Epigenetics refers to the phenomenon that DNA sequences do not change, allowing heritable changes in gene expression, including DNA methylation, histone modifications, and chromatin modifications [20, 21]. LncRNAs could play a key role in epigenetic regulation and promote CCA development by mediating multiple epigenetic modifications [22]. The polycomb repressor complex 2 (PRC2) has histone methyltransferase activity and can silence gene expression by methylating lysine 27 on histone H3 [23, 24]. The enhancer of histidine-lysine N-methyltransferase (EZH2) is the catalytic subunit of the PRC2 complex [24, 25]. It was shown that specific lncRNAs could recruit EZH2 to the promoter region of its target genes and directly repress the expression of target genes by mediating H3K27me3 demethylation modifications, thus promoting the progression of CCA [26, 27, 28, 29, 30]. LncRNA ANRIL was reported to accelerate CCA tumorigenesis by binding to EZH2, recruiting to the promoter region of ERRFI1, the tumor suppressor gene of CCA, and then directly suppressing ERRFI1 expression by mediating H3K27me3 demethylation modifications [26]. LncRNA DANCR promotes CCA cell growth and migration by binding to EZH2 to catalyze the FBP1 promoter region H3K27me3, which can partially inhibit FBP1 expression [27]. Furthermore, in CCA, SNHG1 [28]/AGAP2-AS1 [31], PVT1 [29], and NEAT1 [30] are also able to recruit EZH2 at the promoter regions of CDKN1A, ANGPTL4, and E-cadherin, respectively, resulting in elevated H3K27me3 levels at these loci. In addition to recruiting EZH2 to its target gene promoter, lncRNAs can also recruit SET domain bifurcated 1 (SETDB1) to bind to the survivin gene promoter [32]. Survivin is a bifunctional protein that acts as an inhibitor of apoptosis, and its expression is elevated in most tumors [33]. LncRNA FENDER could recruit SETDB1 to bind to the survivin gene promoter and induce H3K9 methylation to suppress survivin expression, thereby inhibiting proliferation, migration, and invasion of CCA cells [32].

Notably, lncRNAs can also act as scaffolds for at least two different histone modification complexes to mediate epigenetic modifications of genes [34]. Xu et al. [35] confirmed that LncRNA SPRY4-IT1 could act as a scaffold to recruit EZH2, LSD1, and DNMT1 to the tumor suppressor KLF2 and LATS2 promoter regions for epigenetic silencing, respectively. Further studies showed that EZH2 directly binds to the LATS2 and KLF2 promoter regions and mediates H3K27 trimethylation; DNMT1 directly binds to the KLF2 promoter region; LSD1 directly binds to the LATS2 promoter region and induces H3K4 demethylation modification, thus promoting the aggressiveness of CCA cells. These studies confirm that lncRNAs can mediate epigenetic modifications to promote the progression of CCA in multiple ways, as shown in Fig. 1 [26, 27, 28, 29, 30, 31, 32, 35].

Fig. 1.

LncRNAs mediate epigenetic modification in three different ways. (A) In cholangiocarcinoma (CCA) cells, LncRNAs could recruit EZH2 to its target gene promoter region and directly repress target gene expression by mediating H3K27me3 demethylation modifications. (B) In CCA cells, FENDER could recruit SET domain bifurcated 1 (SETDB1) to the survivin promoter region and represses survivin expression by inducing H3K9 methylation. (C) In cholangiocarcinoma (CCA) cells, SPRY4-IT1 could recruit EZH2 and DNMT1 to the KLF2 promoter region and recruit EZH2 and LSD1 to the LATS2 promoter region, respectively, thereby inhibiting transcription of KLF2 and LATS2. Abbreviations: EZH2, enhancer of zeste 2; H3K27, histone H3 lysine 27; ERRFI1, ERBB receptor feedback inhibitor 1; FBP1, Fructose-1, 6-biphosphatase; CDKN1A, cyclin-dependent kinase inhibitor 1; ANGPTL4, angiopoietin-like 4; SETDB1, SET domain bifurcated histone lysine methyltransferase 1; H3K9, histone H3 lysine 9; LSD1, lysine specific demethylase 1; DNMT1, DNA methyltransferase 1; KLF2, Kruppel-like factor 2; LATS2, large tumor suppressor 2.

