TagA, a putative serine protease/ABC transporter of Dictyostelium that is required for cell fate determination at the onset of development

Peptide signals play a wide variety of roles in cell differentiation and multicellular development. Often these signals act at a distance from where they are produced and require membrane transporters either for export from the producing cells or for import into the responding cells. For example, the metazoan signaling peptide Hedgehog, which is involved in specifying axis formation, is exported from cells by the RND transporter Disheveled(Ma et al., 2002). The ATP binding cassette (ABC) superfamily of transporters comprises at least seven families whose members transport diverse substrates across different types of cellular membranes (Ambudkar and Gottesman,1998). The ABCB family members are best known for conferring multiple drug resistance, but members of some ABCB subfamilies are capable of transporting peptides that lack classical export signals(Detmers et al., 2001). Homologs of these subfamily members such as the TAP proteins, as well as other ABC transporters that transport peptides, have been found in microorganisms,plants and metazoans. CSF is a five amino acid peptide, found in Bacillus subtilis, which is derived from the C-terminal end of the 40 amino acid peptide encoded by the phrC gene(Lindsey et al., 2002; Stacey et al., 2002). The CSF peptide appears to regulate both competence and sporulation, since high levels of CSF inhibit competence and promote sporulation. The ABC transporter Spo0K is required to transport CSF into the cell from the extracellular environment(Solomon et al., 1996). The ABC transporter Ste6 transports a-factor, one of two peptides that determine mating compatibility in Saccharomyces cerevisiae(Kuchler and Thorner, 1992). In plants the developmental role of peptide signaling has been relatively well characterized (reviewed by Lindsey et al.,2002; Stacey et al.,2002). CLV3, a 96 amino acid peptide in Arabidopsis is involved in proper sizing of shoot apical meristems which form new leaves, and ENOD40-encoded peptides of the legume family promote formation of nodules for colonization by nitrogen fixing bacteria. Loss of the Arabidopsis ABC transporter, AtMRP5 results in a shift from growth to development in the root. Mammalian ABC transporters have been shown to be capable of transporting hydrophobic peptides, although the biological relevance of this observation remains unclear (Sharom et al.,1998).

In Dictyostelium discoideum, unique members of the ABCB transporter subfamily, TagB and TagC, appear responsible for peptide signal export during development. These Tag proteins are required for cell differentiation in Dictyostelium and have the potential to carry out the processing and transport of peptide signals since they possess an N-terminal serine protease domain and a C-terminal transporter domain. The available genetic data suggest that the function of the Tag proteins is to transport signaling peptides that regulate the timing and nature of cell differentiation. During the development of Dictyostelium, equipotent cells differentiate into prespore cells and distinct populations of prestalk cells that later form the various tissues in the terminally differentiated fruiting body that consists of mature spores held atop a cellular stalk(Kessin, 2001). The TagB protein is required for cell-cell signaling that promotes spore encapsulation,and it is required for the differentiation of prestalk A (PstA) cells in a cell-autonomous fashion (Shaulsky et al.,1995). TagC is structurally similar to TagB and the phenotypic similarities between tagC null mutants and tagB mutants are so striking that it is presumed that TagB and TagC function as a heterodimer. The conserved amino acid residues known to be required for serine protease catalytic activity and for ABC transporter ATPase activity are required for TagB function (G.S., unpublished). In addition, TagC-null mutants fail to release the spore encapsulation-inducing peptide SDF-2(Anjard et al., 1998). Thus, it is likely that Tag proteins work by the proteolytic processing and transport of signaling peptides, in a manner analogous to that in which STE6 transports the proteolytically processed, lipid-modified a-factor pheromone peptide of S. cerevisiae (Taglicht and Michaelis, 1998). Such a cell-cell signaling function is the obvious common feature of all of the transporters characterized as being involved in developmental signaling. A curious feature of the tagBmutant phenotype is the cell-autonomous defect in prestalk cell differentiation experienced by cells that would normally express the tagB gene in wild-type cell development(Shaulsky and Loomis, 1996). This novel feature suggests that the failure of a transporter to export a substrate/signal can affect the signal-producing cell as well.

