Inhibition of DDR1-BCR signalling by nilotinib as a new therapeutic strategy for metastatic colorectal cancer

Abstract

The clinical management of metastatic colorectal cancer (mCRC) faces major challenges. Here, we show that nilotinib, a clinically approved drug for chronic myeloid leukaemia, strongly inhibits human CRC cell invasion in vitro and reduces their metastatic potential in intrasplenic tumour mouse models. Nilotinib acts by inhibiting the kinase activity of DDR1, a receptor tyrosine kinase for collagens, which we identified as a RAS-independent inducer of CRC metastasis. Using quantitative phosphoproteomics, we identified BCR as a new DDR1 substrate and demonstrated that nilotinib prevents DDR1-mediated BCR phosphorylation on Tyr177, which is important for maintaining β-catenin transcriptional activity necessary for tumour cell invasion. DDR1 kinase inhibition also reduced the invasion of patient-derived metastatic and circulating CRC cell lines. Collectively, our results indicate that the targeting DDR1 kinase activity with nilotinib may be beneficial for patients with mCRC.

Keywords: collagen receptor; colorectal cancer; invasion; targeted therapy; tyrosine kinase.

© 2018 The Authors. Published under the terms of the CC BY 4.0 license.

Figures

Figure 1. Nilotinib has anti‐metastatic activity in CRC

1. Nilotinib anti‐invasive activity in a panel of CRC cell lines. Percentage of cell inhibition in Boyden chambers containing 1 mg/ml Matrigel for the indicated CRC cell lines incubated with 100 nM nilotinib (mean ± SEM; n = 3).

2. Nilotinib activity in Boyden chambers and spheroid assays. HCT116 cells incubated with 100 nM nilotinib or DMSO (control) were seeded in the upper compartment of a Boyden chamber containing 1 mg/ml Matrigel for 24 h or embedded as spheroids in collagen I matrix for 72 h and then imaged by phase‐contrast microscopy. The histogram shows the percentage of migrating cells in the collagen I matrix normalized to control condition set at 100% (mean ± SEM; n = 3 independent experiments with six replicates; *P < 0.05 Student's t‐test). The quantification of nilotinib anti‐invasive activity in HCT116 cells is shown in panel (C).

3. Dose–response curve of nilotinib effect on CRC cell invasion. Percentage of cell invasion (relative to control) in Boyden chambers containing 1 mg/ml Matrigel of HT29 (▲) and HCT116 (■) CRC cells treated with the indicated concentrations of nilotinib (mean ± SEM; n = 3).

4. Nilotinib anti‐metastatic activity in nude mice. HCT116 cells were injected in the spleen of nude mice (n = 11/group). After 4 weeks of nilotinib treatment, livers were removed. Representative images of livers (black arrowheads indicate metastasis) and metastatic index of animals treated daily with vehicle or 50 mg/kg nilotinib (oral administration) (mean ± SEM; n = 11/group; *P < 0.05 Student's t‐test).

Source data are available online for this figure.

Figure EV1. Nilotinib inhibits collagen‐mediated DDR1 phosphorylation

1. A

   Collagen I induces DDR1 tyrosine phosphorylation (pY DDR1).

2. B, C

   Nilotinib inhibits DDR1 tyrosine phosphorylation. DDR1 tyrosine phosphorylation level was assessed by Western blotting in protein lysates from the indicated CRC cell lines after stimulation with 40 μg/ml collagen I for 18 h and incubation with the indicated concentrations of nilotinib and for the indicated times with 100 nM of nilotinib.

Figure 2. DDR1 promotes CRC metastasis formation

1. A–D

   DDR1 depletion by shRNA inhibits CRC cell invasion and metastasis. (A) DDR1 expression in CRC cells infected with vectors expressing the indicated shRNA was assessed by Western blotting. Invasion of infected CRC cells in Boyden chamber (B) and in collagen I matrix (C) (mean ± SEM; n = 3 (B) and n = 7 (C); **P < 0.01, ***P < 0.001 Student's t‐test). A representative image of cell migration for each cell line is shown. (D) HCT116 cells infected with the indicated vectors were injected in the spleen of nude mice (n = 12/group). After 4 weeks, livers were removed. A representative liver image for each group, the metastatic index and the ctDNA level (ng/ml of plasma) for each animal are shown (mean ± SEM; *P < 0.05; **P < 0.01 Student's t‐test).

