Targeting Pancreatic Cancer Cell Stemness by Blocking Fibronectin-Binding Integrins on Cancer-Associated Fibroblasts

Abstract

Abstract: Cancer-associated fibroblasts (CAF) generate an extracellular matrix (ECM) which provides a repository for factors that promote pancreatic cancer progression. In this study, we establish that CAF contribution to pancreatic tumor initiation, i.e., stemness, depends on fibronectin (FN) as a scaffold required for assembly of a collagen-containing fibrotic ECM with a critical dependence on the FN-binding integrins, α5β1 and αvβ3. CAF matrix assembly can be prevented by knockdown of FN, integrin α5, or integrin β3 or by a bispecific antibody with dual recognition of α5β1 and αvβ3 that can also destabilize a preexisting matrix. In mice, the ability of CAFs to produce a stiff collagenous matrix and accelerate tumor initiation can be blocked by knockdown of FN or FN-binding integrins or systemic treatment with the α5β1/αvβ3 bispecific antibody. Together, these results reveal that dual targeting of the FN-binding integrins, α5β1 and αvβ3, can block the ability of CAFs and their matrix to enhance pancreatic cancer stemness and progression.

Significance: Simultaneous targeting of two integrins that function as receptors for FN, a protumor ECM protein, can prevent fibroblasts from supporting the malignant behavior of pancreatic cancer cells.

Figure 1

Tumor cells utilize CAF-produced FN to overcome isolation stress. A, CAFs enhance tumor initiation via FN (orthotopic pancreas cancer model). Luciferase-expressing KP4 PDAC cells were orthotopically injected (with or without CAF-1299) into the pancreas of nu/nu mice. After 3 weeks, luciferase imaging was performed to survey tumor establishment in the pancreas. B, FN knockdown prevents CAF assembly of COL fibers. CAF-1299 and CAF-1424 cells were treated with siRNA for a scramble control vs. FN. After 72 hours, IF staining shows FN and COL content. Images are representative of at least three independent experiments. C, Knockdown of FN in CAFs increases soluble COL in media. CAF-1299 cells were cultured for 72 hours, and then the conditioned media and cell lysates were collected and processed for immunoblotting to confirm FN knockdown and assess the level of soluble COL secreted by the cells into the media. si-CTRL, siRNA Universal Negative Control; si-FN, siRNA-mediated knockdown of FN1.

Figure 2

FN-binding integrins mediate ECM production. Knockdown of FN or FN-binding integrins in CAFs prevents the formation of FN and COL fibers. CAF-1299 cells were treated with siRNA for a scramble control vs. FN, ITGA5, or ITGB3. After 72 hours, IF staining shows FN and COL content. The top set of images were not permeabilized. Images are representative of at least three independent experiments. Blots confirm knockdown. si-CTRL, CAFs with siRNA for a Universal Negative Control; si-FN, CAFs with siRNA-mediated knockdown of FN1; si-α5, CAFs with ITGA5 knockdown; si-β3, CAFs with ITGB3 knockdown.

Figure 3

BsAb improves the targeting of FN-binding integrins on CAFs. A, BsAb for dual recognition of αvβ3 and α5β1 heterodimers. Schematic depicts the design of a novel BsAb with monovalent recognition of two antigens, integrins αvβ3 and α5β1. The Fab domains for the BsAb are identical to the “control” bivalent (i.e., monospecific) mAb recognizing αvβ3 (derived from etaracizumab) and α5β1 (derived from volociximab). B, BsAb binding to cells compared with commercial and control antibodies. Flow cytometry plots show the binding of each antibody to CAF-1299 and CAF-1424. C, BsAb shows improved binding to CAFs compared with constituent mAbs. The graph shows median fluorescence intensity (MFI) for antibody binding to CAF-1299 and CAF-1424 cells. Bars and error bars represent SD from n = 3 independent experiments.

