The biological function of DCLK1 in tumorigenesis and metastasis as a marker of tuft cells and CSCs is most thoroughly studied in CRC. In normal human intestinal tissue, stem cells are located at the base of the intestinal crypt epithelium, where they are marked by leucine-rich repeat-containing G protein-coupled receptor 5 (LGR5) without co-expressing gut endocrine markers chromogranin A and somatostatin (Fig. 3) (27). DCLK1 marks fully differentiated intestinal tuft cells located in the crypt and villus (28). Long-lived DCLK1+ tuft cells with characteristic microvilli feature self-renewal ability and potential quiescent stem-like functionality (29). Knockdown of the Wnt regulator APC does not alter this quiescence, but subsequent activation through inflammation induced by dextran sodium sulfate is sufficient to initiate colon cancer in Dclk1-Cre/Apc^flox/flox (Dclk1-Cre/Apc^flox/flox transgenic mice featuring knocking out APC gene in DCLK1⁺ cells) transgenic mice (30). These findings are supported in DCLK1-knockout mice where deficiency results in increased epithelial barrier permeability, higher levels of pro-inflammatory cytokines and chemokines, decreased levels of LGR5 and dysregulated Wnt/β-Catenin pathway genes in Villin-Cre/Dclk1^flox/flox (Villin-Cre/Dclk1^flox/flox mice featuring deletion of DCLK1 expression in villin-positive cells) mice (31). In 2009, Gerbe et al (32) provided conclusive evidence that DCLK1 was in fact a tuft cell rather than stem cell marker, as indicated by the position and marker co-expression (cyclooxygenase enzymes 1/2, advillin and tubulin) of DCLK1-expressing cells. Lineage tracing studies demonstrated that DCLK1-positive cells also express colorectal CSC markers, such as CD133 and CD44 (3,4,33). Of note, DCLK1-positive normal intestinal epithelial cells isolated by fluorescence-assisted cell sorting form spheroids that may assemble into glandular epithelial structures and express multiple markers of gut epithelial lineages when implanted subcutaneously in athymic nude mice (27). Self-renewal and differentiation characteristics of DCLK1-positive cells and low expression in normal tissue both led researchers to speculate that they may mark a type of stem cell (34,35), but these findings are no longer supported in the literature and previous results are likely artifacts resulting from the existence of rare DCLK1/LGR5 double-positive cells.

Certain studies reported that DCLK1 expression is significantly higher in CRC tissue and adenomas compared to normal tissue. In addition, increased levels are seen in distant metastases and it is closely associated with CRC recurrence (36,37). Key evidence demonstrated that DCLK1 specifically marked CSCs in the intestine, which continuously produce tumor progeny to prompt polyp growth, but there is no apparent effect on normal tissue after deletion of these cells (38). Furthermore, overexpression of DCLK1 was observed to enhance the percentage of stem-like human CRC cells in vitro (39,40). It has become clear that DCLK1, while not a bona fide normal stem cell marker, is instead a key marker of differentiated intestinal epithelial tuft cells (6,32), which, in the context of mutation and tumorigenesis, may identify specific CRC stem cells.

A series of reports evidence that DCLK1 regulates EMT, proliferation and CSC maintenance through the Notch, Ras and Wnt pathways via interaction with different microRNAs (miRNAs/miRs). MiR-1291 was observed to directly bind to the 3′-untranslated region sequence of DCLK1 and then inhibited the stemness and cell cycle through the cyclin-dependent kinase inhibitors p21^WAF1/CIP1 and p27^KIP1 (41). The levels of miR-137 and miR-15a were inversely correlated with high levels of DCLK1 detected in CRC with larger tumor size, poor differentiation and lymph node involvement (42). Knockdown of DCLK1 expression led to downregulation of miR-200a, miR-144 and miR-let7a along with downregulation of EMT-associated transcription factors [zinc finger E-box binding homeobox 1 (ZEB1), ZEB2, Snail, Slug and Twist], c-Myc, KRAS and Notch-1 in human colon cancer cells (43). In a non-tumorigenic context, a recent study indicated that miR-195 is able to directly interact with DCLK1 mRNA, resulting in suppressed function for tuft and paneth cells in the small intestinal epithelium by inhibiting DCLK1 translation (44).

