A SAP domain-containing protein shuttles between the nucleus and cell membranes and plays a role in adhesion and migration in D. discoideum

Dynamic cellular movements are central in many developmental, physiological and pathological processes. An elaborate network of signaling pathways are responsible for coordinating cell movement. Through these pathways cells are able to sense environmental signals and respond directionally to those signals, a process referred to as chemotaxis. Some pathways coordinating chemotaxis are well conserved between mammalian cells and the amoeba Dictyostelium discoideum. Because of this conservation, Dictyostelium has been exploited for its use as a model system to study cell movement and chemotaxis (Parent, 2004).

Dictyostelium cell migration is important throughout its life cycle, during both its single-celled growth phase and during development where it forms multicellular fruiting bodies. During growth, cells chemotax toward folic acid produced by their food source, bacteria. In development, starved Dictyostelium cells move to secreted cAMP. This chemotaxis is essential in early stages of multicellular fruiting body development. A chemotactic response is initiated by the binding of a chemoattractant to cellular receptors. An intracellular amplification of the response and cellular polarization then occur, distinguishing front from rear and permitting a reorganization of the cytoskeleton. Polymerized F-actin localizes to the front of a moving cell while an actomyosin organization is present in the rear (Parent and Devreotes, 1999; Devreotes and Janetopoulos, 2003). This cytoskeletal organization allows for F-actin rich pseudopod extensions in the front and a simultaneous suppression of pseudopods and cell retraction in the rear (Parent and Devreotes, 1999; Devreotes and Janetopoulos, 2003). The well-studied PI3K signaling pathway is an example of a conserved pathway involved in chemotaxis that permits polarization of the cell and is important for cell speed and directionality (Parent, 2004). The kinase PI3K phosphorylates at the 3′ hydroxyl of phosphatidylinositol-phosphates while the lipid phosphatase, PTEN, removes this phosphate (Bagorda et al., 2006). PI3K localizes to the front of the cell and PTEN to the rear; thus at the front of the cell PI3K produces an enrichment in membrane phospholipids that allow for the binding of downstream migrational effector proteins, such as those involved in actin dynamics (Parent et al., 1998).

Another pathway shown to have roles in Dictyostelium migration is the ampA pathway (Blumberg et al., 2002; Casademunt et al., 2002; Noratel et al., 2012; Varney et al., 2002a; Varney et al., 2002b). AmpA mutants have been characterized revealing roles for the protein in different aspects of chemotaxis. Overexpression of AmpA protein results in an increase in plaque size due to more rapid cell migration, and a knockout of the ampA gene causes a reduction in plaque size with reduced cell migration (Noratel et al., 2012). AmpA has also been shown to influence the level of actin polymerization and cellular adhesions, both important for cell migration. To try to understand the mode of action of AmpA, suppressor mutants were created that suppress the ampAOE increased migration phenotype. Mutants were obtained and the genes responsible for suppression were determined to have roles in chemotaxis. One of these, described here, is the sma gene, which encodes a protein that contains a SAP DNA-binding domain and nuclear localization signals, as well as a domain similar to phosphatase-tensin (PTEN).

The SAP domain consists of a 35 amino acid motif, which contains conserved hydrophobic and charged amino acids (Aravind and Koonin, 2000). This motif has been found in a number of chromatin associating proteins, such as scaffold attachment factors, DNA repair proteins, RNA processing complexes and proto-oncogene proteins (Ahn and Whitby, 2003; Aravind and Koonin, 2000; Böhm et al., 2005; Kipp et al., 2000). Little is known about the SAP–DNA interaction, but some evidence suggests that only a weak interaction occurs between the SAP domain and DNA. This suggests a more likely role for the SAP motif in targeting or stabilization of the DNA structural organization (Böhm et al., 2005; Kipp et al., 2000). It is also possible that the SAP domains are capable of tethering proteins involved in pre-mRNA processing to active regions of chromatin. SAP domains are found on diverse nuclear proteins and they are found in a wide range of organisms, including yeast, plants, animals, and even bacteriophage (Aravind and Koonin, 2000; Aravind and Koonin, 2001).

Disruption of sma in WT Dictyostelium cells revealed an increased efficiency in chemotaxis; but surprisingly when sma was disrupted in ampAOE cells a severe impairment in cell movement was observed. These opposing effects on chemotaxis depended on the level of AmpA mRNA in the sma null cells. This suggests an interaction between sma and ampA in the pathways controlling chemotaxis. In this report we characterize the novel migrational effector gene, sma, and present an initial characterization of its involvement in the ampA pathways.