The competitive endogenous RNA (ceRNA) mechanism was first proposed by Salmena [36], revealing that lncRNAs can act as natural miRNAs sponge competing miRNAs to regulate miRNA-targeted mRNAs and participate in the physiological processes of cell proliferation, differentiation, and apoptosis, which play an essential role in cancer development and progression [37]. Increased expression of lncRNA DLEU1 has been reported in CCA tissues and cells, and the transcription factor YY1 transcriptionally promotes its expression. Yes-associated protein 1 (YAP1) is a transcriptional co-activator that has an essential regulatory role in cell signaling and functions as a pro-oncogene in various cancers, particularly in CCA [38, 39]. DLEU1 promotes oncogene YAP1 expression by acting as a sponge to bind miR-149-5p competitively. In addition, YAP1/TEAD2 promotes stemness maintenance through transcriptional upregulation of SOX2. YY1/DLEU1/miR-149-5p/YAP1 axis plays a critical pro-cancer role in CCA progression [40]. NRP-1 is a transmembrane protein [41] that functions as a co-receptor for multiple cellular signaling pathways associated with the advancement of cancer and interacts with the HGF/c-Met and TGF-/TGF-RI pathways in CCA cells [42, 43]. The lncRNA TTN-AS1 was reported to sponge miR-320a in an Ago2-dependent manner and down-regulate miR-320a expression in CCA cells. Through the downregulation of miR-320a, TTA-AS1 promotes cell cycle progression, Epithelial-mesenchymal transition (EMT), and angiogenesis via NRP-1. However, after transfection of miR-320a mimics into CCA cells, miR-320a could partially counteract the effect of TTN-AS1 overexpression [42]. Wang et al. [44] confirmed that lncRNA H19 and HULC were increased by oxidative stress and regulated CCA cell migration and invasion by elevating the levels of IL-6 and CXCR4 through sponges let-7a/let-7b and miR-372/miR-373, respectively. Yang et al. [45] demonstrated that lncRNA SNHG12 could sponge miR-199a-5p and upregulate Klotho expression in iCCA cells, promoting iCCA cell growth and metastasis and providing a potential marker and therapeutic target for iCCA patients. In addition, specific lncRNAs can also influence epigenetic and signaling pathways through sponge miRNAs, thus promoting the progression of CCA. Xu et al. [35] demonstrated that SPRY4-IT1 acts as a molecular sponge for miR-101-3p and antagonizes its ability to inhibit EZH2 protein translation, leading to increased EZH2 protein levels, which in turn mediates epigenetic promotion of CCA progression. Patients with CCA had considerably higher PCAT6 levels, whereas PCAT6 was highly expressed in macrophages. PCAT6 contributed to the immunological response of CCA by influencing macrophages, including ROS generation, mitochondrial stress response, and M2 polarization, via sponging miR-326 and activating RohA signaling [46]. Furthermore, the oncogenic lncRNA PAICC activates the Hippo pathway by competitively binding miR-141-3p and miR-27A-3p and downregulating YAP1 expression [47]. According to the most recent findings, Table 1 (Ref. [35, 40, 42, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67]) summarizes the ceRNA regulation mechanisms of some lncRNAs in CCA.

