Alpha5beta1 integrin-fibronectin interactions specify liquid to solid phase transition of 3D cellular aggregates

Abstract

Background: Tissue organization during embryonic development and wound healing depends on the ability of cells on the one hand to exchange adhesive bonds during active rearrangement and on the other to become fixed in place as tissue homeostasis is reached. Cells achieve these contradictory tasks by regulating either cell-cell adhesive bonds, mediated by cadherins, or cell-extracellular matrix (ECM) connections, regulated by integrins. Integrin alpha5beta1 and soluble fibronectin (sFN) are key players in cell-ECM force generation and in ECM polymerization. Here, we explore the interplay between integrin alpha5beta1 and sFN and its influence on tissue mechanical properties and cell sorting behavior.

Methodology/principal findings: We generated a series of cell lines varying in alpha5beta1 receptor density. We then systematically explored the effects of different sFN concentrations on aggregate biomechanical properties using tissue surface tensiometry. We found previously unreported complex behaviors including the observation that interactions between fibronectin and integrin alpha5beta1 generates biphasic tissue cohesion profiles. Specifically, we show that at constant sFn concentration, aggregate cohesion increases linearly as alpha5beta1 receptor density is increased from low to moderate levels, producing a transition from viscoelastic-liquid to pseudo viscoelastic-solid behavior. However, further increase in receptor density causes an abrupt drop in tissue cohesion and a transition back to viscoelastic-liquid properties. We propose that this may be due to depletion of sFn below a critical value in the aggregate microenvironment at high alpha5beta1 levels. We also show that differential expression of alpha5beta1 integrin can promote phase-separation between cells.

Conclusions/significance: The interplay between alpha5-integrin and sFn contributes significantly to tissue cohesion and, depending on their level of expression, can mediate a shift from liquid to elastic behavior. This interplay represents a tunable level of control between integrins and the ECM that can influence tissue cohesion and other mechanical properties, which may translate to the specification of tissue structure and function. These studies provide insights into important biological processes such as embryonic development, wound healing, and for tissue engineering applications.

Figure 1. Flow cytometry analysis of α5β1 expression by selected clones.

Cells were incubated in hanging drop culture to generate 3D aggregates. Single cell suspensions were then assayed for α5β1 integrin expression by flow cytometry. The relative levels of receptor expression for α5-transfected clones (B–E) were compared with that of the parent cell line transfected with an empty vector control (A). Four distinct populations expressing low (B), mid (C), high (D) and very high (E) levels of α5β1 were expanded and used for all subsequent experiments.

Figure 2. Assessment of Fn matrix assembly by immunofluorescence.

Fn matrix assembly by clones expressing low (A–C), High (D–F), and very high (G–I) levels of α5β1 integrin. Clones expressing different levels α5β1 receptor were plated at a density of 5×10⁵ cell/ml. Fibronectin immunostaining was carried out after 2 days in Fn-depleted tissue culture medium (A, D, G) or medium containing 30 µg/ml (B, E, H) or 300 µg/ml (C, F, I) soluble fibronectin. The assembled matrix becomes qualitatively more fibrous as a function of increasing receptor expression and soluble fibronectin concentration. Scale Bar 100 µ.

Figure 3. Evaluation of aggregate mechanical properties.

α5β1 integrin-Fn interaction can elicit both viscoelastic liquid and elastic solid behavior by compressed aggregates (A). Here we show that for CHO cells transfected to express Mid levels of α5 integrin and incubated in 30 µg/ml sFn, some aggregates exhibit liquid-like behavior, where σ₂/σ₁ = 1 and is significantly less than F₂/F₁ (grey bars), whereas others exhibit elastic behavior, where the ratio of σ₂/σ₁approaches that of F₂/F₁ (black bars), (two-proportion z test, α = 0.05). Relative percentage of liquid and elastic aggregates as a function of α5β1 receptor expression and sFn concentration (B). Aggregates incubated in 30 µg/ml sFn, irrespective of α5β1 expression, displayed a preponderance of liquid behavior, (grey bars) whereas those incubated in 300 µg/ml sFn behaved predominantly as elastic solids (black bars). Each data set contained 16–20 aggregates. Linear regression analysis of surface tension vs. volume for aggregates expressing various levels of α5β1 integrin. Figures 3C,D, and E represent clones CHO-X5C5 L, H, and HH, respectively. The circles represent data of individual aggregates within each data set. The line is a regression analysis. The data set is for aggregates displaying liquid-like behavior and was generated for aggregates compressed in standard tissue culture medium. The surface tension of aggregates remained relatively constant over a 5-fold range in volume. Linear regression analysis generated correlation coefficient (r²) values of 0.0009 (Fig. 3C), 0.38 (Fig. 3D), and 0.03 (Fig. 3E) for α5β1 (L), α5β1 (H) and α5β1 (HH), respectively. These r² values indicate that no statistically significant correlation exists between σ and volume (C–E).

Figure 4. Surface tension of aggregates of α5β1 integrin-expressing clones.

Surface tension values of aggregates displaying liquid-like behavior were plotted as a function of α5β1 receptor expression. Aggregates expressing Mid levels of α5β1 integrin possess a higher surface tension than all other clones (ANOVA, p<0.0001, Tukey's MCT, p<0.001). Surface tension appears to increase as a function of receptor expression but only for aggregates expressing low (L) to mid (M) levels of α5β1 integrin. Higher level expression of the receptor results in aggregates whose surface tensions are not significantly different than for that measured for low-expressing aggregates. Data sets contained 20–38 aggregates, representing 40–76 compressions.

Figure 5. Aggregate size as a function of cell number and time in culture.

Compaction assays were performed in which cells for all clones were placed in hanging drop culture at cell seeding densities of 25,000 (A) and 50,000 (B) cells per aggregate and a sFn concentration of 30 µg/ml. Size measurements were initiated on day 2 and repeated daily up to day 5. Each data point represents average size calculated from 10 aggregates per time point for clones expressing Low (blue line), Mid (pink line), High (green line) and Highest (red line) levels of α5β1 integrin.

Figure 6. sFn depletion as a function of α5β1 expression.

(7A) Twenty-five thousand cells of the Mid (grey bars) and HH (white bars) clones were incubated in 30 µg/ml sFn in hanging drop culture for 3 days, at which time the amount of sFn remaining in each hanging drop was assayed by ELISA. Cells expressing Mid levels of α5β1 integrin depleted sFn at a slower rate than did cells expressing higher levels (Student t-test, p = 0.0005). (7B) For the first 72 hours, the rate of depletion of sFn from the hanging drop is significantly higher for HH aggregates than for Mid (unpaired t-test, p<0.05). After 72 hours, the rate of sFn utilization drops markedly and is the same for both cell lines (unpaired t-test, p>0.05).

Figure 7. Fibronectin matrix assembly in 3D aggregates as a function of α5β1 receptor expression.

Representative immunostained tissue sections from aggregates containing 25,000 cells and expressing either Mid (A) or HH (B) levels of α5β1. Images are of aggregates collected after 3 days of incubation. Fibronectin matrix is labeled green and nuclei are blue. The scale-bar is 50 µ.

Figure 8. α5β1 integrin-mediated cell sorting.

Differentially stained cells of M (red) or HH (red) clones sort out from CHO-P3 cells (green) to assume an internal position. This sorting behavior appears to become more apparent as both α5β1 expression and sFn levels increase.