Rap1 GTPase Regulation of Adherens Junction Positioning and Cell Adhesion

See allHide authors and affiliations

Science  15 Feb 2002:
Vol. 295, Issue 5558, pp. 1285-1288
DOI: 10.1126/science.1067549


Cell-cell junctions are distributed evenly around the lateral circumference of cells within an epithelium. We find that the even distribution of adherens junctions is an active process that requires the small guanosine triphosphatase Rap1. Cells mutant forRap1 condensed their adherens junctions to one side of the cell. This disrupted normal epithelial cell behavior, and mutant cell clones dispersed into the surrounding wild-type tissue. Rap1 is enriched at adherens junctions, particularly between newly divided sister cells where it may reseal the adherens junction ring. The regulation of adherens junction positioning could play a role in cell mobility and cell division.

Cells within an epithelium are linked by several types of junctions. Encircling the apical ends of cells are adherens junctions, which link to the actin cytoskeleton intracellularly and can thereby transmit force across the lateral plane of the epithelium (1). Although much attention has been paid to the regulation of apico-basal localization of adherens junctions (2), little is known about the mechanisms that underlie their even distribution around the cell circumference. Rap1 is a small guanosine triphosphatase (GTPase) of the Ras family (3) that has a role in regulating Drosophila morphogenesis (4) through an undetermined mechanism. DuringDrosophila wing development, epithelial cells related by lineage normally stay in a coherent group (5). However, clones of cells mutant for Rap1 (6) dispersed into surrounding wild-type tissue (Fig. 1), indicating that loss of Rap1 function disrupts the normal cell-cell adhesion mechanism that keeps lineage-related cells in a coherent group. This phenotype has not been observed for other mutations studied by clonal analysis, including loss-of-function mutations in related GTPases such as Rho1 and Ras85D (7, 8). Cells lacking Rap1 function still respect the lineage restriction at the anterior-posterior compartment boundary [Web fig. 1 (9)]. Observations of shape defects in Rap1 mutant cells suggested that Rap1 might regulate apical cell-cell adhesion. Pupal wing cells mutant for Rap1 lacked the normal hexagonal shape, and the area of the apical, but not the basal, surface was reduced relative to that of wild-type cells [Fig. 1C (10)]. Dispersed mutant cells were often observed in pairs or groups of four cells [43 and 23%, respectively (11)].

Figure 1

Clones of wing cells mutant for Rap1 lose their normal cohesion. Mitotic recombination was induced in flies heterozygous for a nuclear green fluorescent protein marker (+/nls-GFP). The small number of cells that have undergone recombination divided to produce a cell lacking GFP (+/+) and a sister cell with two copies of nls-GFP (nls-GFP/nls-GFP) (6). Six days later the progeny of each cell were observed in the developing wings of the pupa as groups of cells (clones), either lacking GFP or containing a double dosage (GFP staining is in green, and Phalloidin staining of actin is in red). In wild-type animals (A), both clones have discrete borders; two recombination events have occurred in this wing, producing two pairs of clones. (B) Clones produced in a Rap1 heterozygote (Rap1/nls-GFP) resulted in Rap1 mutant clones lacking GFP, which have intermingled with surrounding wild-type cells (eitherRap1/nls-GFP or nls-GFP/nls-GFP). (C) Higher magnification of Rap1 mutant clones reveals that mutant cells had aberrant shapes and dispersed into wild-type tissue commonly in pairs (arrow) or fours (arrowhead). Bars, 10 μm.

To assess the role of Rap1 in cell-cell adhesion, we examined the subcellular localization of adherens junctions and the adjacent, more basal, septate junctions. In contrast to their even distribution around the apical circumference of wild-type epithelial cells, adherens junction components—including the cell-surface adhesion protein DE-cadherin (Fig. 2, A and D) and two cytoskeletal proteins, α-catenin [visualized with a green fluorescent protein (GFP)–α-catenin fusion protein (Fig. 3, A and B)] and β-catenin (10)—were found predominantly on one side ofRap1 mutant pupal wing cells. In a count of 856 cells containing such clusters of adherens junction components, 702 cells (82%) had adherens junctions condensed into a contact with just one neighboring cell. Clusters were also seen between a mutant cell and 2 other mutant cells (121 cells, 14%), or connecting a mutant cell with 3, 4, or 5 neighboring mutant cells (33 cells, 4%). Clusters of adherens junction proteins were observed only at interfaces between mutant cells, and not between a mutant and a wild-type cell. At interfaces between mutant and wild-type cells, normal levels of adherens junctions were observed.