Many studies have shown that lncRNAs can act as carcinogens or tumor suppressors and regulate tumor progression by participating in multiple signaling pathways. Phosphatidylinositol 3-kinases (PI3Ks) are often over-activated in tumors and usually accompanied by the inactivation of the tumor suppressor gene PTEN. By promoting the PI3K/AKT signaling pathway, dysregulated LncRNA expression can accelerate the development of CCA. Peroxiredoxin 2 (PRDX2) is an essential protein in cancer cell proliferation, invasion, and apoptosis. It is a member of the peroxiredoxin family [68]. LncRNA CASC15 may promote the development of iCCA by binding to PRDX2 and suppressing its degradation, thereby activating the PI3K/AKT/c-Myc signaling pathway [69]. As mentioned, lncRNAs can act as oncogenes and activate the CCA-related signaling pathway by reducing miRNA levels through the ceRNA mechanism. It was found that miRNA-122 inhibits the phosphorylation of ERK/MAPK protein, thus blocking the ERK/MAPK pathway. LncRNA UCA1 promotes the activation of the ERK/MAPK pathway and the metastasis of CAA cells by sponging miR-122 and regulating the expression of its downstream gene CLIC1 [63]. Similarly, SNHG1, a ceRNA for miR-140, increases TLR4 expression and activates the NF-$\kappa$B signaling pathway, influencing CCA development and carcinogenesis [64]. PCAT1 promotes extrahepatic CCA progression by upregulating Wnt1 levels and activating the Wnt/$\beta$-catenin signaling pathway through sponge miRNA-122 [70]. Besides, in CCA, lncRNAs can regulate multiple signaling pathways in addition to one signaling pathway alone, which enhances the proliferation, migration, and invasion of CCA cells in a synergistic relationship. Recent studies have shown that NNT-AS1 overexpression in CCA tissues and cell lines increases CCLP1 and TFK1 cell proliferation and EMT through modulating the activation of PI3K/AKT and ERK1/2 signaling pathways via miR-203 downregulation [71]. Interestingly, this study showed that PI3K/AKT and ERK1/2 signaling pathways were co-regulated by NNT-AS1 and miR-203 and synergistically promoted the proliferation and EMT of CCA cells. In addition, we summarize the mechanisms by which dysregulated lncRNAs regulate cancer-related signaling pathways, as shown in Fig. 2 [47, 63, 64, 65, 66, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78].

Fig. 2.

Mechanism of dysregulated lncRNAs regulating cancer-related signaling pathways, including PI3K/AKT, ERK/MAPK, Wnt/$\beta$-catenin, NF-$\kappa$B, Hedgehog, TGF$\beta$/Smad, and Hippo pathways. From Fig. 2, it can be concluded that most dysregulated lncRNAs promote the proliferation, migration, and invasion of cholangiocarcinoma, except for MEG3 and MIR22HG, which inhibit the progression of CCA. LncRNAs that promote or inhibit CCA progression are shown in blue or green, respectively. An arrow represents activation, a T-shaped arrow represents inhibition, and a V-shaped arrow represents the lncRNA sponge microRNA (miRNA). Abbreviations: PRDX2, peroxiredoxin 2; PTP4A1, phosphatase of regenerating liver 1; CCND1, CyclinD1; CLIC1, chloride intracellular channel 1; TRAF3, TNF receptor-associated factor 3; TLR4, Toll-like receptor 4; MNX1, motor neuron and pancreas homeobox protein 1; IRS, insulin receptor substrates; PI3K, phosphatidylinositol 3-kinase; PTEN, phosphatase and tensin homolog; Akt, protein kinase B; GSK3β, glycogen synthase kinase 3 beta; MEK1/2, MAP kinase/ERK kinase 1/2; ERK1/2, extracellular regulated MAP kinase 1/2; Dvl, segment polarity protein disheveled homolog; LRP, LDL-receptor-related protein; APC, adenomatous polyposis coli; Axin, axis inhibition protein; TGF-β, transforming growth factor-beta; Smo, smoothened, frizzled class receptor; Sufu, Sufu negative regulator of hedgehog signaling; Gli1, gli family zinc finger 1; TNF-α, Tumour Necrosis factor-Alpha; IKK, inhibitor of kappa B kinase beta; NF-$\kappa$B, nuclear factor kappa B; SMAD2/4, drosophila mothers against decapentaplegic protein 2/4; Mst1/2, the STE20-like serine/threonine kinases Mst1 and Mst2; Lats1, large tumor suppressor 1; Mob, mps one binder.

