FOLFOX Therapy Induces Feedback Upregulation of CD44v6 through YB-1 to Maintain Stemness in Colon Initiating Cells

Abstract

Cancer initiating cells (CICs) drive tumor formation and drug-resistance, but how they develop drug-resistance characteristics is not well understood. In this study, we demonstrate that chemotherapeutic agent FOLFOX, commonly used for drug-resistant/metastatic colorectal cancer (CRC) treatment, induces overexpression of CD44v6, MDR1, and oncogenic transcription/translation factor Y-box-binding protein-1 (YB-1). Our study revealed that CD44v6, a receptor for hyaluronan, increased the YB-1 expression through PGE2/EP1-mTOR pathway. Deleting CD44v6, and YB-1 by the CRISPR/Cas9 system attenuates the in vitro and in vivo tumor growth of CICs from FOLFOX resistant cells. The results of DNA:CD44v6 immunoprecipitated complexes by ChIP (chromatin-immunoprecipitation) assay showed that CD44v6 maintained the stemness traits by promoting several antiapoptotic and stemness genes, including cyclin-D1, BCL2, FZD1, GINS-1, and MMP9. Further, computer-based analysis of the clones obtained from the DNA:CD44v6 complex revealed the presence of various consensus binding sites for core stemness-associated transcription factors "CTOS" (c-Myc, TWIST1, OCT4, and SOX2). Simultaneous expressions of CD44v6 and CTOS in CD44v6 knockout CICs reverted differentiated CD44v6-knockout CICs into CICs. Finally, this study for the first time describes a positive feedback loop that couples YB-1 induction and CD44 alternative splicing to sustain the MDR1 and CD44v6 expressions, and CD44v6 is required for the reversion of differentiated tumor cells into CICs.

Keywords: CD44v6; CD44v6 CRISPR/Cas9 knockout; CD44v6-therapy; CIC; MDR1; YB-1; YB-1 CRISPR/Cas9 knockout; colorectal cancer (CRC); stemness genes.

Figure 1

Establishment of FOLFOX resistant colorectal cancer (CRC) cells that exhibit increased CD44v6 expression and signaling. (A) A schematic diagram of the CD44 gene, where constitutive (c) and variable (v) exons, and the PCR primers used to amplify CD44 variable (v) and standard (s) isoforms are shown. The primers for both the CD44v6 and CD44s predominantly generate one PCR product, whereas the primers for the CD44v8 variants amplify three variant PCR products. (B) A time course of FOLFOX (FOLFOX: 50 μg/mL 5-Flurouracil + 10 μM Oxaliplatin + 1 μM leucovorin) stimulation on CD44 isoform mRNA expressions (analyzed by semiquantitative RT-PCR) in SW948 cells was depicted. (C) QPCR assays for variant 6 of CD44 (CD44v6) expression under low-pH (ischemic stress), CoCl2 (hypoxic stress), H2O2 (oxidative stress), 5-FU, OXA, and FOLFOX treatment (chemotherapeutic stress) in SW948 cells are shown. (D) shRNA-mediated knockdown of CD44v6 affects alternative splicing of CD44 and downregulates YB1 and MDR1 expression. (E) Validations of CD44v6 shRNAs were done by the indicated shRNA mediated knockdown and the corresponding shRNA resistant knock-in (KI) gene overexpression. (F) SW948-FR and HT29-FR cells selectively overexpressed CD44v6, MDR1, and YB-1 mRNAs (by QPCR) compared to sensitive (“S”) pairs of cells. The expression of indicated proteins in FR cells compared to sensitive pairs are presented as mean ± SD (n = 3); *, p < 0.01. Student’s t test was used to assess the significance. The experiment was performed three times and representative data are shown. (G) Western blot (WB) analyses for β-catenin and β-tubulin of “S” and “FR” cell lysates of SW948 and HT29 cells are shown. (H) β-catenin luciferase activity of “S” and “FR” lysates of SW948 and HT29 cells treated with or without WNT 3A are shown. The relative luciferase in FR cells compared to sensitive pairs are presented as mean ± SD (n = 3); *, p < 0.01. Student’s t test was used to assess the significance. The experiment was performed three times and representative data are shown. (I) Anchorage-independent growth in soft agar is shown for SW948-FR and HT29-FR cells and compared with their “S” pairs (magnification, 50 ×). (J) Tumor-sphere formation assay was done for the SW948-FR and HT29-FR cells and compared with their “S” pairs (magnification, 100×). (K,L) Tumor formation is shown in nude mice injected with 500,000 SW948-FR cells or with 500,000 SW948-S cells. SW948-FR cells formed tumor nodules in all injected mice (8/8), whereas SW948-S cells did not induce any tumor nodules until 5 months (left, 0/7 mice) (K). Growth curves are shown for these xenograft tumors in BALB/c nude mice (L). Values in C, I, and J represent means ± SD; n = 3–6; * p < 0.05 for FOLFOX resistant cells compared to sensitive cells. Scale bar, 50 μm.