Figure 2

The even distribution of adherens junctions is disrupted in Rap1 mutant clones, but septate junction positioning is not altered. Pupal wings containing Rap1mutant clones marked by the absence of GFP (green) were stained (red) for DE-cadherin (A), canoe (B), ZO-1 (E), and Discs large (C) (6). The level of adherens junction proteins at the interface betweenRap1 mutant and wild-type cells was the same as that between wild-type cells, as can be seen more clearly when DE-cadherin staining from (A) is shown alone (D). Arrowheads in (A) and (D) point to small groups of cells that have very low levels of DE-cadherin or ZO-1 staining. Bars, 10 μm.

Figure 3

The aberrant adherens junction distribution caused by loss of Rap1 is seen in the imaginal disc, but mutant cell dispersal does not begin until wing disc evagination. (A,C, F, H) Clones of Rap1mutant cells are distinguished by the absence of nuclear nls-GFP (22). (B and D) The adherens junctions are visualized with α-catenin–GFP, which showed the same aberrant distribution in 24-hour APF pupal wings (A and B) as seen with the other adherens junction markers in Fig. 2. Altered α-catenin–GFP distribution is also seen inRap1 mutant clones in the larval imaginal disc (C and D). As compared with wild-type clones (E), Rap1 mutant imaginal disc clones do not disperse (F). However, 2 to 3 hours later during wing disc evagination, dispersal of Rap1 mutant clones is seen (H) [compare with the wild-type clone in (G)]. Bars, 10 μm.

Two proteins that may form a molecular link between Rap1 and adherens junctions are the multidomain cytoskeletal linker proteins AF6/canoe and ZO-1. Both AF6 and its Drosophila ortholog canoe bind to activated Rap1 (12, 13), and canoe interacts with ZO-1 (14). Vertebrate ZO-1 binds to the adherens junction component α-catenin (15), thus completing a possible link from Rap1 to adherens junctions. Both canoe and ZO-1 localize to adherens junctions in normal Drosophilaepithelia (14, 16) and like the other adherens junction components, they distributed primarily to one side ofRap1 mutant cells (Fig. 2, B and E). Although ZO-1 also participates in vertebrate tight junctions (17) and may be present in Drosophila septate junctions (14), there was not a comparable alteration in septate junction–associated proteins in Rap1 mutant cells. The MAGUK protein Discs large (Fig. 2C) and the band 4.1 ortholog coracle (10) were evenly distributed around the circumference of Rap1 mutant cells. Thus, loss of Rap1 function specifically impairs even distribution of adherens junctions around the cell circumference. The maintenance of septate junctions could explain how Rap1 mutant cells still retain enough cell adhesion to remain within the epithelium.

The misplacement of adherens junctions in Rap1 mutant clones suggests that dispersion could be due to sorting caused by differential adhesion. L fibroblasts transfected with P-cadherin sort according to their level of cadherin expression (18), and such differential adhesion plays a role in Drosophila oocyte positioning at the posterior of the egg chamber (19,20). The Rap1 mutant cell-dispersal phenotype may be an additional in vivo example of cell sorting according to differential DE-cadherin–mediated adhesion, although in this case, the amount of adhesion is altered by the failure to distribute adherens junctions evenly around the cell circumference, rather than by altered overall cadherin expression. Provided that the quantity of adherens junction components reflects the strength of adhesion, Rap1 mutant cells could have adhered most strongly to mutant cells on the sides of the cell containing adherens junction clusters, very weakly to other mutant cells, and at normal strength to adjacent wild-type cells. Adhesion between mutant and wild-type cells that was stronger than adhesion between most Rap1 mutant cells could have drawn small groups of mutant cells into wild-type tissue (Web fig. 2). These results suggest that regulation of the subcellular distribution of cell-cell junctions could play a role in the mobility and invasiveness of cells within an epithelium.

Because adherens junctions are also misplaced in undispersed Rap1mutant cells, misplacement is likely to be the cause rather than the consequence of cell dispersal. In this case, mislocalization of adherens junctions during wing development should precede cell dispersal. Clonal cells mutant for Rap1 in the late (wandering) third-instar imaginal disc did not disperse (Fig. 3, E and F), yet the adherens junction component α-catenin was already mislocalized (Fig. 3, C and D), indicating that adherens junction mislocalization precedes dispersal. The larvae pupariate within a few hours of this time, and dispersal of Rap1 mutant cells was first observed 2 hours after pupariation (Fig. 3, G and H). Evagination of the disc during this time period requires extracellular protease activity, which is thought to loosen cell-cell and cell-extracellular matrix contacts, allowing cell rearrangements and shape changes to occur (21). Cell rearrangements can be observed as the elongation of marked clones; therefore, cells normally exchange neighbors even if they do not normally mix. Loosening of extracellular contacts likely allows Rap1 mutant cells to mix with their wild-type neighbors. Consistent with this, cell dispersion was initially more pronounced at the distal end of the evaginating wing (10), where cell rearrangements are first initiated (21).

