The cell adhesion molecule DdCAD-1 regulates morphogenesis through differential spatiotemporal expression in Dictyostelium discoideum

The development of ordered structures in multicellular organisms proceeds through a complex series of cellular interactions, in which cells sort out and associate into specific multicellular groups. Cells frequently migrate from one site to another and reorganize at their final destination to establish the various tissue primordia. Therefore, individual cells or groups of cells must continuously change their relative positions and mutual adhesiveness during morphogenesis. It has been proposed that proteins involved in cell adhesion, cell-cell signaling and cell differentiation are genetic tool kits for multicellularity (Vogel and Chothia, 2006; King et al., 2007; Abedin and King, 2010).

The search for the origin of multicellularity often begins with cell-cell adhesion molecules (Bowers-Morrow, 2004; Harwood and Coates, 2004). Recent studies of cell adhesion molecules (CAMs) show that, in addition to cell-cell adhesion, they play key roles in sensing environmental cues and in generating signals that regulate a diversity of cellular processes, including gene expression, cell proliferation, cell polarity, cell motility and apoptosis. Differential temporal and spatial expression of cell adhesion molecules in specific groups of cells or tissues is stringently regulated during development. The formation or dissolution of specific adhesion complexes frequently leads to signals that serve as a major driving force behind migration and cell sorting, as well as differentiation and tissue formation. Therefore, CAMs have been recognized as major morphoregulators and signaling molecules (Edelman and Crossin, 1991; Gumbiner, 1996) because of their participation in many dynamic physiological and pathological processes, such as tissue architecture, organ regeneration and cancer metastasis.

The use of Dictyostelium discoideum to investigate the nature and function of CAMs was popularized by the pioneering work of Gerisch about half a century ago (Beug et al., 1970; Müller-Taubenberger and Bozzato, 2006). During development, cells transit from their solitary state of vegetative growth and embark on a social interactive phase. Cells undergo chemotaxis to form mound structures, which develop into slugs with clearly demarcated anterior and posterior zones. Slugs eventually culminate in the formation of fruiting bodies, which comprise a sorus of spores suspended by a filamentous stalk. The stalk cells display altruism and give up reproduction in order to benefit the spore cells by lifting them above the hazards of the soil, thus increasing their chances of dispersal to a more favorable environment. In addition, aggregative development exposes D. discoideum to chimerism, which includes cheating, whereby individuals have access to group benefits without contributing their fair share. This phenomenon raises the issues of how social cooperation persists in nature and what factors would permit individuals with a compromised phenotype to survive (Foster et al., 2004; Nowak, 2006; Gilbert et al., 2007). Among the many potential factors, cells tend to cooperate with genetically similar individuals and the cell adhesion genes csaA and tgrC1 have been found to confer survival advantages to D. discoideum cells that carry the same or highly related genes (Queller et al., 2003; Benabentos et al., 2009).

Several adhesion systems are known to influence the behavior of D. discoideum cells. Mutations in these genes have major effects on aggregate formation and morphogenesis (Coates and Harwood, 2001; Siu et al., 2004). The Ca²⁺-dependent cell-cell contact sites are mediated by the adhesion molecule DdCAD-1 (also known previously as gp24), which is encoded by the cadA gene and expressed soon after the initiation of development (Brar and Siu, 1993; Wong et al., 1996; Wong et al., 2002; Yang et al., 1997). The Ca²⁺/Mg²⁺-independent contact sites are mediated by csA (gp80) (Harris and Siu, 2002; Kamboj et al., 1989; Siu et al., 1985; Stein and Gerisch, 1996) and TgrC1 (LagC, gp150) (Dynes et al., 1994; Gao et al., 1992; Wang et al., 2000), and they contribute to the formation of stable cell-cell adhesion during the aggregation phase and the post-aggregation phase, respectively.

DdCAD-1 is unique among CAMs because it is synthesized as a soluble protein and transported to the plasma membrane by contractile vacuoles (Sesaki et al., 1997; Sriskanthadevan et al., 2009). DdCAD-1 contains two distinct domains with β-sandwich architecture. Whereas the N-terminal domain is involved in homophilic binding, the C-terminal domain tethers the protein to a membrane anchor (Lin et al., 2006). Studies on cadA⁻ cells have implicated DdCAD-1 in cell sorting and cell-type proportioning (Wong et al., 2002). Because DdCAD-1 is expressed transiently on the plasma membrane during cell aggregation (Sesaki and Siu, 1996), it is not known how DdCAD-1 regulates cell differentiation and cell sorting.

