Matricellular Protein WISP2 Is an Endogenous Inhibitor of Collagen Linearization and Cancer Metastasis

Abstract

Collagen remodeling contributes to many physiologic and pathologic processes. In primary tumors, the linearization of collagen fibers promotes cancer cell invasion and metastasis and is indicative of poor prognosis. However, it remains unknown whether there are endogenous inhibitors of collagen linearization that could be exploited therapeutically. Here, we show that collagen linearization is controlled by two secreted matricellular proteins with antagonistic functions. Specifically, WISP1 was secreted by cancer cells, bound to type I collagen (Col I), and linearized Col I via its cysteine-rich C-terminal (CT) domain. In contrast, WISP2, which lacks a CT domain, inhibited Col I linearization by preventing WISP1-Col I binding. Analysis of patient data revealed that WISP2 expression is lower in most solid tumors, in comparison with normal tissues. Consequently, genetic or pharmacologic restoration of higher WISP2 levels impaired collagen linearization and prevented tumor cell invasion and metastasis in vivo in models of human and murine breast cancer. Thus, this study uncovers WISP2 as the first inhibitor of collagen linearization ever identified and reveals that collagen architecture can be normalized and metastasis inhibited by therapeutically restoring a high WISP2:WISP1 ratio. SIGNIFICANCE: Two secreted factors, WISP1 and WISP2, antagonistically regulate collagen linearization, and therapeutically increasing the WISP2:WISP1 ratio in tumors limits collagen linearization and inhibits metastasis.See related commentary by Barcus and Longmore, p. 5611.

Figure 1.. WISP2 inhibits WISP1-induced collagen linearization.

A, Scanning electron microscopy of Col I lattices formed in the presence of PBS (Col I), 50 μg/mL recombinant WISP1, 100 μg/mL recombinant WISP2, or 50 μg/mL WISP1 + 100μg/mL WISP2 (1:3 WISP1:WISP2 molar ratio). Magenta arrows, examples of “knot-like” structures; yellow arrows, examples of “hairpin- or end-like structures”. Scale bars, 2 μm. B, Curvature ratios of Col I fibrils in lattices from A (n=15, 3 independent experiments, 5 images/lattice). C, Binding of WISP1 to fibrillar Col I (or BSA as negative control) as determined by solid-phase binding assay and detected with an anti-WISP1 antibody (n=6, from 3 independent experiments). D, Binding of WISP2 to fibrillar Col I (or BSA as negative control) as determined by solid-phase binding assay and detected with an anti-WISP2 antibody (n=4, from 2 independent experiments). E, Inhibition of WISP1 (1 μM) binding to fibrillar Col I in the presence of increasing concentrations of WISP2 (n=6, from 3 independent experiments). F, Binding of pre-bound WISP1 (1 μM) to fibrillar Col I in the presence of increasing concentrations of WISP2 (n=6, from 3 independent experiments). G, Binding of pre-bound WISP2 (2 μM) to fibrillar Col I in the presence of increasing concentrations of WISP1 (n=4, from 2 independent experiments). H, Binding of WISP1 to WISP2 (or BSA as negative control) as determined by solid-phase binding assay (n=4, from 2 independent experiments). (B-H) mean ± SEM, (B) one-way ANOVA followed by Tukey’s posttest. ns, P>0.05; ***, P<0.001.

Figure 2.. WISP2 inhibits WISP1-induced cell invasion through Col I.

A-C, WISP2 (A) or WISP1 (B) gene expression levels, or WISP2/WISP1 gene expression ratio (C) in tumors and adjacent normal tissues from patients with different cancer types (BRCA, breast invasive carcinoma, n=113 normal and n=1119 tumors; COAD, colon adenocarcinoma, n=41 normal and n=483 tumors; LUAD, lung adenocarcinoma, n=59 normal and n=541 tumors; LUSC, lung squamous cell carcinoma, n=51 normal and n=502 tumors; PRAD, prostate adenocarcinoma, n=52 normal and n=502 tumors; READ, rectum adenocarcinoma, n=10 normal and n=167 tumors). D, Migratory tracks of 4T1 cells plated on Col I (n=72 cells), Col I + WISP1 (n=77 cells), Col I + WISP2 (n=72 cells), or Col I + WISP1 + WISP2 (n=62 cells) lattices. E, Average cell speed from cells tracked in D. F, Migratory tracks of 4T1 cells plated on plastic in presence of PBS vehicle control (n=69 cells), WISP1 (n=85 cells), WISP2 (n=81 cells), or WISP1 + WISP2 (n=65 cells). G, Average cell speed from cells tracked in F. H, Invasion of 4T1 cells through Col I layered on Transwell inserts, in the presence of WISP1, WISP2, or WISP1 + WISP2 (1:3 WISP1:WISP2 molar ratio; n=6 biological replicates). I, Migration of 4T1 cells through the microporous membrane of Transwell inserts (in the absence of Col I) in the presence of PBS (vehicle), WISP1, WISP2, or WISP1 + WISP2 (n=6 biological replicates). J, Proliferation rate of 4T1 cells plated on Col I, Col I + WISP1, or Col I + WISP2 lattices (n=4 biological replicates). K, Invasion of MDA-MB-231 cells through Col I layered on Transwell inserts, in the presence of WISP1, WISP2, or WISP1 + WISP2 (n=6 biological replicates). L, Migration of MDA-MB-231 cells through the microporous membrane of Transwell inserts (in the absence of Col I) in the presence of PBS (vehicle), WISP1, WISP2, or WISP1 + WISP2 (n=6 biological replicates). M, Proliferation rate of MDA-MB-231 cells plated on Col I, Col I + WISP1, or Col I + WISP2 lattices (n=4 biological replicates). (A-C, E, G-M) mean ± SEM, (A-C) Mann-Whitney test, (E, G-I, K, L) one-way ANOVA followed by Tukey’s posttest. ns, P>0.05; ***, P<0.001.