Figure 4

Dual blockade of integrins αvβ3/α5β1 can prevent and reverse CAF-ECM assembly. A, Function-blocking antibodies targeting FN receptors prevent CAF assembly of FN/COL fibers. CAF-1299 cells were incubated with control IgG vs. indicated antibodies for 72 hours and then processed for immunostaining to examine FN and COL. Images are representative of at least three independent experiments. The graph shows the quantification of staining as mean ± SD for each marker that was measured as % area for each experiment and then normalized to IgG control. *, P < 0.05 using one-sample t test. B, Integrin-targeted antibody can disrupt preexisting CAF-produced ECM. CAF-1299 were plated and allowed to produce ECM for 72 hours before adding control IgG vs. indicated antibodies for an additional 72 hours. Samples were then processed for immunostaining to examine FN and COL. The graph shows the mean ± SD staining for each marker measured as % area in each experiment and normalized to IgG control. *, P < 0.05 using the one-sample t test. C, BsAb prevents upregulation of connective tissue growth factor (CTGF), a protein that regulates cell proliferation, migration, and adhesion. CAF-1299 cells were plated and allowed to produce ECM for 72 hours and then treated with antibody for another 72 hours before cells were removed to leave behind a cell-free ECM atop which PANC1 cells were then plated. After 24 hours, the PANC1 cells were lysed and prepared for immunoblotting to detect the protein expression of CTGF. Blots are representative of at least three independent experiments. Tx, treatment.

Figure 5

Disrupting FN-binding integrins prevents the ability of CAFs to enhance tumor initiation. A, CAFs enhance tumor initiation via FN and FN-binding integrins (subcutaneous xenograft model). PANC1 human PDAC cells were injected subcutaneously (with or without CAF-1299) into the flank areas of nu/nu mice. Mice were monitored twice weekly to detect the earliest emergence of palpable tumors. The graph shows tumor take rate vs. time for 10–12 mice per group, using a volume of 100 mm³ (computed as length × width2) as the threshold for tumor take. At the endpoint of the experiment (day 56), tumors were harvested and prepared for histologic analysis. B, BsAb treatment reduces tumor stroma and increases necrosis. Tumor sections were stained using hematoxylin and eosin. Areas of stroma (S) and necrosis (N) are noted. C, BsAb treatment reduces tumor stiffness. Cryosections of tumors were analyzed using atomic force microscopy to evaluate tissue stiffness. Dots depict the mean value for each 20 × 20 μm region of interest (ROI). Fifteen ROIs were evaluated per tumor, for two tumors per group.

Figure 6

BsAb treatment reduces the fibrotic effect of human CAFs co-injected with tumor cells. A–D, BsAb-treated tumors contain less CAF-produced FN, fibrosis, and angiogenesis. Serial formalin-fixed, paraffin-embedded sections of tumors were processed for IHC detection of hFN (A) or CD31 (D), shown in brown. Mason's trichrome (B) and picrosirius red (PSR; C) histologic stains were used to visualize COL. Graphs depict the quantification of IHC staining using QuPath, with each dot representing the mean value for each tumor slice examined. P values were computed using the Student t test.

Figure 7

CAF-produced FN boosts tumor initiation (summary schematic). CAFs produce a dense and reactive stroma that supports tumor initiation and progression. This project establishes the link between cell surface integrins and FN as a lynchpin for the construction of protumor ECM that can account for the ability of CAFs to support tumor initiation for multiple in vitro and in vivo models. Knockdown or antibody blockade of integrins αvβ3 and/or α5β1 can disrupt FN fiber assembly. FN fibers are known to function as a scaffold that mediates the assembly of other ECM proteins and the incorporation of protumor-secreted factors. Knockdown of αvβ3/α5β1 or their dual blockade using a novel BsAb can disrupt this cascade by preventing the initial assembly of FN fibers. This approach has broad potential as a novel strategy to target the aberrant fibrosis that exacerbates the progression of cancer and fibrotic disease.