In this report, we examined chimeras consisting of cadA⁻ cells and the parental AX4 (cadA⁺) cells. Results show that the gene cadA elicited different social behavior under different environmental conditions. In chimeras of cadA⁺ and cadA⁻ cells, the cadA⁺ cells were favored for spore formation. However, on agar surfaces, cadA⁻ cells differentiated into spores whereas the stalk was made up of mostly cadA⁺ cells. Morphological studies of cells expressing DdCAD-1-GFP ectopically or treated with Alexa 488-conjugated recombinant DdCAD-1 have revealed remarkable changes in the temporal and spatial expression of DdCAD-1 during development. Whereas DdCAD-1 is abundant in the cell-cell contact regions in the anterior zone, it is present primarily in the cytoplasm of posterior cells. Cells that retain DdCAD-1 on the surface sort to the tip and differentiate into stalk cells. These data suggest that DdCAD-1 regulates the sorting behavior of cells through its differential subcellular distribution in prespore and prestalk cells.

PCR products were obtained from cadA cDNA and cadA genomic DNA cloned in pBluescript SK⁺ vectors. To construct the −631::cadA-GFP expression vector, the cadA promoter starting at −631 (GenBank accession number: AAFI02000079.1) was ligated to cadA cDNA and then inserted before GFP in the pIGFP vector, which contains the neomycin marker (kindly provided by Dr David Knecht, University of Connecticut, CT, USA) (see Fig. S1 in the supplementary material). Mutations in cadA were generated as described previously (Sriskanthadevan et al., 2009) and all constructs were sequenced from both ends. Plasmid DNAs were electroporated into cadA⁻ cells (Pang et al., 1999). Cells were collected at mid-log phase, washed twice in cold H50 buffer (20 mM Hepes, 50 mM KCl, 10 mM NaCl, 1 mM MgSO₄, 5 mM NaHCO₃, 1 mM NaH₂PO₄, pH 7.0) and then suspended in H50 at 2×10⁷ cells/ml. Cell samples (100 μl each) were mixed with ~5 μg of plasmid DNA and incubated for 5 minutes on ice in 1-mm cuvettes. Electroporation was carried out by two consecutive pulses of 0.85 kV with a capacitance of 25 mF applied to the cuvette and with a 5-second recovery between pulses. After 5 minutes of incubation on ice, the cells from each cuvette were transferred to 3 ml of HL5 (0.5% Difco proteose peptone No. 2, 0.5% BBL Thiotone E peptone, 1% glucose, 0.5% yeast extract, 2.5 mM KH₂PO₄, 2.5 mM Na₂HPO₄, pH 6.7) in wells of a six-well tissue culture plate (Pang et al., 1999). Transfectants were selected by sequential incubation in 2, 5, 10 and 20 μg/ml G418 (Sigma Chemical Co., St Louis, MO, USA) over a period of 3 weeks. The expression of wild-type or mutant DdCAD-1 was monitored by western blot analysis. Transfectants were maintained in 20 μg/ml of G418.

Dictyostelium strains, AX4 (parental strain of the cadA⁻ strain), cadA⁻ and various transfectants, were cultured either in association with Klebsiella aerogenes or in HL5 axenic medium (Sussman, 1987). Cells were grown and developed at 22°C. To determine the percentage of cadA⁻ spores in chimeric fruiting bodies, axenically cultured AX4 and act15::GFP;cadA⁻ cells were mixed to give a final concentration of 1×10⁸ cells/ml in the developmental buffer DB (5 mM sodium phosphate buffer, pH 6.5, 1 mM CaCl₂, 2 mM MgCl₂) (Fey et al., 2007). Cell mixtures at 20:80, 50:50 and 80:20 ratios were developed on either 2% non-nutrient agar buffered with KK-2 (2.2g/L KH₂PO₄ and 0.7 g/L K₂HPO₄, pH 6.4) or soil plates. Soil plates were prepared using commercially available garden soil (neutral pH) (All Treat Farm Ltd, Wellington North, ON, Canada) which was sieved to obtain particles of homogenous size (≤1 mm in diameter) and then autoclaved (Ponte et al., 1998). Aliquots of 20 g were distributed homogeneously on 90-mm Petri dishes, which were then moistened with 4 ml of sterile water. Aliquots of 0.25 ml (2.5×10⁷ cells) of mixed cells were pipetted onto soil plates to cover an area of ~2 cm². Four samples were placed on the each dish and then incubated at room temperature in a moisture box covered with aluminum foil. The cells were developed for 48 hours. Pictures were taken from two random regions for each example using a dissecting microscope for the quantification of fruiting bodies. Sori were picked randomly, treated with 0.05% SDS in DB, and the percentage of fluorescent cadA⁻ spores was estimated.