Like lncRNAs, circRNAs could directly sponge miRNA and play an essential role in the progression of CCA by competing with mRNA as ceRNA to bind miRNA and regulate its activity. CD73 is a new immunoinhibitory protein that promotes the growth of tumors by inhibiting anti-tumor immune responses [87, 105]. GAL-8 induces the death of activated T cells and promotes the differentiation of immunosuppressive Tregs. It also indirectly affects tumor cells by stimulating the oncogenic-like transformation of epithelial cells via partial and reversible EMT [87, 106]. Xu et al. [87] found that circHMGCS1-016 could sponge miR-1236-3p, down-regulate miR-1236-3p, and upregulate CD73 and GAL-8 expression in iCCA cells. Horizontal upregulation of CD73 and GAL-8 resulted in a suppressed immune environment and, in turn, induced immune escape of iCCA cells, promoting iCCA progression. Interestingly, the study also confirmed that circHMGCS1-016 enhances iCCA resistance to anti-PD1 therapy. Circ-LAMP1 binds directly to miR-556-5p and miR-567, which could regulate YY1 expression. Furthermore, transfection of miR-556-5p and miR-567 mimicked into CCA cells decreased YY1 expression and inhibited the oncogenic effect of circ-LAMP1 [91]. Xu et al. [102] demonstrated that circ-CCAC1 promotes CCA tumorigenesis and metastasis through spongy miR-514a-5p upregulating YY1 expression and further activating GAMLG expression. In a separate study, the transcription factor YY1 binds to the promoter of circ-ZNF609, thereby enhancing its transcription. circ-ZNF609 promotes CCA cell proliferation, migration, and invasion by upregulating LRRC1 through sponge miR-432-5p [90]. Interestingly, YY1 plays a crucial role in circ-LAMP1, circ-CCAC1, and circ-ZNF609 in promoting CCA progression, and whether these three circRNAs have a synergistic role in the mechanism of CCA occurrence and development remains to be further explored. In addition, circRNAs can also act as ceRNAs for miRNAs to inhibit the progression of CCA. It was shown that CircSETD3 competitively binds miR-421 and negatively regulates miR-421 expression, thereby inhibiting the proliferation of CCA cells and inducing apoptosis, but miR-421 mimics can reverse these effects [88].

Dysregulated circRNAs can be closely associated with the progression of CCA by participating in multiple cancer-related signaling pathways. Recent studies have shown that a limited number of translatable circRNAs can encode proteins or functional peptides that activate cancer-related signaling pathways, which regulate cancer progression [97, 107, 108]. cGGNBP2-184aa, a protein encoded by the GGNBP2 induced by IL-6, was reported to promote iCCA cell proliferation and metastasis in vitro and in vivo. Mechanistically, cGGNBP2-184aa interacts directly with the DBD of STAT3 and phosphorylates STAT3 at the Tyr705 site, which is then translocated into the nucleus to activate the transcription of target genes. Thus, IL-6/cGGNBP2-184aa/STAT3 forms a positive feedback loop to maintain constitutive activation of IL-6/STAT3 signaling to promote iCCA progression [97]. In addition, certain circRNAs can act as miRNA sponges and protein scaffolds [109] to localize transcription factors in the nucleus and thus activate cancer-related pathways. CircACTN4 has been reported to sponge miR-424-5p in the cytoplasm, activating the Hippo pathway by upregulating the level of the oncogenic transcriptional cofactor YAP1. In the nucleus, the YBX1 transcription factor is recruited to induce transcription of FZD7, a positive regulator of the Wnt/-catenin signaling pathway that promotes the growth and metastasis of iCCA cells. Importantly, circACTN4 acts as a signaling nexus, enhancing the interaction between YAP1 and $\beta$-catenin, which also confirms the synergistic effect of the Hippo pathway and Wnt/$\beta$-catenin pathway in the growth and metastasis of iCCA cells [84]. Furthermore, there are unanswered questions from this study. Whether circACTN4 acts as a scaffold to recruit YBP1 into the FZD7 promoter by interacting with DNA or other cofactors remains to be determined [110]. The mechanism of circACTN4-mediated interaction between YAP1 and $\beta$-catenin still needs further exploration [84]. Interestingly, circRNAs not only activate cancer signaling pathways but also inhibit CCA progression by suppressing cancer signaling pathways. cNFIB was shown to bind directly to the NTD region of MEK1, blocking the binding of the kinase (MEK1) to the substrate (ERK2) and preventing the phosphorylation of ERK2, thereby inhibiting iCCA proliferation and metastasis by inhibiting MEK1/ERK signaling. Notably, further research is required to determine which binding site on cNFIB mediates the interaction between cNFIB and MEK1 [96]. For the mechanism of circRNA regulation of cancer-related signaling pathways, see Fig. 3 [84, 96, 97, 98].

