ABSTRACT
An interesting but largely unanswered biological question is how eukaryotic organisms regulate the size of multicellular tissues. During development, a lawn of Dictyostelium cells breaks up into territories, and within the territories the cells aggregate in dendritic streams to form groups of ∼20,000 cells. Using random insertional mutagenesis to search for genes involved in group size regulation, we found that an insertion in the cnrN gene affects group size. Cells lacking CnrN (cnrN−) form abnormally small groups, which can be rescued by the expression of exogenous CnrN. Relayed pulses of extracellular cyclic AMP (cAMP) direct cells to aggregate by chemotaxis to form aggregation territories and streams. cnrN− cells overaccumulate cAMP during development and form small territories. Decreasing the cAMP pulse size by treating cnrN− cells with cAMP phosphodiesterase or starving cnrN− cells at a low density rescues the small-territory phenotype. The predicted CnrN sequence has similarity to phosphatase and tensin homolog (PTEN), which in Dictyostelium inhibits cAMP-stimulated phosphatidylinositol 3-kinase signaling pathways. CnrN inhibits cAMP-stimulated phosphatidylinositol 3,4,5-trisphosphate accumulation, Akt activation, actin polymerization, and cAMP production. Our results suggest that CnrN is a protein with some similarities to PTEN and that it regulates cAMP signal transduction to regulate territory size.
We know very little about what regulates the size of most tissues and groups. For some tissues, it appears that secreted factors are involved in size regulation (32). In one of the best-understood size regulation pathways, myostatin secreted by muscle cells regulates muscle size by inhibiting myoblast proliferation. Once the amount of muscle in the body increases, the level of myostatin in serum increases correspondingly, which suppresses myoblast proliferation. This negative feedback loop maintains the appropriate ratio of muscle cells to body mass (64). Mutations in myostatin break the feedback loop, leading to abnormally large muscles (47, 50, 69).
Unlike the above example, tissue size regulation in Dictyostelium discoideum is independent of cell proliferation, since proliferation essentially stops when starvation-triggered development starts (57). Single cells aggregate by chemotaxis, using relayed pulses of extracellular cyclic AMP (cAMP) as a chemoattractant to break a field of cells into aggregation territories. Within territories, cells form dendritic aggregation streams which coalesce into groups that develop into ∼20,000-cell fruiting bodies composed of a stalk supporting a mass of spores (39).
Our understanding of what regulates aggregation territory size is very incomplete. If the cell density is too low for efficient propagation of the relayed cAMP pulses, there will be many very small territories (18). In contrast, disruption of the Gα subunit Gα9 causes abnormally large cAMP pulses and the formation of small territories (10, 11). In addition, cells lacking the F-actin adaptor protein PhdA, the mitochondrial protein TorA, or the secreted inhibitor of the secreted cAMP phosphodiesterase (PDE) also form small aggregation territories, although the mechanisms are still unclear (29, 52, 65).
Within an aggregation territory, excessively large streams break up into multiple groups, and this breakup is regulated by a secreted protein complex named counting factor (CF) (6-9). Disrupting genes encoding CF components or adding antibodies against CF components prevents cells from sensing the presence of excessively large streams. The resulting unbroken streams merge into abnormally large aggregates (6-9). Conversely, adding an excess amount of CF to developing wild-type cells results in excess stream breakup and thus small groups (6-9, 30).
CF regulates group size in part by increasing cell motility and decreasing cell-cell adhesion (30, 54, 63). Treating cells with recombinant countin or CF50, CF components with similar bioactivities to that of CF, results in increased cell motility and decreased cell-cell adhesion and thus a small group size (7, 30, 54). Treating cells with cytochalasin D, a drug disrupting actin polymerization, causes decreased cell motility and a large group size (63). In addition to regulating cell motility, CF also regulates group size by regulating cAMP-induced cAMP production (63). High levels of CF cause increased cAMP-induced cAMP production and thus small aggregates (30, 63), while decreasing cAMP levels by treating cells with PDE causes large groups (63).