Cells were collected for development in 17 mM phosphate buffer (7.35 mM KH₂PO₄ and 2.65 mM Na₂HPO₄, pH 6.4) at 2×10⁷ cells/ml on a platform shaken rotating at 180 rpm at 22°C. After 4 hours, cells were resuspended at ~2.5×10⁶ cells/ml. Cells in 200-μl samples were dispersed by vortexing for 15 seconds and then allowed to re-aggregate by shaking at 180 rpm. At regular intervals, the numbers of non-aggregated cells were scored using a hemocytometer.

Axenically grown −631::cadA-GFP transfectants were washed free of medium and resuspended in DB at 4×10⁶ cells/ml. Aliquots of 0.5 ml were deposited on positively charged coverslips (Fisher Scientific, Pittsburgh, PA, USA) for 30-60 minutes and excess buffer was removed. Aggregates were fixed in 3.7% formaldehyde at different time points and permeabilized with 0.5% Triton X-100 for 15 minutes. Samples were incubated in 1% BSA in MCG buffer (50 mM MES, pH 6.4, 0.2 mM CaCl₂, 2 mM MgCl₂) for 30 minutes. Then, samples were stained with phalloidin (1:400) to mark the cell periphery (Molecular Probes, Eugene, OR, USA) for 30 minutes. Coverslips were mounted in DAKO fluorescent mounting medium (DakoCytomation, Glostrup, Denmark). Confocal images were acquired using the Zeiss LSM 510 microscope.

Cells were cultured either in association with K. aerogenes or axenically in HL5 medium. Cells were collected, washed and resuspended in 17 mM phosphate buffer at 10⁷ cells/ml. The cells were stained in 0.05% Neutral Red for 5 minutes at room temperature (Weijer et al., 1987). Prestalk cells stain more strongly with Neutral Red, which accumulates in the large intracellular acidic vesicles (Gross, 1994). After washing, cells (2×10⁸ cells/ml) were deposited on 1.5% non-nutrient agar (Eichinger et al., 2005). Development was carried out at 22°C in the dark. Slugs were photographed using a stereomicroscope at different time intervals. The distribution pattern of Neutral Red-stained cells within the slug was examined. The lengths of the whole slug (y) and the intensely stained anterior zone (x) were measured using Image J. The x/y ratio (R) was calculated and the frequencies of occurrence of the R values were determined. Experiments were repeated three times and 100-200 slugs were measured in each experiment. Data from all three experiments were pooled for the frequency calculations.

Cells were developed for 4-6 hours before staining with Neutral Red. After staining, cells were incubated at 2×10⁸ cells/ml with wild-type or mutant recombinant DdCAD-1 protein at 0.5 mg/ml or 3 mg/ml for 30 minutes. Then, 100 μl of cells were placed in a straight line ~5 cm long on 1.5% non-nutrient agar plates for development until the slug stage. His-DdCAD-1 was prepared as described previously (Sriskanthadevan et al., 2009) and labeled with sulfodicholorophenol esters (SDP) conjugated with Alexa Fluor 488 at a dye:protein ratio of 5:3 (Molecular Probes). Conjugated proteins were separated from un-reacted labeling reagent using Zeba desalt spin columns (Pierce, Rockford, IL, USA). The protein was resuspended at 6 mg/ml in 20 mM PIPES, pH 6.4. Then, cadA⁻ cells (2×10⁶ cells) were developed for 4-6 hours and incubated with Alexa-488-conjugated His₆-DdCAD-1 proteins at 3 mg/ml for 12 hours on coverslips and then excess buffer was removed. Cells were developed until the culmination stage before fixation with 3.7% formaldehyde for confocal microscopy.

To induce DdCAD-1 cap formation, anti-GFP mAb (1:100; Sigma Chemical Co., St Louis, MO, USA) was added to 1.2×10⁶ DdCAD-1-GFP-expressing cells suspended in 300 μl of 50 mM MES buffer, pH 6.3, and incubated for 30 minutes at room temperature. Alexa-568-conjugated goat anti-mouse antibody (1:400) was added and cells were rotated at room temperature for another 30 minutes. After washing, aliquots of cells were deposited on coverslips and allowed to attach for 15 minutes. The coverslips were washed gently with MCG buffer, fixed in 3.7% formaldehyde, washed and mounted for fluorescence microscopy.

Axenically grown −631:: cadA-GFP transfectants were washed and resuspended at 2×10⁸ cells/ml. Cells were deposited on 1.5% non-nutrient agar and development was carried out at 22°C in the dark. Cells at different stages were collected, dissociated and analyzed by flow cytometry (Becton Dickson LSRII) using excitation wavelengths of 488 nm and 603 nm. Data were analyzed using the WinMDI software.