Developing Dictyostelium aggregates contain several prestalk cell sub-populations that can be distinguished from the prespore cells and can be tracked with cell-type-specific reporter genes(Williams, 1997). Near the end of development, the PstAB cells initiate fruiting body formation by plunging through the prespore mass, forming a cellulosic stalk tube until they make contact with the substratum. The PstA cells enter the stalk tube following the PstAB cells and differentiate into stalk cells. As the prespore cells crawl up the elongating stalk they differentiate into mature spores. Terminal spore differentiation proceeds in a wave from top to bottom of the nascent sorus. This observation is consistent with a number of experiments that suggest a signal emanating from the PstA cells coordinates terminal spore differentiation with fruiting body morphogenesis(Harwood et al., 1993; Richardson et al., 1994; Shaulsky et al., 1995). The TagB and TagC transporters are expressed in PstA cells and so they are in the position to affect spore differentiation during this process(Shaulsky and Loomis, 1996)(G.S., unpublished data). The most plausible model is that TagC exports the encapsulation-inducing SDF-2 peptide, initiating the observed wave of spore differentiation as the PstA cells enter the stalk tube at the top of the prespore mass (Anjard et al.,1998). However, there is also a cell-autonomous requirement for both TagB and TagC in the formation of PstA-derived stalk cells. Cells that do not express a TagB or a TagC transporter cannot become stalk cells(Shaulsky et al., 1995). This aspect of Tag transporter function is not understood, but could be explained if the transport substrate acted as an inhibitor of PstA cell differentiation when retained inside the cell. Given an ABC transporter's capacity to maintain a concentration differential of a small molecule across a membrane, ABC transporter activity could alter cellular physiology in a way that would promote or inhibit a particular differentiation program. In theory, the flux of a signal through a single transporter could control the fates of the signal-producing cells as well as the responding cells. There is some tentative evidence for cell-autonomous functions of ABC transporters in cell differentiation. In Dictyostelium, the prespore-specific rhodamine transporter RhT may be required for the production or maintenance of prespore cells (Good and Kuspa, 2000). In mammalian systems, the ABC transporter encoded by the ABCB1 (MDR1)gene has been implicated in the maintenance of hematopoietic stem cells and in the regulation of programmed cell death(Smyth et al., 1998; Johnstone et al., 1999; Johnstone et al., 2000a; Johnstone et al., 2000b). Recently, the ABCG2 gene has also been suggested to have a role in the maintenance of the undifferentiated state of stem cells(Zhou et al., 2001).

We have identified a novel member of the tag gene family, tagA. TagA mRNA is expressed in the first 2 hours following starvation, it becomes specific to prespore cells and is eventually expressed in mature spores. Inactivation of the tagA gene results in developing structures with enlarged or supernumerary prestalk regions with roughly twice the normal number of prestalk cells. Within these structures, the tagA^- cells that activate the tagA promoter do not become spores but instead adopt a prestalk cell fate. Our results imply that the earliest cells to differentiate require TagA-mediated function to prevent their adoption of a prestalk cell fate, or for promoting their prespore cell fate.

Standard DNA and RNA protocols were performed as described previously(Sambrook et al., 1989). Polymerase chain reaction (PCR) amplification was carried out using primer TBR-5.1 encoding the amino acids VGPSGSG(5′-AAC TGC AGG THG GWC CWT CWG GWA GYG G; where W=A or T, R=A or G, Y=C or T and H=A or C or T) and primer TBR-3.1 encoding GGGLL(S/R)IA (5′-CCG GAT CCG CRA TWC TYT TYT TTT GWC CWC C). These sequences represent the conserved amino acid sequences within the nucleotide-binding domains (NBD) of Dictyostelium TagB and human P-glycoprotein (ABCB1). PCR amplification with wild-type genomic DNA was as follows: 5 cycles of 94°C(30 seconds), 37°C (30 seconds), 72°C (30 seconds), all with 1 degree/second ramping, followed by 25-30 cycles of 94°C (30 seconds),55°C (30 seconds) and 72°C (30 seconds). Products were cloned into BamHI-PstI-linearized pGEM3 (Promega). One of the clones represented the NBD of tagA. Multiple cDNA fragments representing the tagA-coding sequence were isolated from a λ-ZAP (Strategene)cDNA library using the PCR product as a probe(Souza et al., 1998). The library was subsequently screened with probes derived from the identified cDNAs. All positive clones were converted to plasmids as described and their inserts were cloned into pGEM3 (Bai and Elledge, 1997). All cDNAs were confirmed to represent the tagA gene by sequencing and by mapping them using Southern analyses. The genomic tagA locus was isolated by screening size-selected fragments of the genome into pBluescript II KS(+/-). These fragments were confirmed to be tagA by sequencing and Southern analyses. Raw sequence data from the Dictyostelium genome sequencing project(http://dictygenome.bcm.tmc.edu/)were used to confirm the tagA contig and to design PCR primers that confirmed clone overlaps and allowed reconstruction of a complete tagA locus of 7.1 kb, including the 5.1 kb tagA coding sequence and 2 kb of upstream sequence (GenBank accession number AF263455).

The Dictyostelium strains used in this study are described in Table 1. Ax4 cells were grown in HL-5 liquid medium (Sussman,1987) supplemented with streptomycin (50 μg/ml) and penicillin(50 U/ml). Neomycin-resistant strains (neo^r)Ax4[cotB/GFP], Ax4[ecmA/GFP], Ax4[act15/GFP] and all respective derivatives, were grown in HL-5 liquid medium supplemented with 20 μg/ml G418 (Geneticin, Gibco). All strains were removed from drug-containing medium 36 hours prior to assay. Cells were plated for synchronous development on nitrocellulose filters as described previously(Sussman, 1987).