Restriction enzyme mediated integration (REMI) mutagenesis was used to create insertional mutations in an ampA overexpressing (ampAOE) cell line. Dictyostelium cells grown on lawns of bacteria ingest the bacteria, migrating further into the lawn as they clear the bacteria. The clearing in the lawn is referred to as a plaque and its size reflects the ability of cells to migrate, grow, or phagocytose bacteria. The level of expression of the ampA gene influences the rate of migration of cells in plaques (Noratel et al., 2012). On bacterial lawns, ampAOE cells produce significantly larger plaque sizes than wild-type cells, and ampA^- cells produce much smaller plaques (Fig. 1A–C,I). A suppressor of the larger ampAOE plaques would be expected to produce smaller plaques, likely comparable in size to wild-type or ampA^- cell plaques. Following REMI mutagenesis more than six thousand plaques were screened for a suppression phenotype. Three mutants that produced plaques significantly reduced in size compared to ampAOE cells were recovered, identified, and characterized. One of these was the suppressor mutant JSK3 described here. The two other suppressor mutants have been described previously (Kelsey et al., 2012a; Kelsey et al., 2012b). Sequencing DNA flanking the REMI disruption site in JSK3 indicated that the REMI insertion resulting in the phenotypic suppression was immediately upstream of an uncharacterized gene, DDB_G0281803 (supplementary material Fig. S1). As depicted in Fig. 1D, the JSK3 suppressor mutant produced very small plaques compared to ampAOE. In order to determine whether it was inactivation of the adjacent DDB_G0281803 gene that was responsible for the suppression phenotype, independently generated knockouts of the adjacent DDB_G0281803 in ampAOE cells were created. A knockout plasmid was constructed that would remove 500 base pairs of the DDB_G0281803 gene coding region and replace them with a blasticidin resistance cassette (supplementary material Fig. S2A). This new knockout plasmid was utilized to produce DDB_G0281803 gene disruptions in ampAOE, wild-type and ampA⁻ backgrounds. Fig. 1H shows PCR results confirming correct integration of this knockout plasmid into the WT strain to generate the knockout strain subsequently used for the characterization of the gene. Supplementary material Fig. S2B demonstrates that there is no mRNA for the DDB_G0281803 gene present in any of the 3 knockout strains generated with this plasmid (sma⁻, ampAOE/sma⁻and ampA⁻/sma⁻). The introduction of the knockout plasmid into ampAOE cells reproduced the significantly decreased plaque size phenotype (Fig. 1F,I). Based on characteristics of the gene discussed below we have named it sma (SAP domain-containing modulator of adhesion).

Surprisingly, the disruption of sma in a wild-type or ampA⁻ background produced an extremely large plaque phenotype (Fig. 1E,G), the opposite of what was observed when sma was disrupted in ampAOE. The plaque sizes in sma⁻ and ampA⁻/sma⁻ cells were more than four times the size of the original parental WT strain, making their large plaque phenotype even more pronounced than that produced by the ampAOE strain (Fig. 1I).

To further confirm that the phenotypes observed in sma⁻ cells were due to a knockout of the sma gene, rescue plasmids were expressed in wild-type and sma⁻ cells. A GFP-Sma rescue plasmid was constructed, which contains the full length sma gene with an N-terminal GFP tag (supplementary material Fig. S3). Cell lines carrying the plasmid expressed a GFP fusion protein of the correct size to be the GFP-Sma protein (∼170 kDa; 144 kDa for Sma and 27 kDa for GFP). Wild-type cell lines expressing the fusion protein appeared normal; they produced plaques on bacterial lawns that were the same size as WT and they completed development normally (supplementary material Fig. S3). Introduction of the GFP-Sma plasmid into sma⁻ cells rescued the null mutant phenotype and produced normal WT size plaques on bacterial lawns, rather than the enormous plaques seen with the sma⁻ mutant (Fig. 1I, sma⁻+GFP-Sma), indicating that sma gene knockout was responsible for the altered plaque size.

Changes in plaque sizes such as those seen in sma knockout cells (Fig. 1) can be indicators of a defect in cell growth, phagocytosis, or cell migration. No changes in cell growth or phagocytosis were seen in sma⁻ cells (supplementary material Fig. S4); however, significant changes in cell motility were identified in sma mutant cells. Migration of the different sma and ampA mutants to folic acid and to cAMP is analyzed in Fig. 2. AmpAOE cells produced larger plaque sizes compared to WT cells. An increase in the velocity of these cells is shown in Fig. 2A. Also, the chemotaxis plot of ampAOE cells reveals longer path lengths (Fig. 2B). This confirms our previous observation that the increased velocity in ampAOE cells permits an increase in plaque size compared to WT (Noratel et al., 2012). Sma⁻ cells also produced significantly larger plaques compared to WT. As indicated in Fig. 2A, sma⁻ cells moved faster, were significantly more directional, and displayed an increased chemotactic index (Fig. 2A). This is also evident in the chemotaxis plots in Fig. 2B. Together, these chemotaxis effects permit a quicker, more directional movement to the chemoattractant, which could explain the increased plaque sizes in sma⁻ cells. As expected based on plaque sizes, ampA⁻/sma⁻ cells also display an increased efficiency in chemotaxis compared to WT cells (Fig. 2A). Unlike the ampA⁻ cells, which show a decreased rate of migration (Noratel et al., 2012), the ampA⁻/sma⁻ cells show the same velocity as WT cells but they show greater directionality and chemotactic index. The sma⁻ defect in chemotaxis appears to be specific for chemotaxis to folic acid, as rates and directionality of chemotaxis to cAMP in sma⁻ cells were not significantly changed (Fig. 2A).

By contrast to the situation in WT cells, overexpression of AmpA protein in sma⁻ cells (ampAOE/sma⁻) caused formation of small plaques. Consistent with this, ampAOE/sma⁻ cells displayed a significantly decreased rate of migration (Fig. 2A). This defect is clearly illustrated in the chemotaxis plots where ampAOE/sma⁻ cells are shown moving very little from the origin (Fig. 2B).