The chemotactic migration assay was performed according to Wallace and Frazier (Wallace and Frazier, 1979). Cells were collected at 4 hours of development and resuspended at 2.5×10⁸ cells/ml in 17 mM sodium phosphate buffer. The cAMP stock solution (10 mM) was diluted to 100 μM or 250 μM. Non-nutrient 1.5% agar plates were prepared fresh and fully hydrated with buffer before the experiment. Four wells were created in the center of the quadrants of an agar plate. Then, the wells were filled with 30 μl of cAMP and 1 μl aliquots of the cell suspension were placed 5 mm away from the well. The plates were incubated at room temperature in a moisture box covered with aluminum foil. Cell migration was monitored using a dissecting microscope and the distance of migration was measured after 16 hours.

DdCAD-1 appears to function transiently at the aggregation stage of development as it redistributes from the cell-cell contacts when the cell adhesion protein csA moves into these regions (Sesaki and Siu, 1996). However, disruption of the cadA gene leads to aberrant cell sorting and defects in cell-type proportioning (Wong et al., 2002), suggesting a role for DdCAD-1 in the post-aggregation phase of development. Changes in the environmental conditions of development have been used successfully to investigate subtle functions of cell adhesion proteins (Ponte et al., 1998; Queller et al., 2003), we have taken a similar approach to investigate the role of DdCAD-1 in the post-aggregation stages of development. Chimeras of wild-type and mutant cells were analyzed under different environmental conditions. When the parental AX4 cells and GFP-expressing cadA⁻ cells were mixed for development on soil plates, disproportionate spore yields were observed in fruiting bodies collected at 48 hours of development (Fig. 1A). The yield of cadA⁻ spores was ~40% lower than the expected value if cells from both strains were equally represented in the chimeras, consistent with the developmental phenotype of cadA⁻ cells (Wong et al., 2002). These results show that the cadA⁺ cells promoted the survival of other cadA⁺ cells, but not the cadA⁻ cells. It is possible that the presence of DdCAD-1 on the cell surface allows AX4 cells to adhere preferentially to each other and facilitates the recruitment of cells into aggregates. The presence of soil particles might have compromised further the ability of cadA⁻ cells to form aggregates. Indeed, when AX4 cells and cadA⁻ cells were developed separately on soil plates, cadA⁻ cells developed much more slowly than AX4 cells. Whereas most of the AX4 cells completed development by 24 hours, only a few fruiting bodies were observed for cadA⁻ cells (Fig. 1B). At 48 hours, the number of cadA⁻ fruiting bodies reached only 40% of that of the AX4 cells, indicating that the loss of DdCAD-1 expression hampered the aggregation process.

Interestingly, cadA⁻ cells displayed opposite behavior when chimeras were developed on non-nutrient agar plates. Contrary to the data obtained with soil plates, the cadA⁻ cells differentiated primarily into spores (Fig. 2Aa,b). Only a few of the fluorescent cadA⁻ cells were found occasionally in the stalk. The quantitative data showed that almost all of the cadA⁻ cells in the chimeras formed spores when chimeras were made up of 20% cadA⁻ cells (Fig. 2Ac). When chimeras were made up of 80% cadA⁻ cells, essentially all spores were derived from the cadA⁻ strain. The cadA gene allowed the cadA⁻ to cheat and survive as they preferentially differentiated into whereas the cadA⁺ cells differentiated into stalk cells and eventually died.

In order to investigate how DdCAD-1 exerts an influence on cell survival during development, cadA⁻ cells were transfected with plasmid DNA containing the act15::cadA-GFP construct (see Fig. S2A,B in the supplementary material) to ensure even expression of the fusion protein in cells (see Fig. S2C,D in the supplementary material). Normal sorting pattern was restored in slugs by the ectopic expression of DdCAD-1 (see Fig. S2E,F in the supplementary material), confirming our previous observation (Wong et al., 2002). Transfected cells were mixed with cadA⁻ cells at different ratios and the location of the DdCAD-1-GFP-expressing cells in the chimeras was monitored at different stages. The green cells appeared mostly in the periphery at the mound stage. Most of them were present in the anterior prestalk region by the slug stage. To quantify the distribution of green cells, images of slugs were collected and divided into three segments for the measurement of fluorescence intensity (Fig. 2B). The data confirmed the preferential localization of green cells in the anterior zone of slugs. Spore yield was also identical to that of AX4 cells in similar chimeras.