Transformation of Dictyostelium cells was performed according to the method of Manstein and Hunt (Manstein and Hunt, 1995) using a BTX 600 electroporation device, or by calcium-phosphate precipitation and glycerol shock(Nellen and Firtel, 1985). Insertional inactivation of the tagA gene was achieved by homologous recombination at the genomic locus after electroporation of linearized DNA fragments. The knockout vector was created by inserting a 1.5-kb fragment containing the blasticidin S resistance (bsr) gene, under the control of the actin 15 (act15) promoter and the actin 8 terminator(Adachi et al., 1994) into the ClaI site within the portion of the tagA cDNA predicted to encode the nucleotide-binding domain. The bsr cassette was excised from pBSRΔBglΔEco with NarI and AccI and inserted into the ClaI site of tagA cDNA,destroying the AccI, NarI and ClaI restriction sites. Integration at the native locus was determined by Southern analyses.

An expression plasmid with the tagA coding region, under the control of its native promoter was constructed by substituting a 7.1 kb XbaI/HindIII tagA fragment, described above, for the lacZ/act8 cassette in pDdGal16(H+)(Harwood and Drury, 1990). The resulting plasmid, ptagA/tagA, was transformed into tagA^- cells by calcium-phosphate precipitation(Nellen and Firtel, 1985). The ptagA/lacZ expression plasmid was made with genomic DNA 2 kb upstream of the tagA coding region including the 5′ end of the coding sequence up to the first BamHI site. This 2.2 kb EcoRI-BamHI fragment was inserted into pDdGal16(H+) between its EcoRI and BglII sites. Staining of developmental structures and spores for β-galactosidase was carried out as described previously (Shaulsky and Loomis,1993).

Sporulation was measured by harvesting cells from filters into 20 mM potassium phosphate buffer, pH 6.2, and treating them with 0.4% non-ionic detergent NP-40 for 10 minutes at 22°C. Cells were then washed with potassium phosphate buffer twice, and disaggregated by trituration with an 18-gauge needle. Refractile spores were counted by phase-contrast microscopy and plated on SM agar plates with bacteria. The number of colonies was used as an estimate of the number of viable spores in each sample. At least three independent determinations were carried out for each strain and are reported as the mean±s.e.m.

Induction of prestalk gene expression was measured in submerged culture(Harwood et al., 1995) as modified previously (Wang and Kuspa,2002). Vegetative cells were harvested at a density of 1-2×10⁶/ml, washed once in 20 mM sodium phosphate buffer (pH 6.4) and three times in stalk buffer [10 mM Mes, 2 mM NaCl, 10 mM KCl, 1 mM CaCl₂, 50 μg/ml streptomycin, 50 U/ml penicillin (pH 6.2)]. Cells were plated at 2.5×10⁴ cells/cm² in stalk buffer supplemented with 5 mM cAMP. After 24 hours, cell cultures were washed free of cAMP with stalk buffer and the original buffer volume was replaced with stalk buffer with or without supplements. Cells were assayed for prestalk gene expression 24 hours later by fluorescence microscopy, via an ecmA/GFP reporter construct and scored for the production of stalk-like cells using phase-contrast microscopy. Cellulose deposition by the stalk-like cells was confirmed by staining with calcafluor(Harrington and Raper,1968).

Antibodies against TagA protein were raised at Bethyl Laboratories, Inc.(Montgomery, Texas). The peptide LPSNSRNTRNADKLRNRSET, representing amino acids 1627-1646 of the predicted TagA protein, was synthesized and conjugated to keyhole limpet hemocyanin via a cysteine residue added at the amino terminal end of the peptide. The conjugated peptide was used to immunize rabbits and the resulting polyclonal antibodies were affinity purified on a column conjugated with the peptide antigen.

Cells were harvested at various times during development and resuspended in 50 mM Tris-HCl (pH 8.0), 5 mM EDTA, 150 mM NaCl, 0.5% NP-40, 1 mM PMSF. Protein concentrations were determined with the BioRad protein determination kit (BioRad Laboratories, Richmond CA) and equal amounts of protein were resolved on 6% polyacrylamide gels. Protein was electrotransferred to a nitrocellulose membrane and detected with the affinity-purified anti-peptide antibody described above, followed by goat anti-rabbit antibody and visualized with the ECL kit according to the manufacturer's protocols (Amersham Life Sciences).

RNA was isolated from the wild type and the transformed strains during vegetative growth and development. Spores and stalks were purified as described by Van Driessche et al. (Van Driessche et al., 2002). Spores and stalks purified by this procedure were estimated to be >99% pure by direct microscopic observation. Cells were harvested and suspended in TRIzol Reagent (Gibco BRL). RNA extraction was performed according the manufacturer's protocol. RNase protection assays were performed using a RPA III kit (Ambion) according to the manufacturer's protocol with probes synthesized from various fragments of tagA cDNA, or genomic DNA, templates using a riboprobe in vitro transcription system (Promega), as described in the text.