Figure 2

Flow cytometric analyses of EpCAM+/CD44v6+/ALDH+/CD133+ cells in SW948-FR cells isolated from SW948-FR/subcutaneous (SQ) tumors. Alexa fluor-tagged antibodies at the indicated laser lines were used to isolate: (A) EpCAM+/ECadherin- cells; (B) CD44v6+/ALDH1+ cells from EpCAM+/ECadherin-cells; (C) CD133+ cells from CD44v6+/ALDH1+ cells. (D) Immunofluorescence staining shows colocalization of CD44v6 (Red) and YB-1 (Green) in EpCAM+/CD44v6+/ALDH1+/CD133+ (CICs), scale bar, 50 μm. (E) Percentages of EpCAM+/E-cadherin-, CD44v6+/ALDH1+, and CD44v6+/ALDH1+/CD133+ on sorted cells were assessed by flow cytometry on freshly purified CRC cells isolated from subcutaneous (SQ) SW948-FR tumor cells. (F) Western blots are shown for EpCAM+/CD44v6-/ALDH-/CD133- (Non-CICs) and EpCAM+/CD44v6+/ALDH+/CD133+ (CICs) by probing with E-Cadherin, EpCAM, CD44v6, ALDH1, YB-1, and CD133 antibodies. FACS, immunofluorescence and WB data are representative of three experiments. Enrichment of CICs in FR cells compared to sensitive pairs in Figure 2E are presented as mean ± SD (n = 3); *, p < 0.01. Student’s t test was used to assess the significance. The experiment was performed three times and representative data are shown.

Figure 3

Mechanism of CD44v6 induced YB-1 expression. (A) Time course results are shown of CD44v6 and YB-1 protein expressions after transfecting the v6 rescue plasmid (described in the Methods section) into v6 Mu1 SW948-FR CICs. Expression of proteins at 0 h was not shown because of the absence of CD44v6 in the protein lysate. (B) Effects of knocking out CD44v6 in in v6 Mu1 CICs of sensitive and resistant SW948 cells on PGE2 production (analyzed by ELISA as described in Methods) in the presence and absence of FOLFOX treatment are shown. (C) 17-P-T-PGE2 (synthetic PGE2) induces YB-1 in CICs that were treated with PGE2 at 5 μM for the indicated times. YB-1 expression levels were determined by immunoblotting with anti-YB-1 antibody; β-tubulin as loading control. (D) Effects are shown of CD44v6 Mu1 knockout on PGE2 and FOLFOX induced YB-1 expression. CICs isolated from SW948-S and SW948-FR that were previously transfected with either v6 Mu1 or vector for 48 h and then treated with or without synthetic PGE2 or FOLFOX. YB-1 expression levels were determined by immunoblotting with anti-YB-1 antibody; β-tubulin as loading control. (E) Effects are shown of PGE2/EP1 receptor, and of mTOR signaling on YB-1 expression. CICs were either transfected with in v6 Mu1 or vector for 48 h and then treated with or without EP1 inhibitor (5 µM AH6809) or mTOR inhibitor (10 nM PP242) for 2 h. They were then cultured in serum-free medium for 16 h and treated with synthetic PGE2 at 5 μM, or 1 × FOLFOX for 24 h. The cell lysates were processed for YB-1 and β-tubulin (as loading control). PGE2 secretion data in Figure 3B represent means ± SD; n = 4–6; * p < 0.05 compared to either vehicle control, vector + FOLFOX treatment group, or vector control group.

