Abp1 regulates pseudopodium number in chemotaxing Dictyostelium cells

During chemotaxis, cells polarize and coordinately move toward an extracellular signal. To effectively move, cells must detect shallow gradients of extracellular signal and translate this signal into changes in cell shape and cell adhesion (Parent, 2004). Particularly important for cell motility is the proper extension and retraction of actin-filled pseudopodia (Van Haastert and Devreotes, 2004; Williams and Harwood, 2003). Thus actin polymerization, depolymerization, and the branching and crosslinking of actin filaments must be under tight spatial and temporal control for effective cell movement (Iijima et al., 2002). To date, an increasing number of actin and actin-associated proteins have been identified that could regulate the organization of the actin cytoskeleton during directed chemotactic migration.

Dictyostelium discoideum is an excellent model to study chemotaxis and cell motility. When starved, Dictyostelium amoebae secrete pulses of extracellular cAMP and use this cue to stream together to form a multicellular aggregate. As they chemotax toward the cAMP signal, Dictyostelium amoebae rapidly extend and retract pseudopodia enriched in both actin and many actin-associated proteins. How these proteins coordinately regulate pseudopodium formation is not completely understood, but at least part of pseudopodium behavior is accomplished by myosins. Two myosin I proteins, MyoA and MyoB, accumulate at the leading edge of streaming cells to regulate retraction of pseudopodia and to suppress non-productive pseudopodium formation (Novak et al., 1995; Wessels et al., 1996). The conventional myosin, myosin II, concentrates at the trailing edge of chemotaxing cells, and functions to suppress lateral pseudopodia (Wessels et al., 1988). In addition to these myosins, actin-associated proteins also regulate actin dynamics within pseudopodia. The Arp2/3 complex, as well as its activators such as the WASP/SCAR family of adaptor proteins, functions in actin network dynamics at the leading edge (Machesky and Insall, 1998; Mullins et al., 1998; Zalevsky et al., 2001). PIR21, a component of the SCAR complex, regulates actin polymerization and affects pseudopodium formation (Bear et al., 1998; Blagg et al., 2003). While some of the important regulators of pseudopodium formation are known, how their diverse activities are coordinated is not well understood.

One important regulator of the actin cytoskeleton in eukaryotic cells is the protein Abp1. Abp1 was one of the first actin binding proteins identified in the yeast S. cerevisiae (Drubin et al., 1988). Overexpression of Abp1 in yeast alters actin organization and affects polarized cell growth (Drubin et al., 1990; Fazi et al., 2002). The mammalian Abp1 homologue functions in receptor-mediated endocytosis (Kessels et al., 2000; Kessels et al., 2001). Both yeast Abp1 and mammalian Abp1 contain an ADFH domain at their amino termini that binds to actin. Both yeast and mammalian Abp1 also have an SH3 domain at their carboxyl termini that binds ligands that function in endocytosis. This domain structure supports a role for Abp1 as a functional link between the actin cytoskeleton and other proteins (Qualmann and Kessels, 2002).

Although Abp1 is found in a diverse array of organisms, the contribution of Abp1 activity to cellular functions remains an open question. Here we report on the contribution of Abp1 to cellular function in Dictyostelium. We find that Dabp1 (Dictyostelium actin binding protein 1) is concentrated in dynamic regions of the cell cortex rich in F-actin. Moreover, Dictyostelium cells engineered to either abolish expression or to increase expression of the Dabp1 protein displayed a distinct phenotype. Altering levels of Dabp1 in cells profoundly impaired the ability of cells to limit pseudopodium number in early development, and limited rates of directed cell movement. Consequently, cells overexpressing Dabp1 formed smaller aggregation centers and small fruiting bodies during development. These results reveal a specific and critical role for Dabp1 in regulation of pseudopodium number during directed cell migration.