2. E–G

   DDR1 overexpression promotes CRC cell invasion and metastasis formation. (E) DDR1 expression in CRC cells infected with the indicated viruses. (F) Invasion in Boyden chambers of CRC cells that were infected with the indicated viruses (mean ± SEM; n = 3; *P < 0.05 Student's t‐test). (G) SW620 cells infected with the indicated viruses were injected in the spleen of nude mice (n = 12 for the mock group and n = 20 for the DDR1 group). After 4 weeks, livers were removed. A representative liver image for each group, the metastatic index and the relative ctDNA level of each animal are shown (mean ± SEM; *P < 0.05 Student's t‐test).

Source data are available online for this figure.

Figure 3. Nilotinib anti‐tumour activity is mediated by DDR1 kinase inhibition in CRC

1. A

   Biochemical analysis of the nilotinib‐resistant DDR1 T701I mutant. Tyrosine phosphorylation and DDR1 expression level in HEK293T cells transfected with the indicated DDR1 constructs and stimulated with 40 μg/ml collagen I for 18 h in the presence or not of the indicated concentration of nilotinib.

2. B–D

   Nilotinib anti‐tumour activity is abrogated by DDR1 T701I expression in CRC cells. (B) Western blot analysis of DDR1 levels in HCT116 cells expressing the indicated shRNAs and infected with the indicated DDR1 constructs (wild type, WT; T701I; kinase dead, KD). (C) Invasion assays in Boyden chambers of the indicated cell lines after incubation with DMSO or 50 nM nilotinib (mean ± SEM; n = 4; **P < 0.01 Student's t‐test). (D) DDR1‐depleted HCT116 cells infected with the indicated DDR1 constructs were injected in the spleen of nude mice (n = 5/group) and treated daily with 50 mg/kg nilotinib, starting at day 1 post‐injection (oral administration). After 4 weeks, livers were removed. A representative image of liver for each group and the metastatic index of each animal are shown (mean ± SEM; *P < 0.05 Student's t‐test).

3. E, F

   Nilotinib inhibits DDR1‐mediated invasive and metastatic activity of CRC cells. (E) Invasion assays in Boyden chambers of the indicated SW620 cell lines that were incubated with DMSO or 100 nM nilotinib (mean ± SEM; n = 3; *P < 0.05; **P < 0.01 Student's t‐test). (F) After inoculation of SW620 cells that overexpress DDR1 in the spleen, nude mice (n = 21 for the DMSO group and n = 14 for the nilotinib group) were treated daily with DMSO or 50 mg/kg/d nilotinib as indicated, starting at day 7 post‐injection (oral administration). After 4 weeks, livers were removed. A representative image of liver for each group, the metastatic index and the relative ctDNA level of each animal are shown (mean ± SEM; *P < 0.05; **P < 0.01 Student's t‐test).

Source data are available online for this figure.

Figure EV2. RAS‐independent nature of DDR1 signalling

1. A, B

   Depletion of DDR1 abolishes (A) and overexpression of DDR1 induces (B) BCR but not MAPK or AKT phosphorylation. Tyrosine phosphorylation (pTyr) and phosphorylation levels of DDR1, BCR, MAPK and AKT were assessed by Western blotting in protein lysates from the indicated CRC cell lines stimulated or not with 40 μg/ml collagen I (Col I) for the indicated times.

2. C

   MAPK inhibition (treatment with 10 μM of PD98059 during 24 h) does not impact on DDR1 signalling.

3. D

   DDR1 signalling in CRC liver metastases. DDR1, BCR, MAPK and AKT activities from protein lysates of liver metastases described in Fig 3F.

Figure 4. Quantitative phosphoproteomic analyses identifies BCR as a novel DDR1 substrate

1. Workflow of the phosphoproteomic analysis of the DDR1 kinase signalling cascade in HCT116 CRC cells.

2. Western blot analysis of DDR1 tyrosine phosphorylation (pY) in HCT116 cells stimulated with 40 μg/ml collagen I for 18 h and incubated or not with 100 nM nilotinib for 1 h before cell lysis and immunoprecipitation (IP) with an anti‐DDR1 antibody.

3. Heatmap of the phosphotyrosine peptides in HCT116 cellular extracts for which the level was consistently modulated upon collagen I stimulation and by nilotinib treatment (two biological replicates).

4. Workflow of the phosphoproteomic analysis of the DDR1 kinase signalling cascade in liver metastases of the nude mice inoculated with DDR1 overexpressing SW620 cells described in Fig 3F.

5. Heatmap of phosphotyrosine peptides in liver metastasis protein extracts (nude mice described in Fig 3F) that were consistently reduced upon nilotinib treatment (three biological replicates).

Source data are available online for this figure.

