Review

Knowing the Boundaries: Extending the Differential Adhesion Hypothesis in Embryonic Cell Sorting

See allHide authors and affiliations

Science  12 Oct 2012:
Vol. 338, Issue 6104, pp. 212-215
DOI: 10.1126/science.1223953

Abstract

Successful embryogenesis requires proper sorting and compartmentalization of different cell types. Mechanical interactions between cells help govern these processes. In the past, physics-based theories have guided in vitro studies of cell sorting and tissue surface tension. Recent experiments have challenged this approach, indicating that adhesive molecules also act as signaling molecules that initiate local reorganization of actomyosin and demonstrating that cells at the boundary of a colony of initially identical cells become “mechanically polarized.” Extending physical models to account for mechanical polarization helps solve a long-standing paradox about magnitudes of tissue surface tensions and potentially explains discrepancies between recent in vivo and in vitro cell-sorting experiments. New experiments are needed to further explore the connection between mechanical polarization and tissue boundary formation in vivo.

How do cells in a developing embryo know where to go and how to behave once they get there? Sometimes a programmed chemical signal, called a morphogen, guides cells, whereas in other cases, cells self-organize. Cell sorting and tissue layering behaviors are remarkably robust: Two types of dissociated embryonic tissues that are homogeneously mixed will spontaneously sort into layers. But how do cells know which way to sort?

The Differential Adhesion Hypothesis

Physical reasoning has provided a partial answer. More than 50 years ago, Steinberg proposed the differential adhesion hypothesis (DAH), which postulates that (i) each type of embryonic tissue has a unique “tissue surface tension” (TST) that governs how tissues sort, just as fluid surface tension governs how immiscible liquids demix, and (ii) differences in TST arise from differences in cell-cell adhesion (1). Because interactions between cells are often difficult to probe inside an embryo, investigations of the DAH have largely been carried out in vitro. By measuring forces applied to cellular aggregates in vitro, one can extract a surface tension–like quantity and demonstrate that “effective surface tension” accurately predicts sorting behavior. However, key questions about the DAH remain unresolved: How do single-cell properties such as adhesion and cortical tension impact TST? How do in vitro analyses relate to tissue layering and compartmentalization in a developing embryo?

The answer to the first question is simple if cells in a tissue are just like molecules in a fluid, where the surface tension depends on the depth of the interaction potential between the molecules (their “adhesion energy”) and the area over which the molecules interact. In this scenario, the TST is roughly equal to the adhesion energy of a cell per unit area. Because adhesion in tissues is predominantly mediated by cadherin molecules, the TST should equal the energy required to rupture a cadherin bond times the number of cadherins per unit cell surface area. Experimental studies have shown that TST is indeed proportional to the number of surface cadherins (2).

However, the magnitude of TST measured in cell aggregates—about 1 × 10−3 N/m—does not support this idea. The rupture energy of cadherin bonds is about 10 kT or 1 × 10−19 Joules, and there are about 10 cadherins per square micron on the surface of typical cells (3), which together result in an adhesion energy per unit area of about 1 × 10−7 N/m, which is four orders of magnitude smaller than typical TSTs. Other experimental results (3, 4) support theories (5, 6) that hypothesize that cell sorting correlates more strongly with the magnitude of single-cell “cortical tension” generated by active actomyosin contractility rather than cell adhesion.

Mechanical Polarization

These apparently contradictory conclusions are based on the assumption that all cells in an undifferentiated tissue are roughly identical and can be characterized by properties measured for a single cell. Several recent papers point toward an intriguing possibility for resolving these contradictions because they demonstrate that cells at tissue boundaries are different from those in the interior. They suggest that boundary cells actively change their mechanical properties or “mechanically polarize.”

Building on previous work (7, 8), Gardel and co-workers have developed an exciting new traction force imbalance method (TFIM) that determines exact forces exerted across cell-cell contacts in small groups of adherent cells on two-dimensional (2D) substrates in unconstrained geometries (9). Unlike laser ablation (10) or pipette aspiration (11) experiments, TFIM makes no assumptions about the mechanical properties of cells. TFIM was used to quantify mechanical polarization in cell pairs and triplets: Cells reorganized their adhesive and cytoskeletal machinery so that tension forces and actin density were substantially higher along external interfaces of doublets or triplets than on internal interfaces.