Searching the Dictyostelium gene database (www.dictybase.org) revealed a single gene for Abp1, (accession number: AY437927, Dictyostelium abpE) on chromosome 2. The predicted reading frame for the Dabp1 gene, abpE, encoded a protein of 481 amino acids with a molecular mass of 59 kDa. Similar to Abp1 proteins from other species, the amino acid sequence of the Dabp1 protein included an ADFH (actin depolymerizing factor homology) domain at the amino terminus (amino acid 1-130), and an SH3 (Src homology 3) domain at the carboxyl terminus (amino acid 422-481). Relative to the rest of the protein, these domains shared a high degree of amino acid identity with other abp1 proteins. Across its entire length, Dabp1 shared 21% identity with mouse Abp1 and 18% identity with Abp1 in Saccharomyces cerevisiae. The ADFH domain in Dabp1 shared 28.5% identity with mouse Abp1 and 19.2% identity with the S. cerevisiae Abp1. The SH3 domain was more conserved among Abp1 members. The SH3 domain in Dabp1 shared 37% identity with mouse Abp1 and 34% identity with Abp1 in S. cerevisiae (Fig. 1). Using a PCR-based approach, we cloned a cDNA for the gene and generated an antibody against the gene product. Western blots probed with affinity-purified antibodies raised against the E. coli-expressed fusion protein revealed that the Dabp1 protein migrated anomalously on SDS gels with a molecular mass of 70±3 kDa (Fig. 2). Whereas cells overexpressing Dabp1 tagged with GFP showed multiple bands in western blots, cells that overexpressed Dabp1 without the GFP tag did not display multiple bands (data not shown).

To determine the intracellular location of Dabp1, we stained Dictyostelium cells with affinity-purified anti-Dabp1 antibodies. Inspection of growing cells by fluorescence microscopy revealed an extensive association of Dabp1 with the cell cortex (Fig. 3, top panels). In polarized cells undergoing directed migration during early development, Dabp1 was especially concentrated at the leading edge (Fig. 3, bottom panels). To investigate a possible association of Dabp1 with the actin cytoskeleton, we stained cells simultaneously with phalloidin to label F-actin and with antibodies against Dabp1. In growing cells, Dabp1 colocalized with F-actin at some, but not all, regions of the cortex (Fig. 3, top panel). In streaming cells, Dabp1 colocalized with F-actin to a much higher extent and completely overlapped with actin at the leading edge (Fig. 3, bottom panel). We observed that the amount of Dabp1 varied in growing cells. One possible reason is that Dabp1 associates only with portions of the actin cytoskeleton that are particularly dynamic, a state that might differ between cells. By contrast, Dabp1 consistently associated with the leading edge of migrating cells, a region of active actin remodeling, suggesting that Dabp1 could associate preferentially with areas of dynamic F-actin in Dictyostelium.

The localization of Dabp1 at the cell cortex suggested that the actin cytoskeleton could be required for the cortical localization of Dabp1. To test this idea, we disrupted the filament-rich actin cortex in wild-type cells with cytochalasin A, a drug that depolymerizes F-actin. Subsequently, staining F-actin cells with Texas Red (TXRED)-labeled phalloidin confirmed that cytochalasin dramatically altered cortical actin. Labeling the cytochalasin-treated cells with an antibody against Dabp1 showed that the association of Dabp1 with the cell cortex was abolished concurrent with the loss of cortical actin (Fig. 4A). These results indicated that cortical actin was required for the cortical localization of Dabp1.

To examine a possible requirement of Dabp1 for the cortical localization of actin, we generated cell lines with altered expression of the Dabp1 protein. Using homologous recombination, we deleted the abpE gene to generate Dabp1 null mutants (Dabp1^- cells). Cell lines with increased levels of Dabp1 (Dabp1⁺ cells) were also made by transforming both a wild-type strain and the Dabp1 null mutants with an extrachromosomal plasmid that overexpressed the GFP-Dabp1 fusion protein. Western blots probed with an anti-Dabp1 serum confirmed the absence of the Dabp1 protein in Dabp1^- cells, and increased levels of Dabp1 protein in Dabp1⁺ cells (Fig. 2). Initial phenotypic analysis of these strains showed that the absence of Dabp1 did not influence cell growth, while the overexpression of Dabp1 retarded cell growth slightly.