Figure 3.. The C-terminal domain of WISP1 drives WISP1-induced Col I linearization and cell invasion but is dispensable for WISP1-Col I binding.

A, Scheme of WISP constructs stably transduced in 4T1 cells. IGFBP, insulin-like growth factor domain; VWC, von Willebrand factor type C repeat domain; TSP-1, thrombospondin type-1 repeat domain; CT, cysteine-rich knot-like C-terminal domain. B, Detection of WISP1, WISP1ΔCT, WISP2, and WISP2+CT binding to fibrillar Col I, using conditioned media from 4T1 cells stably overexpressing these constructs or stably transduced with an empty vector (EV) control. Col I-bound proteins were detected with anti-WISP1 (left) or anti-WISP2 (right) antibodies. Binding was normalized to the molarity of each construct in the conditioned medium (left, n=9, from 3 independent experiments; right, n=8, from 4 independent experiments). C, Binding of WISP1-myc protein to fibrillar Col I in the presence of WISP1, WISP2, WISP1ΔCT, or WISP2+CT, using 1:1 (v:v) ratios of conditioned medium from 4T1 cells overexpressing these proteins. Conditioned medium from 4T1-EV (4T1 cells stably transduced with an empty vector) was used as negative control. Col I-bound proteins were detected with an anti-Myc antibody (n=4, from 2 independent experiments). D, Scanning electron microscopy of Col I lattices formed in the presence of concentrated conditioned medium from 4T1 stable cell lines. Scale bars, 2 μm. E, Curvature ratios of Col I fibrils in lattices from D (n=15, 3 independent experiments, 5 images/lattice). Statistical significance vs Col I + EV is shown. F, Average cell speed of 4T1-EV (n=123), 4T1-Wisp1 (n=75), 4T1-Wisp2 (n=69), 4T1-Wisp1ΔCT (n=106), and 4T1-Wisp2+CT (n=66) cells plated on Col I lattices. G, Invasion of 4T1-EV, 4T1-Wisp1, 4T1-Wisp2, 4T1-Wisp1ΔCT, or 4T1-WISP2+CT cells through Col I lattices layered on Transwell inserts (n=9, except 4T1-EV and 4T1-Wisp1, n=15 biological replicates). H, Average cell speed of 4T1-EV (n=115), 4T1-Wisp1 (n=89), 4T1-Wisp2 (n=54), 4T1-Wisp1ΔCT (n=53), and 4T1-Wisp2+CT (n=59) cells plated on plastic. I, Migration of 4T1-EV, 4T1-Wisp1, 4T1-Wisp2, 4T1-Wisp1ΔCT, or 4T1-WISP2+CT cells through uncoated Transwell inserts (n=9, except 4T1-EV, n=18 biological replicates). (B, C, E-I) mean ± SEM, one-way ANOVA followed by Tukey’s posttest. ns, P>0.05; **, P<0.01; ***, P<0.001.

Figure 4.. WISP2 and WISP1ΔCT block TGFβ1-induced cell invasion through Col I by acting as WISP1 antagonists.