In CRC cell lines, Notch pathway-regulated markers of CRC CSCs [DCLK1, LGR5, aldehyde dehydrogenase 1 family member A1 (ALDH1) and CD44] by JAK2, STAT3 and ERK1/2 phosphorylation and increased expression of Jagged 1 to promote stemness (45). On the basis of the above findings, researchers have proposed potential therapeutics to suppress proliferation, colony formation and reduce the number of DCLK1+ cells via the Notch pathway (46,47). The Wnt pathway promotes elongator acetyltransferase complex subunit 3 expression and SOX9 translation, which in turn support LGR5(+)/DCLK1(+) intestinal cancer stem cells in response to intestinal regeneration after radiation-induced injury (28). Furthermore, β-catenin nuclear translocation is increased by overexpression of RNA binding motif protein 3, which induces stemness in DCLK1(+)/LGR5(+)/CD44(+) CRC cells (48). Wnt/β-catenin pathway and pluripotency transcription factors c-Myc, KLF transcription factor 4, OCT4 and SOX2 are activated by commensal-polarized macrophages through a microbiome-induced bystander effect in E. faecalis-colonized IL10 knockout mice, leading to increases in the number of DCLK1(+)/CD44(+) cells through gene mutation, chromosomal instability and endogenous transformation to promote tumorigenicity (35). Basic research indicated that compounds, such as γ-mangostin present in the mangosteen (Garcinia mangostana) fruit, WNT5A agonist FOXY5 and niclosamide, are able to regulate chemotherapy resistance and cancer stemness by decreasing the number of DCLK1-positive cells (21,49,50). KRAS mutation was observed to upregulate DCLK1 protein levels, which was reversed by inhibiting KRAS expression (51).

DCLK1 marks a small subpopulation of morphologically and functionally distinct pancreatic cancer cells, which promote tumorigenesis in multiple mouse models (52,53). In normal adult pancreas, DCLK1 is expressed in ductal epithelial cells and islet cells (54) and it is upregulated in murine and human pancreatic intraepithelial neoplasia (55). It is co-expressed with neurogenin-3 and somatostatin, and pancreatic stem cell markers, but not with insulin and glucagon, which mark pancreatic α cells (24). DCLK1-positive cells isolated by flow cytometry injected into nude mice give rise to nodules with a hyperplastic appearance (56). Acetylated tubulin (AcTub), a marker of differentiation of specific pancreatic intraepithelial neoplasia, is frequently co-expressed with DCLK1 and regulates epithelial-mesenchymal transformation of pancreatic cancer cells. AcTub and DCLK1-marked cells demonstrate a typical tuft cell morphology with prominent microvilli at the apical surface of the cell and lead to increased size and number of spheroids in cancer self-renewal assays (53,57). Furthermore, these cells express high levels of ABL proto-oncogene 1, non-receptor tyrosine kinase and insulin-like growth factor 1 receptor, which are drug targets in clinical cancer therapy (58,59). DCLK1 was also reported to be a marker of a population of pancreatic cancer-initiating cells with morphological and molecular features of gastrointestinal tuft cells (53), which drive pancreatic tumor growth by immune cell-derived IL17, which in turn regulates POU class 2 homeobox 3, ALDH1A1 and IL17 receptor C (60). In the pancreas, DCLK1 marks pancreatic tuft and acinar, but rarely islet cells. DCLK1+ tuft cells expand in response to chronic injury or chronic inflammation, and DCLK1+ epithelial cells are a source of acinar-ductal metaplasia after Kras-G12D mutation. These findings indicated that DCLK1+ pancreatic cells may act as pancreatic intraepithelial neoplasia stem cells, but whether or not these arise from pancreatic DCLK1+ tuft cells or DCLK1+ acinar cells is a matter of debate (5,61).

