The cancer glycocalyx mediates intravascular adhesion and extravasation during metastatic dissemination

Abstract

The glycocalyx on tumor cells has been recently identified as an important driver for cancer progression, possibly providing critical opportunities for treatment. Metastasis, in particular, is often the limiting step in the survival to cancer, yet our understanding of how tumor cells escape the vascular system to initiate metastatic sites remains limited. Using an in vitro model of the human microvasculature, we assess here the importance of the tumor and vascular glycocalyces during tumor cell extravasation. Through selective manipulation of individual components of the glycocalyx, we reveal a mechanism whereby tumor cells prepare an adhesive vascular niche by depositing components of the glycocalyx along the endothelium. Accumulated hyaluronic acid shed by tumor cells subsequently mediates adhesion to the endothelium via the glycoprotein CD44. Trans-endothelial migration and invasion into the stroma occurs through binding of the isoform CD44v to components of the sub-endothelial extra-cellular matrix. Targeting of the hyaluronic acid-CD44 glycocalyx complex results in significant reduction in the extravasation of tumor cells. These studies provide evidence of tumor cells repurposing the glycocalyx to promote adhesive interactions leading to cancer progression. Such glycocalyx-mediated mechanisms may be therapeutically targeted to hinder metastasis and improve patient survival.

Fig. 1. Effect of GCX degradation on extravasation of TCs.

A Fluorescent staining of endothelial junctions using ZO-1 and hyaluronic acid, chondroitin sulfate, and heparin sulfate before and after enzymatic removal from the microvascular network endothelium. B Changes in MVN permeability to 70 kDa dextran for untreated control networks and after enzymatic removal of HA, CS, and HS (n = 3 devices, three regions of interest averaged for each). C Immunofluorescent staining of HA, CS, and HS on TCs before and after enzymatic degradation. D–F Changes in extravasation efficiency of MDA-MB-231 cells after treatment of the MVNs, TCs, or both after enzymatic digestion of D hyaluronic acid, E chondroitin sulfate, and F heparin sulfate (n = 6 devices, 5 ROIs each). The scale bars in both (A) and (C) are 60 µm. Significance was assessed by student’s t test assuming normally distributed data, p < 0.05 *, p < 0.001 ***.

Fig. 2. Characterization of TC arrest under flow in the MVNs.

A Schematic diagram of the microfluidic device used to perfuse MVNs under a constant pressure difference Δp and cells being carried by luminal flow into A narrow channels and B impacting the endothelium at bifurcations or C large vessels (partially realized with Biorender.com). B Speed of inert beads carried by luminal flow driven by different pressure differences (n = 50 beads tracked in three devices, average and standard deviation shown) and arrest efficiency of TCs (n = 4 devices, average and standard deviation shown) as a function of the pressure difference. C Confocal image of two arrest mechanisms showing MDA-MB-231 cells within MVNs arrested by (1) physical trapping and (2) adhesion. The scale bar is 200 µm. D Bead speed through the MVNs (n as above, the error bars indicate the standard deviation) and vessel diameter as a function of specific GCX component removal (the error bars indicate the standard deviation between the averages of n = 3 devices, 3 regions of interest each). E Arrest efficiency of TCs as a function of GCX enzymatic treatment of the MVNs alone, TCs alone, or both MVNs and TCs (n = 4 devices). F Cumulative percentage of TCs either physically trapped in small vessels or adhered to large vessels in the MVNs (n > 40 cells). Statistical significance was assessed by student’s t test assuming normally distributed data, p < 0.05 *, p < 0.01 **, p < 0.0001 ****. A normal distribution of the data in (B) and (D) was confirmed by Kolmogorov–Smirnov test.

Fig. 3. Absence of endothelial activation following GCX treatment.

A Normalized (to CD31) expression of EC adhesion molecules associated with endothelial activation (n = 3 devices). B Normalized (to β-actin) expression of ICAM-1 on ECs (HUVEC) and stromal cells (nHLF) cultured in well plates after treatment with hyaluronidase (n = 3 wells). C Imaging of ICAM-1 after MVN treatment. The scale bar is 200 µm. D ICAM-1 is not highly expressed on ECs in the vicinity of TCs arrested for 6 h prior to fixing (white arrow), as compared to ICAM-1 expressed on stromal cells (red arrows). The scale bar is 60 µm. Statistical significance was assessed by student’s t test assuming normally distributed data, p < 0.05 *, p < 0.01 **, p < 0.001 ***.

Fig. 4. Mechanisms of arrest and adhesion mediated by the HA–CD44 complex.

A Confocal imaging of arrested TC in the microvasculature and CD44 anchoring. The scale bar is 20 µm. In (B), the TCs are bound through CD44 (red arrows) to streaks of HA. In C, the streaks (red arrows) continue past the TCs, possibly indicating previous deposition. The scale bars for both (B) and (C) are 60 µm. The dashed arrows indicate the direction of flow. D Diagram of GCX-mediated arrest mechanism of TCs, involving deposition of HA upon impact with the endothelium under flow and anchoring to HA through CD44 (partially realized with Biorender.com). E Normalized (to β-actin) expression of CD44 isoforms in TCs after treatment with hyaluronidase in well plates (n = 3 wells). F Expression of HA on TCs after antibody-blocking of CD44 (n = 4 wells) and confocal image of treated TC arrested in a small MVNs capillary, not showing CD44 anchoring or HA. The scale bar is 20 µm. G Normalized (to total protein concentration) expression of CD44 isoforms in various TCs, and H HA expression and I arrest efficiency in the MVNs of the same TCs (n as above, shown for all). Statistical significance was assessed by student’s t test assuming normally distributed data, p < 0.05 *, p < 0.01 **, p < 0.001 ***.

Fig. 5. Role of CD44 during TC extravasation.

A Time-lapse confocal live imaging (scale bar = 20 µm, post-TC perfusion time) of extravasating TC showing the ECM-binding role of CD44 (red arrows) in the process, which includes protrusion through the endothelium, permanence in the basement membrane, and migration into the interstitium, schematically represented in (B), partially realized with Biorender.com. C Extravasation efficiency at 6 h after treatment with a CD44-blocking antibody (n = 6 devices) of TCs, MVNs, or both and D extravasation efficiency after siRNA-mediated CD44 targeting (n = 6 devices, five devices for the siRNA control). E, F Relationship between normalized (to total protein concentration) expression of CD44 isoforms e CD44s and f CD44v and extravasation efficiency (n = 6). Statistical significance was assessed by student’s t test assuming normally distributed data, p < 0.05 *, p < 0.01 **, p < 0.001 ***, p < 0.0001 ****.