Fig. 3.

CircRNAs regulate cancer-related signaling pathways, including AKT3/mTOR, Hippo, Wnt/$\beta$-catenin, ERK/MAPK, and JAK/STAT3 signaling pathways. CircRNAs that promote or inhibit CCA progression are shown in blue or green, respectively. An arrow denotes activation, a T-shaped arrow denotes inhibition, and a V-shaped arrow denotes circRNA sponge miRNA. Abbreviations: PI3K, phosphatidylinositol 3-kinase; AKT3, AKT Serine/Threonine Kinase 3; mTOR, mammalian target of rapamycin; YAP1, Yes-associated protein 1; YBX1, Y-box binding protein 1; FZD7, Frizzled-7; MEK1, mitogen-activated protein kinase 1; ERK2, extracellular regulated MAP kinase 2; IL6, interleukin- 6; STAT3, signal transducers and activators of transduction-3.

Extracellular vesicles (EVs) are vesicle-like vesicles with a bilayer membrane structure, ranging from 40 nm to 1000 nm in diameter, that are detached from the cell membrane or secreted by the cell into the extracellular matrix [111, 112]. EVs, which are mainly composed of microvesicles (MVs) and exosomes [113], are abundantly found in various body fluids and cell supernatants [112] and transport signaling molecules such as non-coding RNA, DNA fragments and proteins in a stable manner [114]. They could play an essential role in tumor development as transmitters of transduction signals between cancer cells and as crucial mediators of intercellular communication by shifting their contents and altering biological responses in other cells [115, 116]. However, few studies have focused on the mechanism of action of EVs and EV-circRNAs in CCA. Recent studies have shown that CCA-derived EVs can act as cancer migration and invasion mediators by transferring oncogenic circRNAs to normal bile duct cells [101]. It has been reported that circ-0000284 promotes the development and progression of CCA by competitively binding to miR-637 and upregulating the expression of LY6E. This study also found that the expression of circ-0000284 in exosomes was approximately three times higher than that in producing cells. Importantly, circ-0000284 may transfer directly from CCA cells to surrounding normal cells via exosomes, evoke the malignant phenotype of CCA cells through cellular communication, and regulate the biological functions of peripheral cells [101]. Notably, this is the first EVs-mediated circRNA identified in CCA, suggesting that EV-circRNAs can act as novel players in regulating the progression of CCA. Furthermore, exosomal circRNAs can also play an essential role in cellular communication as a significant mediator. A recent study showed that circ_0020256 was delivered to CCA tumor cells via exosomes secreted by tumor-associated macrophages (TAMs) to enhance the biological activity of CCA cells by regulating the expression of transcription factor E2F3 through interaction with miR-432-5p [100]. Several studies have shown that circRNAs can be transported by tumor-secreted EVs to reshape the tumor microenvironment by inducing immunosuppression and angiogenesis and providing a supportive microenvironment for metastatic cancer cells [102, 117, 118]. Xu et al. [102] confirmed that circ-CCAC1 was transported from CCA cell-derived EVs to endothelial monolayer cells, enhancing endothelial monolayer permeability and causing angiogenesis by breaking the vascular endothelial barrier, which in turn promoted CCA progression. Significantly, circ-CCAC1 in the EVs of this study was highly expressed in bile samples from CCA patients. The abundance and ease of detection of bile-derived EV-circRNA compared to cellular free nucleic acids detected by conventional liquid biopsies, which can be repeatedly sampled for large-scale testing, opens up new perspectives on EV-circRNAs as a noninvasive biomarker for CCA.