In zebrafish and mouse models, it has been confirmed that DCLK1⁺ cells are enriched in pancreatic intraepithelial neoplasia and their expansion is an early event in KRAS-induced pancreatic tumorigenesis (52,62). In an established KRAS transgenic mouse model of pancreatic cancer, DCLK1-positive CSC-like cells increased, while at the same time, knockdown of DCLK1 expression in human pancreatic cells reduced c-Myc and KRAS through a let-7a miRNA-dependent mechanism (43). In clinical tissue samples of PDAC, KRAS and TP53 mutations were indicated to be associated with DCLK1 gene overexpression, which may contribute to the migration, proliferation and colony formation abilities of pancreatic cancer cells (63). Mutation of KRAS in DCLK1+ pancreatic cells does not affect cell quiescence or longevity but contributes functionally to the pathogenesis of pancreatic cancer (24). In 2019, Qu et al (64) reported that DCLK1-α-long increases invasion and drug resistance by activating the PI3K/AKT/MTOR signaling pathway through increasing KRAS activity in pancreatic and duodenal homeobox Pdx1^CreKRAS^G12D transformation-related protein 53^R172H mouse models and in the human pancreatic cancer cell lines AsPC-1 and MiaPaCa-2. The hepatocyte growth factor/c-MET axis is necessary for the expression of DCLK1 in tumor cells and the recent two papers indicated that it is strongly associated with tumor immune escape, including the promotion of M2-macrophages and decrease of CD4+ and CD8+ T cells in PDAC (26,65). It facilitates pancreatic cancer progression by mediating the interaction between PCSCs and stromal pancreatic stellate cells (66). Overall, despite these complex and emerging findings, overwhelming evidence has indicated that DCLK1 is a vital pancreatic CSC-like marker, which is upregulated and closely associated with precancerous lesions, tumorigenesis and invasion (67–70).

In human normal stomach tissue, stem cells are located in the isthmus of gastric glands and DCLK1-positive cells were originally located in the gastric stem cell zone. These DCLK1-positive cells were not able to be labeled by bromodeoxyuridine, which was consistent with static stem cells lacking typical cell proliferation ability, suggesting that DCLK1 may be a marker of quiescent stem cells (71). The expression of DCLK1 in gastric cancer tissues was significantly higher than that in adjacent normal tissue and significantly correlated with lymph node metastasis and prognosis (72–75). A recent study suggested that long non-coding RNA small nucleolar RNA host gene 1 promoted the effects of DCLK1/Notch1 on the EMT process through regulating miR-15b expression (76). Small extracellular vesicle (exosome) isolated from a DCLK1-overexpressing human gastric cancer cell line promoted the migration of non-transfected gastric cancer cells in a kinase-dependent manner (77). DCLK1 is also a potential biomarker to predict the survival of patients with gastric cancer (78).

DCLK1 expression progressively increases from Barrett's esophagus to dysplasia and then to esophageal adenocarcinoma (2,79). In human esophageal squamous cell carcinoma (ESCC) cells, DCLK1-S induced MMP2 expression via MAPK/ERK signaling to activate the EMT (80). Knockdown of DCLK1 inhibited the progression of ESCC by regulating proliferation, migration and invasion by suppressing the β-catenin/c-Myc pathway (81). These results indicated that DCLK1 levels are associated with the occurrence and development of esophageal cancer. In Barrett's esophageal adenocarcinoma, the expression of DCLK1 and LGR5 are significantly increased in squamous epithelial cells located at the gastric spout, which indicates that Barrett's esophageal adenocarcinoma probably comes from gastric cancer (82,83).

Serum estradiol levels are an important factor of increased risk of postmenopausal breast cancer. Haakensen et al (84) detected differentially expressed genes by analysis of gene microarrays and indicated that DCLK1 was one of six influenced by serum estradiol. DCLK1 gene expression was downregulated in breast carcinoma samples compared with normal tissue samples but did not exhibit any significantly differential expression between invasive breast cancer and ductal carcinoma in situ. DCLK1 was not significant as an independent factor associated with serum estradiol in a linear regression model. A series of subsequent studies on DCLK1 expression in breast carcinomas were developed and the clinical results indicated DCLK1 was associated with clinicopathological features, estrogen receptor status and neuroendocrine markers (85). A cohort study including 1,132 cases reported that DCLK1 levels varied in several molecular subtypes. Luminal cancers had higher DCLK1 expression than HER2-overexpression and triple negative breast cancers (TNBCs). Elevated DCLK1 was associated with a lower histologic grade, absence of lymphovascular invasion, fibrotic foci, necrosis and lower pN stage. DCLK1 did not correlate with other breast CSC markers and stem cell features, but significant correlations were found with estrogen receptor and neuroendocrine markers. Zhao et al (86) used DCLK1 to devise a clinically practical method based on immunohistochemistry for the molecular subtyping of the mesenchymal subtype TNBC. Specifically, DCLK1 marked a mesenchymal subtype enriched in stem cell-related gene signatures and activated JAK/STAT3 pathway, which is highly correlated with CSC-like breast cancer cells (86). In support of these findings, Ramamoorthy et al (87) reported that DCLK1 is downregulated with ALDH and CD133 downstream of the Notch signaling pathway, which results in inhibition of TNBC stemness. In breast cancer cell lines, silencing of DCLK1 decreased the levels of Wnt/β-catenin pathway proteins such as β-catenin, c-Myc and cyclin D1 to decrease cell migration and invasion (88). Further basic studies indicated that DCLK1 is a molecular regulator of breast cancer proliferation, migration, invasion and a degradome-related metastatic stem-like profile (88–90). Furthermore, miR-424-5p was indicated to act as a tumor suppressive miRNA regulating breast cancer cell proliferation, migration and invasion via binding DCLK1 in vitro (90). In combination, these findings suggest that DCLK1 is a potential therapeutic target in breast cancer, but further mechanistic studies are required.

