DCLK1 promotes colorectal cancer stemness and aggressiveness via the XRCC5/COX2 axis

Abstract

Rationale: Doublecortin-like kinase 1 (DCLK1) is a serine/threonine kinase that selectively marks cancer stem-like cells (CSCs) and promotes malignant progression in colorectal cancer (CRC). However, the exact molecular mechanism by which DCLK1 drives the aggressive phenotype of cancer cells is incompletely determined. Methods: Here, we performed comprehensive genomics and proteomics analyses to identify binding proteins of DCLK1 and discovered X-ray repair cross-complementing 5 (XRCC5). Thus, we explored the biological role and downstream events of the DCLK1/XRCC5 axis in human CRC cells and CRC mouse models. Results: The results of comprehensive bioinformatics analyses suggested that DCLK1-driven CRC aggressiveness is linked to inflammation. Mechanistically, DCLK1 bound and phosphorylated XRCC5, which in turn transcriptionally activated cyclooxygenase-2 expression and enhanced prostaglandin E₂ production; these events collectively generated the inflammatory tumor microenvironment and enhanced the aggressive behavior of CRC cells. Consistent with the discovered mechanism, inhibition of DCLK1 kinase activity strongly impaired the tumor seeding and growth capabilities in CRC mouse models. Conclusion: Our study illuminates a novel mechanism that mediates the pro-inflammatory function of CSCs in driving the aggressive phenotype of CRC, broadening the biological function of DCLK1 in CRC.

Keywords: Cancer stem cells; Doublecortin-like kinase 1; Inflammatory tumor microenvironment; Prostaglandin E2.

Figure 1

Increased DCLK1-B expression infers CRC aggressiveness. (A) Schematic illustration of DCLK1-B promoter-green fluorescent protein (GFP)-tagged cells. DCLK1-B^high and DCLK1-B^low cells were sorted based on GFP tagging, and the difference in stemness between the two groups was investigated via an in vitro LDA. (B) Heatmap comparing the relative expression of CSC genes and malignancy-related genes in the DCLK1-B bulk, DCLK1-B^high, and DCLK1-B^low cell populations, as determined by RT-qPCR (n = 3 biological replicates). (C) Immunoblot of DCLK1-A and DCLK1-B expression in DCLK1-B WT and OE HCT116 cell lines, accompanied by a positive control (mouse brain). (D and E) Effect of DCLK1-B overexpression on liver metastasis. Luciferase-labeled DCLK1-B WT and DCLK1-B OE HCT116 cells were inoculated into the spleens of NSG mice. Mice were tracked for 35 days after splenic injection (n = 10 mice per group). (D) Representative in vivo bioluminescence images (left) of mice injected with luciferase-labeled DCLK1-B WT and DCLK1-B OE HCT116 cells, accompanied by a corresponding graph showing the quantitative analysis of the region of interest (right). (E) Representative hematoxylin and eosin-stained livers with metastasis. N: normal, T: tumor. (F) Survival analysis of the DCLK1-B WT- and DCLK1-B OE-inoculated groups (n = 10 mice per group). Survival curves were plotted using the Kaplan-Meier method, and statistical significance was determined by the log-rank test. (G) Immunofluorescence analysis of DCLK1-B expression in tumor and matched normal adjacent intestinal tissues from CRC patients (n = 123 patients). Immunofluorescence staining is shown for EpCAM (green), DCLK1-B (red), and DAPI staining (blue) with the corresponding merged and magnified images. Scale bars, 100 µm. Graphs show the integrated optical density (IOD) indicating the DCLK1-B protein level in the normal and tumor epithelium (left) and the Kaplan-Meier survival curves of the CRC patients (right). RFS, relapse-free survival; OS, overall survival. Statistical significance was determined by paired Student's t-tests for IOD quantification and by the log-rank test for Kaplan-Meier analysis. The data are presented as the means ± SEMs. *** indicates p < 0.001. The statistical significance of differences in tumor growth and the survival of mice with liver metastasis was determined by two-way repeated-measures ANOVA followed by the Bonferroni post hoc test.