Figure 5. BCR phosphorylation on Tyr177 mediates DDR1 invasive signalling

1. A–D

   BCR is a novel DDR1 substrate in CRC cells. (A) Incubation with 100 nM nilotinib and (B) shRNA‐mediated silencing of DDR1 reduce BCR phosphorylation at pTyr177 (pTyr177‐BCR) induced by collagen I stimulation (40 μg/ml for 18 h). (C) DDR1 overexpression in SW620 cells increases pTyr177‐BCR level induced by collagen I stimulation (40 μg/ml for 18 h). (D) pTyr177‐BCR induced by collagen I stimulation does not require ABL‐like activities. pTyr177‐BCR levels were assessed in HCT116 cells in which DDR1 was silenced by shRNAs and transfected with the indicated DDR1 constructs, stimulated or not with collagen I (40 μg/ml for 18 h) and incubated or not with 100 nM nilotinib.

2. E, F

   BCR modulates HCT116 cell invasion. (E) The level of BCR depletion upon infection with viruses expressing the indicated shRNAs was checked by Western blotting. (F) Invasion assays in Boyden chamber (left) and in collagen I matrix (right) of the indicated HCT116 cell lines (mean ± SEM; n = 3 and n = 7, respectively; **P < 0.01; ***P ≤ 0.001 Student's t‐test). A representative image of cell migration for each condition is shown.

3. G, H

   Tyr177‐BCR regulates DDR1 invasive activity. (G) BCR and pTyr177‐BCR levels were assessed in CRC cells expressing the indicated shRNAs and infected with viruses that express the indicated BCR constructs (wild type, WT; Y177F, YF) and stimulated or not with collagen I (40 μg/ml for 18 h). (H) Invasion assays in Boyden chambers of the indicated cell lines (mean ± SEM; n = 4; *P < 0.05; **P < 0.01 Student's t‐test).

Source data are available online for this figure.

Figure EV3. BCR phosphorylation is DDR1 kinase‐dependent

Western blotting to assess pY792 DDR1 and pY177 BCR levels in SW620 cells that overexpress wild‐type DDR1 (DDR1), a kinase‐dead DDR1 mutant (DDR1 KD) or the gatekeeper mutant DDR1 (DDR1 TI) or not (mock), after stimulation or not with 40 μg/ml collagen I for 18 h.

Figure 6. DDR1 kinase signalling increases β‐catenin transcriptional activity

1. A, B

   DDR1 kinase activity increases β‐catenin transcriptional activity. (A) HEK293 cells and (B) HCT116 cells were transfected with TOPflash, mutant control FOPflash, Renilla luciferase and the indicated constructs and incubated (+) or not (−) with 100 nM nilotinib. The luciferase activity normalized to control condition is shown (mean ± SEM; n = 2 independent experiments with three replicates; *P < 0.05; **P < 0.01 Mann–Whitney test). Western blot analyses confirmed the expression of the different DDR1 variants and of β‐catenin.

2. C

   DDR1 kinase activity increases the transcript level of selected β‐catenin target genes in CRC cells. Relative transcript level of the indicated genes in SW620 cells infected with viruses that express the indicated DDR1 variants and incubated or not with 100 nM nilotinib as shown (mean ± SEM; n = 2 independent experiments with three replicates; ns: not significant; *p < 0.05 **P < 0.01; ***P < 0.001 Mann–Whitney test).

3. D

   DDR1 regulates the transcript level of selected β‐catenin target genes in CRC cells. Relative transcript level of the indicated genes in HCT116 cells infected with viruses that express that indicated shRNAs (mean ± SEM; n = 2 independent experiments with three replicates; ns: not significant; *P < 0.05; **P < 0.01; Mann–Whitney test).

4. E

   DDR1 signalling increases β‐catenin nuclear activity in CRC cells. Immunofluorescence analysis of active β‐catenin in SW620 cells infected with indicated viruses and stimulated overnight or not with collagen I (50 μg/ml), Wnt3A (200 ng/ml) and treated or not with nilotinib (100 nM) as indicated. A representative example (left panel) and quantification (right) of cells with active nuclear β‐catenin (mean ± SEM; n = 3; ns: not significant; *P < 0.05 ***P < 0.01; Mann–Whitney test). Scale bar: 100 μm.

5. F

   DDR1 signalling increases β‐catenin nuclear activity in metastatic CRC. IHC analysis of active β‐catenin level in experimental metastatic tumours described in Fig 3F. A representative example (left) and quantification (right) of CRC cells with active nuclear β‐catenin (mean ± SEM; n = 5 tumours per group with 4 fields per tumour; ns: not significant; *P < 0.05 ***P < 0.001; Mann–Whitney test). Scale bar: 250 μm.