Only a small number of known markers of CSCs in kidney cancers is available. Among these are the commonly reported broad CSC markers ALDH, CD44 and CD133. Ge et al (91) reported that DCLK1 stimulated essential molecular and functional characteristics of renal CSCs, including expression of ALDH, self-renewal and resistance to approved tyrosine kinase inhibitors sunitinib, sorafenib, everolimus and temsirolimus, suggesting that DCLK1 is a potential renal CSC marker. Furthermore, they indicated that overexpression of DCLK1 was a direct regulatory factor in renal clear carcinoma progression, supporting the notion that DCLK1 is a potential CSC target to inhibit RCC metastasis in early stages (3). Of note, treatment with a DCLK1-targeted monoclonal antibody was able to inhibit tumorigenesis in ACHN renal cancer xenografts, suggesting a potential therapeutic strategy for this highly chemoresistant cancer (91). In addition, a small-molecule kinase inhibitor of DCLK1, DCLK1-IN-1, demonstrated obvious inhibition of immune checkpoint ligand programmed death ligand 1 and an apparent increase in immune-mediated cytotoxicity alone or in combination with anti-programmed death 1 therapy by suppressing DCLK1 phosphorylation and downregulating pluripotency factors and CSC- or EMT-associated markers, including c-MET, c-MYC and N-cadherin in RCC cell lines. These experimental results were consistent with the analysis of clinical populations in which DCLK1 predicted RCC survival. In addition, its expression was correlated with reduced CD8+ cytotoxic T-cell infiltration and increased in M2 immunosuppressive macrophage populations (92).

To date, DCLK1 has not been identified to be a hepatocellular CSC marker. However, the expression of DCLK1 in chronic hepatitis, cirrhosis and hepatocellular carcinoma was significantly increased (93). DCLK1 is mainly expressed in epithelial and stromal cells, lymphocytes and bile duct cells of liver tissue of patients with chronic hepatitis C virus infection. Furthermore, the level of DCLK1 is related to the expression of S100A9 protein (94). S100A9 is a key protein of pro-inflammatory signaling by binding to advanced glycation end product receptor and toll-like receptor 4 to activate the NFκB pathway (95). Upregulation of DCLK1 may promote the expression of S100A9 protein, while downregulation of DCLK1 directly reduces the expression of S100A9 protein and reduces signal cell infiltration of inflammatory cells (96). Ali et al (94) reported that DCLK1 was overexpressed in liver cells infected with hepatitis C virus and further results indicated that DCLK1 was involved in the replication of hepatitis C virus. According to recent findings, tuft cells express CD300lf (a murine norovirus receptor) and are virally induced to proliferate through this receptor to improve murine norovirus infection. Although research in this area is limited, it is worth considering if tuft cells in the intestine may similarly take part in the replication of hepatitis C and other viruses (97). Liver tissues from patients with cirrhosis and HCC exhibited overexpression of DCLK1, β-catenin and cleaved E-cadherin. DCLK1-overexpressing hepatoma cells induced high levels of β-catenin, α-fetoprotein and SOX9, which led to clonogenicity and dedifferentiated phenotypes (98). In HCC tumors, DCLK1-positive cells have characteristics of CSCs and co-express marker proteins CD133, LGR5, Lin28, AFP and c-Myc (99,100). DCLK1 may be a new target for the treatment of hepatitis C virus-induced tumorigenesis. However, the stem cell characteristics of DCLK1 in hepatocellular carcinoma require confirmation by further research.

All related signaling pathways of DCLK1 in different types of cancer are illustrated in Fig. 4.