Non-coding RNAs (ncRNAs) is well-known as a class of RNA molecules that do not encode proteins, including microRNAs (miRNAs), long non-coding RNAs (lncRNAs), Circular RNAs (circRNAs), Piwi-interacting RNAs (piRNAs), ribosomal RNAs (rRNAs), and transfer RNAs (tRNAs) with known biological functions, as well as those with unknown functions [1]. And, ncRNAs have been widely reported to lack clear potential to encode proteins or functional peptides, but regulate the expression of other genes [2, 3]. Notably, ncRNAs are closely involved in many human diseases [4,5,6], especially cancer, because they act their essential roles in several important biological processes, including gene expression, epigenetic regulation, and autophagy [7,8,9,10].

Hitherto, it has been generally accepted that proteins are the targets of most drugs in human diseases [11,12,13]. After all, protein is the material basis of life, and almost all life activities are closely related to proteins. The strategy of targeting pathogenic proteins has brought many new drugs in the past decade [14, 15]. However, starting from the blueprint of the human genome, with the in-depth understanding of genomics, proteomics and other biological molecules, the limitations of targeting proteins have begun to be realized [16]. First, only 2% of human DNA encodes proteins. In addition, 10–15% of the approximately 20,000 expressed human proteins are considered to be disease-related, that is, 0.2% of the genome codes for disease-related proteins [17, 18]. Secondly, targeting proteins with small molecules requires the identification of specific binding sites on the spatial structure of proteins, which is challenging because most proteins are “undruggable” [19].

Of note, ncRNAs may account for 70% of the whole human genome [20], making the modulation of ncRNAs as a therapeutic strategy more attractive. Moreover, because ncRNAs are located upstream of proteins, they may be able to target such undruggable proteins at an early stage. For example, ribocil binds to the ncRNA structural unit of riboflavin, FMN riboswitch, and inhibits it cellular production and metabolites [21]. Designing structure-specific ligands is reported to target a common structure in the Dicer processing sites of miRNAs cluster and thereby inhibiting their biogenesis [22]. Using high-throughput screening, the small-molecule compound AC1NOD4Q (ADQ) has been identified as a leading candidate compound for targeting lncRNAs, and ultimately inhibiting cancer cell metastasis via the Wnt/β-Catenin signaling pathway [23]. In addition, some candidate small-molecule drugs targeting miRNAs have entered clinical trials for cancer treatment [24, 25].

With the emergence of ncRNAs and innovations in high-throughput screening technology, the mechanisms of ncRNAs in cancer is being further explored. Currently, the number of small molecules for regulating ncRNAs have been increasing. Therefore, we attempt to elucidate the reasons why ncRNAs are modulated by small-molecule compounds, and highlights the designing strategies for discovery of small-molecule compounds for modulating ncRNAs in cancer, which may provide a new clue on exploiting potential small-molecule drugs.

Non-coding RNAs as emerging druggable targets in cancer

Non-coding RNAs (ncRNAs) can form different secondary structures through the typical Watson–Crick base pairing principle and atypical bond-mediated interactions, including helices, hairpins, loops, bumps, and pseudoknots. The further interaction of these structural elements to form more complex three-dimensional structures is expected to lead to enough drug conformations for small-molecule binding and recognition. For example, Linezolid and Ribocil are excellent small-molecule compounds, which are in line with the classic drug Lipinski’s rule of five, have small polar total surface area, and have good cell membrane permeability and no apparent toxicity. Crucially, they perform their pharmacological effects by binding to structural “pockets” of RNA targets, unlike most small-molecule drugs that target disease-causing proteins. Due to the simple composition of ncRNAs, compared with the complex spatial structure of proteins, small-molecule compounds can easily enter the key sites that affect their functions [19]. Additionally, the diversity of ncRNA folding space makes up the disadvantage of having only four nucleotides, and also provides an opportunity for small molecular compounds to target ncRNAs [26]. ncRNAs are involved in intracellular protein assembly, and diseases with “no cure” may be improved at the RNA level before protein production [27, 28].

The therapeutic use of ncRNAs has been drawn a rising attention since they were first discovered, especially in the last decade, ncRNAs such as miRNA and lncRNA have become the focus of research [29,30,31,32]. ncRNAs, which were initially ignored as transcriptional noise or artefactual transcripts, have their complex biological functions. They play the key roles in carcinogenesis or tumor inhibition, and are also regarded as crucial biomarkers for cancer diagnosis and prognosis. For example, overexpressed miR-122 has been reported to reduce Hes1 and NOTCH1 in A549 stem cells, and by blocking the NOTCH1 pathway, miR-122 can reduce the resistance of A549 stem cells to gefitinib, and thereby suppressing cancer cell proliferation and migration [33]. Moreover, miR-34a has been shown to have an anti-tumor activity in breast cancer, which blocks the NOTCH1 pathway by functionally targeting Notch1, thereby regulating cell proliferation, migration, and invasion [34].

Of note, lncRNAs have a transcript length of more than 200 nucleotides [35], which participated in a variety of cellular processes, including chromatin structure regulation, transcriptional regulation, reinforcement or suppression of protein activity, regulation of nuclear bodies and play important roles in cancer, including epigenetic regulation, interactions with miRNAs, cell cycle regulation, and mediating hormone-induced cancer [36, 37]. DANCR has been found to the 3 'untranslated region of CTNNB1 mRNA and block miRNA-mediated CTNNB1 translation inhibition by competitive binding to miR-214 and other sites, resulting in increased CTNNB1 protein expression and activation of the Wnt signaling pathway [38]. Myocardial infarction-associated transcript (MIAT), another lncRNA, has been shown to promote gastric cancer growth and metastasis by regulating the miR-141/DEAD-box RNA helicase 5 (DDX 5) pathway [39]. Further, The HOX transcript antisense intergenic RNA (HOTAIR), a well-known lncRNA, was shown to HOTAIR deletion can inhibit the proliferation, migration, invasion and PI3K/ATK signaling pathway of gastric cancer cells [40].

CircRNAs, which are a class of ncRNA molecules, are characterized by the absence of a 5' terminal cap and a 3' terminal poly (A) tail which form a ring structure with covalent bonds [41]. Compared with miRNAs and lncRNAs mentioned above, there are fewer studies on circRNAs, but this does not affect the key role of its in cancer. For example, circCDK13 has been reported to be a novel circRNA that inhibits the progression of hepatocellular carcinoma, which are possibly mediated via the PI3K/AKT and JAK/STAT pathways [42]. Hsa_circ_0000515 is a novel circRNAs participating in breast cancer by regulating the miRNA-296-5p/CXCL10 axis [43]. Circ YAP1 can act as a tumor suppressor in gastric cancer cells by targeting miR-367-5p/p27Kip1 axis [44]. Together, these findings suggest that circular RNA is a potential target for future diagnosis and treatment [45] (Fig. 1).