A, Relative Wisp1 mRNA expression levels in 4T1-EV (4T1 cells stably transduced with an empty vector), 4T1-Wisp1ΔCT, and 4T1-Wisp2 cells cultured in the presence of vehicle or TGFβ1 (2 ng/mL). Primers targeting sequences coding in the CT domain of Wisp1 were used to detected endogenous Wisp1 expression levels but not overexpressed Wisp1ΔCT (n=3 biological replicates). B, Average cell speed of 4T1-EV + vehicle (n=44), 4T1-EV + 2 ng/mL TGFβ1 (n=37), 4T1-Wisp2 + vehicle (n=36), 4T1-Wisp2 cells + TGFβ1 (n=44), 4T1-Wisp1ΔCT + vehicle (n=36), and 4T1-Wisp1ΔCT + TGFβ1 (n=35) cells plated on Col I lattices. C, Invasion of 4T1-EV, 4T1-Wisp2, and 4T1-Wisp1ΔCT cells treated with TGFβ1 (2 ng/mL) or vehicle control, through Col I lattices layered on Transwell inserts (n=6 biological replicates). D, WISP1 protein concentration in conditioned medium from 4T1-gCont (4T1-indCas9 cells stably transduced with a pool of control non-targeting gRNAs) and 4T1-Wisp1KO (4T1 cells with CRISPR/Cas9 knockout of Wisp1) treated with TGFβ1 (2 ng/mL) or vehicle control (n=4, from 2 independent experiments). E, Average cell speed of 4T1-gCont + vehicle (n=72), 4T1-gCont + TGFβ1 (n=64), 4T1-Wisp1KO (n=68), 4T1-Wisp1KO + TGFβ1 (n=48) cells plated on Col I lattices. F, Invasion of 4T1-gCont and 4T1-Wisp1KO cells, treated with TGFβ1 (2 ng/mL) or vehicle control, through Col I lattices layered on Transwell inserts (n=6 biological replicates). Mean ± SEM, two-way ANOVA followed by Tukey’s posttest. ns, P>0.05; *, P<0.05; **, P<0.01; ***, P<0.001.

Figure 5.. WISP2 limits collagen linearization in tumors and inhibits breast cancer metastasis.

A, Primary tumor mass, 28 days after orthotopic inoculation of 4T1-EV, 4T1-Wisp1, 4T1-Wisp2 or 4T1-Wisp1ΔCT cells into the 4^th mammary fat pad of BALB/c female mice (n=10 mice/group, except 4T1-Wisp2, n=9 mice). B, Numbers of lung metastases in mice from A. C, Representative images of fibrillar collagen in primary tumors from A visualized by picrosirius red staining followed by polarized light microscopy. Corresponding bright-field images show tissue integrity. Scale bars, 200 μm. D, Average polarized light intensity in picrosirius red staining images of tumors from A. E, Average stiffness (Young’s modulus) of tumors from A, measured by atomic force microscopy. F, Scheme of experimental design. Recombinant WISP2 (rWISP2; 100 μg in 200 μL PBS) was administered via intraperitoneal injection starting on day 8 post-injection of 4T1 cells. Tissues were collected 2 h after the last dose of rWISP2. G, Primary tumor mass, 28 days after orthotopic inoculation of 4T1 cells into the 4^th mammary fat pad (PBS, n=10; rWISP2, n=8 mice/group). H, Numbers of lung metastases in mice from G. I, Representative images of fibrillar collagen in primary tumors from G visualized by picrosirius red staining followed by polarized light microscopy. Corresponding bright-field images show tissue integrity. Scale bars, 200 μm. J, Average polarized light intensity in picrosirius red staining images of tumors from G. (A, B, D, E, G, H, J) mean ± SEM, (A, B, D, E) one-way ANOVA followed by Tukey’s posttest. (G, H, J) unpaired two-sided t-test. ns, P>0.05; *, P<0.05; **, P<0.01, ***, P<0.001.

Figure 6.. WISP1 promotes whereas WISP2 inhibits human breast cancer metastasis.

A, Invasion of MDA-MB-231 cells overexpressing WISP1 (MDA-WISP1) or WISP2 (MDA-WISP2), or stably transduced with an empty vector control (MDA-EV) through Col I lattices layered on Transwell inserts (n=9 biological replicates). B, Migration of MDA-EV, MDA-WISP1, or MDA-WISP2 cells through uncoated Transwell inserts (n=6 biological replicates). C, Proliferation rate of MDA-EV, MDA-WISP1, or MDA-WISP2 cells plated on Col I lattices. (n=12 biological replicates). D, Primary tumor mass, 70 days after orthotopic inoculation of MDA-EV, MDA-WISP1, or MDA-WISP2 cells into the 4^th mammary fat pad of NSG female mice (n=7 mice/group). E, Numbers of lung metastases in mice from D. F, Representative images of fibrillar collagen in primary tumors from D visualized by picrosirius red staining followed by polarized light microscopy. Corresponding bright-field images show tissue integrity. Scale bars, 200 μm. G, Average polarized light intensity in picrosirius red staining images of tumors from D. (A-E, G) mean ± SEM, (A, B, D, E, G) one-way ANOVA followed by Tukey’s posttest. ns, P>0.05; *, P<0.05; **, P<0.01; ***, P<0.001.