Both cell motility and cAMP-induced cAMP production are regulated by phosphatidylinositol 3-kinase (PI3K)-dependent pathways in Dictyostelium (26, 40, 44). Upon cAMP stimulation, PI3Ks are activated by both Gβγ and activated Ras (35, 55). Activated PI3Ks lead to the production of phosphatidylinositol 3,4,5-trisphosphate (PIP3), which in turn activates Rac/cdc42 and thus induces actin polymerization (16, 53, 70). Cell movement is driven by pseudopodia filled with the polymerized actin (15). Meanwhile, newly synthesized PIP3 also recruits and activates a number of PH domain proteins, including protein kinase B (Akt) and cytosolic regulator of adenylyl cyclase (CRAC). Akt regulates cell motility by regulating cytoskeleton remodeling (14, 17, 48), and CRAC is required for the activation of adenylyl cyclase (AC), which then catalyzes cAMP production (19, 21, 42). Cells lacking PI3K1 and PI3K2 have defective responses to chemoattractant (48). In contrast, cells lacking phosphatase and tensin homolog (PTEN), a phosphatase that dephosphorylates PIP3, have extended responses to cAMP stimulation (35, 36).
To elucidate group size regulation, we conducted a second-site suppressor screen by restriction enzyme-mediated insertion (REMI) mutagenesis (56) on smlA− cells, which overaccumulate CF and form very small fruiting bodies (5, 58). cnrN, one of the genes identified in the REMI screen, encodes a PTEN-like protein. In this report, we show that CnrN negatively regulates PI3K-dependent pathways, including AC-mediated cAMP production. Loss of CnrN results in excessively large cAMP pulses, which in turn cause the formation of small aggregation territories and thus small fruiting bodies. In agreement with the observation that CnrN regulates territory and stream formation while CF regulates stream breakup, we also found that CnrN-mediated size regulation is independent of CF.
MATERIALS AND METHODS
Cell culture and REMI mutagenesis.Cell culture was done as previously described (6), with the exception that HL-5 medium was obtained from Formedium. REMI mutagenesis of clone HDB7YA smlA− cells (5) was done as described previously (56). The insertion location was identified by inverse PCR as described previously (38).
Generation of knockout and overexpressor cells.To make cnrN− cells, a cnrN 5′-homologous arm with a SacII site at the 5′ end and an XbaI site at the 3′ end was amplified by PCR on Ax2 genomic DNA, using the primers 5′-CTCTCCGCGGTTGCATTGCTTAAAAATTGCA-3′ and 5′-CTGGTCTAGACTATGTCTATCTGCTGATACT-3′. A cnrN 3′-homologous arm with a HindIII site at the 5′ end and a KpnI site at the 3′ end was similarly amplified using the primers 5′-CCTCAAGCTTGATCAAGCAAATTCACTTTCA-3′ and 5′-CTCTGGTACCGGTAGTAGTTGTTGTAGTTAA-3′. After digestion with the corresponding restriction enzymes, the two arms and the 1.4-kb XbaI-HindIII blasticidin resistance cassette from pUCBsrΔBam (1, 60) were ligated into pBluescript II SK (Stratagene) by a direct four-way ligation. The resulting plasmid was digested by SacII and KpnI to generate a linear knockout construct containing both homologous arms and the blasticidin resistance cassette. This was then transformed into Ax2 cells by electroporation (56). Transformants were selected by blasticidin, and PCR was performed to verify the deletion of the cnrN gene in blasticidin-resistant transformants. Four individual clones had a disruption of cnrN and were named cnrN− cells. All of the cnrN− clones had the same phenotype. Clone cnrN−-4 was used in this study. A Northern blot was probed with cnrN to confirm the loss of cnrN in both vegetative and starved (for 6 h) cnrN− cells.