Expression profiling was carried out with DNA microarrays as described previously (Van Driessche et al.,2002). Cells (10⁸) developing on two filters were harvested and processed to produce total RNA for each time point. The raw data from the microarray hybridizations were processed according to the procedure described in Van Driessche et al. (Van Driessche et al., 2002). Briefly, raw image files were quantified and the resulting data was subjected to a single array normalization procedure to remove spatial and intensity artifacts and to put the data on a common measurement scale (Yang et al.,2002). Replicate arrays from the same biological preparation were averaged, and the averages for two biological preparations were then averaged to yield the final data set. We compared the time patterns of gene expression in the tagA mutant cells and our previous wild-type data(Van Driessche et al., 2002). The 2,021 genes whose expression levels were altered dramatically during development were determined and ordered in the `blue-yellow' plots as described previously (Van Driessche et al., 2002). Each gene's studentized score was compared against the linear contrast function y=x/12–1 to evaluate how strongly and consistently the gene appeared to change in relative expression across developmental time. The same gene order is used in all plots shown.

To compare microarray data for all genes from each time point of tagA mutant development with each time point of wild-type development we determined the time point for the wild type at which the expression pattern was most similar to that found in the tagA^- mutant by using the Pearson correlation distance to make comparison. To measure the distance we used data from all of the genes that were not excluded from the analysis for quality control reasons (∼6,000 genes).

Genomic sequence analysis has identified 68 ABC transporter genes from Dictyostelium (Anjard et al.,2002). We carried out a PCR screen of the Dictyosteliumgenome designed to identify genes of the ABCB transporter subfamily. A fragment of the tagA gene was among the PCR products that were cloned and this was used to screen cDNA and genomic DNA libraries to isolate a full-length copy of the gene. The predicted amino acid sequence of the 5.1-kb tagA coding region indicates that it is a member of the taggene class of the ABCB subfamily (Fig. 1). This class includes the previously characterized tagB,tagC and tagD genes of Dictyostelium that are predicted to encode serine protease/ABC transporter proteins(Shaulsky et al., 1995; Anjard, et al., 2002). While the overall gene structure suggests it is a tag gene, the predicted amino acid sequence of the transporter domain is more similar to mammalian ABCB subfamily transporters than it is to the other tag transporters(Fig. 1D)(Anjard et al., 2002). In the most conserved region of the transporter domain, the NBD, the deduced amino acid sequence of TagA is 53% identical to that of human ABCB1 (MDR1 or P-glycoprotein) and 32% identical across the entire transporter domain.

To determine the timing of tagA expression during development an RNase protection assay was used to determine mRNA levels since we were unable to detect tagA mRNA on northern blots. A fragment of the tagA cDNA (PR 1 or PR 2, Fig. 1A) was used to protect tagA mRNA from digestion by RNase. This showed that the tagA gene is expressed at low levels in vegetative cells and accumulates to its highest levels in the first 2 hours of development (Fig. 2A, upper panel). Anti-peptide antibodies were raised to a predicted intracellular epitope within TagA and affinity purified using the immunizing peptide. Western blots stained with these antibodies detected a protein of an apparent molecular mass of 190 kDa, the size predicted for TagA(Fig. 2B). This protein was not detected in the tagA disruption mutant, but was detected in the same mutant transformed with a tagA expression plasmid (see below)indicating that the antibodies detect TagA protein. TagA accumulated to its maximum level between 6 and 10 hours and then persisted at lower levels to the end of development (Fig. 2B).

The tagA gene was inactivated in wild-type (Ax4) cells by inserting a selectable gene into the ClaI restriction site within the predicted NBD of the ABC transporter domain(Fig. 1A). The resulting mutant produced no detectable tagA mRNA by our RNase protection analysis and no detectable TagA protein (Fig. 2). When starved on nitrocellulose filters, the developmental morphology of tagA^- cells appeared relatively normal until 12 hours after the onset of development. At 12 hours, the structures produced by the Ax4 strain were all hemispherical mounds, whereas the tagA^- cell mounds had already formed a tip of prestalk cells. At 14 hours, when the Ax4 cells had formed tipped mounds, tagA^- cells formed elongated finger-like structures with spiral-shaped tips or mounds with supernumerary tips(Fig. 3A). These structures went on to form fruiting bodies in an asynchronous fashion between 24 and 36 hours while Ax4 cells completed development by 26 hours (not shown).

The developing structures of tagA mutants suggested aberrant prestalk cell differentiation. To visualize the major cell types and examine possible defects in proportioning or morphogenesis, the tagA mutation was reproduced in strains expressing green fluorescent protein (GFP) under the control of either the prestalk-specific ecmA promoter or the prespore-specific cotB promoter. Slugs formed by the tagA^-[ecmA/GFP] mutants displayed an extended prestalk zone and a general increase in fluorescence resulting from additional GFP-positive cells in the prespore zone, consistent with increased numbers of ecmA-positive prestalk cells (Fig. 3B). The most dramatic difference between wild-type and tagA mutant fruiting bodies is in the lower cup region of the spore head. In wild-type cells, very little expression of the ecmA reporter construct is seen in the lower cup, presumably because of the lower number of ecmA-expressing PstB cells that contribute to this structure, whereas the tagA mutant fruiting bodies consistently displayed more GFP fluorescence in this region (Fig. 3C). The developmental morphology and the ecmA/GFPexpression pattern suggested that the tagA mutant makes more prestalk cells than normal. We confirmed this by harvesting developing structures at various times, dissociating the cells and determining the percentage of ecmA-positive cells (Fig. 3D). This demonstrated that prestalk cell differentiation in the tagA mutant began at the same time as the wild type, but that the number of prestalk cells increased faster in the mutant and remained 2-3 times higher than wild-type throughout the second half of development. The expression pattern of the cotB/GFP reporter construct was found to be complementary, both spatially and proportionally, to that of ecmA(data not shown). 96% of all cells expressed one of these two constructs. We tested whether the excess ecmA-positive cells in the tagAmutant could be identified as PstO cells. Using wild-type cells expressing the lacZ reporter under the control of a PstO cell-specific promoter(Ax4[ecmO/lacZ]) and in the corresponding tagA mutant(tagA^-[ecmO/lacZ]) we observed no increase in the number of PstO cells in the tagA mutant (data not shown).