As a control, cadA⁻ cells were transfected with an act15::GFP construct. The GFP;cadA⁻ cells were mixed with the AX4 (cadA⁺) cells for development. Most of the GFP;cadA⁻ cells were found in the middle and posterior regions of the slug, whereas the anterior region was occupied primarily by the AX4 cells (Fig. 2C). Even within the 80:20 chimeras, a distinctly ‘dark’ area was evident at the tip region. In a second control, GFP;cadA⁻ cells were mixed with cadA⁻ cells. As expected, the green cells distributed more or less evenly along the whole length of the slug (Fig. 2D). At the early culmination stage, all three types of chimeras showed patterns of green cell distribution similar to those observed at their respective slug stage (Fig. 2E). These results suggest that cadA⁺ cells preferentially sort to the anterior zone of chimeric slugs and differentiate into prestalk cells, whereas cadA⁻ cells occupy primarily the posterior two-third of slugs and eventually differentiate into spores.

The notion that DdCAD-1 plays an important role in the sorting of prestalk cells from prespore cells would predict the presence of DdCAD-1 on the cell surface during morphogenesis. To investigate how DdCAD-1 might be involved in cell sorting during development, cadA cDNA was fused to GFP and the fusion construct was ligated to the endogenous cadA promoter −631 (Yang et al., 1997) (see Fig. S3A in the supplementary material) for transfection into cadA⁻ cells. The use of the endogenous cadA promoter ensured that expression occurred according to the normal temporal program of development. G418-resistant clones were screened by direct visualization under a fluorescence microscope. The proper temporal expression of DdCAD-1-GFP in transfectants was confirmed by probing protein blots with antibodies against DdCAD-1and GFP (see Fig. S3B in the supplementary material). Whereas both transfectants and the parental cadA⁻ cells expressed the EDTA-sensitive or EGTA-resistant adhesion sites, cell cohesion assays showed that the EDTA/EGTA-sensitive or Ca²⁺-dependent cell-cell adhesion sites were observed only in the transfectants (see Fig. S3C in the supplementary material).

As DdCAD-1-mediated cell-cell adhesion requires Ca²⁺, we tested whether the Ca²⁺-binding capability of DdCAD-1 is required for proper cell adhesion and cell sorting during development. cadA⁻ cells were transfected with two constructs of DdCAD-1 that contained mutations in the Ca²⁺-binding site I and S(I+II). Expression of the mutant proteins were confirmed with western blots (Fig. 3Bb). When cell sorting experiments were performed using Neutral Red-stained cells, cadA⁻ cells expressing DdCAD-1-GFP ectopically showed the normal sorting pattern similar to AX4 cells (Fig. 3Ab-d). However, ectopic expression of DdCAD-1 containing mutated Ca²⁺-binding sites failed to rescue the cell sorting defects of the cadA⁻ cells. Most of the mutant slugs showed non-sorted patterns (Fig. 3Bc,d). These results indicate that the Ca²⁺-binding capability of DdCAD-1 is required for the proper sorting process of prestalk and prespore cells at the slug stage.

Transfectants that expressed DdCAD-1-GFP at levels comparable to the DdCAD-1 level in AX4 cells were selected for development and the temporal and spatial expression of DdCAD-1-GFP was monitored by confocal microscopy. Developmental structures were fixed and F-actin was stained with phalloidin to demarcate the cell boundaries (Fig. 4). Confocal microscopy revealed that DdCAD-1-GFP was synthesized inside the cytoplasm shortly after the initiation of development and became associated with the plasma membrane and filopodial structures as previously reported (Sesaki and Siu, 1996). At the mound stage, DdCAD-1-GFP was present mostly in the cytoplasm whereas enrichment in the cell-cell contact regions was not observed. A similar pattern of DdCAD-1 expression was evident at the tipped mound stage. Additionally, DdCAD-1 appeared to be absent from a subset of cells scattered throughout the aggregate. At the migrating slug stage, cells with higher levels of DdCAD-1 appeared in the anterior region. This difference in DdCAD-1 expression between the anterior cells and the posterior cells became more obvious in culminants.

When cells were subjected to fluorescence-activated cell sorting (FACS) analysis, a single group of fluorescent cells was observed at the mound stage (see Fig. S4 in the supplementary material). However, two groups of fluorescent cells became evident at the culmination stage. About 30% of the cells had a much reduced level of DdCAD-1, whereas the other group showed a broad range of higher fluorescence intensity. Taken together, the data indicate heterogeneity in DdCAD-1 expression as development progressed and that cells with a higher level of DdCAD-1 expression had a propensity to differentiate into prestalk cells.