To make overexpressor cells, the cnrN coding sequence with a KpnI site at the 5′ end and a BamHI site at the 3′ end was amplified by reverse transcription-PCR (RT-PCR) on Ax2 total RNA, using the primers 5′-CTCGGTACCATGAACCAAGTATTTTTTTCAAAAATTCGA-3′ and 5′-TCTGGATCCATCTTGTATTATCTCTACATTTTTAATATC-3′, and then ligated into pDXA-3D (23) to generate an overexpression plasmid with a Myc tag at the C terminus of CnrN. The resulting plasmid was transformed into cnrN− cells along with the helper plasmid pREP (46). Individual G418-resistant clones with the same phenotype were isolated and were named cnrN−/cnrNOE cells. Clone cnrN−/cnrNOE-7 was used in this study. The expression of Myc-tagged cnrN in overexpressor cells was verified by Western blotting using an anti-Myc antibody (Bethyl Laboratory).
To generate smlA− cells lacking or overexpressing cnrN, smlA− cells (5) were transformed and checked for genotype as described above (the smlA− cells were made using a ura cassette to disrupt the smlA gene in DH1 uracil auxotrophs, so the resulting smlA− cells were sensitive to blasticidin). This resulted in smlA−/cnrN− cells and smlA−/cnrNOE cells. Clones smlA−/cnrN−-1 and smlA−/ cnrNOE-6 were used in this study.
Phosphatase activity assay.The cnrN coding sequence with a BamHI site at the 5′ end and an XhoI site at the 3′ end was amplified by RT-PCR on Ax2 total RNA, using the primers 5′-CTGGGGATCCATGAACCAAGTATTTTTTTCAAAAATTCG-3′ and 5′-CCTGCTCGAGTTAATCTTGTATTATCTCTACATTTT-3′, and then ligated into pGEX-4T-1 (GE Healthcare). The expression of recombinant glutathione S-transferase (GST)-cnrN in Escherichia coli BL21 cells (Novagen) was induced using 0.5 mM isopropyl-β-d-thiogalactopyranoside (IPTG; Calbiochem). GST-cnrN was purified as described previously (27). Western blots to detect GST fusion proteins were done as described previously (27). A phosphatase activity assay was conducted by incubating 300 μl of purified proteins with 300 μl of 1 mg/ml pNPP (Sigma) as a substrate in the reaction buffer (10 mM Tris, pH 7.0, 2 mM MgCl2, 5 mM KCl, 2 mM dithiothreitol [DTT]) for 30 min at 37°C. After the reaction was stopped by adding 300 μl 3 N NaOH, the absorbance at 405 nm was measured using a spectrophotometer.
Group size and number assay.Preparation of conditioned medium (CM) and starvation of cells on filter pads were done as described previously (5, 6), except that cells were resuspended to a density of 6 × 106 cells/ml in PBM (20 mM KH2PO4, 1 mM MgCl2, 0.01 mM CaCl2, pH 6.5). Aggregates were counted and photographed 12 h after starvation. For PDE treatment, cells were starved on filters soaked with PBM for 4 h, and the filters were then transferred to pads soaked with 0.01 unit/ml PDE (Sigma) in PBM. For microscopy, cells were starved on a PM (3 mM Na2HPO4, 7 mM KH2PO4, 2 mM MgSO4, pH 6.5)-1% agar (Oxoid) plate or in an eight-well chamber slide (Nalge Nunc) at a density of 2.5 × 105 cells/cm2. Images were taken using a Cohu charge-coupled device camera on a Nikon SMZ-10 microscope or a Nikon Diaphot microscope, respectively.