The ceRNA hypothesis is a currently well-studied mode of gene expression regulation. Currently, some circRNAs have been shown to act as sponges for miRNAs and regulate the ability of miRNAs to target mRNAs by competitively binding to miRNAs [36]. However, the abundance of circRNAs is low in most mammals. In order to function as sponge miRNAs, circRNAs need to be highly expressed in the cytoplasmic matrix or EVs and contain many miRNA binding sites, so most circRNAs cannot act as miRNA sponges [119]. For example, although the level of cNFIB is upregulated in both iCCA tissues and cell lines, there are no binding sites and interactions detected between cNFIB and miRNAs, so it cannot act as a miRNA sponge. Therefore, the adaptation of the ceRNA hypothesis for circRNA is becoming increasingly controversial. Furthermore, a study demonstrated that brain circRNA sponge miRNA did not exhibit a more robust ability than linear mRNA. Instead, it was able to interact with RNA binding protein (RBP) [120]. Interestingly, some researchers have proposed that some circRNAs bind, sequester, or store molecules such as transcription factors into specific subcellular locations by “sponging” RBP, which act as dynamic scaffolds for assembling other components [121]. For example, circ-Foxo3 can prevent CDK2 function and block cell cycle progression by interacting with CDK2 and p21 to form the circ-Foxo3-p21-CDK2 ternary complex [122]. Xu et al. [102] demonstrated that circ-CCAC1 enters endothelial cells via EVs, binds strongly to EZH2, and sequesters it in the cytoplasm without affecting its overall expression. Because circ-CCAC1 restricts the nuclear localization of EZH2, it increases the expression of SH3GL2 to inhibit the expression of intercellular linker proteins ZO-1 and Occludin, ultimately increasing cell permeability. Thus, after circRNA binding to RBP, circRNAs could dissociate protein interactions or alter intracellular protein distribution, which in turn regulates the progression of CCA.

Interestingly, Du et al. [96] proposed an effective circRNA sponge miRNA-like mode of action in a study exploring the inhibition of iCCA proliferation and metastasis. Mechanistically, cNFIB could bind directly to the NTD region of MEK1 and disrupt the interaction between MEK1 and ERK2 by binding competitively to MEK1, thereby downregulating ERK phosphorylation to inhibit iCCA translocation. In addition, this study knocked down ERK2, performed cNFIB pull-down, and discovered that cNFIB enriched MEK1 more [96]. Notably, does this mode of action of circRNA in iCCA have similarities to circRNA sponge-like miRNAs?