Figure 2

DCLK1-B promotes cell survival, apoptosis resistance, and migration. (A) Immunoblot of DCLK1-A and DCLK1-B expression in DCLK1-B WT, OE, and KO HCT116 cell lines, accompanied by a positive control (mouse brain). (B) Series of biological functional assays showing the effects of DCLK1-B expression on cancer cell survival, proliferation, apoptosis, migration, and invasion. Clonogenic assays (survival), MTT assays (proliferation), Annexin-PI FACS analysis (apoptosis), wound closure assays (migration), and invasion assays (invasion) were performed with DCLK1-B WT, OE, and KO HCT116 cells (n = 3 biological replicates). (C) Immunoblot of DCLK1-A and DCLK1-B expression upon independent KD of DCLK1-A and DCLK1-B by siRNA transfection. After 48 h of siRNA transfection, cells were lysed for protein analysis. (D) Series of biological functional assays showing the effects of DCLK1-A and DCLK1-B KD on cancer cell survival, apoptosis, and migration. (E) Schematic illustration showing the lengths and the shared protein kinase domain of the DCLK1 isoforms referenced in UniProt [O15075] (left). The kinase activity of DCLK1 was measured at increasing concentrations of DCLK1-IN-1 (right, n = 5 biological replicates). The IC₅₀ value, 143 nM, is shown. (F) Cell viability rates and cytotoxic IC₅₀ values were determined by an MTT assay after 48 h of treatment with DCLK1-IN-1 in both HCT116 and hCRC#1 cells (n = 5 biological replicates for both cell lines). (G and H) Series of biological functional assays showing the effects of DCLK1-IN-1 on cancer cell survival (3 µM), apoptosis (3 µM), and migration (1 µM) in HCT116 (G) and hCRC#1 (H) cells (n = 3 biological replicates). The data are presented as the means ± SEMs. *, ** and *** indicate p < 0.05, p < 0.01, and p < 0.001, respectively. Statistical significance was determined by unpaired two-tailed Student's t-tests for comparisons between two groups and one-way ANOVA with Dunnett's multiple comparison test for comparisons among three groups.

Figure 3

DCLK1 enhances COX2 expression via XRCC5 phosphorylation. (A) GSEA of pathways enriched in DCLK1^high patients compared to DCLK1^low patients from open-source CRC patient data (GSE21510) (top) and in DCLK1-B WT cells compared to DCLK1-B KO cells from RNA sequencing profiles (bottom). (B) Heatmap showing the relative expression of inflammation-related genes upon DCLK1-B KO, as determined by RT-qPCR (n = 3 biological replicates). (C) Upstream analysis suggesting a potential relation of COX2 with a related disease and function, i.e., inflammation and gastrointestinal tumors, caused by DCLK1-B expression alteration. (D) Alteration of COX2 expression upon DCLK1-B OE and KO (top) and DCLK1-B kinase domain inhibition (1 µM, bottom). COX2 expression was analyzed at the protein and mRNA levels. (E) Immunoblot of DCLK1-B, XRCC5, and phosphorylated (p-) XRCC5 in HCT116 (top) and hCRC#1 (bottom) cells subjected to immunoprecipitation with an anti-DCLK1-B antibody. (F) Immunoblot of p-XRCC5 and XRCC5 upon DCLK1-B expression regulation (left) and DCLK1-B kinase domain inhibition (1 µM, right) in both HCT116 (top) and hCRC#1 (bottom) cells. (G) Relative transcription of COX2 and its resultant translation upon DCLK1-B OE and siXRCC5 transfection. A luciferase vector containing the COX2 promoter region (-1236 to +230) was transfected into cells, and luminescence was measured and normalized to β-galactosidase (n = 3 biological replicates). (H) Relative transcription of COX2 upon OE of XRCC5 WT, an active mutant form of XRCC5 (T715D), and an inactive phosphomimetic form of XRCC5 (T715A). Total transcription was normalized to β-galactosidase transcription and is presented as the fold change with respect to non-transfected HEK293T cells (n = 3 biological replicates). (I) Potential binding site for p-XRCC5 in the COX2 promoter region identified by a chromatin immunoprecipitation assay. The COX2 promoter region was fragmented into 6 segments and immunoprecipitated with an anti-p-XRCC5 antibody. (J) Schematic illustration of the mechanism by which the DCLK1/p-XRCC5/COX2 axis sustains cancer stemness and aggressiveness. The data are presented as the means ± SEMs. *** indicates p < 0.001. Statistical significance was determined by unpaired two-tailed Student's t-tests for comparisons between two groups and one-way ANOVA with Dunnett's multiple comparison test for comparisons among three or more groups.