Extending these ideas to larger cell populations, Mertz et al. (12) demonstrate that coherent colonies of initially identical cells on 2D surfaces can be characterized by an effective surface tension. Just as in the Gardel experiments, traction forces were localized near the boundary of the colony, and cells at the boundary were mechanically polarized. Another recent study by Krens et al. (13) showed that cells on the surface of 3D zebrafish embryonic explants develop strong apical-basal actin polarization, although these cells express epithelial markers at later times and may also be differentiated.

These results show that many different cell types in different environments have machinery to sense and respond mechanically to “being at a boundary.” What are the components of this machinery? An intriguing possibility is that cadherins act as signaling molecules, altering local actomyosin dynamics and generating cell polarization (Fig. 1). Earlier work on cell doublets has shown that cadherin-mediated adhesion regulates Rho family guanosine triphosphatases to reduce actomyosin contractility along a cell-cell contact interface relative to the edges of the contact (14), which is consistent with the TFIM experiments (9). Generalizing to larger aggregates, one would expect cortical tension to be reduced at all cell-cell contacts for cells in the middle of a confluent tissue. In contrast, cells at the boundaries would have a reduced cortical tension only along internal “contacting” interfaces, which could lead to mechanical polarization of boundary cells similar to that seen in experiments (12). So, although the mechanical energy of cadherin bonds is very small, cadherin signaling might still be responsible for boundary polarization.

Fig. 1

Mechanical polarization in cell doublets. Initially, two cells coming into contact will adhere via the external domains of cadherin molecules (red dots). Over longer time scales, internal domains of cadherins interact with actomyosin, which leads to the reorganization of actin cortical network (green lines) in a mature cell-cell contact. Recent work demonstrates that this often leads to up-regulation of cortical tension along the external boundary and/or down-regulation along internal interfaces. It may be that upstream extracellular biochemical signals, such as Hedgehog, influence this cascade.

ILLUSTRATION: K. SUTLIFF/SCIENCE

An Extended Differential Adhesion Hypothesis

These results also demonstrate that “surface tension” is a robust property even in tissues or colonies where cells are locally and actively changing their mechanical properties. Simple models that express a cell’s mechanical energy as a function of cortical tension, adhesion expression, and elasticity can easily account for mechanical polarization at boundaries. They accurately describe shapes of groups of cells in culture and in epithelial layers in vivo, even in the presence of fluctuations and cell divisions (11, 15). For these tissue types, it is reasonable to hypothesize that TST is the mechanical energy penalty incurred in increasing tissue surface area by moving cells to the boundary.

From this type of model, it has recently been shown that, if cells at the boundary mechanically polarize, the net mechanical effect on the tissue is the same as if there were extra adhesion between all of the cells (16). Within an extended DAH framework, TST is then proportional to an “effective adhesion” that is the sum of cadherin adhesion energy, about 1 × 10−7 N/m, and a term equal to the difference between cortical tensions along internal and external interfaces, about 1 × 10−3 N/m (11, 17). This suggests that TST is dominated by boundary mechanical polarization instead of the mechanical energy of adhesive bonds. And, if mechanical polarization is regulated by cadherin signaling, this could explain why TST scales with the number of surface cadherins and yet why it is strongly correlated with actomyosin-dependent cortical tension.

Mechanical Polarization in Vivo

Evidence for boundary mechanical polarization from in vitro studies provides a starting point for formulating hypotheses about tissue layering in an embryo. Boundary polarization certainly occurs in vivo. For example, recent work in Drosophila has demonstrated that actomyosin cables are critical for maintaining compartment boundaries in the embryo (18), and clonal cells with increased adhesion generate an actomyosin cable at the clone interface (19). But are the same mechanisms that regulate mechanical polarization between cells in culture involved in establishing tissue boundaries in embryos?

Little has been done to determine how properties that are important for cell sorting in vitro affect cell sorting in vivo. Recent work by Ninomiya et al. demonstrates that tissue intercalation and compartmentalization in vivo does not always correlate with in vitro cell sorting or TST measurements (20). Their results suggest that although in vitro sorting is dominated by short–time scale interactions between the external domains of adhesion molecules, in vivo intercalation is dominated by long–time scale spatial reorganization of cadherins and cortical tensions. Ninomiya’s in vivo work in the Xenopus embryo focused on dense, intercalating tissues, where cell-cell contacts are mature and long-lived. Other researchers have focused on in vivo processes, where cells are less densely packed and change neighbors much more quickly, such as germ-layer progenitor cells during the shield stage in zebrafish (21, 22), where the short–time scale adhesive interactions that govern in vitro sorting are more likely to be important.

