The extracellular matrix and focal adhesion kinase signaling regulate cancer stem cell function in pancreatic ductal adenocarcinoma

Abstract

Cancer stem cells (CSCs) play an important role in the clonogenic growth and metastasis of pancreatic ductal adenocarcinoma (PDAC). A hallmark of PDAC is the desmoplastic reaction, but the impact of the tumor microenvironment (TME) on CSCs is unknown. In order to better understand the mechanisms, we examined the impact of extracellular matrix (ECM) proteins on PDAC CSCs. We quantified the effect of ECM proteins, β1-integrin, and focal adhesion kinase (FAK) on clonogenic PDAC growth and migration in vitro and tumor initiation, growth, and metastasis in vivo in nude mice using shRNA and overexpression constructs as well as small molecule FAK inhibitors. Type I collagen increased PDAC tumor initiating potential, self-renewal, and the frequency of CSCs through the activation of FAK. FAK overexpression increased tumor initiation, whereas a dominant negative FAK mutant or FAK kinase inhibitors reduced clonogenic PDAC growth in vitro and in vivo. Moreover, the FAK inhibitor VS-4718 extended the anti-tumor response to gemcitabine and nab-paclitaxel in patient-derived PDAC xenografts, and the loss of FAK expression limited metastatic dissemination of orthotopic xenografts. Type I collagen enhances PDAC CSCs, and both kinase-dependent and independent activities of FAK impact PDAC tumor initiation, self-renewal, and metastasis. The anti-tumor impact of FAK inhibitors in combination with standard chemotherapy support the clinical testing of this combination.

Fig 1. Type I collagen- β1 integrin signaling enhances the clonogenic growth of PDAC cells.

(a) Colony formation by BxPC-3, Capan-1 and MIA PaCa-2 cells following growth on type I collagen, fibronectin, or laminin for 96 hours. Data represent the mean ± SD (n = 4) compared to control cell growth on plastic; **P < 0.001; ***P < 0.0001 by ANOVA. (b) Colony formation assay by cells from 2 distinct patient derived xenografts cells cultured on plastic or type I collagen for 96 hours. (c) Secondary colony formation by MIA PaCa-2, and Capan-1 cells. Data represent the mean ± SD (n = 4) compared to control; *P < 0.05, **P < 0.001. (d) Primary and secondary colony formation by MIA PaCa2 cells expressing a scrambled control (shCtrl) or β1 integrin (shBeta1) shRNA after culture on type I collagen for 96 hours. Data represent mean ± SD (n = 4). **P < 0.001.

Fig 2. FAK overexpression increases PDAC clonogenic growth and tumor initiating capacity.

(a) Ratio of phospho-FAK to total FAK expression by MIA PaCa2 cells expressing a scrambled control (shCtrl) or β1 integrin (shBeta1) shRNA and cultured on plastic or type I collagen for 96 hours. (b) Colony formation by MIA PaCa-2 and Capan-1 cells overexpressing FAK-FL following growth on plastic or type I collagen for 96 hours. Data represent mean ± SD (n = 4) of control versus FAK-FL; *P < 0.05; **P < 0.001; ***P < 0.0001. (c) Tumor growth of MIA PaCa2 cells overexpressing FAK-FL following subcutaneously injection in to NSG mice. *P = 0.04 compared to control (vector).

Fig 3. Type I collagen-β1 integrin-FAK signaling impacts ALDH⁺ PDAC CSCs.

(a) Frequency of ALDH⁺ MIA PaCa-2, Capan-1, and BxPC-3 cells following culture on plastic or type I collagen for 96 hours. Data represent mean ± SD (n = 3) compared to cell growth on plastic; *P < 0.05, **P < 0.001, ***P < 0.0001. (b) Frequency of α2β1⁺ cells within ALDH⁺ or all tumor cells from 3 distinct patient derived xenografts. (c) Frequency of phospho-FAK⁺ cells within ALDH⁺ or all tumor cells from 2 distinct patient derived xenografts as determined by flow cytometry.

Fig 4. FAK kinase-inhibition decrease clonogenic PDAC growth in vitro and inhibits tumor growth in vivo.

(a) In vitro colony formation by MIA PaCa-2 and Capan-1 cells following treatment with vehicle control (DMSO) or VS-4718 on type I collagen for 96 hours. Data represent mean ± SD (n = 3) of DMSO versus VS-4718; *P < 0.05. (b) Colony formation by 2 distinct patient derived xenograft cells following treatment with vehicle control (DMSO) or VS-4718 on type I collagen for 5 days. (c) Subcutaneous tumor growth of a patient derived xenograft (JH102) following treatment with vehicle control, VS-4718, gemcitabine plus nab-paclitaxel (Gem-Pac), or all three drugs together (Gem-Pac-VS). Five mice were included in each group. **P = 0.0076, *P = 0.03 by ANOVA. (d) Colony formation by MIA PaCa-2 and Capan-1 cells overexpressing FAK-Y397F following growth on type I collagen for 96 hours. Data represent mean ± SD (n = 4) of control versus FAK-Y397F; *P < 0.05, **P < 0.001.

Fig 5. The loss of FAK inhibits self-renewal.

(a) Frequency of ALDH⁺ MIA PaCa-2 and Capan-1 cells expressing scrambled control (shCtrl) or FAK shRNA (shFAK) following treatment with doxycycline for 96 hours. Data represent mean ± SD (n = 3) of shCtrl versus shFAK; *P < 0.05. (b) In vitro colony formation by MIA PaCa-2 and Capan-1 cells expressing scrambled control (shCtrl) or FAK shRNA (shFAK) following treatment with doxycycline for 96 hours on type I collagen. Data represents ± SD (n = 4) of control versus hairpin; **P < 0.001. (c) In vivo subcutaneous tumor growth by MIA PaCa-2 cells expressing scrambled control (shCtrl, n = 4) or FAK shRNA (shFAK, n = 7) following treatment with doxycycline. Error bar represents SD; *P = 0.02.

Fig 6. The loss of FAK inhibits PDAC cell migration in vitro and metastasis in vivo.

(a) In vitro migration by MIA PaCa-2 and Capan-1 cells expressing scrambled control (shCtrl) or FAK shRNA following treatment with doxycycline on type I collagen for 96 hours. Data represent mean ± SD (n = 4) of control versus hairpin; **P < 0.001. (b) In vivo metastases by orthotopic MIAPaCa-2 tumors expressing scrambled control (shCtrl) or FAK shRNA (shFAK) following treatment with doxycycline. Seven mice were used in each group. Control versus hairpin; *P = 0.008.