Oncogenic KRAS mutation plays a key role in pancreatic ductal adenocarcinoma (PDAC) tumorigenesis with nearly 95% of PDAC harboring mutation-activated KRAS, which has been considered an undruggable target. Doublecortin-like kinase 1 (DCLK1) is often overexpressed in pancreatic cancer, and recent studies indicate that DCLK1+ PDAC cells can initiate pancreatic tumorigenesis. In this study, we investigate whether overexpressing DCLK1 activates RAS and promotes tumorigenesis, metastasis, and drug resistance. Human pancreatic cancer cells (AsPC-1 and MiaPaCa-2) were infected with lentivirus and selected to create stable DCLK1 isoform 2 (alpha-long, AL) overexpressing lines. The invasive potential of these cells relative to vector control was compared using Matrigel coated transwell assay. KRAS activation and interaction were determined by a pull-down assay and coimmunoprecipitation. Gemcitabine, mTOR (Everolimus), PI3K (LY-294002), and BCL-2 (ABT-199) inhibitors were used to evaluate drug resistance downstream of KRAS activation. Immunostaining of a PDAC tissue microarray was performed to detect DCLK1 alpha- and beta-long expression. Analysis of gene expression in human PDAC was performed using the TCGA PAAD dataset. The effects of targeting DCLK1 were studied using xenograft and Pdx
Pancreatic ductal adenocarcinoma (PDAC) has the worst prognosis of any major malignancy with less than an 8% 5-year survival rate and is the third leading cause of cancer-related deaths in the United States [
Cells with cancer stem cell-like (CSC) properties have been identified in PDAC. These cells are often resistant to conventional chemotherapy and radiation therapy and as such may explain why current treatments do not cure PDAC or prevent recurrences. Doublecortin-like kinase 1 (DCLK1) is often overexpressed in pancreatic cancer and is coexpressed with other PDAC CSC markers, and recent studies indicate that DCLK1+ PDAC cells can initiate pancreatic tumorigenesis in the presence of mutation and inflammation [
Many studies have reported targeting KRAS for PDAC treatment but it remains an undruggable target [
All animal experiments were performed with the approval and authorization from the Institutional Review Board and the Institutional Animal Care and Use Committee, University of Oklahoma Health Sciences Center and University of Pennsylvania Perelman School of Medicine. Mice were housed under controlled conditions, including a 12-h light-dark cycle, with ad libitum access to food and water.
Athymic nude mice were purchased from The Jackson Laboratory (Bar Harbor, Maine).
For DCLK1 mRNA expression levels in various cancer types, the Cancer Genome Atlas (TCGA) esophageal (ESCA), stomach (STAD), liver (LIHC), pancreas (PAAD), and colorectal (COADREAD) datasets were used. For DCLK1 protein expression levels in various cancer types and in normal tissues, the Human Protein Atlas (THPA) datasets were used [
The standard data run of The Cancer Genome Atlas PAAD dataset was downloaded and sorted for DCLK1 expression. Mann-Whitney U test was used for analysis and comparison of other gene expressions between these two groups (n=45 for each group).
Only publicly available, deidentified data were accessed from TCGA for the analysis reported here. Basic characteristics of the PDAC patients used in the survival analysis are provided in Supplementary Tables
A pancreatic adenocarcinoma tissue microarray (US Biomax, HPan-Ade 180 Sur-02) containing 180 microsections including 60 paired tumor and normal adjacent tissues was immunostained with anti-DCLK1 antibody (Abcam, ab31704) following our previously described protocol [
Human pancreatic cancer cell lines, AsPC-1 and MIA PaCa-2 (MP2), were obtained from ATCC and grown in Dulbecco’s Modified Eagle’s Medium with 4.5 g/L glucose and L-glutamine, without sodium pyruvate (Cellgro) supplemented with 10% fetal bovine serum (Sigma) at 37°C and 5% CO2. Lentivirus containing human DCLK1-AL cDNA sequence was constructed as described previously [
Cells (5000 cells per well) were seeded into a 96-well tissue culture plate in triplicate. The cells were cultured in the presence of Gemcitabine (0, 12.5, 25, 50, 100, and 200 nM), everolimus (37.5
Matrigel coated transwell assays (BD Biosciences) were prepared by soaking in serum-free media for 2 h at 37°C in a 24-well plate. MP2-RFP and MP2-DCLK1 cells (5000 cells/well) were seeded into each transwell in serum-free media in triplicate. Cell culture medium containing 10% FBS was added to the bottom of each well as chemoattractant and the cells were incubated for 22 h at 37°C. A cotton swab was used to scrape noninvasive/migratory cells off the top of transwell assays and the remaining cells were fixed with 100% methanol, stained with 0.1% crystal violet, and allowed to dry. After drying all invading cells were counted from each transwell at 4x magnificence.