The propensity of tagA mutants to differentiate as ecmA-positive prestalk cells was examined in chimeras with wild-type cells to explore the possibility that this phenotype results from a perturbation of intercellular signaling. Chimeras were made with wild-type and tagA mutant cells at various ratios in which one of the strains was marked with the ecmA/GFP reporter to follow the prestalk population of each strain. Prestalk cells derived from the tagA^-cells were consistently overrepresented in the prestalk cell population of the chimeras (Table 2). For instance, when only 10% of the cells were tagA^- 18% of the prestalk cells came from the mutant population. We also carried out an analogous set of experiments with the prespore reporter (cotB/GFP)and found that the tagA mutants were slightly underrepresented in the prespore population, as expected (data not shown). Finally, we demonstrated that tagA mutant cells participate in development of the chimeras by examining mixtures of control strains that expressed GFP under the control of the actin15 promoter. Wild-type (Ax4[act15/GFP]) or mutant(tagA^-[act15/GFP]) strains were mixed with unmarked cells to determine their representation within aggregates and both strains contributed at the expected percentage to the developing populations within each type of cell mixture described in Table 2 (data not shown). Since tagA mutant cells are able to enter aggregates as well as wild-type cells, these results suggest that the additional prestalk cells result from a cell-autonomous defect in cell differentiation.

The excessive prestalk cell differentiation in tagA mutants could result from a change in the way the mutant cells respond to signals in the mound, or by a cell-autonomous mechanism as suggested above. To explore this further, we examined the production of stalk cells by tagA^- cells at low cell density in submerged culture. In this assay, Ax4 cells have been shown to produce ∼15% prestalk cells in response to DIF (Good and Kuspa,2000). However, in our previous studies 4% of the cells consistently differentiated into prestalk cells (as defined by ecmAexpression) without added DIF-1 and when most endogenous DIF-1 production was inhibited by addition of cerulenin (Kay,1998; Good and Kuspa,2000). As with development on filters, a higher percentage of tagA^- cells expressed ecmA in submerged culture compared to Ax4 cells (Fig. 3E). Without added DIF-1 nearly three times more tagA^- cells expressed ecmA. A similar percentage of ecmA-positive cells were formed when tagA^-cells were treated with cerulenin, suggesting that the endogenous DIF-1 synthesis is not required for this process(Fig. 3E). These results reinforce the idea that tagA^- cells have a cell-autonomous propensity to form excess prestalk cells.

PstB cells defined by the ecmB gene are scattered throughout the finger and slug structures prior to terminal differentiation(Williams, 1997). PstB cells at the bottom of the prespore region of the slug can contribute to the basal disk. In the absence of slug migration, a population of `rearguard' cells will form the basal disc proper while the PstB cells will form the outer basal disc and the lower cup at the base of the sorus. Expression of the ecmBgene appears to be precocious and elevated in the tagA mutants(Fig. 4A). Additional ecmB-positive cells are also apparent in tagA mutant slugs but they do not appear to expand the volume of ecmB-positive tissues in the final fruiting body as we observed with ecmA expression(Fig. 4B).

Information on the level of gene expression on a genomic scale can be used to measure the progression of cells through the developmental program and as a means of assessing the differences between wild-type and mutant cells(Hughes et al., 2000; Kim et al., 2001; Van Driessche et al., 2002). We compared the level of gene expression of 2,021 developmentally regulated genes during 24 hours of development in wild-type and in tagA mutant cells. The 2,021 genes that we focused on can be considered a `universal phenotype' since they are not altered in their developmental regulation in different wild-type strains or in cells with different nutritional histories(Van Driessche et al., 2002). We need not be concerned whether the observed gene expression changes are a direct or indirect effect of the lack of TagA since we are only interested in deriving insights into TagA function by observing alterations in the transcriptional program. In Fig. 5A,B, the level of gene expression at each time point is presented relative to the mean level of each gene's expression throughout development in the respective strain. We found that the tagA mutant cells exhibited fairly normal regulation of gene expression. To further elucidate the differences between the strains, we compared the level of expression for each gene in the tagA samples with the mean level of that gene in the wild-type samples (Fig. 5C). This comparison revealed that the tagA mutant cells express the developmentally induced genes at a lower level than wild-type cells and they express the developmentally repressed genes at a higher level than the wild-type cells. In addition, the induction of developmental genes and the repression of growth phase genes were not as rapid in the mutant as in the wild type. These results indicate that tagA mutant cells fail to make the sharp transitions from growth to development and from the unicellular stage to the multicellular stage that are observed in wild-type cells(Van Driessche et al., 2002). We also compared the expression of the cell-type enriched genes defined by Van Driessche et al. (Van Driessche et al.,2002) in the tagA mutant with the wild-type expression. When the tagA mutant expression data is normalized to the wild-type gene expression, most of the stalk-enriched genes appear to be overexpressed in tagA mutants while the spore-, prespore- and prestalk-enriched genes appear to be under-expressed (data not shown).