Figure 4

Requirement of CD44v6 and YB-1 for the stemness of SW948-FR CICs after acquisition of FOLFOX resistance. (A) The effects are shown of CD44v6 knockout on cell viability (using ATP Glo method) of CD44v6 Mu1, CD44v6 WT, and on CD44v6-rescue SW948-FR/CICs (1 × 10³ cells/well of 96-well plate) which were cultured for the indicated times. The time point 0 h represents the number of inoculated cells (data not shown). (B) The influence of CD44v6 on cell cycle of CD44v6 Mu1, CD44v6 WT, and CD44v6-rescue SW948-FR CICs are shown. Cells (1 × 10⁴) were cultured for 48 h, and the percentage of cells in G1, S, and G2 phases of cell cycles were examined with flow cytometry. (C,D) Indicated SW948-FR CICs were seeded into a 6-well plate at 1 × 10⁵ cells/well and cultured for 48 h. The extents of apoptosis of CICs in the cultures were examined by flow cytometry. (E) A ChIP assay was performed with chromatin from SW948-FR CICs using an anti-CD44v6 antibody. The immunoprecipitated DNA was amplified by PCR and subcloned. A total of 13 clones were sequenced. Computer-based analysis revealed the presence of various consensus binding sites for common transcription factors in these DNA sequences (see Supplemental Table S2). QPCR analyses show the expressions of these 13 transcription factors in CICs of “S’ and “FR” cells of SW948. (F) Expressions are shown for antiapoptosis/stemness-related genes in v6 WT CICs, v6 Mu1 CICs, YB-1 WT overexpressed v6 Mu1 CICs, and in CD44v6 rescue plasmids overexpressed v6 Mu1 CICs. QPCR was conducted to detect the expression levels of the genes. (G,H) Influence of v6-rescue plasmid into v6 Mu1 CICs for stemness-related gene expressions (G) and stemness-related TFs (c-Myc, TWIST1, OCT4, and SOX2) (H) were measured by QPCR. (I,M) The CD44v6-WT plasmid, or the constructs of the “CTOS” (c-Myc, TWIST1, OCT4, and SOX2) TFs either alone or with CD44v6-rescue plasmid, were transfected into the v6 Mu1 SW948-FR CICs. At 72 h after transfection, QPCR (I) or Western blot (M) analyses were done to examine the expression levels of these transcription factors. (J,K,L,N,O) Effects of simultaneous expressions of the “CTOS” with the CD44v6-rescue plasmid are shown on sphere formation (J), on the expressions of indicated stemness-related genes and protein (K,L), and on differentiation-related genes and protein (N,O) in v6 Mu1 SW948-FR CICs. The experiments were biologically repeated for three times. QPCR data represent means ± SD; n = 4–6; * p < 0.05 compared to vector control, WT control, or vehicle cell control group. Western blot data are representative of three experiments.

Figure 5

CD44v6-YB-1 defines CIC-like SP cells. (A) SW948-FR cells labeled with Hoechst 33342 showed 3.3% of SP cells in the SP gated region. Following treatment with verapamil, the SP cells were reduced to 0.26%. (B) v6 Mu1 and YB-1 Mu3 regulate the side population. The SP cells were <10% of vector cells in the v6 Mu1 cells, and the YB-1-Mu3 SW948-FR cells. (C) Cell proliferation rates were measured by ATP GLO assay for SP and non-SP cells. SP cells underwent rapid proliferation compared with non-SP cells. (D) SP cells exhibited high resistance to 1 × FOLFOX whereas the non-SP cells were sensitive to 1 × FOLFOX. (E) CD44v6 and YB-1 knockdown in SW948-FR cells decreased the drug resistance. Control, v6 Mu1, and YB-1 Mu3 SW948-FR cells were treated with various doses of FOLFOX for 10 days in 3% FBS DMEM. Cell viability was assessed using the clonogenic assay. The clonogenicity of CD44v6-Mu1 cells was significantly decreased compared with YB-1-Mu3 cells. (F) Expressions of core stemness genes in SW948-FR/SP and non-SP cells by QPCR are shown. (G) Expressions of anti-apoptosis/stemness-related genes in vector and v6 Mu1 transfected SW948-FR/SP cells are shown. (H) Expressions of CRC differentiation genes in SW948-FR/SP and non-SP cells are shown. Each bar represents the means of three determinations ± SD. * p < 0.05 among the indicated groups compared to respective control group. FACs data are representative of three experiments.