6. G

   DDR1 invasive activity in CRC cells requires proper β‐catenin expression. Invasion assays in Boyden chambers of SW620 cells infected with the indicated viruses and incubated with DMSO or with 0.1 mM of the β‐catenin pharmacological inhibitor IWR‐1‐endo (mean ± SEM; n = 3; *P < 0.05 Student's t‐test).

Source data are available online for this figure.

Figure EV4. DDR1 regulates β‐catenin signalling in CRC cells

1. A

   Relative transcript level of the indicated genes in HCT116 cells expressing DDR1 or Ctrl shRNA (mean ± SEM; n = 2; *P < 0.05 **P < 0.01; Mann–Whitney test).

2. B, C

   Representative images (left panels) and quantifications (right panels) of active β‐catenin nuclear staining in control or DDR1‐overexpressing SW620 cells (B) and derived liver metastasis (C). Are shown, respectively, means ± SEM; n = 2 (B) and n = 5 tumours per group with 2 fields per tumour (C); *P < 0.05 **P < 0.01 ***P < 0.001 Mann–Whitney test. Scale bars: 250 μm.

Figure EV5. Anti‐migratory activity of BCR

1. BCR and pTyr177‐BCR levels were assessed in CRC cells expressing both indicated shRNAs and BCR constructs (wild type, WT; Y177F, YF) and stimulated or not with collagen I (40 mg/ml for 18 h).

2. Migration assays in Boyden chambers of the indicated cell lines (mean ± SEM; n = 5; *P < 0.05 Student's t‐test).

Figure 7. Nilotinib inhibits DDR1 invasive activity of patient‐derived CRC cells

1. A

   Patients with CRC showing high DDR1 expression have shorter progression‐free survival (PFS) and overall survival (OS). Kaplan–Meier analysis using data from 143 patients with stage IV CRC subdivided according to the tumour DDR1 expression level (high/low).

2. B

   Increased DDR1 activity in metastatic nodules from patients with CRC. DDR1 activity was evaluated based on the relative DDR1 tyrosine phosphorylation level measured by immunoprecipitation (IP) followed by Western blotting of protein lysates of matched healthy tissue, primary tumour and metastatic lesions from 18 patients with CRC. A representative example for one patient (upper panels) and the relative level of DDR1 tyrosine phosphorylation in each patient quantified by ImageJ (lower histogram) are shown (mean ± SEM; *P < 0.05 Student's t‐test).

3. C

   DDR1 activity in patient‐derived CRC lines stimulated or not with collagen I (40 μg/ml for 18 h) and incubated or not with 100 nM nilotinib (red asterisks indicate DDR1 activation).

4. D, E

   DDR1 invasive activity in patient‐derived CRC lines. Cell invasion was assessed in Boyden chambers after incubation with the indicated doses of nilotinib (D) or transfection with the indicated siRNAs (mean ± SEM; left panels n = 3 and right panels n = 6; *P < 0.05; **P < 0.01 Student's t‐test). (E) Level of DDR1 depletion upon siRNA transfection.

5. F

   DDR1 signalling in patient‐derived CRC cell lines stimulated or not with collagen I (40 μg/ml for 18 h) and treated or not with nilotinib (100 nM) as shown.

6. G, H

   DDR1 invasive activity in patient‐derived CRC cell lines. Cell invasion was assessed in Boyden chambers after incubation with the indicated doses of nilotinib (G) or transfection with the indicated siRNAs (mean ± SEM; n = 4; *P < 0.05; **P < 0.01 Student's t‐test). (H) Level of DDR1 depletion upon siRNA transfection.

7. I

   Nilotinib inhibits metastatic activity of CPP30 cells. After inoculation of CPP cells in the spleen, nude mice (n = 10/group) were treated daily with DMSO or 50 mg/kg nilotinib (i.p.) as indicated, starting at day 6 post‐injection. After 49 days, livers were removed. A representative image of liver for each group and the metastatic index of each animal are shown (mean ± SEM; *P < 0.05 Student's t‐test).

Source data are available online for this figure.

Figure 8. Proposed model for kinase‐dependent DDR1 invasive activity in CRC cells

1. Collagen‐stimulated DDR1 activation induces the phosphorylation of BCR on Tyr177, which disrupts BCR/β‐catenin interaction. This signalling cascade results in an increased β‐catenin nuclear activity leading to expression of target genes necessary for cell invasion.

2. Inhibition of DDR1 kinase activity with nilotinib decreases CRC cell invasion by reducing this β‐catenin‐dependent signalling cascade necessary for cell invasion.