Figure 4

DCLK1/XRCC5 axis shapes the pro-tumor microenvironment via COX2 signaling. (A) Schematic illustration of the protocol for injection of shXRCC5-conjugated adeno-associated virus (AAV) into 3.5-week-old Apc^Min/+ mice (top). Both the AAV-Ctrl and AAV-XRCC5^KD groups were monitored until the age of 20 weeks and sacrificed for analysis. Delivery of the virus was confirmed by GFP expression in the targeted organ, the intestine (bottom; n = 8 for both AAV-Ctrl and AAV-XRCC5^KD). (B) The frequency of polyp formation in both the AAV-Ctrl and AAV-XRCC5^KD groups is presented. The number of polyps in each individual mouse was determined and used to calculate the total tumor burden. (C) Representative immunoblots of COX2 and XRCC5 expression in the AAV-Ctrl and AAV-XRCC5^KD groups. (D) In vitro ELISA showing the amount of PGE₂ secreted into the cell culture medium upon DCLK1-B expression regulation and XRCC5 knockdown (n = 3 biological replicates). (E) Levels of secreted PGE₂ in plasma and tumor tissue. Plasma was collected from B6, AAV-Ctrl, and AAV-XRCC5^KD mice. Tumor tissues were collected from intestinal polyps of AAV-Ctrl and AAV-XRCC5^KD mice (n = 8 biological replicates). The estimated normal range of the plasma PGE₂ level was determined by the values obtained in B6 mice (shaded in gray). (F) Heatmap comparing the relative expression of pro-inflammation, M2 polarization-related, cytokine production-related, and immune tolerance-related genes in B6, AAV-Ctrl, and AAV-XRCC5^KD mice, as determined by RT-qPCR (n = 8 biological replicates). (G) Schematic illustration of in vivo experiments investigating the therapeutic effect of DCLK1-IN-1 in CRC tumorigenesis and resultant growth curves of tumors (n = 5 for both the vehicle and DCLK1-IN-1 treated group). Luciferase-labeled MC-38 cells (5x10⁴cells mixed with Matrigel/mouse) were injected subcutaneously into the inguinal folds of C57BL/6J (B6) mice prior to treatment. Once the mean volume of the xenograft tumors reached 100 mm³, 10 mg/kg of DCLK1-IN-1 was administered daily. Mice were tracked for a total of 21 days (14 days post treatment). (H) Level of PGE₂ in plasma and tumor tissue. Plasma was collected from naïve B6 and tumor-bearing B6 mice. Tumor tissues were collected from vehicle and DCLK1-IN-1-treated tumor-bearing mice (n = 5 biological replicates). (I) RT-qPCR heat map of relative expression of pro-inflammation, M2 polarization, cytokine production, and immune tolerance-related genes in vehicle and DCLK1-IN-1-treated mice. The data are presented as the means ± SEMs. *** indicates p < 0.001. Statistical significance was determined by unpaired two-tailed Student's t-tests for comparisons between two groups and one-way ANOVA with Dunnett's multiple comparison test for comparisons among three or more groups. Two-way repeated-measures ANOVA followed by the Bonferroni post hoc test for comparison of tumor metastasis over time.