As the functions of circRNAs continue to be explored, circRNAs have been demonstrated to act as protein recruiters and as scaffolds to regulate protein-protein interactions. It has been reported that circMRPS35 could serve as a recruiter to recruit histone acetyltransferase KAT7 to the promoters of the FOXO1 and FOXO3a genes, thereby altering the expression of the downstream genes p21, p27, Twist1, and E-calmodulin and inhibiting the proliferation and invasion of gastric cancer cells [123]. In another study, circACTN4 was confirmed to initiate FZD7 transcription by recruiting the transcription factor YBX1 to the FZD7 promoter region in the nucleus, ultimately promoting iCCA cells proliferation and invasion [84]. These studies illustrate that circRNAs recruit them to chromatin for chromatin remodeling or regulate the transcription of target genes by acting as protein recruiters. Through this mechanism, circRNAs are strongly associated with the occurrence and development of CCA. However, there are relatively few studies on this aspect, and it is hoped that shortly, this mode of action of circRNAs as recruiters to recruit functional proteins may provide new ideas for the targeted treatment of cholangiocarcinoma.

CircRNA has long been considered a class of endogenous non-coding RNA. However, as research continues, this concept seems increasingly skeptical. Legnini et al. [124] and Pamudurti et al. [125] found that circ-ZNF609 and circMbl3 could translate proteins in mouse myoblasts and fly heads, respectively. Importantly, circ ZNF609 contains an infinite open reading frame (ORF) that spans the start codon and is translated into the protein in a splicing-dependent and cap-independent way. It has been shown that circ-SMO encodes the carcinogenic protein SMO-193a.a., which de-repressed SMO from PTCH1 during Shh stimulation via enhancing SMO cholesterol modification. Moreover, Shh/Gli1/FUS/SMO-193a.a. establishes a positive feedback loop to maintain Hedgehog signaling activation in glioblastoma [126]. Indeed, there is growing evidence that some circRNAs can be involved in carcinogenesis through internal ribosome entry sites, N6-methyladenosine modification, or rolling circle amplification for non-cap-dependent translation of circRNAs encoding proteins. In another study, Li et al. [97] found a highly conserved ORF that can encode a 184-amino-acid protein when investigating the protein-coding potential of cGGNBP2. This study further identifies a key role for IRES in recruiting ribosomes to initiate the translation of circRNAs lacking the 5${}^{\prime }$ cap. Notably, a predicted protein consisting of 184 amino acids (cGGNBP2-184aa) can be translated by cGGNBP2 driven by this IRES. Furthermore, the cGGNBP2-edited protein cGGNBP2-184aa was shown to promote iCCA growth and metastasis by activating the JAK-STAT signaling pathway. With the deepening of research, the discovery of the hidden functions of CircRNAs encoding proteins or functional peptides has aroused the curiosity of many researchers. As a relatively new research field, it not only enriches the connotation of translationomics and proteomics but also provides a new perspective for researchers to study the mechanism of circRNAs in CCA.