Cell-cell adhesion and Western blots.Adhesion was assayed as previously described (54), except that cells were cultured to 5 × 105 cells/ml in HL-5 medium, resuspended to 5 × 106 cells/ml in PBM, and then starved on PBM-soaked filter pads at a density of ∼5 × 105 cells/cm2. Two, 4, and 6 h after starvation, cells were washed off the filter pads with PBM. The total cell number was counted immediately after vortexing of cells for 10 s, and singlets and doublets were counted as dissociated cells after 2 min of gentle rotation. gp24 levels were measured by Western blotting as described previously (54). Western blots to detect CF components were done as described previously (6-8).
Cell motility assay.Cell motility was assayed as described previously (7), except that 1.5 × 104 cells were starved in a well of an eight-well chambered slide (Nalge Nunc) for 6 h. Cell movement was captured at 30 s per frame for 20 frames, using a Cohu high-performance charge-couple device camera. Cell motility was measured by ImageJ.
cAMP-induced cAMP accumulation.The level of cAMP-induced cAMP production was determined as described previously (66), except that cells were developed in development buffer and treated with 2 mM caffeine for 20 min. After cells were stimulated with 10 μM 2′-deoxy-cAMP (Sigma) in the presence of 10 mM DTT, levels of cAMP accumulation were determined using a [3H]cAMP Biotrak assay kit (GE Healthcare).
GTPγS-stimulated in vitro AC activity.Guanosine-5′-O-(3-thio)triphosphate (GTPγS)-induced AC activity in cell lysates was determined as previously described (41, 66). Briefly, developed cells were lysed in the presence of 50 μM GTPγS (Upstate), and the AC activity was determined by measuring the production of cAMP from ATP.
PDE activity and PIP3 levels.Secreted PDE activities in CM and intracellular PDE activities in cell lysates were determined by measuring the hydrolysis of [3H]cAMP, as previously described (2). PIP3 levels were measured as described previously (35). Briefly, cells were developed in phosphate-free buffer with a 75 nM cAMP pulse every 6 min for 6 h and then metabolically labeled with 32PO4 (Perkin-Elmer) for 1 h. Total lipids extracted from cAMP-stimulated cells were analyzed by thin-layer chromatography (TLC). After autoradiography, bands corresponding to PIP3 were scraped off the TLC plate and mixed with scintillation cocktail, and the radioactivity was directly counted. We found that densitometry of the autoradiographs correlated well with the direct radioactivity counts.
Akt membrane translocation and Akt phosphorylation.Akt membrane translocation assays were performed as described previously (31), except that starved cells were pulsed with 75 nM cAMP every 6 min for 6 h and were then treated with 2 mM caffeine for 20 min. Membrane-bound Akt was analyzed by Western blotting using affinity-purified rabbit anti-Akt antibodies as previously described (31). Akt phosphorylation was assayed as described previously (43). In this assay, phosphorylated Akt was detected by Western blotting using anti-phospho-threonine antibodies. The same membranes were stripped with stripping buffer (100 mM β-mercaptoethanol, 2% sodium dodecyl sulfate [SDS], 62.5 mM Tris-Cl, pH 6.8) for 30 min at 50°C and reprobed with anti-Akt antibodies.
F-actin staining and actin polymerization assay.F-actin was stained with phalloidin as described previously (51), with the modification that cells were starved at 2.5 × 105 cells/cm2 in an eight-well chambered glass slide (Nalge Nunc) for 5 to 6 h until aggregation streams started to form. Cells were then fixed with 3.7% formaldehyde and stained with Alexa fluor 488-phalloidin (Invitrogen). Images were taken using an Axioplan fluorescence microscope (Carl Zeiss). F-actin assays were performed as described previously (63), except that cells were starved in the presence of 75 nM cAMP pulses every 6 min and treated with 2 mM caffeine after starvation.