The sma gene also appears to play a role in cell–cell adhesion. As depicted in Fig. 3A, sma knockout cells tended to clump more during vegetative growth than WT cells. The graph in Fig. 3B gives values for the percent of cells found in clumps of three or more. While only about 10% of wild-type cells were found in clumps, over 30% of sma⁻ cells were found in clumps.

The effects of the sma gene knockout on cell migration and cell–cell adhesions provide evidence of a role for sma in two pathways known to be influenced by the ampA gene (Noratel et al., 2012; Varney et al., 2002a). The results also suggest a complex interaction between ampA and sma. Below, we further characterize the sma gene in WT cells. The ampA/sma interaction is also investigated and discussed later.

The sma gene encodes a 1216 amino acid, 144 kDa protein, with no known homologs identifiable from BLAST searches. A number of putative protein domains were detected using Motif Scan (Pagni et al., 2007). These include a G-binding formin 3-homology domain (GBD-FH3), a phosphatase tensin domain (PTEN), a SAP DNA-binding domain (SAP), and two nuclear localization signals (NLS) (Fig. 4A). G-binding formin 3 homology domains are N-terminal domains found in some formins (Rivero and Somesh, 2002). Formins are actin-binding proteins that promote the formation of linear actin filaments. The FH3 domain of formins is thought to be important for targeting the proteins (Rivero et al., 2005). A G-binding domain is present in the FH3 domain where interactions with Rho GTPases have been described. Rho-GTPases are essential in a number of actin dynamic processes (Rivero and Somesh, 2002). Phosphatase tensin domains are known to be involved in signaling pathways regulating actin polymerization and cell migration (Leslie et al., 2005). PTEN (phosphatase tensin homolog deleted on chromosome 10) is probably the most well-known protein in that family. In Dictyostelium it acts as an inhibitor of the PI3K signaling pathway that permits polarization and regulates cell migration and directionality (Leslie et al., 2005).

The SAP domain is a DNA-binding motif named after the proteins in which it was originally identified: SAF-A/B, Acinus and PIAS (Aravind and Koonin, 2000). Alignment of the Sma SAP domain with other SAP domains from mammalian and yeast proteins indicates that Sma contains the same key features of this domain including the two stretches of conserved hydrophobic amino acids separated by a region that includes a conserved glycine (Fig. 4B) (Aravind and Koonin, 2000). Sma also is predicted to have the conserved alpha helices in the two hydrophobic regions found in all SAP domains (Aravind and Koonin, 2000). These hallmark helical regions, along with the conserved positively charged amino acids, are proposed to make contact with DNA (Aravind and Koonin, 2000; Aravind and Koonin, 2001). Like the Sma protein described here, most proteins carrying a SAP domain are multi-domain proteins. Interestingly, SAP domain-containing proteins are present in all Dictyostelia subgroups. Dictyostelium discoideum has about 20 SAP domain-containing proteins, none of which have been characterized (Gaudet et al., 2011; Gaudet et al., 2008). The different domains present in each of the 20 SAP domain-containing proteins in Dictyostelium discoideum appear unique for each protein, with few similarities among them (supplementary material Fig. S5). It appears the SAP domain is not specific to the N-terminus, where it is in Sma. Also, no other SAP domain-containing protein contains exactly the same additional domains as found in Sma. There is one that also contains a PTEN domain and another that contains a cortactin domain, raising the possibility that other SAP domain-containing proteins in Dictyostelium may also interact with the cytoskeleton (supplementary material Fig. S5). This represents a departure for SAP domain-containing proteins, as most are associated with proteins domains involved in DNA repair, RNA processing, or that suggest a function as chromatin scaffold proteins.

A predicted SAP domain suggests an interaction with DNA. This is in agreement with the presence of two predicted nuclear localization signals found in Sma.

In order to localize the Sma protein in cells a portion of the sma gene coding region was fused to an mRFPmars protein and used to generate a gene replacement strain in which the Sma-mRFP protein is expressed under the control of the endogenous sma promoter. Expression of the fusion protein under its own control regions should mimic the normal expression of the Sma-protein. The Sma-mRFP expressing cells were demonstrated to have the wild-type gene replaced with the Sma-mRFP fusion gene and were shown to express the correct size protein product (supplementary material Fig. S6). The Sma-mRFP gene replacement strain retained the WT phenotype both with regard to plaque size and developmental progression indicating that the Sma-mRFP gene is functional (supplementary material Fig. S6). It was necessary to use indirect immunofluorescence microscopy with fixed cells to detect the mRFP fusion protein in Dictyostelium cells because the sma gene is expressed at a sufficiently low level that the amount of Sma-mRFP fusion protein produced resulted in fluorescence too weak to detect with laser power that did not kill the cells.

Sma protein levels increased as a function of cell density in vegetative cells. The protein appeared localized to the nucleus but at high densities increasing amounts of the protein localized to the cell periphery (Fig. 5). Fig. 5A,B reveal an increase in the fluorescent intensity of the Sma-mRFP signal as cells progressed from low to high density. An increase in Sma-mRFP protein product with increasing cell densities was also seen in western blots (Fig. 5C) confirming the expression trend seen by immunofluorescence (Fig. 5A,B).