MP2-RFP and MP2-DCLK1 cells (250 cells/well, n=6 per group) were seeded into an ultra-low attachment 96-well plate in RPMI containing 0.5% FBS and incubated at 37°C under 5% CO2 for 5 days. Medium without FBS was added on day 3 to prevent evaporation. On day 5, spheroids were manually counted under a light microscope at 10x magnification, and representative images were taken. Spheroids were defined as having at least 10 cells. Efficiency of spheroid formation was calculated by dividing the number of spheroids formed by the number of cells seeded.
Both AsPC-RFP and AsPC-DCLK1 cells were cultured in serum-free medium overnight, followed by full growth medium (10% FBS) for 15 min in the presence of either DMSO or XMD8-92 (15
Both AsPC-RFP and AsPC-DCLK1 cells were lysed with Pierce IP Lysis Buffer (Thermo Fisher Scientific). The cell lysates were used for immunoprecipitation by incubating with anti-RAS antibody for 2h at room temperature, spinning down the precipitates with Protein A conjugated anti-mouse secondary antibody, washing 3 times with Pierce IP washing buffer, and eluting with gel loading buffer. The eluates were separated on a SDS-PAGE and subjected to western blot analysis with anti-DCLK1 antibody (Abcam, ab31704).
Total proteins of cell lysates were subjected to Western Blot analysis. The concentration of total proteins was determined by BCA protein assay. Equivalent amounts of total proteins were separated on a SDS-PAGE and transferred onto a nitrocellulose membrane. The membrane was blocked with 5% nonfat dry milk and probed with the primary antibody. The membrane was then incubated with IRDye 800CW-conjugated secondary antibody. The proteins were detected using Li-Cor Odyssey system.
DCLK1-targeted therapeutic monoclonal antibody (CBT-15mAb) and isotype control mAb were supplied in PBS (COARE Biotechnology). In addition, total RNA was isolated from monoclonal hybridoma cells secreting DCLK1 antibody (CBT-15); cDNA was synthesized using a primer downstream of the last variable region for heavy chain (HC) constant and light chain (LC) kappa constant. Each RT-reaction was subject to PCR using degenerate primer sets (USBIO, 11904-10A) to amplify all likely rearrangements. To create the human/mouse IgG chimeric antibody, PCR fragments from the above reaction were inserted into pFUSEss-CHIg-hG1 to express heavy chain and pFUSEss-CLIg-hK to express light chain kappa. Heavy chain was further cloned into pLenti CMV PURO DEST and light chain kappa was further cloned into pLenti CMV BLAST DEST. The expression plasmids constructed above were cotransfected along with packaging plasmids pMD2.G (Addgene), pMDL/RRE g/p (Addgene), and pRSV-Rev (Addgene) into 293T cells. Generation of the concentrated lentivirus was done as described previously [
SW1990 or AsPC-1 pancreatic cancer cells (0.5x106) in Matrigel were injected into the flanks of 8-week old athymic nude mice (n=6 for CBT-15 vs. isotype control groups and n=7 for CBT-15X vs. isotype control groups for both SW1990 and AsPC-1 cells) and allowed to grow to an average tumor volume of 100 mm3. Mice with xenografted tumors were injected intraperitoneally (
KPC mice with tumors measuring 50-100 mm3 were identified using ultrasonography. These mice were injected
All statistical analyses and figures were prepared using R v3.2, GraphPad Prism 6.0, SPSS Statistics 22, and Microsoft Excel. For nonparametric data the Mann-Whitney U test was used, and for parametric data Student’s t-Test was used. Kaplan-Meier survival analyses were performed in GraphPad Prism 6.0. Cox regression analyses were performed using IBM SPSS Statistics 22. Heatmaps were generated using Genesis. A p-value of less than 0.05 was considered statistically significant for all analyses.