Figure 5

PGE₂ produced by the DCLK1/PGE₂ axis mediates CRC aggressiveness. (A) Series of biological functional assays showing the effects of the DCLK1-B/PGE₂ axis on cancer cell survival, proliferation, apoptosis, and migration (n = 3 biological replicates). DCLK1-B WT and KO HCT116 cells were treated with PGE₂ (2 µM and 3 µM). (B) Heatmap showing the relative expression of CSC genes and malignancy-related genes upon alteration of the DCLK1/PGE₂ axis, as determined by RT-qPCR (n = 3 biological replicates). (C) Stemness was investigated with various methods, e.g., measurement of sphere size, sphere growth, and sphere-forming potential. Sphere size was analyzed by measuring the sphere diameter and fold change relative to the mean value of DCLK1-B WT spheres (n = 4 biological replicates). Growth was analyzed by a quantitative Cell Titer-Blue assay (n = 4 biological replicates). Sphere-forming potential was analyzed by an in vitro limiting dilution assay (n = 4 biological replicates). p values indicate statistical significance of differences in stem cell frequencies between any of the groups. (D) Series of biological functional assays showing the effects of the DCLK1-B/PGE₂ axis on cancer cell survival, proliferation, apoptosis, and migration (n = 3 biological replicates). DCLK1-B WT and OE HCT116 cells were treated with the PGE₂ receptor (EP4) inhibitor L-161,982 (1 µM and 3 µM). (E) Heatmap showing the relative expression of CSC genes and malignancy-related genes upon alteration of the DCLK1/PGE₂ axis, as determined by RT-qPCR (n = 3 biological replicates). DCLK1-B WT and OE HCT116 cells were treated with L-161,982 (3 µM). (F) Stemness was assessed in DCLK1-B WT and OE HCT116 cells treated with L-161,982 (3 µM). p values indicate statistical significance of differences in stem cell frequencies between any of the groups. See panel (C) for details. The data are presented as the means ± SEMs. *, ** and *** indicate p < 0.05, p < 0.01, and p < 0.001, respectively. Statistical significance was determined by one-way ANOVA with Dunnett's multiple comparison test for comparisons among three or more groups and two-way repeated-measures ANOVA followed by the Bonferroni post hoc test for comparisons of sphere growth.

Figure 6

DCLK1-IN-1 efficiently inhibits DCLK1/XRCC5/COX2 signaling and stemness in CRC. (A-C) Alteration of stemness upon DCLK1-IN-1 treatment (1 μM) was investigated with various methods (n = 4 biological replicates), e.g., measurement of sphere growth and sphere size. Sphere growth was analyzed by a quantitative Cell Titer-Blue assay, while sphere size was analyzed by measuring the sphere diameter and fold change relative to the mean value of Ctrl spheres. (A) HCT116, (B) HT29 and (C) hCRC#1. (D) Schematic illustration of the protocol for in vivo treatment with the DCLK1 inhibitors, DCLK1-IN-1. HCT116 cells (1x10⁶ cells mixed with Matrigel/mouse) were injected subcutaneously into the inguinal folds of NSG mice prior to treatment. Once the mean volume of the xenograft tumors reached 100 mm³, 10 mg/kg of DCLK1-IN-1 was administered daily. Mice were tracked for 32 days after DCLK1 inhibitor injection (n = 8 mice per group). (E) Growth curves of the first tumor generation (left) and in vivo limiting dilution assay (right) comparing the tumor-repopulating potential of the second generation between the Ctrl and DCLK1-IN-1 treatment groups. (F) Levels of PGE₂ in plasma and tumor tissue. Plasma and tumor tissues were collected from Ctrl and DCLK1-IN-1-treated NSG tumor-bearing mice. The estimated normal range of the plasma PGE₂ level was determined by the values obtained in naïve mice (shaded in gray). The data are presented as the means ± SEMs. *** indicates p < 0.001. Statistical significance was determined by unpaired two-tailed Student's t-tests for two-group comparisons, one-way ANOVA with Dunnett's multiple comparison test for three-group comparisons, and two-way repeated-measures ANOVA followed by the Bonferroni post hoc test for tumor growth comparisons.