A role for DdCAD-1 in cell sorting would imply the presence of DdCAD-1 on the cell surface. However, the high level of cytoplasmic DdCAD-1-GFP prevented the morphological analysis of membrane-associated DdCAD-1. Therefore, transfectants at the slug stage were treated with detergent for a longer period of time to allow intracellular DdCAD-1-GFP to leak out from cells. Confocal images of these cells revealed differential association of DdCAD-1-GFP with the plasma membrane along the long axis of the slug. An abundance of DdCAD-1-GFP was observed in cell-cell contact regions in the anterior zone, with close coincidence with the phalloidin-staining pattern, whereas the intracellular green fluorescence was relatively low (Fig. 5A). However, DdCAD-1 was present mostly in the cytoplasm in the posterior zone where membrane association of DdCAD-1 became less evident (Fig. 5B). During culmination, DdCAD-1 was present predominantly in cell-cell contacts in the upper cup region surrounding the descending stalk tube (Fig. 5C). In fruiting bodies, cells expressing high levels of DdCAD-1-GFP were found concentrated in the upper cup and lower cup regions of the sorus (Fig. 5D).

To determine whether DdCAD-1-GFP was associated with the surface of anterior cells, the anterior and posterior zones of slugs were cut and the cells were dissociated for further analysis. Cell-cell contacts of anterior cells, but not posterior cells, were found to be enriched with DdCAD-1-GFP (Fig. 5E). When antibody-induced clustering of DdCAD-1 was performed on living cells, DdCAD-1 ‘caps’ were observed on the elongated anterior cells, indicating that DdCAD-1 was present on the cell surface (Fig. 5Fa). By contrast, antibody treatment of posterior cells failed to induce cap formation (Fig. 5Fb), confirming that DdCAD-1-GFP was present at a very low level or absent from the cell surface.

DdCAD-1 is synthesized as a soluble protein, which adheres to the cell surface via interaction with an as yet unidentified membrane anchoring protein (Lin et al., 2006). As this membrane anchoring protein is expected to be present in cadA⁻ cells, exogenously added DdCAD-1 should bind to cadA⁻ cells. We therefore examined whether exogenously added recombinant DdCAD-1 could rescue the mutant phenotype of cadA⁻ cells. Recombinant His₆-DdCAD-1 (wild type) and His₆-S(I+II) DdCAD-1 (containing mutations in the Ca²⁺-binding pockets SI and SII) were purified from Escherichia coli (Fig. 6A). Protein blots showed that both proteins bound to cadA⁻ cells with similar efficiency (Fig. 6B). To monitor the cell sorting phenomenon, cells were incubated with Neutral Red for 5 minutes before the addition of recombinant protein. The wild-type His₆-DdCAD-1 protein rescued the cell sorting defect in cadA⁻ cells, whereas the mutant protein failed to do so (Fig. 6C). It became evident that exogenously applied DdCAD-1 was sufficient to restore the wild-type cell sorting pattern in null cells and its rescue function was dependent on Ca²⁺.

To monitor the fate of the bound recombinant DdCAD-1 during development, DdCAD-1 was conjugated with the fluorescent probe Alexa-488 before adding to cells and the distribution patterns of the labeled protein was followed by confocal microscopy. At the aggregation stage, DdCAD-1 was associated primarily with the cell membrane (Fig. 6Da,b). The honeycomb patterns of early mound and tight mound structures showed that the recombinant DdCAD-1 was present primarily in the cell-cell contact regions (Fig. 6Dc,d). However, in the core region of the tight mound (Fig. 6Dd), diffuse fluorescence in the cytoplasm became visible, providing the first sign of DdCAD-1 internalization. Also, cells in the outermost layer of the mound structure showed a higher level of fluorescence intensity associated with the cell membrane. At the slug stage, a lot of cells began to show fluorescence in their cytoplasm, indicative of protein internalization (Fig. 6De). Also, cells in the anterior zone appeared to be more elongated, suggesting that they might be moving toward the tip region. At the early culmination stage, the contact regions among cells in the anterior upper cup region were highly enriched with DdCAD-1, whereas DdCAD-1 was mostly in the cytoplasm in the posterior cells (Fig. 6Df). These results confirmed the extracellular role of DdCAD-1 in cell sorting. Additionally, membrane-bound DdCAD-1 was either shed or internalized by many cells at the slug stage.

Differential chemotactic response to cAMP has been implicated in the sorting of prestalk from prespore cells (Jiang et al., 1998). When AX4 cells and cadA⁻ cells were subjected the chemotaxis assay, AX4 cells migrated at a rate ~20% slower than cadA⁻ cells (Fig. 7A). These results would predict that AX4 cells took more time to reach the aggregation center than cadA⁻ cells. Indeed, when cadA⁺ and cadA⁻ cells were mixed at equal numbers for development on agar, the cadA⁺ cells were found primarily in the periphery of the mound structure (Fig. 7B).