To examine the influence of Dabp1 on cortical actin, we stained Dabp1^- cells and Dabp1⁺ cells with TXRED-labeled phalloidin. Inspection of growing cells by fluorescence microscopy revealed that neither the absence nor increased levels of Dabp1 protein altered the cortical organization of the actin cytoskeleton (Fig. 4B). Moreover, after treatment with cytochalasin and subsequent washout, Dabp1^- and Dabp1⁺ cells reestablished localization of actin to the cortex with kinetics similar to wild-type cells (data not shown).

The association of Dabp1 with the leading edge of developing cells suggested that Dabp1 could function in areas containing dynamic actin. Active remodeling of the actin cytoskeleton is particularly important when Dictyostelium amoebae develop into multicellular fruiting bodies in response to starvation. During early development, Dictyostelium cells respond to extracellular cAMP signals and coordinately stream into aggregation centers (Gerisch, 1987). Actin and actin-associated proteins form a dynamic gel organized into a dominant pseudopodium at the leading edge of polarized cells (Gerisch et al., 1993). We therefore tested a possible contribution of Dabp1 to the ability of cells to organize their actin into pseudopodia, adopt a polarized morphology, and move efficiently during development. Wild-type cells, Dabp1^- cells, and Dabp1⁺ cells were placed under starvation buffer to induce chemotaxis into aggregation centers.

Under these conditions, wild-type cells readily adopted a polarized shape (Fig. 5, left panels). By 12-13 hours, streams of wild-type cells moving into an aggregation center were readily apparent. By 18 hours, most wild-type cells were integrated into either streams or large aggregation centers. Dabp1^- mutants followed a similar developmental pattern: by 8 hours, cells were elongated, and by 13-18 hours Dabp1^- cells were incorporated into streams centered around an aggregation center (Fig. 5, middle panels). By contrast, Dabp1⁺ cells were dramatically delayed in early development. Dabp1⁺ cells were unable to adopt a polarized morphology efficiently; elongated cells were not seen until 13 hours under starvation buffer (Fig. 5, right panels). Small aggregation centers surrounded by broken streams of cells were only apparent after 18-19 hours.

To examine later developmental stages, we placed equal numbers of wild-type cells, Dabp1^- cells and Dabp1⁺ cells on nonnutrient agar plates, a surface on which Dictyostelium cells complete development to form multicellular fruiting bodies consisting of a spore-filled sorus on top of an elongated stalk. After 14 hours on nonnutrient plates, wild-type cells and Dabp1^- cells had formed multicellular migrating slugs. At 14 hours, most Dabp1⁺ cells had only aggregated into mounds, an earlier stage of development (Fig. 6). By 28 hours, both wild-type cells and Dabp1^- cells were fully developed into robust fruiting bodies consisting of a stalk topped with a sorus full of spores. By contrast, Dabp1⁺ cells took about 12 hours longer to develop and formed much smaller fruiting bodies (Fig. 6). Thus overexpression of Dabp1 resulted in delayed aggregation of streaming cells, formation of smaller aggregation centers and smaller fruiting bodies.

Light microscopy was used to monitor wild-type cells, Dabp1^- cells and Dabp1⁺ cells during chemotaxis (Fig. 7). As they streamed together, wild-type cells adopted a highly elongated and polarized morphology. The average length of polarized wild-type cells was 28.1±3.1 μm (n=30). Dabp1^- cells were also elongated in early development, and similar in average length to wild-type cells (26.2±3.3 μm; n=12). However Dabp1⁺ cells failed to adopt an elongated shape during early development. The average length of the streaming Dabp1⁺ cells was 18.9±2.0 μm (n=37), only 67% of the length of streaming wild-type cells.