Boundary mechanical polarization might provide a framework for explaining Ninomiya’s results and uniting in vivo and in vitro studies. If surface tension is dominated by long–time scale boundary polarization, the cadherin density should not correlate with surface tension when cadherins are missing internal domains that interact with the cytoskeleton, as observed. Furthermore, it is tempting to speculate that in vivo interfaces with intercalating cells are ones that have not mechanically polarized, even though each individual tissue boundary polarizes in cell culture medium (Fig. 2). This could explain the discrepancy between differential TST and cell intercalation. Finally, cells in vivo often interact with complex extracellular matrix structures, which might influence boundary polarization and cell sorting.

Fig. 2

Mechanical polarization at tissue-culture and tissue-tissue boundaries. Green lines indicate a higher-than-average density of cortical actomyosin, and the thickness of the lines indicates the amount of tension generated. Red dots indicate multiple cadherin bonds. (A) Many tissues have cells that mechanically polarize at interfaces with culture medium. This effect can explain the magnitude of TST in vitro. (B and C) Nuclei are labeled blue or orange to designate two different tissue types. Even if boundary polarization occurs for individual tissues in vitro, it may (B) or may not (C) occur at the interface between two tissues; this may explain differences between in vivo cell intercalation and in vitro cell sorting.

ILLUSTRATION: K. SUTLIFF/SCIENCE

Looking Forward

To test these hypotheses, new experiments and techniques are needed to (i) analyze cytoskeletal organization along tissue boundaries, (ii) quantify interfacial tensions and forces in vivo, and (iii) study signaling upstream and downstream of mechanical polarization.

It would be useful to demonstrate conclusively that mechanical polarization occurs in 3D tissues composed of identical, undifferentiated cells, just as in the 2D experiments. Fluorescent probes and confocal microscopy could be used to image actin and myosin reorganization at the boundary of a cell aggregate. The same tools could be used to investigate mechanical polarization at the boundaries between two tissue types in vivo and in vitro. Are actin and myosin more dense at an interface between two tissue types? If so, is cell intercalation less likely at a two-tissue interface that has mechanically polarized? How is actomyosin reorganization at two-tissue interfaces different from that at tissue–cell culture medium interfaces?

We also need to develop tools to precisely quantify interfacial tensions inside tissues. Laser ablation is commonly used to estimate tensions in 2D epithelial layers, but it can only do so up to an unknown constant and faces significant challenges in 3D tissues. An alternate approach would be to use TFIM to quantify interfacial tensions between two tissue types and to determine whether these tensions can be predicted on the basis of actomyosin or cadherin dynamics imaged by using confocal microscopy. Other promising new techniques include molecular force sensors (23) and membrane fluctuation analyses (24).

Experiments are needed to see whether extracellular signals regulate mechanical polarization (Fig. 1). Of the several signaling pathways known to be involved in tissue boundary formation in vivo (2528), Hedgehog signaling is especially intriguing because it is known to regulate myosin activity and to induce cadherin expression at tissue boundaries in Drosophila. Does modulating Hedgehog signaling affect actomyosin accumulation during boundary formation? It will also be critical to characterize downstream effects of mechanical polarization. Polarized cytoskeletal activity will likely result in changes in cell morphology and could provide a cue for cells to differentiate in response to their location at tissue boundaries.

Although physical reasoning can be useful in developmental biology, embryonic tissues test the limits of physical theories for collective phenomena and pattern formation. For example, it is clear that localization and transport of adhesion molecules play an important role in vivo (20). We need to develop a model that predicts how mobile adhesive molecules affect cell shapes and/or forces and vice versa. We also must extend these models to account for a broader range of mechanical effects at boundaries, such as differences in cell protrusivity and oriented cell divisions. Another challenge is to interpolate between short–time scale mechanical interactions that are important in tissues with highly mobile cells and long–time scale interactions that are important for dense tissues with mature contacts.

Collective mechanical interactions provide a robust mechanism for dictating cell movements, and new work suggests that actomyosin reorganization at boundaries is important for these mechanical interactions in vitro and possibly in vivo. Signaling pathways upstream of this reorganization may provide the tight regulation that is critical for proper embryonic development.

References and Notes

  1. Acknowledgments: The Amack lab is supported in part by a grant from NIH (R01HL095690). The Manning lab acknowledges support from the College of Arts and Sciences at Syracuse University.
View Abstract

Navigate This Article