Viscoelastic properties of human pancreatic tumors and in vitro constructs to mimic mechanical properties

Abstract

Pancreatic ductal adenocarcinoma (PDAC) is almost universally fatal, in large part due to a protective fibrotic barrier generated by tumor-associated stromal (TAS) cells. This barrier is thought to promote cancer cell survival and confounds attempts to develop effective therapies. We present a 3D in vitro system that replicates the mechanical properties of the PDAC microenvironment, representing an invaluable tool for understanding the biology of the disease. Mesoscale indentation quantified viscoelastic metrics of resected malignant tumors, inflamed chronic pancreatitis regions, and histologically normal tissue. Both pancreatitis (2.15 ± 0.41 kPa, Mean ± SD) and tumors (5.46 ± 3.18 kPa) exhibit higher Steady-State Modulus (SSM) than normal tissue (1.06 ± 0.25 kPa; p < .005). The average viscosity of pancreatitis samples (63.2 ± 26.7 kPa·s) is significantly lower than that of both normal tissue (252 ± 134 kPa·s) and tumors (349 ± 222 kPa·s; p < .005). To mimic this remodeling behavior, PDAC and TAS cells were isolated from human PDAC tumors. Conditioned medium from PDAC cells was used to culture TAS-embedded collagen hydrogels. After 7 days, TAS-embedded gels in control medium reached SSM (1.45 ± 0.12 kPa) near normal pancreas, while gels maintained with conditioned medium achieved higher SSM (3.38 ± 0.146 kPa) consistent with tumors. Taken together, we have demonstrated an in vitro system that recapitulates in vivo stiffening of PDAC tumors. In addition, our quantification of viscoelastic properties suggests that elastography algorithms incorporating viscosity may be able to more accurately distinguish between pancreatic cancer and pancreatitis.

Statement of significance: Understanding tumor-stroma crosstalk in pancreatic ductal adenocarcinoma (PDAC) is challenged by a lack of stroma-mimicking model systems. To design appropriate models, pancreatic tissue must be characterized with a method capable of evaluating in vitro models as well. Our indentation-based characterization tool quantified the distinct viscoelastic signatures of inflamed resections from pancreatitis, tumors from PDAC, and otherwise normal tissue to inform development of mechanically appropriate engineered tissues and scaffolds. We also made progress toward a 3D in vitro system that recapitulates mechanical properties of tumors. Our in vitro model of stromal cells in collagen and complementary characterization system can be used to investigate mechanisms of cancer-stroma crosstalk in PDAC and to propose and test innovative therapies.

Keywords: Cancer associated fibroblasts; Collagen hydrogels; Indentation; Pancreatic ductal adenocarcinoma; Pancreatic stellate cells; Pancreatitis; Tissue mechanics.

Figure 1

Reported moduli for collagen gels range three orders of magnitude for small concentration variations. Different papers reporting modulus to denote elastic component of mechanical properties use various names, e.g. Young’s Modulus, Elastic Modulus, Apparent Modulus, and Storage Modulus, and cover wide-ranging values for similar concentrations and formulations. Precise test conditions are described in respective reports [–22], but notably, effective modulus is different at low (0.1 mm/min, 16A) and high (1 mm/min, 16B) displacement rates [. For comparison, reported shear storage moduli were converted to elastic-like moduli assuming Poisson’s ratio is 0.495 [23].

Figure 2

Samples of resected pancreatic tissue were indented using custom indentation device. Large samples as in the example tissue shown were sliced to obtain a flat top surface (A) and placed into a custom silicone well (B). Well was flooded with room temperature media to fully submerge sample during indentation (C).

Figure 3

(A) Xenograft PDAC tumors are grown in mice to expand and obtain pancreatic cancer (PDAC) cells, while tumor-associated stromal (TAS) cells are directly isolated from resected tumor. Conditioned medium is collected from PDAC cells and used to maintain Living Mechanical MicroEnvironments (LiMMEs), collagen hydrogels embedded with TAS. (B) LiMMEs are indented by moving the piezo-electric stage (1) down. Contact between the spherical probe (2) and the sample causes the cantilever (3) to deflect, while the capacitive sensor’s (4) deflection measurements are used to calculate indentation forces. (C–D) TAS and PDAC cells cultured on plastic sufaces and labeled to visualize F-Actin (green) and nuclei (blue). Scale bars = 100 μm

Figure 4

Graph showing indentation depth (top dashed orange line) and the normal force exerted by the tissue (bottom solid blue line) for a representative tumor sample. The central relaxation portion (labeled between 30–240 s) was used to calculate steady-state modulus as described in Section 3.

Figure 5

Hertz fit to quasi-static loading (A) yields comparable elasticity results to SSM obtained through relaxation experiments (B). However, unlike quasi-static loading, relaxation curves fitted to different constitutive models also allow quantification of time-dependent properties like viscosity.

Figure 6

Elastic properties of diseased tissue follow different trends than viscous properties. (A) Steady-state modulus (SSM) of pancreatic ductal adenocarcinoma tumors (PDAC, “cancer”) is significantly higher than pancreatitis (p < 0.001) and normal (p = 0.002) resected tissues. (B) Viscosity, or liquid-like property, of pancreatitis tissue is significantly lower than PDAC (p < 0.001) and normal (p = 0.003) resected tissue. (C) Characteristic time of relaxation follows the same trend as viscosity. For definition of viscosity and characteristic time, see text and notes for Eqn. 3. P-values were calculated using a Wilcoxon nonparametric multiple comparisons test for each property. Box plots indicate mean, quartiles, and range. ^Outlier of 13.7 kPa not shown; ^†Outlier of 3400 kPa·s not shown

Figure 7

LiMMEs reach SSM values within the ranges of normal and tumor tissue when maintained with control and conditioned medium, respectively (A). Characteristic Relaxation Times (B) show a similar increasing trend when treated with cancer cell supernatant, while controls remain mostly unchanged. ^Outlier of 13.7 kPa not shown; ^†Outlier of 157 s not shown.

Figure 8

SSM values after 7 days of culture for TAS LiMMEs. LiMMES were treated with PDAC cell supernatant (conditioned media, CM) for 24, 72, or 168 hours after fabrication as indicated and maintained in control media until characterization on day 7 (indicated by open circles). Untreated control LiMMEs on Day 7 are repeated here from Figure 7 for reference (open boxes), as are LiMMES treated with CM for 72 hours and characterized at 72 hours (grey boxes). Transient CM exposure for 72 hours seemed to induce lasting stromal transformation that accelerated remodeling for the entire time period.

Figure 9

Visualization of tumor-associated stroma (TAS) cells in Living Mechanical MicroEnvironments (LiMMEs) on Day 7. (A, B) Calcein AM immunofluorescent plasma label suggests an increase in TAS proliferation and spreading in treated LiMMEs. (C–F) Labeled actin and nuclei of TAS in LiMMEs after 7 days in culture. Confocal images (C–D) reinforce 3D nature of gels, with nuclei appearing out-of-plane, and fluorescence images (E–F) illustrate actin fibers spanning various focal depths.