstrating clinical utility for liquid biopsy applications. Therapeutically, ASO-mediated snoRNA suppression reduces tumor burden in vivo, and engineered probiotics generating tumor-suppressive snoRNAs or antisense antagonists may represent novel precision oncology approaches. Consequently, these diminutive snoRNA molecules exert disproportionately substantial impacts on tumor initiation and progression through dynamically interwoven intra- and extracellular networks, urgently calling for concerted efforts to decrypt the fundamental mechanisms underlying snoRNA dysregulation, and map the domino-like oncogenic cascades triggered by aberrant snoRNAs, thereby illuminating novel pathways for clinical diagnostics and targeted therapeutics.

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Funding

This research was funded by the National Key Clinical Specialties Construction Program.

Disclosure

The authors declare no conflicts of interest.

References

1. Weinberg RA, Penman S. Small molecular weight monodisperse nuclear RNA. J Mol Biol. 1968;38(3):289–22. doi:10.1016/0022-2836(68)90387-2

2. Zieve G, Penman S. Small RNA species of the HeLa cell: metabolism and subcellular localization. Cell. 1976;8(1):19–31. doi:10.1016/0092-8674(76)90181-1

3. Lafaille FG, Harschnitz O, Lee YS, et al. Human SNORA31 variations impair cortical neuron-intrinsic immunity to HSV-1 and underlie herpes simplex encephalitis. Nat Med. 2019;25(12):1873–1884. doi:10.1038/s41591-019-0672-3

4. Chauhan W, Setra Janardhana Shetty S, Ferdowsi S, et al. Rpl13a snoRNAs U34 and U35a: new targets for sickle cell disease complications. Circ Res. 2025;137:e40–61.

5. Saeed S, Siegert AM, Tung YCL, et al. Biallelic variants in SREK1 downregulating SNORD115 and SNORD116 cause a Prader-Willi-like syndrome. J Clin Invest. 2025;135(16). doi:10.1172/JCI191008

6. Bogard B, Bonnet H, Boyarchuk E, et al. Small nucleolar RNAs promote the restoration of muscle differentiation defects in cells from myotonic dystrophy type 1. Nucleic Acids Res. 2025;53(6). doi:10.1093/nar/gkaf232

7. Deng Z, Li C, Hu S, et al. sdRNA-D43 derived from small nucleolar RNA snoRD43 improves chondrocyte senescence and osteoarthritis progression by negatively regulating PINK1/Parkin-mediated mitophagy pathway via dual-targeting NRF1 and WIPI2. Cell Commun Signal. 2025;23(1):77. doi:10.1186/s12964-024-01975-2

8. Falaleeva M, Pages A, Matuszek Z, et al. Dual function of C/D box small nucleolar RNAs in rRNA modification and alternative pre-mRNA splicing. Proc Natl Acad Sci U S A. 2016;113(12):E1625–34. doi:10.1073/pnas.1519292113

9. Song J, Dong L, Sun H, et al. CRISPR-free, programmable RNA pseudouridylation to suppress premature termination codons. Mol Cell. 2023;83(1):139–155.e9. doi:10.1016/j.molcel.2022.11.011