A growing number of studies have discovered that most lncRNAs are significantly expressed in the tissues, cells, and bile of CCA and have tremendous diagnostic and prognostic potential as CCA biomarkers. Ge et al. [127] found that two lncRNAs, ENST00000588480.1, and ENST00000517758.1, were highly expressed in bile-derived exosomes of CCA patients were highly expressed in CCA patients. When these two lncRNAs were combined for diagnosis, their area under the curve (AUC), sensitivity, and specificity were 0.709, 82.9%, and 58.9%, respectively. Their sensitivity was superior to serum CA19-9 (82.9% vs. 74.3%). It was confirmed that the higher the expression of these two lncRNAs in CCA patients, the worse their survival and can be used as predictors for monitoring CCA [127]. It was reported that lncRNA-NEF was downregulated in the plasma of iCCA patients and had good diagnostic properties, significantly distinguishing iCCA patients from healthy controls (AUC = 0.8642). However, unfortunately, this study did not mention the sensitivity and specificity of lncRNA-NEF as a diagnostic marker for iCCA. In addition, overall survival (OS) was significantly lower in iCCA patients with low lncRNA-NEF expression (p = 0.0198) [128]. DLEU1 was shown to be associated with advanced Tumor Node Metastasis (TNM) stage and lymph node infiltration, and CCA patients with high DLEU1 expression had poorer OS. More importantly, DLEU1 can be used as a prognostic marker for CCA (AUC = 0.747, specificity = 65.4%, sensitivity = 72.4%) to predict the prognosis of CCA [40]. In the clinical correlation study of circRNAs and CCA, the expression of circRNA Cdr1as was markedly elevated in tumor tissues relative to neighboring normal tissues and was highly related to lymph node infiltration, progressed TNM, and postoperative recurrence. In addition, Cdr1as could be used as a novel independent prognostic biomarker to predict overall survival in CCA patients (sensitivity = 83.3%, specificity = 58.3%) [129]. Notably, researchers found that circRNAs were highly expressed in bile and serum-derived EVs from CCA patients, implying that EV-cirRNAs could serve as potential non-invasive biomarkers for CCA. For example, Xu et al. [102] found upregulated levels of circ-CCAC1 in bile and serum-derived EVs of CCA patients, both of which have good diagnostic properties. The diagnostic value of serum EVs (AUC = 0.759) was almost equal to that of serum CA19-9 (AUC = 0.757), but bile EVs (0.857) was superior to CA19-9. Interestingly, the combination of both bile- or serum-derived EV-circ-CCAC1 and serum CA19-9 had better diagnostic performance than alone. Furthermore, this study confirmed that high expression (p = 0.001) of circ-CCAC1 was an independent prognostic marker for iCCA and that circ-CCAC1 expression was a predictor of postoperative recurrence of iCCA (p = 0.002). In Table 3 (Ref. [40, 61, 62, 96, 97, 102, 128, 129, 130, 131, 132, 133]), we summarize some relevant studies of ncRNAs as markers of CCA.

In non-surgical biliary malignancies, gemcitabine in combination with cisplatin is considered the standard chemotherapy regimen. However, they have poor chemoresistance and limited efficiency, with a median survival of only 11.7 months [5]. Therefore, finding more effective drug treatment options and improving the sensitivity of CCA to chemotherapeutic agents is crucial to improve the survival rate of CCA patients. As a chromatin regulator, BRCA-1 associated protein-1 (BAP1) has been reported to regulate the expression of lncRNA NEAT-1 and, thus, the sensitivity of gemcitabine in CCA cells through epigenetic regulation. Interestingly, by understanding the expression of BAP1, the researchers found that both olaparib and GSK126 have synergistic effects with gemcitabine in CCA cells and could be combined to enhance sensitivity to gemcitabine. However, this synergistic combination needs to be validated by extensive clinical studies before it can be applied to the clinic [134]. Lu et al. [135] reported that LINC00665 was substantially expressed in gemcitabine-resistant CCA cell lines and was linked to a poor prognosis in CCA patients. When the expression level of LINC00665 was downregulated in gemcitabine-resistant CCA cell lines, the drug resistance effect of CCA cells was reduced; conversely, LINC00665 overexpression increased gemcitabine resistance in CCA-sensitive cells, suggesting the ability of LINC00665 to increase the chemoresistance of CCA cells. Mechanistically, LINC00665 regulates BCL9L expression by sponging miR-424-5p, which activates Wnt/$\beta$-Catenin signaling and ultimately promotes Wnt/$\beta$-Catenin signaling-mediated EMT and stemness properties [135]. In addition, HOTTIP could enhance the resistance of gemcitabine and cisplatin in CCA by sponging miR-637 [136]. The expression level of Circ-SMARCA5 was reported to be downregulated in tumor tissues, and its upregulation increased the sensitivity of cisplatin and gemcitabine in iCCA cells [104]. Upregulation of cNFIB or trametinib (MEK inhibitor) treatment alone has been reported to inhibit tumor progression. The combination of cNFIB overexpression and trametinib enhanced this inhibition by in vivo experiments [96]. This suggests that synergistic effects between cNFIB and trametinib may enhance pit tumor efficacy and that iCCA cells with high levels of cNFIB may potentially enhance the sensitivity of trametinib, providing a new idea for the treatment of iCCA.