Interestingly, the GFP-Sma fusion protein, which was expressed under the control of the high level actin promoter and introduced into cells on a high copy number plasmid, was detected mostly near the nucleus (supplementary material Fig. S3). The overexpressed GFP-Sma, even at high density, seemed to remain mostly nuclear although a small amount of GFP-Sma was seen at the cell periphery. This is in contrast to the increased distribution of Sma at the cell periphery in the Sma-mRFP knockin construct, which was expressed under the control of the endogenous sma promoter that is a much lower level promoter and is present at only 1 copy per cell. This suggests the nucleus may be the default localization for Sma. Possibly it requires an interaction partner that is limiting to transport it to the cell periphery.

Sma-mRFP cells at increasing densities were fractionated, and Sma-mRFP was predominately in membrane fractions and in the soluble nuclear fraction (Fig. 5D). Interestingly, the amount of Sma in the soluble nuclear fraction remained fairly constant, while the amount in the membrane fraction increased with increasing cell density.

Because Sma clearly plays a role in cell migration it was of interest to determine whether it showed any interaction with the actin cytoskeleton and whether its localization changed in migrating cells. As expected based on the fractionation results shown in Fig. 5, a positive correlation of Sma-mRFP localization with the DAPI stained nucleus was found in vegetatively-growing cells (Fig. 5, Fig. 6D for the Pearson's coefficient). A small but positive correlation between Sma-mRFP and F-actin was also determined (Fig. 6A,D). There are small patches where there is colocalization between Sma and F-actin (yellow arrow in Fig. 6A). A perfect colocalization would give a Pearson's coefficient value of 1.0. The values for nuclear (0.35) and F-actin (0.08) colocalization with Sma-mRFP during vegetative growth in medium-density cells were much lower, suggesting only some of the Sma-mRFP localized with these organelles/proteins. The remainder of the Sma protein would appear to be in the cellular membrane fractions (Fig. 5D).

Fig. 6B shows images of elongated cells migrating to folic acid. Sma-mRFP and polymerized F-actin are labeled. Upon migration to folic acid much of the Sma protein was seen concentrated to the poles of the moving vegetative cells and it appeared to completely leave the nucleus. In some cells a small portion of the Sma protein was seen concentrated in F-actin rich extensions (Fig. 6B, top). In other cells Sma appears polarized and localized to regions adjacent to F-actin (Fig. 6B, bottom). Fig. 6D reveals a colocalization value for Sma and F-actin of 0.13. This value is low and reflects the fact that Sma is mostly seen localized just outside regions of actin polymerization. In migrating cells chemotaxing to folic acid, Sma appeared to be totally absent from the nucleus (Fig. 6B,D). The colocalization values in Fig. 6D show a Pearson's coefficient of zero for the Sma-mRFP signal and the DAPI signal in folic acid chemotaxing cells indicating complete absence from the nucleus. The same increased localization of Sma to the poles of cells migrating to folic acid was observed using the actin promoter driven GFP-Sma fusion construct (supplementary material Fig. S3E, bottom). The excess GFP-Sma fusion protein expressed under the control of the actin promoter remains in association with the nucleus, in contrast to the complete loss of Sma protein from the nucleus when the endogeneous sma promoter, which is a much lower level promoter, is used to drive expression of the fusion protein.

When cells were starved for development for 6 hours and migrating to cAMP, a very different result was observed. The Sma-mRFP protein remained almost entirely associated with the nucleus. The amount of colocalization between Sma-mRFP and DAPI was significantly increased showing a Pearson's coefficient of 0.8, reflecting almost complete localization to the nucleus (Fig. 6D). Fig. 6C shows the almost complete nuclear localization of Sma-mRFP in starved cells migrating to cAMP although a small amount of the Sma is present in the cytoplasmic membranes but not at the poles of cells. Thus, in growing cells migrating to folic acid Sma leaves the nucleus and a portion of it becomes polarized to the ends of the cell with some Sma found associated with the F-actin cytoskeleton. By contrast, in starving cells migrating to cAMP, Sma largely remained in the nucleus.

A disruption in sma caused suppression in the ampAOE pathway controlling plaque sizes. This was evident in the considerably smaller plaques produced by ampAOE/sma⁻ cells, compared to ampAOE cells (Fig. 1). Sma has a SAP DNA-binding domain suggesting the possibility of gene regulation or RNA interactions. Could Sma regulate ampA expression? To determine if a disruption in sma interferes with AmpA mRNA accumulation, transcript levels of ampA were monitored when sma was disrupted. The results (Fig. 7A) show a decrease in ampA mRNA transcripts when sma was disrupted. When Sma was knocked out, the level of ampA transcripts in ampAOE cells decreased to near WT levels, and WT ampA transcript levels appeared to decrease even further in the sma⁻ cells. It appears that Sma is necessary for full levels of ampA mRNA accumulation, and in sma⁻ cells an almost compete loss in ampA transcript is observed.

