Cancer-associated fibroblasts support vascular growth through mechanical force

Abstract

The role of cancer-associated fibroblasts (CAFs) as regulators of tumor progression, specifically vascular growth, has only recently been described. CAFs are thought to be more mechanically active but how this trait may alter the tumor microenvironment is poorly understood. We hypothesized that enhanced mechanical activity of CAFs, as regulated by the Rho/ROCK pathway, contributes to increased blood vessel growth. Using a 3D in vitro tissue model of vasculogenesis, we observed increased vascularization in the presence of breast cancer CAFs compared to normal breast fibroblasts. Further studies indicated this phenomenon was not simply a result of enhanced soluble signaling factors, including vascular endothelial growth factor (VEGF), and that CAFs generated significantly larger deformations in 3D gels compared to normal fibroblasts. Inhibition of the mechanotransductive pathways abrogated the ability of CAFs to deform the matrix and suppressed vascularization. Finally, utilizing magnetic microbeads to mechanically stimulate mechanically-inhibited CAFs showed partial rescue of vascularization. Our studies demonstrate enhanced mechanical activity of CAFs may play a crucial and previously unappreciated role in the formation of tumor-associated vasculature which could possibly offer potential novel targets in future anti-cancer therapies.

Figure 1

CAFs support vascularization in 3D microtissues. (a) When co-cultured with ECs in Fibrin or combination Fibrin-Collagen (FN + Coll) gels, CAFs support significantly more vascular growth compared to NBFs. NHLF also demonstrate significantly higher vascularization potential compared to NBFs. Data are presented as total vessel length per unit area, normalized to NBF in Fibrin: 0.0014 ± 0.0002 μm⁻¹; or NBF in FN + Coll: 0.0018 ± 0.0006 μm⁻¹. *p < 0.01 vs. NBF; ^p < 0.01 vs. CAF for same gel type. (Right) Immunofluorescent images of CD31 staining of 3D vessel systems show interconnected vascular networks in CAF & NHLF samples, but not in NBF samples. (b) CAFs in co-culture with ECs (CAFs/ECs) demonstrate higher steady state levels of soluble VEGF than NBFs/ECs. NHLF/EC co-cultures exhibit significantly lower levels of VEGF compared to CAF/EC samples. *p < 0.01 vs. NBF; ^p < 0.01 vs. CAF. (c) Inhibition of VEGFRs suppresses CAF- and NHLF-supported vascular growth compared to vehicle treated controls but shows significantly larger average vessel growth compared to NBF vehicle controls. Data are presented as average vessel length in μm. *p < 0.01 vs. NBF vehicle; ^p < 0.01 vs. NBF + inhibitor (Right) Immunofluorescent images of CD31 staining show vascular fragments of >100 μm in length present in CAF samples with inhibited VEGFR. (d) Conditioned media from CAF/EC cultures (CM-CAF) minimally rescues vascular formation in fibrin gels. Conditioned medial from NHLF/EC (CM-NHLF) cultures showed similar results when added to NBF samples. Data are presented as average vessel length in μm. (Right) Immunofluorescent staining of gels for CD31 demonstrate NBF + CM-CAF or + CM-NHLF groups exhibit only short fragments of blood vessels, <50 μm in length. *p < 0.01 vs. NBF; ^p < 0.01 vs. CAF. Scale bars = 250 μm.

Figure 2

CAFs demonstrate mechanosensitive vascularization potential. (a) CAFs support enhanced levels of vascular growth over increasing fibrin gel compositions, with a significant increase seen from 2.5 mg/mL to 5.0 mg/mL samples. NHLF samples demonstrate the opposite effect, with significantly decreased vascularization potential as fibrin concentration increases over the same range. Data are presented as total vessel length per area, normalized to NBF samples in 2.5 mg/mL fibrin gels: 0.0017 ± 0.0001 μm⁻¹. (Right) Immunofluorescent images of CD31 staining show vascular networks formed in the presence of CAFs and NHLFs, but not NBFs. *p < 0.01 vs. NBF at same fibrin concentration; ^p < 0.01 vs. CAF at 2.5 mg/mL fibrin; ^$p < 0.01 vs. NHLF at 2.5 mg/mL. Scale bars = 250 μm. (b) CAFs generate larger deformations in 3D gels compared to NBFs, as seen in 3D vector plots of bead displacements. Each arrow represents a tracked movement of a fiducial marker; color corresponds to magnitude as indicated by the scale bar (0–8 μm). Inset scale bars = 20 μm. (c) Data from vector maps is pooled from a minimum of 2 technical replicates, binned and plotted as histograms to show population behavior, where the y-axis represents the percent of data collected at a particular deformation value. Inset numbers are average deformation magnitudes ± standard deviation for each cell line. Each plot represents n > 40 bead movements tracked. CAFs demonstrate significantly larger deformations at all gel compositions compared to NBFs. CAFs have significantly larger averaged deformations compared to NHLFs at 2.5 and 5.0 mg/mL fibrin only. NHLFs exhibit larger deformations compared to NBFs in 5.0 mg/mL and 10 mg/mL fibrin gels,. *p < 0.01 vs. NBF at same fibrin concentration for average values; # vs. CAF at same fibrin concentration; ^p < 0.01 vs. CAF at 2.5 mg/mL fibrin; ^&p < 0.01 vs. CAF at 5.0 mg/mL; ^$p < 0.01 vs. NHLF at 2.5 mg/mL fibrin; ^%p < 0.01 vs. NHLF at 5.0 mg/mL fibrin. More statistical data can be found in Table S1.