In order to compare the temporal progression of the mutant and the wild-type cells through development, we calculated the similarity in expression levels between each of the genes at every time point in the mutant samples with each of the genes at every time point in the wild-type samples. If there were few differences between the transcriptional programs of the two strains, one would expect this comparison to result in a diagonal line(Fig. 5D). The actual data indicate that the tagA mutants develop with normal timing only during the first 2 hours of development (Fig. 5D). The transcriptional program of the tagA mutant cells then appears to pause, as the samples collected at 4 and 6 hours of development are most similar to the sample collected from wild-type cells at 2 hours of development. After 6 hours of development, the overall trend of the plot suggests an accelerated rate of development up to 16 hours. At this time the transcriptional program appears to be arrested in the tagAmutants since all times from 16 to 24 hours in the tagA mutant are most similar to the 16-hour wild-type sample. The apparent 16-hour arrest in the developmental program of tagA mutants is consistent with their extended period of fruiting body formation (from 24 to 36 hours) and is reflected in the delayed spore production and poor spore viability. At 24 hours of development, wild-type cells produced 2.4±0.6×10⁷ spores (87% of which were viable) while the tagA mutants produced 2.6±1.6×10⁶ spores(13% of which were viable). These results indicate that tagA^- mutants have a marked attenuation of development at the onset of TagA expression and again at culmination, suggesting that TagA performs critical functions at these two times.

Tissue-specific expression of the tagA gene was determined by expressing lacZ under the control of the tagA promoter in wild-type and mutant cells. This promoter is likely to be complete as it rescued TagA protein expression and it corrected the development of tagA mutant cells (Fig. 2B). Histochemical staining of the developing Ax4[tagA/lacZ] cells for β-galactosidase showed expression of the tagA gene in the prespore region of developing fingers and in the sori of fruiting bodies (Fig. 6A and data not shown). We isolated spores from mature fruiting bodies after 36 hours of development and stained them for β-galactosidase activity. We found that 82±8.0% of wild-type spores appeared to express the tagA/lacZ reporter construct as judged by their blue color in bright-field microscopy. In all developing structures observed, the tips of fingers and anterior (PstA) region of the slug showed no detectable lacZ expression. In the tagA^-[tagA/lacZ]strain, very faint staining could be observed in the prespore region during development, but only after 24 hours of staining. Upon culmination, weakβ-galactosidase expression was evident only in cells of the outer basal disc and in the lower cup of the fruiting body(Fig. 6A). These tagA-expressing cells occupied the same position in the mutant fruiting bodies as the extra ecmA-positive cells described above. Interestingly, there was no detectable staining in the tagA mutant sori. Very few spores isolated from these fruiting bodies displayed anyβ-galactosidase activity as determined by staining for 48 hours(0.58±0.62%). To confirm this result we attempted an RNase protection assay on spore and stalk RNA with a probe that lies between the promoter and the insertion mutation in the tagA gene. The samples were harvested at a time that the tagA mRNA levels are predicted to be extremely low in wild-type cells (24 hours; Fig. 2) and less than ten percent of the tagA mutant cells are expected to express the gene (Fig. 6A). In spite of this, the assay consistently revealed an enrichment of tagA mRNA in the spores of the wild type, as expected,and a consistently higher signal in the stalk tissue relative to the spores produced by tagA mutant cells(Fig. 6B).

In an attempt to detect cell-cell signaling requiring TagA, we looked for the rescue of tagA expression in tagA^- prespore cells by mixing tagA^-[tagA/lacZ] cells with unmarked wild-type cells in a 1:1 ratio. Spores were isolated from the resulting fruiting bodies and 3.4±2.2% stained positive for lacZ expression. This modest six-fold increase in the staining of spores (from 0.58% when the mutant develops alone) indicates that tagA promoter could be activated above the threshold of detection by histochemical staining within the prespore cells of tagA mutants. This limited rescue of tagA expression by wild-type cells is suggestive of a positive feedback in tagA expression.