In addition to decreased cell length, cells that overexpressed Dabp1 also displayed an increase in pseudopodium number. As they chemotaxed, wild-type cells generally formed two protrusions from the cell body, a dominant pseudopodium at the leading edge and a trailing uropodium at the rear. Approximately 92% wild-type cells exhibited this morphology (Table 1). Of the remaining 7.5% of wild-type cells, none had more than five cellular protrusions. The number of pseudopodia increased somewhat in three independent Dabp1^- cell lines: 26.5% of the null mutants projected multiple pseudopodia; none had more than five pseudopodia. By contrast, Dabp1⁺ cells displayed a dramatic increase in pseudopodium number. The majority of cells (∼90%) overexpressing Dabp1 had multiple pseudopodia (Table 1), with some cells exhibiting as many as 10 pseudopodia. The stellate shape of Dabp1⁺ cells made it difficult to discern a distinct leading pseudopodium and/or a trailing uropodium. Imaging chemotaxing cells showed that wild-type cells generally extended a single pseudopodium persistently in the direction of movement. Most Dabp1⁺ cells, however, extended pseudopodia in multiple directions. Moreover, we also noticed that the extension and retraction of pseudopodia was slower in the Dabp1⁺ cells, resulting in slower turnover of pseudopodia relative to wild-type cells (supplementary material Movie 1). Increased numbers of pseudopodia extending in multiple directions were also observed in cells overexpressing Dabp1 without the GFP tag, indicating that the phenotype was not due to the addition of the GFP tag to abp1 (supplementary material Movie 2). The impaired ability to extend a dominant pseudopodium in a single direction may explain why Dabp1⁺ cells take longer to stream into aggregation centers.

The pseudopodia formed by wild-type cells are enriched in filamentous actin. The deletion of abp1 had a small effect on pseudopodium number whereas overexpression of abp1 resulted in a significantly increased number of pseudopodia in chemotaxing cells. To determine whether these pseudopodia were also enriched in F-actin, we stained cells with TXRED-labeled phalloidin (Fig. 8). Chemotaxing wild-type cells formed pseudopodia at their leading edges enriched in F-actin. Although Dabp1^- and Dabp1⁺ cells formed multiple pseudopodia, all had normally enriched F-actin.

The delayed aggregation and altered morphology exhibited by Dabp1⁺ cells prompted us to examine whether these cells also displayed defects in motility. To initiate chemotaxis, we submerged wild-type cells, Dabp1^- cells and Dabp1⁺ cells under starvation buffer. When the cells were actively streaming, we imaged independent cells moving toward aggregation centers. Wild-type cells moved persistently in one direction (Fig. 9A; supplementary material Movie 3). Approximately 90% of movements were greater than 8 μm/minute, with an average velocity of 12.5±6 μm/minute (n=159, 15 independent cells) (Fig. 9B). Dabp1^- cells also moved persistently in one direction (Fig. 9A; supplementary material Movie 4), with around 80% moving at greater than 8 μm/minute, and an average velocity of 11.1±6 μm/minute (n=129, 11 independent cells) (Fig. 9B). In comparison, Dabp1⁺ cells frequently stalled in one location without productive translocation (Fig. 9A; supplementary material Movie 1). When Dabp1⁺ cells translocated, only 34% of their movements were greater than 8 μm/minute, and their average velocity was only half that of wild-type cells (6.8±4 μm/minute; n=166, 12 independent cells) (Fig. 9B).

The Dabp1 protein has an ADFH domain at its amino terminus and an SH3 domain at its carboxyl terminus. To test the functional contribution of these domains to Dabp1, we designed two plasmids to express either the ADFH domain or the SH3 domain (Fig. 10A). Each domain was expressed as a fusion protein with GFP at the amino terminus. The expression plasmids were introduced into wild-type cells. Western blots probed with anti-GFP antibodies showed high expression levels for the two GFP-tagged proteins (data not shown).

To examine the influence of these domains on cell phenotype, cells expressing the ADFH domain (ADFH⁺ cells) and cells expressing the SH3 domain (SH3⁺ cells) were induced to begin chemotaxis by starvation. ADFH⁺ and SH3⁺ cells formed normal fruiting bodies when plated on non-nutrient agar. Nonetheless, overexpression of the SH3 domain dramatically altered the morphology of cells during streaming in early development (Fig. 10B). In contrast to ADFH⁺ cells, which appeared similar in morphology to wild-type cells, most streaming SH3⁺ cells exhibited multiple pseudopodia (Table 1). Cells overexpressing the SH3 domain exhibited up to 7 pseudopodia during chemotaxis. Thus the SH3 domain appeared to be a key functional determinant in controlling pseudopodium number during the aggregation phase of early development.