Figure 3

Mechanotransductive pathways regulate CAF supported vascular growth. (a) Inserting caRho into NBFs promotes enhanced blood vessel growth compared to NBF empty vector (EV) controls. Furthermore, inhibiting ROCK2, SN1, or YAP in CAFs suppresses formation of blood vessels in 3D cultures compared to CAF-EVs. Inhibition of ROCK1 in CAFs produced no significant changes in vascularization measured. Data are presented as total vessel length per area normalized to NBF-EV: 0.0032 ± 0.0001 μm⁻¹. *p < 0.05 versus NBF-EV; ^p < 0.05 versus CAF-EV. (b) Histograms showing distribution of deformations induced by genetically-modified fibroblasts in 3D fibrin gels. Inset numbers are average deformation magnitudes ± standard deviation for each cell line. NBF-EV and CAF-EV data are shown on all plots to facilitate ease in comparing inhibited lines to controls. The addition of caRho shifts the peak of observed deformations to the right, indicating higher levels of mechanical activity in these cells compared to NBF-EVs. Alternatively, inhibiting ROCK, SN1, and YAP shifts peak deformations to the left, demonstrating that the modified CAFs generate smaller deformations in 3D compared to CAF-EVs. *p < 0.05 vs. NBF-EV; ^p < 0.05 vs. CAF-EV.

Figure 4

VEGF production and utilization is impacted by mechanotransduction inhibition. (a) VEGF produced during the vasculogenic ring assay demonstrates differences between EV controls and modified cells lines. *p < 0.05 vs. NBF-EV/EC; ^p < 0.05 vs. CAF-EV/EC. Increased levels of VEGF do not necessarily correspond to increased vascular growth, as demonstrated by NBF-EV/EC samples having significantly higher steady state levels of VEGF than CAF-EV/EC samples. (b) To determine how EC presence affected VEGF measurements in the ring assay, we measured VEGF production in fibroblast only cultures in the same concentration 3D fibrin gels. These results suggest that modified CAFs may produce more VEGF than CAF-EVs, but that the ECs in these cultures are less efficient at utilizing the molecule for the creation of vessels. Further, there is no difference in NBF-EV and NBF-caRho fibroblast only samples measurements of VEGF, while NBF-caRho/EC samples show more VEGF than NBF-EV/EC samples. This indicates that VEGF consumption alone cannot explain enhanced vessel growth in NBF-caRho/EC samples. *p < 0.05 versus NBF-EV; ^p < 0.05 vs. CAF-EV.

Figure 5

Magnetic induction of ECM deformations partially rescues vascularization. (a) Thrombin-coated magnetic beads move in response to magnetic stimulation; magnet was located on left of images shown (i). In bright field images, magnetic bead movement can be easily tracked. White circles indicate an initial position; red circles indicate final position. Inset number is average movement ± standard error of the mean (ii). For same region shown in (i), blue 1 μm fluorescent beads were tracked using the MATLAB bead tracking algorithm described in this paper. Color bar represents magnitude of displacement (0–8 μm). Inset number is average movement ± standard error of the mean. There is statistically no difference in the measured displacements (p = 0.775). Scale bars = 25 μm. (b) In a static study, a magnet was placed 2.5 cm to the left of fibrin gels containing thrombin-coated mangetic beads cultured with ECs only or with ECs and fibroblasts. No significant enhancement of blood vessel growth was observed in any of the samples, demonstrating that static strains induced the ECM are not sufficient to rescue vascularization. Vascular growth numbers are presented as total vessel length pre area normalized to NBF-EV control: 0.0038 ± 0.0013 μm⁻¹. (c) In a dynamic stimulation study, thrombin-coated magnetic beads in fibrin gels containing ECs only or ECs and fibroblasts were moved by placing a N52 magnet on an orbital shaker located 3 cm below the gels. Magnetic bead movement stimulated enhanced blood vessel growth in all samples, except NBFs. This demonstrates mechanical activity in the ECM can moderately promote or rescue blood vessel growth in some samples that do not support vascularization in control studies. Vascular growth numbers are shown as total vessel length per area normalized to NBF-EV control: 0.0033 ± 0.0006 μm⁻¹. *p < 0.05 versus NBF-EV control; ^p < 0.05 versus control group for each specific line.