These results suggest that a fate change occurs in tagA-expressing cells within the tagA^- cell population so that cells that would have become prespore cells develop into a cell type similar in character to PstB cells. Since the expression profiling suggested that the majority of tagA^- cells are affected by the loss of TagA we examined the potential for additional alterations of cell specification by surveying the expression of archetypal cell-specific genes. The spore coat protein gene cotB is a reliable marker of prespore and spore cells and is coordinately regulated with several other spore coat protein genes(Fosnaugh and Loomis, 1991). The cotB/lacZ reporter construct used to visualize cotBexpression showed a normal staining pattern in wild-type cells, with all the spores staining blue with X-gal and no staining of the stalk cells(Fig. 6C). TagA mutant spores stained as expected, but many of the vacuolated stalk cells were also stained,revealing that these cells had expressed the cotB gene at some time in development (Fig. 6C). The spiA gene is normally expressed exclusively within encapsulating prespore cells and is required for the long-term stability of dormant spores(Richardson and Loomis, 1992; Richardson et al., 1994). The spiA mRNA displayed slightly lower spore/stalk enrichment in the tagA mutant samples compared to the wild type(Fig. 6D). Intriguingly, the ecmB mRNA appears to be present in the spore RNA purified from tagA mutants in a much higher proportion than in the wild type. This contrasts with the expression pattern observed with the ecmB/lacZreporter gene that showed little expression in tagA mutant prespore cells or spores (Fig. 4). This difference suggests that the promoter present in the ecmB/lacZconstruct is not active in tagA mutant prespore cells or that the native ecmB gene has additional promoter elements that are missing in the artificial lacZ construct. Alternatively, the native ecmB mRNA may be more stable than lacZ mRNA in tagAmutant prespore cells. Nevertheless, the unexpected patterns of cotB,ecmB and tagA expression in tagA mutants indicate that tagA mutants produce terminally differentiated cells in spite of substantial mis-expression of cell-specific genes.

We have identified a new member of the tag gene family that is required for cell fate determination early in Dictyosteliumdevelopment. TagA appears to prevent a discrete population of cells from becoming prestalk cells at the earliest stages of development. TagA expression defines a population of cells that normally differentiate into spores, but in the absence of TagA activity those cells become some type of stalk cell and form part of the outer basal disc and lower cup of the fruiting body. These cells behave like true PstB cells as determined by their localization in slugs and fruiting bodies. In addition, we observed a significant alteration in the cell-specific gene expression patterns. Three out of the four cell-type-specific genes that we examined showed significant expression in the other cell type in tagA mutants. These findings support the notion that TagA protein is fundamentally important for the specification or maintenance of the differentiated state of the cell types. However, the fact that not all tagA mutant cells become stalk cells and most still sporulate, suggests that there are overlapping mechanisms for the establishment and maintenance of cell type proportions that can compensate for the lack of TagA. Overlapping mechanisms that promote the production of spores would help to ensure the propagation of the species.

The analysis of the tagA gene suggests that overt prespore cell differentiation may occur as early as 2 hours of development. TagA mRNA and protein begin to accumulate to significant levels within the first 2 hours of development. The results obtained with the tagA/lacZ reporter construct indicate that TagA expression becomes spore specific. It is possible that the tagA promoter is active in all cells early, theβ-galactosidase is turned over during aggregation and tagAexpression becomes spore specific only later, but the more than 8-hour half-life of the β-galactosidase produced from this construct argues against this possibility (Detterbeck et al., 1994). Thus, it is possible that some cells require the expression of tagA in the first 2 hours of development in order to differentiate as prespore cells. In this regard, it is interesting that the first significant increase in the curve that describes the increase in ecmA-expressing cells in the tagA mutant extrapolates back to about 6 hours of development (Fig. 3). In addition, the global gene expression profiling of the tagA mutant revealed a delay in the developmental program beginning between 2 and 4 hours of development. Thus, the cell fate and gene expression changes that we observed in the mutant support the idea that the first critical time for TagA function is prior to 6 hours of development.

It has generally been accepted that cell-type-specific gene expression begins at about eight hours, as the mound forms during aggregation(Williams et al., 1989; Haberstroh and Firtel, 1990; Fosnaugh and Loomis, 1993). However, recent reports suggest that some form of cell-type divergence may occur much earlier. Iranfar and co-workers uncovered six genes with cell-type specific expression at the slug stage that initiate expression between 2 and 5 hours of development (Iranfar et al.,2001). Van Driessche and co-workers identified dozens of cell-type-enriched mRNAs that reach their highest expression levels from 2 to 6 hours of development (Van Driessche et al., 2002). More detailed analyses may reveal that these mRNAs become cell-type enriched much later in development through, for instance,their differential stability within the different cell types. It is also possible that these transcripts reflect an early divergence in the physiological state of cells within the starving population that influences cell-type specification.