The RT-PCR results shown in Fig. 7A suggest that a Sma knockout is equivalent to an AmpA knockout with respect to ampA mRNA levels, so we investigated the effect of sma disruption on the different ampA pathways to try to confirm this relationship. The phenotypic differences of ampA⁻ or ampAOE cells compared to WT are shown in Figs 7 and 8. AmpA⁻ cells polymerize less actin compared to WT cells, are smaller in size, and display increased cell–substrate adhesion. The opposite is true for ampAOE cells, which produce increased levels of actin polymerization, show larger cell sizes, and decreased cell–substrate adhesions. Based on the RT-PCR results, we would expect to see sma⁻ phenotypes similar to ampA⁻ and ampAOE/sma⁻ phenotypes similar to WT. As indicated below, this does appear to be the case. Fig. 7B shows images of alexa fluor 488-phalloidin staining of polymerized F-actin in vegetative cells. A disruption of sma in wild-type and ampAOE cells resulted in a decrease in the amount of F-actin. There was no further decrease in actin in ampA⁻/sma⁻. The bar graph in Fig. 7C shows the amount of alexa-fluor 488-phalloidin binding to F-actin in the different cell types. The decrease in the amount of F-actin when sma was disrupted in WT was equivalent to levels seen in ampA⁻ cells, WT cells had 2× more F-actin than did sma⁻ and ampA⁻ cells. Also, the amount of polymerized actin in ampAOE/sma⁻ cells was near WT levels, ampAOE cells contained 1.5× the amount of polymerized actin found in WT and ampAOE/sma⁻ cells. The same trend was found in the cell size and cell–substrate adhesion assays. The bar graph in Fig. 8A shows a significant decrease in cell volumes when the sma gene was disrupted in WT and ampAOE cells. The differences in cell sizes can be viewed in the images in Fig. 3A. The cell volume of ampA⁻ cells was similar to the cell volume of sma⁻ cells and there was no further decrease in cell volume in the ampA⁻/sma⁻ double mutant suggesting that the loss/decrease in ampA mRNA in the sma⁻ cells was sufficient to account for the decrease in cell volume. Likewise, the knockout of sma in the ampAOE strain reduced the enlarged volume of the ampAOE cells to close to the volume of WT cells.

To compare cell–substrate adhesion, the percentage of adhesive cells that remained attached to Petri dishes after shaking was determined (Fig. 8B). A significant increase in cell–substrate adhesion was seen in sma⁻ cells compared to WT and in ampAOE/sma⁻ cells compared to ampAOE. In ampAOE/sma⁻ 60% of the cells were highly adhesive, similar to the WT level of 63%. Sma⁻ cells displayed 71% of their cells as highly adhesive, very similar to the ampA⁻ level of 70%. The assays in Figs 7 and 8 display a clear trend where the disruption of sma causes a change in ampAOE phenotypes to WT, and a disruption of sma in WT cells mimics or phenocopies the ampA⁻ phenotype. A sma disruption in ampA⁻ cells appeared to have no additive effect on the assays shown in Figs 7 and 8. A double knockout of genes in different pathways usually results in an additive effect, which should produce phenotypes more severe than either mutant alone. The phenotypes of the double knockout mutant, ampA⁻/sma⁻, were not significantly different than either mutant alone, suggesting that sma and ampA are involved in the same pathway controlling actin polymerization, cell size, and substrate adhesion. Potentially the sma gene acts upstream of ampA controlling its level of mRNA accumulation.

Both ampA and sma knockouts resulted in increases in cell–cell adhesion as assayed by counting cells in clumps of 3 or more (Fig. 3A,B). However, the effect of sma on cell–cell adhesion was much stronger than the effect of ampA. Sma null cells showed over 30% of growing cells in clumps (3× the wild-type level) while ampA null cells showed only 20% or twice the wild-type level of cell clumping (P<0.05) (Fig. 3B). Overexpression of ampA resulted in reduced cell–cell clumping to about half of the wild-type levels (Fig. 3B). The loss of sma in the ampA overexpressing strain (ampAOE/sma⁻) increased clumping to 1.5 times the wild-type level. The sma⁻/ampA⁻ double mutant showed the same higher level of clumping that sma⁻ alone shows. This is in contrast to the results with actin polymerization, cell volume and cell–substrate adhesion where the level of effect of the sma null mutation is identical to the effect of the ampA null mutation alone, suggesting that the role of sma is confined to its effects on ampA mRNA levels for these events. For cell–cell adhesion by contrast sma has a role in addition to being necessary for ampA mRNA accumulation.

Sma expression increased during development, with a peak level at about 12 hours (supplementary material Fig. S7). Disruption of sma also resulted in the inability of cells to produce the later fruiting body structures (supplementary material Fig. S8). Interestingly, the developmental defect in the sma⁻ cells was more severe than the defect in ampA⁻ cells. AmpA null cells showed a 6-hour delay in development but eventually formed fruiting bodies (Varney et al., 2002b). By contrast, the sma⁻ cells formed rough flattened mounds with only an occasional fruiting body. This is consistent with the fact that while sma may regulate ampA, it also appears to have additional functions. Sma-mRFP expressing cells are distributed throughout the early finger structures where a faintly dispersed signal was seen (supplementary material Fig. S8). This suggests Sma is not restricted to a specific cell type during development. This also differs from ampA because ampA expression during development is confined to the Anterior-Like Cells.

We have characterized a gene involved in a signaling pathway important for chemotaxis. An interaction between the newly identified sma pathway and the ampA pathway in influencing chemotaxis is demonstrated. Sma appears to have a role in cell migration, as mutants with disruptions in the sma gene are more efficient in chemotaxis. Interestingly, the presence of AmpA strongly interferes with the more efficient sma⁻ mutant strain's chemotaxis and impairs cell movement. This is in contrast to the fact that overexpressing AmpA in sma⁺ wild-type cells increases cell migration. We have presented an initial characterization of the sma gene and of its unexpected interaction with ampA.