In order to assess DCLK1’s gene expression pattern across gastrointestinal cancer types, we analyzed the TCGA esophageal (ESCA), stomach (STAD), liver (LIHC), pancreas (PAAD), and colorectal (COADREAD) datasets and found that pancreatic cancer tissue has the highest DCLK1 mRNA expression levels among the gastrointestinal cancer types (Figure
To further evaluate DCLK1 protein expression in PDAC tumors, we performed immunohistochemistry using anti-DCLK1 antibody on a commercially available tissue microarray with tumor and normal adjacent tissues (NAT) from stages I/II pancreatic cancer patients. We found higher expression of DCLK1 in most of the tumor samples and assessed the effect of DCLK1 expression on patient survival. The expression levels of DCLK1 in the tumor tissues did not predict survival (data not shown). However, patients with high levels of DCLK1 in the NAT had significantly reduced overall survival compared to patients with low levels (median 6-7 months and 12-13 months, resp.) (Figures
In order to determine whether DCLK1 expression level correlates with KRAS related pathways in human PDAC patients, we analyzed RNA-Seq expression data from TCGA (PAAD). We grouped patients into DCLK1-low (bottom 25th percentile) and DCLK1-high (top 25th percentile) groups and compared expression of genes downstream of RAS activation. We found that DCLK1-AL and BL are associated with increased EMT based on genetic signature analysis. In addition, DCLK1-high patients have increased expression of PI3K/AKT/MTOR and downstream signaling pathways which support stemness, antiapoptosis, and tumorigenesis (Figure
To assess the effects of DCLK1-AL on PDAC, we established stable cell lines overexpressing DCLK1-AL-RFP fusion protein in AsPC-1 and MP2 cells using RFP as control (Figure
Drug resistance is a mechanism by which quiescent tumor stem cells maintain viability while the bulk of the tumor is destroyed by chemotherapies targeting rapidly dividing tumor cells. To assess whether overexpression of DCLK1-AL increases drug resistance, we treated MP2-RFP and MP2-DCLK1 cells with various concentrations of gemcitabine for 48 h and performed an MTT assay. MP2-DCLK1 cells significantly resisted gemcitabine treatment compared to MP2-RFP cells at most doses (p<0.05) (Figure
Using a coimmunoprecipitation assay, we found that DCLK1-AL forms a complex with RAS (Figure
Since high DCLK1 expression in pancreatic cancer patients is correlated with activation of pathways downstream of RAS (PI3K/MTOR) (Figure
We recently reported that monoclonal antibody CBT-15 targeting DCLK1-AL/BL inhibits renal cancer tumorigenesis
Although athymic nude mice maintain a partially functional immune system, mAb therapies are best assessed in models with full immune function. Given our recent promising findings in renal cell cancer [
Despite advances in the understanding of pancreatic cancer biology and in surgical and medical therapy in recent years, little impact has been made on the mortality associated with this cancer. Therefore, there is an unmet need to find new therapeutic approaches against PDAC. Zhang et al. reported recently that DCLK1 levels in PDAC tumor tissues predict poor survival [
There are two DCLK1 isoforms transcribed from the
KRAS activating mutations are present in 95% of PDAC tumors, but targeting KRAS directly has been unsuccessful so far and many inhibitors have failed in clinical trials [
In order to evaluate the effect of targeting DCLK1
In summary, the studies reported here illustrate the role of DCLK1 in KRAS activation, PDAC tumor cell invasion, drug resistance, pancreatic tumor growth
DCLK1 promotes KRAS-driven PI3K/AKT/mTOR signaling in PDAC leading to increased invasive, antiapoptosis, stemness, and tumorigenic properties. DCLK1-targeted therapies may overcome this signaling and improve PDAC outcomes.
The data used to support the findings of this study are included within the article.
Dongfeng Qu and Nathaniel Weygant are equally contributing authors.
Courtney W. Houchen is a cofounder of COARE Biotechnology Inc. Dongfeng Qu, Nathaniel Weygant, and Randal May have ownership interests in COARE Biotechnology Inc. Other authors have declared that no conflicts of interest exist.
We would like to thank Dr. Sripathi M. Sureban and Edwin Bannerman-Menson, COARE Biotechnology, for supplying CBT-15 monoclonal antibody for KPC studies. We would like to thank Dr. Weihong Li, St. Luke Hospital, Maumee, Ohio, for assistance on some pathological questions. We acknowledge Stephenson Cancer Center at the University of Oklahoma Health Sciences Center, Oklahoma City, OK, and an Institutional Development Award (IDeA) from the National Institute of General Medical Sciences of the National Institutes of Health under Grant number P20 GM103639 for the use of Histology and Immunohistochemistry Core, which provided immunohistochemistry services. This research was funded by a NIH R01 Grant to Courtney W. Houchen (5R01CA182869).
Figure S1: CBT-15X chimeric mAb inhibits pancreatic cancer xenograft tumor growth. A. Excised tumor volume and tumor mass from SW1990 pancreatic cancer cells originated xenograft. B. Excised tumor volume and tumor mass from AsPC-1 pancreatic cancer cells originated xenograft. Figure S2: overexpression of DCLK1-AL in MP2 cells enhances tumor spheroid formation. A-B. Spheroids formation is significantly enhanced in MP2-DCLK1 cells (P<0.0001). C. Representative images display differences between MP2-RFP and MP2-DCLK1 spheroid formation. Table S1: Patient Characteristics. Publicly available, deidentified data were accessed from TCGA, and basic characteristics of the PDAC patients are presented. Table S2: Univariate and Multivariate Analyses. Publicly available, deidentified data were accessed from TCGA for the analysis reported here.