Physiological differences, such as prior growth conditions and cytosolic pH, between vegetative cells have been observed to influence cell type divergence later in development (e.g., Leach et al., 1973; Maeda and Maeda, 1974; Gross et al., 1983). It is well documented that cell cycle phase at the time of starvation influences cell differentiation later in development (reviewed by Maeda, 1997). In fact, this influence can be observed as cell-autonomous cell-type specification in low cell density cultures and is under regulation by the RtoA protein(Gomer and Firtel, 1987; Wood et al., 1996). It is likely that the influence of the cell cycle is to provide a bias in cell fate determination that is realized by later signaling through, for example, cAMP,DIF or calcium (e.g. Clay et al.,1995; Thompson and Kay,2000; Azhar et al.,2001). A prespore-specific transporter has been characterized,RhT, that appears to be involved in prespore cell differentiation and whose activity can be detected prior to the expression of the spore coat protein gene, cotB, as the prestalk and prespore regions are coalescing in the early mound (Good and Kuspa,2000). All of these physiological differences amongst cells can predispose a particular cell to one cell fate or another, but it is generally accepted that these biases do not determine cell fate and are reversible in different experimental contexts in vivo. The tagA gene is an example of a prespore-specific gene that is expressed at the onset of development and is also required for the sporulation of those cells that express it. Thus, tagA provides genetic evidence that some prespore cell differentiation occurs well before the aggregation stage and may provide a link between the physiological status of growing cells and the cell fate determination that occurs in the first few hours of development.

We used transcriptional profiling as a way of obtaining a global view of the physiological state of the tagA mutants during development. Monitoring global changes in gene expression allows the detection of mutation-induced deviations from an otherwise robust transcriptional program. The early pause in the transcriptional program that we observed between 2 and 6 hours of development in the tagA mutant coincides well with the onset of TagA RNA and protein expression in the wild type. The ablation of the normal transcriptional program precisely when TagA is first expressed reinforces the notion that tagA plays an important role in early development. It is important to note that we would not have observed this pause unless the majority of cells in the population had experienced a 4-hour delay in the transcriptional program. At least 70% of wild-type cells express tagA at some time during development, but only about ten percent of the cells express detectable levels of tagA in the mutant as judged by β-galactosidase staining. These facts together with the delay in the transcriptional program and the inappropriate expression of the cotBand ecmB genes suggests that most cells are affected by the loss of tagA, but most of them compensate for the loss and go on to make spores and stalk while a small percentage of cells are directed to an anomalous PstB-like state. The fact that the early delay in global gene expression in tagA mutants lasts for 4 hours suggests that the defect in the unicellular to multicellular transition stems from a failure to make an initial population of TagA-expressing cells in a timely fashion. The later delay in the transcriptional program in the tagA mutant, between 16 and 24 hours, suggests that TagA functions late or that the mutants lose synchrony as development proceeds. This fits with the observed morphological asynchrony, the 12-hour delay in the completion of development and low viability of tagA mutant spores.

Although we have yet to develop an assay for the biological function that is mediated by TagA, we were able to obtain indirect evidence of a putative TagA signaling event by monitoring deviation from the wild type in the expression of thousands of genes early in development. This is important given that the mutant organism can compensate for the lack of TagA, making it difficult to explore TagA function using cellular or morphological criteria. TagA is most similar in its predicted structure to TagC and TagB. TagC has been implicated in the cellular export of the peptide signal, SDF-2, thought to stimulate the terminal differentiation of prespore cells; there is also genetic evidence that TagB is required for this signaling event(Shaulsky et al., 1995; Anjard et al., 1998). The protease/transporter homology of TagA and the cell-autonomous phenotype of tagA mutants suggest that TagA exports a peptide that must be cleaved and removed from the cell for prespore cell differentiation to occur prior to aggregation. The export of a differentiation inhibitor was also proposed to explain the cell-autonomous specification of PstA cells by TagB and the maintenance of the undifferentiated state of stem cells in mammals(Shaulsky et al., 1995; Zhou et al., 2001).

It will be important to determine the regulatory pathways that tagA impinges on within the prespore cells that are most affected by loss of tagA function. There are two other genes whose inactivation results in the cell-autonomous production of PstB-like cells: one encodes the Dictyostelium homolog of glycogen synthase kinase 3 (GSK-3), gskA, and the other, stkA, encodes the Stalky protein that resembles a GATA family transcription factor(Harwood et al., 1995; Chang et al., 1996). When gskA mutant cells are co-developed with wild-type cells they produce PstB cells that occupy the lower cup and outer basal disk and they over-express the ecmB gene(Harwood et al., 1995). Both of these phenotypes are reminiscent of what we have observed in tagAmutants. Thus, one possibility is that TagA is needed to maintain active GSK-3 in a small cohort of prespore cells early in development. Stalky mutants also overproduce stalk cells, but stkA appears to function much later in development than either tagA or gskA. The function of stkA also appears to be independent of gskA(Chang et al., 1996). Future work will focus on identifying the signaling pathways controlled by TagA and the identification of the TagA substrate.

We thank Glaucia Souza for making the Ax4 act15/GFP strain and Miroslava Ibarra and Trushar Sarang for excellent technical assistance. We thank Richard Gibbs, the Human Genome Sequencing Center at Baylor College of Medicine and Dictyostelium international sequencing consortium for providing unpublished DNA sequence data. This work was supported by National Institutes of Health (NIH) RO1 GM52359 and PO1 HD39691. R.G. was supported, in part, by a developmental biology training grant from the National Science Foundation (BIR-9413237). M.C. was supported, in part, by an NIH Minority Student Development Grant (R25GM56929, G. Slaughter, P.I.).