Sma protein levels increased as the density of growing cells increased. A low level of Sma localization to the nucleus appeared consistently throughout the growing cell densities using both cell fractionation techniques and immunofluorescence microscopy. As cells reached higher densities an increasing amount of the Sma protein localized to the cytoplasmic membrane fractions, especially at the cell periphery. Consistent with this location at the cell periphery at high cell densities, a knockout of Sma resulted in severe clumping of the cells indicating a role for Sma in reducing cell–cell adhesions. Together these results point to a possible role in regulating cell–cell adhesions in a density-dependent manner.

The predicted SAP DNA-binding domain and nuclear localization signals in the Sma protein, as well as the fact that some of the Sma protein is localized to the nucleus, suggests a role for Sma in DNA binding, chromatin interaction, or some aspect of gene expression. Indeed, Sma plays a role in regulating ampA mRNA levels. The fact that Sma is present in the soluble nuclear fraction and not in the chromatin fraction, however, suggests that Sma has only an indirect role in DNA binding or transcription. Alternatively it could play a role in RNA processing, transport out of the nucleus or RNA stability. The ampA gene does not have any introns so any controls at the level of RNA processing are not likely to involve RNA splicing. Further investigation is required to fully understand the role of Sma in controling ampA mRNA levels. Interestingly, however, the default location of Sma appears to be the nucleus because when the Sma protein is overexpressed from an actin promoter on a high copy number plasmid the extra Sma protein localizes to the nucleus and in fact the very intense nuclear staining in the overexpressing Sma strains almost obscures the Sma that is localized to the cytoplasmic membranes at the cell periphery. This raises the possibility that a factor whose concentration is limiting is required for Sma to exit the nucleus and localize to the cytoplasmic membrane fraction.

In migrating cells chemotaxing to folic acid Sma appears to completely leave the nucleus and translocate to the poles of the cells and to localize immediately adjacent to F-actin rich regions with some Sma actually colocalizing with F-actin. This could imply a role for Sma in actin dynamic processes. Supporting this role was the predicted GBD-FH3 domain, a domain involved in G-protein interactions regulating actin dynamics (Rivero et al., 2005; Rivero and Somesh, 2002). The possibility of a SAP domain-containing protein interacting with actin is not completely novel. The SAP domain-containing transcription factor MKL (myocardin-like protein) was found to change its subcellular localization as a function of the Rho-actin signaling pathway (Du et al., 2004; Miralles et al., 2003). Rho signaling promoting actin polymerization resulted in a shift of MKL localization, from the nucleus to the cytoplasm, where it was seen colocalizing with actin. It is possible that a similar signaling event recruits Sma to the cytoplasm, to regions adjacent to actin polymerization.

Another putative domain in Sma, the PTEN-like domain, is important in the regulation of signaling pathways contributing to cell polarization in migrating cells (Leslie et al., 2005). Sma⁻ cells appear to be very efficient in chemotaxis, more so than wild-type cells. It is the chemotactic index that is particularly increased in the sma gene knockout. Sma localization in folic acid migrating cells was predominately at the cell poles. This raises the question of a possible role for Sma in the cell polarization process. The Sma presence at the poles of the cell could help regulate specific signaling pathways during chemotaxis. A key question is how the absence of Sma results in a cell more effective in chemotaxis.

Sma was mostly absent from the nuclear region in cells chemotaxing to folic acid but remained in the nucleus in starved cells migrating to cAMP. In fact, in starved cells migrating to cAMP, almost no Sma was detected in the cytoplasm. Also, the sma⁻ increased migration defect was detected only in cells chemotaxing to folic acid and not in those chemotaxing to cAMP. This suggests a role for Sma in migration only during the vegetative-growth phase, when cells migrate in search of bacteria, which they locate via folic acid.

Comparison of folate chemotaxis with cAMP chemotaxis indicates that four pathways essential for signal amplification during chemotaxis to cAMP are activated by folate but do not appear to be essential for chemotaxis to it (Kortholt et al., 2011). As a result the chemotactic response to folate requires a steeper gradient and shows a reduced chemotactic index compared to cells migrating up a steep optimal cAMP gradient (Kortholt et al., 2011). It has also been suggested that because of the need to frequently change direction in the search for bacteria, growing cells remain sensitive to chemoattractants across their entire periphery, which would result in reduced polarization and a reduced chemotactic index relative to cells migrating very directionally to a cAMP source (Janetopoulos and Firtel, 2008). Sma clearly plays a role in reducing the chemotactic index in growing cells but how it does this or if it might interact with any of the 4 pathways that function to amplify the chemotactic signal is clearly an issue for the future.