Despite the previous finding that DdCAD-1 appears only transiently during early cell aggregation and becomes hardly detectable on the cell surface in the post-aggregation stages of development (Sesaki and Siu, 1996), studies presented in this paper have provided evidence that DdCAD-1 can exert opposite influences on the social behavior of Dictyostelium depending on the environment encountered by the organism. When development is carried out on soil plates, cadA promotes the survival of cells that express the same gene product on the cell surface. However, on non-nutrient agar plates, cadA confers opposite behavior on cells, allowing the cadA⁻ cells to behave as cheaters and survive while the cadA⁺ cells undergo terminal differentiation to form stalk cells and eventually die.

DdCAD-1 is expressed at high levels at the onset of development. During cell streaming, DdCAD-1 is abundant especially on membrane protrusions such as lamellipodia and filopodia, which are dynamic structures that are known to mediate initial contacts among Dictyostelium cells (Choi and Siu, 1987; Sesaki and Siu, 1996) as well as mammalian cells (Sandig et al., 1997; Vasioukhin et al., 2000). After entry into the cell stream, DdCAD-1 redistributes from the contact regions but remains associated with membranes on the outer edge of the stream to make contacts with incoming cells (Sesaki and Siu, 1996). The loss of DdCAD-1 would hinder the recruitment of cells into streams owing to compromised cell-cell adhesion. Development of cadA⁻ cells is known to be delayed for 5-6 hours and their ability to form spores is compromised (Wong et al., 2002). These defects are exacerbated by the presence of the rough and uneven surfaces of soil particles, resulting in the further delay of cell aggregation and subsequent developmental processes (Fig. 1). By contrast, wild-type cells complete development in 24 hours on soil plates, just as efficiently as development on agar surfaces. As the ability of the knockout cells to enter cell streams is compromised on soil plates, chimeric cell aggregates would contain mostly wild-type cells, resulting in a significant reduction of cadA⁻ spores. In this way, the cadA gene favors the survival of individuals that carry the same gene.

However, cadA switches to the opposite behavior in chimeras developed on agar surfaces and promote the survival of the knockout cells. An explanation of this unusual effect would involve two intertwining developmental processes, cell sorting and cell differentiation, both of which are defective in the cadA⁻ cells (Wong et al., 2002). Cells in the center of an aggregate are known to have a greater propensity to differentiate into prespore cells (Huang et al., 1997). In order that the cadA⁻ cells may form spores, the wild-type cells have to sort out from the knockout cells to occupy the anterior zone and differentiate into stalk cells. Both differential cell cohesiveness and chemotactic migration have been implied in the cell sorting process (Jiang et al., 1998). Our results show that the cadA⁻ cells undergo faster chemotactic migration than do wild-type cells. Although the difference is small, it is significant and should allow the knockout cells to reach the aggregation center earlier than the wild-type cells (Fig. 7). This might account for the abundance of wild-type cells in the periphery of chimeric aggregates, as illustrated in the schematics in Fig. 8.

Normally, cells that show greater cohesiveness for one another would stay in the core and the less cohesive ones would sort to the periphery in chimeric aggregates (Steinberg, 1975; Steinberg and Gilbert, 2004; Steinberg and Takeichi, 1994). But, how could the knockout cells remain in the core of the chimeric mounds? Interestingly, a similar sorting pattern has been observed when wild-type cells and cadA⁻ cells are mixed and gyrated in suspension cultures (Wong et al., 2002). Further analysis by these authors has revealed that a higher level of csA is expressed at the later stages of cell aggregation by the knockout cells. The increased level of csA compensates for the loss of DdCAD-1 expression, allowing the knockout cells to stay in the aggregation core owing to higher mutual adhesion. However, wild-type cells inside the core of the mound structure eventually sort out to the periphery owing to their lower level of csA expression. Cells in the periphery of the mound have a propensity to differentiate into prestalk cells and migrate to the anterior zone of the slug (Siu et al., 1983). Eventually, the cadA⁺ cells undergo terminal differentiation and die in order to confer altruistic benefits to the knockout cells. The influence of the cadA gene on cell behavior in chimeras appears to rely on an intricate balance in the expression of the various cell adhesion molecules. As the loss of csA expression elicits the precocious expression of TgrC1 in csaA-knockout cells (Wang et al., 2000), it is conceivable that this phenomenon might also account for the behavior of the csaA gene in chimeras of csaA⁻ cells and wild-type cells (Queller et al., 2003).

The influence of cadA on cell sorting is also corroborated by rescue studies. Ectopically expressed DdCAD-1 rescues the cell sorting defect observed in the knockout cells. The sorting defect can also be rescued by exogenous recombinant DdCAD-1. Rescued cells display the normal ratio in the anterior-posterior pattern of slugs. In both types of experiments, the cell sorting function of DdCAD-1 is dependent on the integrity of its Ca²⁺-binding sites.