Figure 6

Nuclear YB-1 binds with CD44v6 and modulates CD44v6 and MDR1 transcription by binding to CD44v6 and MDR1 promoters. (A) Nuclear extracts were isolated from SW948-S and SW948-FR cells first transfected with vector, or v6 Mu1, and then treated with or without 1 × FOLFOX for 8 h. Nuclear extracts were immunoprecipitated by YB-1 antibody followed by Western blotting of the indicated proteins. (B) The scheme shows the MDR1 promoter constructs with YB-1 binding sites (mdr1(1) and mdr1(2)). (C) MDR1 Luciferase activities are shown for SW948-FR/CICs overexpressing CD44v6 Mu1, or YB-1 Mu3, or vector (Control) for 24 h. (D,F) MDR1 is transcriptionally regulated by YB-1 in SW948-FR/CICs. (D) The sketch map shows the predicted YB-1 binding sites (CAAT or ATTG) within the MDR1 promoter (MDR1(A) and MDR1(B)). PCR primers designated for MDR1(A) and MDR1(B) were used for amplification of the potential YB-1 binding sites of the MDR1 gene by ChIP semiquantitative PCR assays using anti-CD44v6, anti-YB-1, or an irrelevant IgG antibody (control). Total genomic DNA was used as input for the ChIP PCR. (E) ChIP QPCRs representing the PCR products in CD44v6, YB-1, or IgG immunoprecipitated DNA versus 10% input DNA in SW948-FR/CICs using primers for MDR1(A) and MDR1(B) sites are shown. (F) ChIP QPCRs representing the PCR products in CD44v6, YB-1, or IgG in SW948-FR/CICs overexpressing CD44v6 Mu1, or YB-1 Mu3, or vector for 24 h are shown. (G) The sketch map of predicted YB-1 binding sites (CD44v6 [1] and CD44v6 [2]) within the CD44v6 promoter is shown. (H) CD44v6 luciferase assays are shown for SW948-FR/CICs overexpressing v6 Mu1, or YB-1 Mu3, or vector for 24 h. (I) Semiquantitative PCR products using ChIP QPCR primers are shown for designated YB-1 binding sites as CD44v6(A) and CD44v6(B). (J) A representative ChIP QPCR representing the PCR product in immunoprecipitated CD44v6 is shown. (K) ChIP QPCR using PCR primers for designated CD44v6(A) sites were used for amplification of the YB-1 binding sites of the CD44v6 gene in ChIP assays in untreated CICs, or CICs overexpressing CD44v6 Mu1, or YB-1 Mu3, or vector (Control) for 24 h. Values represent means ± SD; n = 3–5; * p < 0.05, compared to whole cell lysate and nuclear fractions isolated from sensitive cells, sensitive cell groups, vehicle control, IgG control, vector control, and appropriate control groups. Western blot and semiquantitative PCR data are representative of three experiments.

Figure 7

Role of CD44v6 in the SQ tumorigenesis of SW948-FR/CICs in vivo. (A,B) Effects of CD44v6, or YB-1, or a combination of CD44v6 + YB-1 knockout on tumor growth in nude mice are shown. CD44v6 Mu1, YB-1 Mu3, or v6 Mu1 + YB-1 Mu3 knockout FOLFOX resistant CRC CICs and wide-type CICs were injected into nude mice. The tumor volumes in mice were measured every five days (A). Sixty days later, the mice were sacrificed. A solid tumor was collected from each mouse. (B) The impacts of v6 Mu1, YB-1 Mu3 knockout FOLFOX resistant CRC CICs on tumor weights are shown. (C) The CD44v6, MDR1, and YB-1 protein levels in tumors of mice injected with v6 Mu1, YB-1 Mu3 knockout FOLFOX resistant CRC CICs or WT CICs are shown. β-Tubulin was used as an internal control. (D) Expressions of proliferation/antiapoptosis/invasion/stemness related genes (by QPCR) in these solid tumors are shown. Data are presented as mean ± SD (n = 7); * p < 0.05. ANOVA followed by Bonferroni’s post-hoc test was used to assess the significance. Western blot data are representative of three experiments. QPCR data are presented as mean ± SD (n = 4); * p < 0.05. Scale bar, 50 μm.

Figure 8

Proposed model for a positive feedback loop coupling YB-1 activation and CD44 alternate splicing and CD44v6 then sustains cancer initiating cell proliferation and stemness.