Sma appears to control the level of ampA mRNA accumulation. While the effect of sma knockouts on cell size, cell–substrate adhesion and actin polymerization can be accounted for largely by the loss of ampA mRNA in the sma knockout cells, the effect of the sma knockout on cell migration does not simply phenocopy the ampA null mutation. Cells carrying knockouts of ampA cannot migrate (Noratel et al., 2012), but sma knockouts that reduce ampA mRNA to almost undetectable levels, similar to an ampA knockout, migrate better than wild type. This indicates that Sma potentially has a role in cell migration that is independent of AmpA. By contrast, knockout of sma in the ampA overexpressing strain that reduces ampA mRNA to wild-type levels shows a severe defect in migration. Thus, in the absence of sma, ampA has an antagonistic effect on cell migration. At this point we do not have a clear idea of how AmpA influences cell migration. We know that it influences the amount of cell–substrate adhesion and that it traffics to the cell surface, is endocytosed and localizes to a perinuclear compartment that has been identified as a slow recycling compartment (Noratel et al., 2012). There is no indication that AmpA has any kind of catalytic or nucleic acid-binding role. We have suggested a possible role for ampA in recycling a cell–substrate adhesion molecule (Noratel et al., 2012). Sma clearly plays a role in reducing cell–cell adhesion and sma null cells show significantly more cell–cell adhesion than even ampA null cells, which are increased in both cell–cell and cell–substrate adhesion relative to wild type. Just how these two pathways intersect is at this point far from clear.

An increase in Sma expression was found in developing cells, and a role during development was confirmed by the defects observed in sma⁻ developing cells, which failed to form fruiting bodies and arrested at mound stage. While ampA null cells delay development at mound stage, the role of Sma in development has to be larger than its role in regulating ampA mRNA levels. The ampA gene is only expressed in the Anterior-Like Cells during development whereas the sma gene is expressed in all cell types. Sma does not appear to have a role in cAMP chemotaxis during the aggregation stage of development as evidenced both from measuring its effects on chemotaxis of cells to cAMP and the fact that the sma null cells develop normally to mound stage. Could the mostly nuclear localization of Sma during migration to cAMP (early development) predict a more transcription-centered role during development while having a more cytoskeleton-actin-adhesion based role during growth?

Whatever the role of the sma gene may be, it is clearly an interesting new player in cell migration, cell adhesion and some aspects of gene expression. Dictyostelium, with its ease of molecular genetics and a full spectrum of different SAP domain-containing proteins, could prove an ideal model system in which to pursue studies of this interesting class of proteins. SAP domain-containing proteins have been shown to be involved in adenocarcinomas and leukemias emphasizing the importance of further research on this group of proteins (Choong et al., 2001; Hashii et al., 2004).

WT (AX3), ampAOE (G418 resistant), or ampA⁻ (originally blastcidin resistant, but the blasticidin resistance cassette was removed via cre-loxP recombination (Kimmel and Faix, 2006)) cells were grown as described previously (Kelsey et al., 2012a). Low-density growing cells refers to a concentration less than 1×10⁶ cells/ml. Mid-density (exponential log growth) cells were 2×10⁶–5×10⁶ cells/ml. High-density cells were greater than 7×10⁶ cells/ml. Cells were starved on filters to induce development (Eichinger and Rivero, 2010).

The DNA preparation was performed as described previously (Nellen et al., 1987). Restriction enzyme mediated integration mutagenesis was from Kuspa and Shaulsky, as described by Kelsey (Kuspa, 2006; Shaulsky et al., 1996; Kelsey et al., 2012a).

Fractionation used the Thermo Scientific subcellular protein fractionation kit according to the manufacturer's instructions, as described previously (Kelsey et al., 2012a).

A sma knockout plasmid (supplementary material Fig. S2) was constructed and used to create knockout strains in WT, ampAOE and ampA⁻ backgrounds. It contained a 1 kb coding region (AB) and another 1 kb coding region (CD) downstream of the first. A 500 bp sequence was left out between the two coding regions. The AB and CD regions were inserted at EcoRI, HindIII and SpeI, SacII restriction sites in a pBluescript plasmid containing a floxed blasticidin resistance cassette previously inserted at a SmaI site (Kimmel and Faix, 2006). The fragments were inserted on either side of the cassette.

AB segment primers: forward 5′-CAGAAGCTTCAAATTTATTAGCATTA-3′, reverse 5′-CAGGAATTCAAATCTTTTTGAGCAAATTCA-3′.

CD segment primers: forward 5′-CAGACTAGTTTGCCAATGTTAAATCAAAGG-3′, reverse 5′-CAGCCGCGGGTCGTTGTTGTTGCTGTTGG-3′.

Primers used to check for correct insertion: forward 5′-AATAGTAATGAACAAGGTTGTG-3′, reverse 5′-TTGCATTCTATTGTACAACG-3′.

The Sma-mRFP plasmid (supplementary material Fig. S6) was constructed to monitor Sma expression. The Sma-mRFP plasmid expresses the Sma protein region followed in frame by the mRFPmars coding sequence at the carboxy terminus of the protein (Müller-Taubenberger et al., 2006). Following the mRFPmars coding region is a 270 bp terminator sequence from the 3′ non-coding region of the ampA gene. A 1200 bp region of the sma 3′ coding sequence was inserted at ApaI and SalI sites into pBluescript+Bsr, which already contained a floxed blasticidin resistant cassette previously inserted at the SmaI site (Kimmel and Faix, 2006). A 1500 bp region of the 3′ sma non-coding sequence immediately downstream of the termination site was inserted at NotI and SacII sites. Sma 3′ coding primers: forward 5′-GAACCTAGGGCCCTCAAGAGCAATCAGTAATCCA-3′, reverse 5′-GCACGCGTCGACCACAAATTATTATAGTAGAG-3′. The 3′ sma non-coding region primers were: forward 5′-CGCGCAATTAGCGGCCGCCCAATACTTGAATTGGTAATA-3′, reverse 5′-ACTCGATCCGCGGCCTCTATCTTTATTCGCTA-3′.