To investigate whether Rap1 recruitment to adherens junctions is involved in aberrant junction distribution in mutant cells, we expressed a transgene encoding a GFP-Rap1 fusion protein. This fusion protein is under the control of the endogenous Rap1 promoter and was expressed ubiquitously throughout development (22, 23). In normal wing imaginal disc cells, GFP-Rap1 was broadly distributed in the cytoplasm and basolateral membrane and highly concentrated at the position of the adherens junctions (Fig. 4C), consistent with the possible interaction of Rap1 with adherens junction proteins canoe and ZO-1. Despite its own polarized distribution, Rap1 was not required for normal apico-basal distribution of adherens junctions; α-catenin was located apically in Rap1 mutant imaginal disc clones (Fig. 4D). β-Catenin and DE-cadherin also did not mislocalize along the apico-basal axis in Rap1 mutant pupal wing clones (10).

Figure 4

GFP-Rap1 is enriched at the site of adherens junctions, but is not required for apico-basal adherens junction positioning (22). (A) Diagram of the plane of focus for (B), (C), and (D). The gray bar represents the position at which sections were taken across a characteristic fold in third-instar wing imaginal discs. Apical cell edges face the inside of the fold. (B) Disc expressing nls-GFP and α-catenin–GFP, showing the two apical rows of adherens junctions (arrowheads). (C) Disc expressing GFP-Rap1. GFP-Rap1 is in the cytoplasm, associated with the basolateral membrane, and enriched at the position of adherens junctions (arrowheads). (D) α-Catenin–GFP in a Rap1 clone marked by the absence of nls-GFP. Apico-basal positioning of α-catenin–GFP is not altered in Rap1 cells. Bars, 10 μm.

The distribution of GFP-Rap1 in dividing cells suggests a mechanism by which Rap1 might normally act to ensure even adherens junction distribution. Dividing cells in the wing imaginal disc retain their adherens junctions with surrounding cells (24), and the localization of GFP-Rap1 was not altered during division (Fig. 5A). However, GFP-Rap1 was consistently enriched at the junction between newly formed sister cells (Fig. 5B). A transient enrichment of GFP-Rap1 between sister cells in the epidermis of living embryos was also observed (Web fig. 3). Hence, Rap1 may reorganize the adherens junction ring subsequent to or during late cytokinesis to ensure that appropriate amounts of adherens junctions are maintained around the circumference of new cells.

Figure 5

GFP-Rap1 is enriched at the junction between newly divided sister cells (22). (A) In the third-instar wing imaginal disc, cortical GFP-Rap1 is distributed evenly in cells in metaphase (arrow) and early cytokinesis (arrowhead). (B) GFP-Rap1 is enriched at the junction between sister cell pairs (arrowhead). Bars, 10 μm.

One model explaining how loss of Rap1 function during cytokinesis leads to adherens junction clustering is as follows. Maintenance of adherens junction distribution throughout cell division requires a mechanism to convert the single adherens junction ring into two rings, involving breaking and resealing of the ring during cytokinesis. Rap1 could be essential for this process. Failure to reseal the adherens junction ring could allow it to recoil to one side of the cell (Web fig. 4), driven by contraction of the actin and myosin present in the ring (21). This would cause rearrangement of cadherin contacts into clusters on sides adjacent to mutant cells with a similar defect, but not on the sides of the cell contacting wild-type cells, where cadherin distribution is stabilized at a normal density. Clusters would most likely form at the interface between sister cells, because both cells' rings recoil at the same time. However, clusters could also form between two adjacent mutant cells that are not sisters if they were in a similar state at the same time. Accordingly, the 14% of clusters between one mutant cell and two others demonstrates that clusters were present at interfaces between cells that are not sisters from their most recent division. Further rounds of division could lead to segregation of clusters into just one daughter cell, producing cells with few adherens junctions, as seen within some Rap1 mutant clones (Fig. 2, A and E).

Rap1 maintains circumferential adherens junction distribution in cells and thus shares with Rho GTPase family members the ability to regulate the cytoskeleton and cell adhesion. Thus, its demonstrated role in morphogenetic processes that are driven by adhesion-dependent cell shape changes and movements (4) may involve regulation of the link between the cytoskeleton and adherens junctions.

  • * Present address: Royal Society of New Zealand, 4 Hallswell Street, Wellington, New Zealand.

  • To whom correspondence should be addressed at Wellcome/CRC Institute, Tennis Court Road, Cambridge CB2 1QR, UK. E-mail: nb117{at}


View Abstract

Navigate This Article