RESULTS
Identification and characterization of cnrN.A second-site suppressor screen by REMI mutagenesis of smlA− cells, which overaccumulate CF and form extremely small fruiting bodies (5, 58), was conducted to search for genes affecting CF signal transduction. From approximately 40,000 independent mutants, we selected 37 clones forming large fruiting bodies, and using inverse PCR, we identified the insertion site for 16 REMI mutants. Twelve of these disrupted genes were previously uncharacterized Dictyostelium genes, and we named them cnr genes in Dictybase (data not shown); one of these was cnrN (NCBI GeneID 3388487). The cnrN cDNA sequence was determined by RT-PCR and rapid amplification of cDNA ends (RACE) PCR. The open reading frame of cnrN encodes a 72-kDa protein with 23 to 25% sequence identity to PTENs (Fig. 1A and data not shown), including Dictyostelium PTEN (DdPTEN), which acts as an antagonist of PI3K-dependent pathways by dephosphorylating PIP3 (20, 40). The putative phosphatase domain of cnrN, from amino acids 20 to 190, has 41% identity to that of DdPTEN, from amino acids 5 to 185 (Fig. 1B).
cnrN has similarity to PTEN phosphatases. (A) Schematic protein structure of CnrN. Putative domains include the phosphatase and tensin homolog PPASE-TENSIN (positions 20 to 190), the tyrosine-specific protein phosphatase/dual-specificity protein phosphatase PTP/DSP (positions 104 to 178), and the C2 calcium/lipid-binding region C2_CaLB (positions 213 to 351). (B) Sequence alignment of the putative phosphatase domain of cnrN with that of DdPTEN. Black bars represent identical residues, and gray bars represent similar residues. (C) A GST-cnrN fusion protein was expressed in E. coli, and affinity-purified proteins were analyzed by SDS-PAGE and Coomassie staining (left) or Western blotting with an anti-GST monoclonal antibody (right). Vector, the expression vector without an insert. (D) Phosphatase activity of purified GST-cnrN. Values are means ± standard errors of the means (SEM) for four independent experiments (*, P < 0.05; t test). (E) The levels of cnrN mRNA were analyzed by Northern blotting. The 0-hour point represents vegetative cells.
To determine if cnrN is a phosphatase, a GST-cnrN fusion protein was expressed in E. coli (Fig. 1C). Using pNPP as a substrate, GST alone had essentially no phosphatase activity, while similarly purified recombinant GST-cnrN showed a phosphatase activity of 23 pmol/minute/μg protein (Fig. 1D), which is higher than the pNPP phosphatase activity (0.7 pmol/minute/μg protein) of the Dictyostelium phospholipid-inositol phosphatase (49). The recombinant cnrN thus has phosphatase activity.
The accumulation of cnrN mRNA in wild-type cells was examined by Northern blotting. The level of cnrN mRNA is relatively low in vegetative cells, increases after starvation, peaks between 5 and 10 h, and then decreases (Fig. 1E). Aggregation territories and streams form after 5 to 8 h of starvation (13). This pattern of cnrN accumulation indicates that cnrN mRNA is present when aggregation territories and streams are forming.
Loss of cnrN results in abnormally small fruiting bodies.To elucidate the function of cnrN, we generated cnrN− cells by homologous recombination in wild-type cells, using a blasticidin resistance cassette to replace the first 209 residues of the cnrN coding region, which contains the phosphatase domain. Disruption of cnrN was verified by PCR (data not shown) and Northern blotting (Fig. 2A). Compared to wild-type cells, cnrN− cells formed smaller fruiting bodies (Fig. 2B and C). To verify the phenotype of cnrN− cells, cnrN was expressed in cnrN− cells. The resulting cnrN−/cnrNOE cells formed fruiting bodies with a size larger than that of cnrN− fruiting bodies and similar to that of wild-type fruiting bodies (Fig. 2D). This suggests that the phenotype of cnrN− cells is due to the loss of cnrN.