Several mechanisms have been proposed for the formation of the anterior-posterior pattern in Dictyostelium. This pattern might arise by the sorting out of prestalk cells primarily to the anterior zone and the prespore cells to the posterior zone of the slug as a consequence of differential cell adhesiveness (Araki et al., 1994; Leach et al., 1973; Siu et al., 1983; Steinberg, 1975; Steinberg and Gilbert, 2004). Studies from several laboratories have suggested that prestalk and prespore cells arise in a spatially random manner throughout the aggregate and that sorting results from cell-type specific alterations in adhesion (Abe et al., 1994; Nicol et al., 1999). However, it has also been proposed that pattern can be generated by position-dependent differentiation of the different cell types (Krefft et al., 1984; Wolpert, 1981). The positional information model states that the value related to a position in a coordinate system determines cell differentiation (Wolpert, 1989; Wolpert and Szathmary, 2002). During Dictyostelium development, cells that are starved in different cell cycle phases tend to reach the aggregate at different times and therefore occupy different positions within the aggregate. Cells in the periphery, which are usually starved at S phase or early G2 phase, differentiate mostly into prestalk cells, suggesting that a combination of cell cycle stage and positional information dictates cell fate (Gomer and Firtel, 1987; Maeda et al., 2003; Weijer, 2009; Weijer et al., 1984; Zimmermann and Weijer, 1993). Although both wild-type and cadA⁻ cells are capable of differentiating into the different cell types required for fruiting body formation, the cadA⁻ cells achieve 100% cheating and form spores exclusively in chimeras. These results are consistent with position-dependent differentiation during development.

In addition to cell cycle position, a variety of factors are known to control cell fate determination. The differentiation-inducing factors (DIFs) are a family of small chlorinated diffusible molecules that induce stalk cell differentiation but suppress spore differentiation (Kay and Jermyn, 1983; Kay et al., 1999; Thompson and Kay, 2000). Recently, the nutritional state of cells has also been found to influence cell fate. Cells that starve first and have lower energy reserves tend to form the stalk, whereas the better-fed ones migrate faster to the aggregate center and differentiate into spores (Kuzdzal-Fick et al., 2010).

Evidently, a number of factors can influence the peripheral localization of cells and subsequent differentiation. Results shown in this paper have revealed heterogeneity in the temporal and spatial expression of DdCAD-1. Whereas most of the cells have internalized or shed their membrane DdCAD-1, a distinct subset of cells retain the expression of DdCAD-1 on the surface whether DdCAD-1 is expressed ectopically or added exogenously. These cells are localized first in the periphery of the aggregate and eventually move to the anterior zone occupying the tip region of the slug. Peripheral cells usually follow a circular path to move around the mound to the tip region (Nicol et al., 1999; Siegert and Weijer, 1995). In chimeras, the presence of DdCAD-1 on the surface of peripherally located wild-type cells would help them pull other DdCAD-1-expressing cells out towards them during the rotational movement, whereas prespore cells in the core would re-establish contact and continue to rotate uninterrupted.

In chimeras, cells in the anterior zone with high levels of DdCAD-1 on the cell surface display altruism and differentiate into stalk cells and die, whereas the posterior cells with DdCAD-1 localized mainly in the cytoplasm survive as spores. Remarkably, similar social behavior has been observed during development of wild-type cells. Additionally, in vitro rescue experiments have demonstrated that cells that retain DdCAD-1 in the contact region move into the anterior zone and differentiate into prestalk cells. These observations thus explain why genetically compromised cells might be able to survive and behave as cheaters in chimeras.

Our results also provide evidence for a direct extracellular role of DdCAD-1 in cell sorting, which in turn might account for the social behavior conferred by the cadA gene. Cell sorting during development is achieved by differential subcellular distribution of DdCAD-1, although the mechanism involved and its relationship with the differentiation signal remain obscure. D. discoideum cells express several adhesion systems to govern the social behavior of cells during morphogenesis. Future studies on the mechanisms that regulate the spatiotemporal expression of these proteins should lead to a better understanding of how cells communicate and influence behavior, as well as providing insights into the complexity of development.

We thank Drs David E. Isenman and Mitsu Ikura for invaluable advice and members of the Siu laboratory for discussion. This work was supported by Operating Grant FRN-6140 from the Canadian Institutes of Health Research to C.-H.S. S.S. was supported in part by an Ontario Graduate Scholarship. C.Y. is the recipient of an International Scholarship from the China Scholarship Council.