In early development, chemotaxing cells rapidly rearrange their actin cytoskeleton in order to adopt an effective shape for cell movement (Pollard and Borisy, 2003; Williams and Harwood, 2003). Remodeling the actin cytoskeleton results in extension of a dominant pseudopodium at the leading edge and retraction of the uropodium at the trailing edge, coordinated shape changes that allow cells to move in response to a extracellular chemical gradient (Varnum-Finney et al., 1987). Key to productive motility is the tight control of pseudopodium number for cells moving toward an aggregation center (Chung and Firtel, 2002). Here we have shown that overexpression of Dabp1 resulted in formation of multiple pseudopodia and impaired cell motility during early development. These defects probably account for the delay in aggregation center formation in cells overexpressing Dabp1. Because of their inability to form a single dominant pseudopod, Dabp1⁺ cells frequently stalled while extending pseudopodia in multiple directions, displayed reduced cell velocities and failed to move persistently to an aggregation center. As a consequence of impaired motility, cells overexpressing Dabp1 formed smaller aggregation centers on starvation plates and made small fruiting bodies.

A concern with interpreting overexpression experiments is that phenotypes resulting from overexpression of a protein may not reflect the normal function of the protein. Nonetheless, the similarity in phenotypes exhibited by Dabp1⁺ cells and Dabp1^- cells supports the interpretation that the phenotypes associated with overexpression of Dabp1 reveal its physiological role. Both Dabp1⁺ cells and three independent Dabp1^- cell lines projected excess pseudopodia during chemotaxis. Overexpression of Dabp1 yielded stronger phenotypes: relative to Dabp1^- cells, Dabp1⁺ cells exhibited an increased number of pseudopodia, and also displayed a dramatic reduction in cell motility. The milder phenotype associated with Dabp1^- cells could reflect redundancy in proteins that regulate the actin cytoskeleton: when Dabp1 is depleted, other proteins could partially substitute for its function. Functional redundancy during development has been noted previously for other proteins associated with the actin cytoskeleton (Witke et al., 1992).

Formation of extra pseudopodia was limited to developing Dabp1⁺ cells; growing cells that overexpressed Dabp1 were normal in appearance. This intracellular role for Dabp1 was also specific for pseudopodium formation and not general for other actin-based processes. Most other actin-based processes were intact in the Dabp1 null mutants, including cytokinesis, pinocytosis and phagocytosis (data not shown). Pseudopodia in streaming cells contain abundant amounts of F-actin and the pseudopodia in Dabp1⁺ cells were similarly rich in F-actin. Thus Dabp1⁺ cells were able to construct pseudopodia that appeared normal, but were defective in limiting the number of pseudopodia during development.

A repertoire of proteins important for regulation of pseudopodium number is emerging. One class of proteins known to regulate pseudopodium number in Dictyostelium is unconventional myosins. MyoA and MyoB localize to the leading edge of streaming cells and play critical roles in regulating where and when a cell forms pseudopodia (Morita et al., 1996; Titus et al., 1993; Wessels et al., 1996). These two myosin Is play independent roles in suppressing lateral pseudopodium formation during chemotaxis (Falk et al., 2003). Null mutants for either MyoA or MyoB extend a greater number of pseudopodia. Overexpression of MyoB inhibits pseudopodium formation and impairs cell motility (Novak and Titus, 1997). Like myosin Is, Dabp1 was enriched in pseudopodia, and conceivably could influence pseudopodium number by participating in a regulatory pathway involving MyoA or MyoB. Dabp1 could also contribute to other pathways that regulate polymerization of actin in pseudopodia. For example, Dabp1 could influence the activity of a protein that promotes actin polymerization, such as Arp2/3, also localized in pseudopodia.