Generation of cnrN− cells. (A) Total RNAs were isolated from both vegetative and starved (6 h) cnrN− cells, and the cnrN mRNA was examined by Northern blotting. (B) Fruiting bodies formed by wild-type (WT) cells. Bar, 0.5 mm. (C) cnrN− cells generated by homologous recombination form smaller fruiting bodies than those of parental wild-type cells. (D) The abnormal phenotype was rescued by expression of cnrN in cnrN− cells, and the resulting rescued cells were named cnrN−/cnrNOE cells.
cnrN − cells form small aggregation territories and short aggregation streams.Both wild-type and cnrN−/cnrNOE cells form normal aggregation territories and streams at approximately 6 h and undergo stream breakup 8 to 10 h after starvation (Fig. 3A). In contrast, cnrN− cells form many small territories and very short streams and start to form aggregates without obvious stream breakup 6.5 to 8 h after starvation (Fig. 3A). The number of aggregation territories formed by cnrN− cells is approximately 4 times more than that for wild-type or cnrN−/cnrNOE cells, but the total stream length of each cnrN− territory is about 25 times shorter (Fig. 3B and C). This suggests that cnrN is required to generate and/or maintain streams and normally sized aggregation territories.
Development of wild-type, cnrN−, and cnrN−/cnrNOE cells. (A) Cells developed on nonnutrient agar plates were photographed at the times indicated on the individual panels. Compared to wild-type (WT) and cnrN−/cnrNOE cells, cnrN− cells formed smaller aggregation territories with fewer and shorter aggregation streams, and then a large number of small aggregates, without undergoing stream breakup. cnrN− cells eventually formed small fruiting bodies. Bar, 1 mm. (B) The number of aggregation territories in a field of view, as shown in panel A, was determined by counting the number of aggregation centers. Values are means ± SEM for three independent experiments (*, P < 0.005; one-way analysis of variance [ANOVA] with Bonferroni's posttest). (C) The total length of aggregation streams in a field of view was measured. Total stream length was then divided by the number of territories to obtain the average total amount of stream length per territory. Values are means ± SEM for three independent experiments (*, P < 0.01; one-way ANOVA with Bonferroni's posttest).
CF does not affect cnrN− group size, nor does cnrN affect smlA− group size.Compared to wild-type cells, cnrN− cells accumulate higher levels of countin (data not shown). To test if high countin levels contribute to the small cnrN− group size, we treated cells with 1 μg/ml anti-countin antibodies to deplete active countin. Wild-type cells treated with anti-countin antibodies formed larger and fewer aggregates than those of untreated cells, whereas similarly treated cnrN− cells formed aggregates similar in size and number to those of untreated cnrN− cells (Fig. 4A and B). To further determine whether CF affects cnrN− group size, we tested the sensitivity of cnrN− cells to high levels of CF by mixing cnrN− cells with 10% smlA− cells (5). A total of 10% smlA− cells were able to induce either wild-type or cnrN−/cnrNOE cells to form a large number of small aggregates but had no effect on cnrN− aggregate size (Fig. 4C and D).
CF does not affect cnrN− group size. (A) Cells were starved on filter pads soaked with either PBM buffer or 1 μg/ml anti-countin antibodies in PBM. Aggregates were photographed after 12 h. Bar, 0.2 mm. (B) The number of aggregates was counted. Values are means ± SEM for three independent experiments (*, P < 0.005; t test). (C) Cells were starved with or without 10% smlA− cells. Aggregates were photographed after 12 h. Bar, 0.2 mm. (D) The numbers of aggregates from three independent experiments were counted (*, P < 0.05; t test).
The insensitivity of cnrN− cells to CF suggests that cnrN might be required for CF signal transduction. However, neither disrupting nor overexpressing cnrN in smlA− cells (which overaccumulate CF) affected smlA− group or fruiting body size on filter pads or bacterial lawns, which suggests that cnrN is not part of the CF signaling pathways (Fig. 5A and B). Furthermore, starving cnrN− cells in smlA− CM led to a higher average motility, while starvation in countin− CM led to a lower average motility (Fig. 5C and D), although CF appeared to have less significant effects on cell-cell adhesion (data not shown). Together, the data suggest that cnrN is not required for CF to regulate cell motility, which is a key factor for group size regulation (54, 63), and that cnrN− cells form small groups by using a CF-independent mechanism.