The Sma-mRFP plasmid was linearized by ApaI and SacII for transformation.

The primers used to check for Sma-mRFP homologous recombination into the sma gene were: forward 5′-CTCATTAGTAGATTCACTCTAC-3′ and reverse 5′-GCAAATCTCACTATTACCAATTC-3′.

A GFP-Sma-containing rescue plasmid (supplementary material Fig. S3) was constructed containing an N-terminal GFP coding sequence and a Gly–Pro–Gly linker sequence fused to the sma cDNA. The plasmid also contained an actin 15 promoter and terminator. The full length sma cDNA replaced the arp2/3 gene at BamHI and SacI restriction sites in the GFP-Arp2/pBIG plasmid (Dicty Stock Center) (Asano et al., 2004). The primers for sma PCR were: forward 5′-CAGGGATCCAATGAATAAAAGAACAAGAGATAAG-3′ and reverse 5′-CAGGAGCTCTTACCACAAATTATTATAGTAGAGTG-3′.

Western analysis, indirect immunofluorescence, F-actin and DAPI staining were performed as described previously (Kelsey et al., 2012a; Kelsey et al., 2012b). It was necessary to use indirect immunofluorescence microscopy with fixed cells to detect the mRFP fusion protein in Dictyostelium cells because the mRFP fluorescence was too weak to detect with laser power that did not kill the cells.

To calculate total cell volumes, Alexa fluor-594 conjugated deoxyribonuclease I (Molecular Probes) was added to label unpolymerized G-actin (Kelsey et al., 2012a; Kelsey et al., 2012b).

A Leica SP5 scanning confocal light microscope was utilized as described previously (Kelsey et al., 2012a; Kelsey et al., 2012b). Volocity software 5.5 (PerkinElmer) was used to assemble images, calculate the Pearson's coefficient of correlation (Barlow et al., 2010) and cell volume measuring the G-actin signal. Agar plates were viewed as described previously (Kelsey et al., 2012a; Kelsey et al., 2012b). Images used for calculating plaque sizes were analyzed using the Metamorph Imaging series version 7.0r1 (Universal Imaging, West Chester PA, Molecular Devices Corporation). For time-lapse experiments, cells were imaged on the Leica SP5 confocal microscope with a low laser setting (5% Argon laser line 488 nm) using brightfield illumination with either a 40× .55 NA long working objective or a 63× 1.4 NA oil objective. Z-slices of cells were captured at 20-second intervals for 5 minutes and images were processed with the Volocity or Metamorph software.

The procedure was adapted from Hadwiger and Srinivasan, and is described by Kelsey (Hadwiger and Srinivasan, 1999; Kelsey et al., 2012a). Cell movement was analyzed using DIAS (Wessels et al., 2006) and ImageJ (W.S. Rasband, ImageJ, U.S. National Institutes of Health, Bethesda, Maryland, USA, http://imagej.nih.gov/ij, 1997–2011) software. Velocity and directionality values were obtained using the manual tracking (F. Cordelieres, Manual Tracking, Institute Curie, Orsay France, http://rsb.info.nih.gov/ij/plugins/track, 2004) and chemotaxis tool (G. Trapp and E. Horn, Chemotaxis and Migration tool, Munich, Germany, http://ibidi.com) on ImageJ. The chemotactic index (CI) is equal to the cosine of the angle formed between the line of movement of the cell and the line representing direct movement to the chemoattractant (Iijima and Devreotes, 2002).

As described previously (Kelsey et al., 2012b; Noratel et al., 2012).

As described previously (Noratel et al., 2012).

As described previously (Kelsey et al., 2012a; Kelsey et al., 2012b). The ampA primers were: 5′-TCAAATTAACACCAGTGCAAC-3′ and 5′-ATGGTTCACAACGACCACAA-3′, which yield an 870 bp product. The sma primers were: 5′-ATTGCATATGGCACTCAAATGA-3′ and 5′-TTACCACAAATTATTATAGTAGAGTG-3′, which yield a 1626 bp product. The control, Ig7, primers were: 5′-TTACATTTATTAGACCCGAAACCAAGCG-3′ and 5′-TTCCCTTTAGACCTATGGACCTTAGCG-3′, which yield a 370 bp fragment (Hopper et al., 1993). A gel doc digital imaging system (Alpha Innotech Corporation IS-100) was used to quantify band intensity under conditions where the signal was linearly dependent upon the amount of material loaded. Unequal protein loading or unequal RNA levels were taken into account with the Coomassie and Ig7 controls. Then the relative intensity of the bands to one another was calculated, with the most intense band set equal to 100%.

P-values were calculated using the paired two-tailed Student's t-test in Excel.

We thank Annette Muller-Taubenberger for advice on antibodies for detecting the mRFP-mars protein and Alice Rutatangwa and Julie Wolf for help with plasmid construction. We also thank the Dictyostelium Stock Center for the GFP-Arp2/pBIG plasmid and the mRFP-mars plasmid and Alan Kimmel for the floxed blasticidin cassette plasmids. We are grateful to Chere' Petty and the UMBC Keith R. Porter Imaging Facility for help with microscopy and Elizabeth F. Noratel for discussions. This work was supported by NSF grant numbers MCB-0444883 to D.D.B. and MRI-0722569 to D.D.B. and Theresa Good.