How might Dabp1 influence the activity of another regulatory protein? The phenotype resulting from overexpression of domains of Dabp1 supports the idea that Dabp1 binds a regulatory protein via the carboxyl-terminal SH3 domain. Overexpressing the ADFH domain of Dabp1 had little effect on the morphology of chemotaxing cells. By contrast, overexpression of the SH3 domain resulted in an increased number of pseudopodia in chemotaxing cells, similar to the defect seen in cells overexpressing full-length Dabp1. This dominant-negative phenotype suggests that the Dabp1 protein influences pseudopodium number through binding partners for the SH3 domain. Binding partners for the SH3 domain of Abp1 homologues in other species have also been found to be important for its function. The SH3 domain of mammalian Abp1 binds to the GTPase dynamin, and overexpression of this SH3 domain in cultured cells causes defects in receptor-mediated endocytosis (Kessels et al., 2001). The SH3 of yeast Abp1 binds to six ligands, all proteins involved in endocytosis (Fazi et al., 2002). Conceivably, overexpression of the SH3 domain of Abp1 could sequester partners for all of the SH3 domains involved in multiple pathways. However, overexpression of the SH3 domain of Abp1 probably sequesters a limited set of discrete ligands. Different SH3 domains are sufficiently specific to distinguish subtle differences in the primary structure of potential ligands (Rickles et al., 1995; Sparks et al., 1996). Thus, it seems more likely that the dominant negative effect of overexpression of the SH3 domain is due to binding a proline-rich ligand important for limiting pseudopodium number in chemotaxing cells.

The wild-type strain Ax2, Dabp1^- (Dabp1 null) cells and Dabp1⁺ cells (cells overexpressing Dabp1) of Dictyostelium discoideum were used. Cells were cultured in HL-5 medium on Petri dishes at 18°C. Dabp1⁺ cells were maintained in HL-5 with 10 μg/ml G418 (geneticin; Gibco-BRL); Dabp1^- cells were cultured in HL-5 with 5 μg/ml blasticidin (ICN, Biomedicals).

The DNA sequence of yeast Abp1 (accession number X51780) was used to search the Dictyostelium genome database (http://www.dictybase.org) for the best match using reciprocal BLAST. A single gene ortholog was obtained, Dictyostelium abpE. A second database, InParanoid (http://inparanoid.cgb.ki.se) searched with BLAST also identified the abpE gene as the Dictyostelium abp1 ortholog. A complete cDNA clone was obtained from a cDNA library using a polymerase chain reaction (PCR)-based strategy using primers 5′CCGGATCCATGGCATCATTAGATATTAGTGATCCAGATATTAC3′ and 5′CCGAATTCCTCGAGTTACAATTGTTGTACGAAATTAGATGGGAAG3′. Predicted protein sequences were analyzed using the Megalign program (DNAStar, Inc., Madison, WI) with the default ClustalV parameters.

A cDNA for Dabp1 was cloned into the BamHI and EcoRI sites of the plasmid PGEX-2T. In this plasmid, the cDNA for Dabp1 was inserted downstream of the glutathione S-transferase (GST) gene, resulting in expression of a GST-Dabp1 fusion protein. This plasmid was transformed into Escherichia coli DH5α for large-scale protein purification. To purify the GST-Dabp1 fusion protein, cell cultures and cell lysates were prepared as described previously (Vithalani et al., 1998). The purified GST-Dabp1 protein was used to generate polyclonal anti-Dabp1 antibodies in rabbits (Cocalico Biologicals, Reamstown, PA). The polyclonal anti-Dabp1 antibodies were affinity-purified on a GST column using a GST orientation kit (Pierce, Rockford, IL).

The vector used to generate Dabp1^- cells was constructed by first using PCR to amplify sequences corresponding to regions that flanked the 5′ and the 3′ ends of the entire coding sequence for the abpE gene. The primers 5′-GAGCTCTTGTAGTTCCCCTTACCAAATCATTGTG-3′ and 5′-GGATCCGTTTTGGTTCAAAGAATAATATTTGTTGG-3′ were used to amplify a 5′ fragment flanked by SacI and BamHI sties. The primers 5′ AAGCTTAACTATTTCCATTTGTTTTCCTTATTTATCC 3′ and 5′-CTCGAGAGGGGTGTCTCTGGCTGTGTCG-3′ were used to amplify a 3′ fragment flanked by HindIII and XhoI sites. Both fragments were subsequently cloned into the plasmid pSP72-BSR (Wang et al., 2002) so that the blasticidin gene was flanked by these two fragments. A linear DNA fragment containing this abpE gene replacement cassette was excised using the flanking SacI and XhoI restriction sites. After treating with phosphatase, 5-10 μg of this linear DNA fragment was used to transform Ax2 wild-type cells by electroporation. Cells were diluted into 96-well plates, and cultured with blasticidin to generate clonal transformants. Clonal transformants in which the entire coding region of Dabp1 was replaced by integration of the blasticidin marker were identified using PCR analysis. The absence of Dabp1 expression was confirmed in these cells by western blot analysis with anti-Dabp1 antibodies.

To make a construct for Dabp1 expression, a cDNA for Dabp1 expression was PCR-amplified from a Dictyostelium cDNA library with primers 3′-CCGAATTCCTCGAGTTACAATTGTTGTACGAAATTAGATGGGAAG-5′ and 5′-GGATCCGCAGCAGCAGCAGCAATGGCATCATTAGATATTAGTGATCCAGATATTAC-3′. This fragment was cloned into the BamHI and XhoI sites of the plasmid pTX-GFP (Levi et al., 2000), placing green fluorescent protein (GFP) at the amino terminus of the Dabp1 protein. A linker containing five alanines was included between GFP and Dabp1.

To make an expression vector for Dabp1 without GFP tag, we modified the pTX-GFP-Dabp1 plasmid by removing the GFP cassette at the BamHI and HindIII sites, blunting and re-ligating the plasmid.

The ADFH and SH3 domains of Dabp1 were identified by comparing the sequence of Dabp1 with conserved domains using NCBI software (http://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi). To make an expression vector for the ADFH domain, the first 130 amino acids of Dabp1 were cloned by PCR from a Dictyostelium cDNA library. This sequence was cloned into the pTX-GFP plasmid with GFP placed at the amino terminus of the ADFH domain. A linker of five alanines separated the GFP and the ADFH domain.

To make an expression vector for the SH3 domain, the carboxyl-terminal 69 amino acids of Dabp1 (the 59 amino acids of the entire SH3 domain and ten additional amino acids amino-terminal to the SH3 domain) were amplified by PCR from a Dictyostelium cDNA library. This DNA fragment was cloned into the pTX-GFP plasmid with GFP fused at the amino terminus of the SH3 domain.

Ax2, Dabp1^- and Dabp1⁺ cells were harvested at mid-log phase, adjusted to 2×10⁶ cells/ml and allowed to settle on glass coverslips for 20 minutes at 18°C. Cells were fixed for 5 minutes at -20°C by incubation with methanol containing 1% formaldehyde. To improve imaging, some cells were gently flattened with a square of 2% agar NA (Amersham Biosciences, Uppsala, Sweden) before fixing for 5 minutes in methanol containing 1% formaldehyde at -20°C (Fukui et al., 1987). Subsequently, fixed cells were processed for immunostaining.

For immunostaining, fixed cells on coverslips were incubated with affinity-purified anti-Dabp1 antibodies (15 μg/ml) at 37°C for 40 minutes. After washing four times with phosphate-buffed saline (PBS), the coverslips were incubated with BODIPY FL-conjugated goat-anti-rabbit IgG (30 μg/ml; Molecular Probes, Eugene, OR) at 37°C for 40 minutes. TXRED-labeled phalloidin (0.3 unit/ml; Molecular Probes) was then used to detect actin in fixed cells.

For cytochalasin treatment, cells adherent to coverslips were incubated with 10 μM cytochalasin A at room temperature. After 30-60 minutes, cells were flattened and fixed for 5 minutes in methanol with 1% formaldehyde at -20°C. For double-labeling of both Dabp1 and actin, immunostaining with anti-Dabp1 antibodies was followed with TXRED-labeled phalloidin. Images were taken using an inverted Nikon microscope TE200 (Nikon Instruments, Dallas, TX) with a 100× 1.4 NA PlanFluor objective and a Quantix 57 camera (Roper Scientific, AZ) controlled by Metamorph software (Universal Image, PA), and then processed using Adobe Photoshop software.

For studying the morphology of streaming Ax2, Dabp1^- and Dabp1⁺ cells, the cells were imaged on an inverted Nikon TE200 microscope with either the 20× objective or the 100× objective. For studying development, images were captured on a Zeiss Semi SR microscope with 0.8× or 1.2× objectives controlled by NIH image software.

To study the motility of streaming, Ax2, Dabp1^- and Dabp1⁺ cells were harvested at late-log phase, washed once with the starvation buffer, PDF (2 mM KCl, 1.1 mM K₂HPO₄, 1.32 mM KH₂PO₄, 0.1 mM CaCl₂, 0.25 mM MgSO₄, pH 6.7), and then resuspended into the same PDF buffer at a density of 2×10⁶ cells/ml. 400 μl of the cell suspension were added to a one-well coverslip-chamber (Nulge-Nunc Int., Naperville, IL) and incubated at 18°C until cells were actively streaming. Images of cells moving toward aggregation centers were recorded at 6-second intervals using an inverted Nikon TE200 microscope (Nikon Instruments) with a 60× 1.4 NA PlanFluor objective and a Quantix 57 camera (Roper Scientific, AZ) controlled by Metamorph (Universal Image Corp., PA).

The velocity of streaming cells was measured by studying the movement of a single cell outside of the aggregation center in the images of moving cells. The longest extension of the uropodium of the streaming cell was used as a starting point. The x and y position of the uropodium in each frame was noted. The distance moved by the tail at a given time was calculated from the z value derived from the x and y values (z²=x²+y²) and converted from pixels to μm (1 μm=4.52 pixels). Velocity was calculated from the z value and time (v=z/t).

To study the localization of Dabp1 or actin in the starved cells undergoing chemotactic aggregation, cells were harvested at late log phase, washed once with PDF, and then resuspended at a density of 2×10⁶ cells/ml in PDF buffer. Cell suspension (200 μl) was spotted on a coverslip and incubated in a humidified chamber at 18°C until cells were actively streaming. Cells were then gently flattened with an agarose square, and fixed and stained with affinity-purified anti-Dabp1 antibodies or TXRED-labeled phalloidin as described above.

To study the development of starved cells on agar plates, 1×10⁸ Ax2, Dabp1^- or Dabp1⁺ cells were harvested at late log phase. Cells were washed once with PDF and resuspended into 3 ml PDF. The cell suspension was spread on a PDF agar plate (20 g agar/l PDF, 30 ml/plate, prepared the day before use) and allowed to settle for 40 minutes at room temperature. After aspirating excess liquid, cells were allowed to develop at 18°C for 30 hours. Images were taken with a Zeiss Semi SR microscope with a 0.8× or 1.2× objectives controlled by NIH image software. Western blot analysis demonstrated an equivalent amount of Dabp1 protein throughout development in wild-type and in Dabp1⁺ cells (data not shown).

Developing Ax2, SH3⁺, ADFH⁺, Dabp1^- and Dabp1⁺ cells were prepared and imaged as described for the velocity measurement assays. For Dabp1^- cells, three independent cell lines were examined. For each cell line, approximately 80-100 cells were selected and an outline of the cell perimeter was drawn. Protrusions extending from the cell body were counted as pseudopodia.

We would like to thank Arturo De Lozanne for helpful comments on the manuscript and Tom Egelhoff for the gift of the pTX-GFP. This work was supported by NIH RO1 GM048625 to T.J.O.