cnrN does not affect smlA− group size, but CF affects cnrN− cell motility. Similar to the generation of cnrN− and cnrN−/cnrNOE cells, smlA−/cnrN− cells or smlA−/cnrNOE cells were generated by disrupting cnrN or by expressing exogenous cnrN in smlA− cells. (A) Fruiting bodies formed on bacterial lawns. Bar, 0.5 mm. (B) Aggregates formed on filter pads. Bar, 0.2 mm. (C) The motility of cnrN− cells was measured after 6 hours of starvation in the presence of wild-type, smlA−, or countin− CM. Values are means ± SEM for four independent experiments (*, P < 0.001; one-way ANOVA with Bonferroni's posttest). (D) A motility histogram of more than 100 individual cells for each cell line indicates that compared to wild-type (WT) CM treatment, smlA− CM treatment leads to increased motility, while countin− CM treatment leads to a decreased motility of cnrN− cells.
Loss of cnrN results in large cAMP pulses.Altered levels of cAMP accumulation may cause small aggregation territories. For example, compared to parental Ax3 wild-type cells, gα9− cells accumulate very high levels of cAMP and form very small territories and fruiting bodies (10, 11). We thus measured levels of cAMP-stimulated cAMP accumulation in different cell lines. Cells were exposed to exogenous cAMP pulses and subsequent caffeine treatment to ensure that all cells were at an equivalent state. Cells were then stimulated with an oversaturating dose of 2′-deoxy-cAMP to eliminate the responses to the intracellular as well as newly synthesized cAMP (11). In the presence of DTT, which inhibits most PDE activity (2, 33), cells accumulate newly produced cAMP upon 2′-deoxy-cAMP stimulation (11). We observed that wild-type cells accumulated cAMP (Fig. 6A) to levels that we observed previously (62). As shown in Fig. 6A, the level of accumulated cAMP in cnrN− cells within the first 3 min after stimulation was much higher than that in wild-type or cnrN−/cnrNOE cells. In the absence of DTT, the level of cAMP production in cnrN− cells in the first 2 min after cAMP stimulation was also higher (data not shown). We also treated cells with exogenous cAMP PDE 4 hours after starvation on filter pads, when territory and stream formation occurs, and found that PDE treatment caused an increased size and decreased number of aggregates and that the effect of PDE treatment was greater for cnrN− cells than for wild-type or cnrN−/cnrNOE cells (Fig. 6B and C). The size and number of aggregates formed by cnrN− cells treated with PDE were comparable to those of wild-type cells without PDE treatment, indicating that PDE treatment rescued the small-group phenotype of cnrN− cells. These results suggest that the small groups formed by cnrN− cells may result from excessive cAMP accumulation during development.
cnrN − cells accumulate high levels of cAMP. (A) Wild-type (WT), cnrN−, or cnrN−/cnrNOE cells were developed for 6 h, treated with 2 mM caffeine for 20 min, and shaken at 200 rpm for 10 min at room temperature. Cells were stimulated with 10 μM 2′-deoxy-cAMP in the presence of 10 mM DTT, and the cAMP levels at the indicated time points were determined. The 0-h time point represents the cAMP level in cells prior to stimulation. Values are means ± SEM for three independent experiments. The levels of cAMP accumulation in cnrN− cells at 0.5, 1, 2, and 3 min were significantly higher than those of wild-type or cnrN−/cnrNOE cells (one-way ANOVA with Bonferroni's posttest; P < 0.05). (B) Wild-type, cnrN−, or cnrN−/cnrNOE cells were starved on PBM-soaked filter pads for 4 h and then transferred to filter pads containing 0.01 unit/ml bovine brain PDE. Aggregates were photographed 12 h after the starvation. Bar, 0.2 mm. (C) The number of aggregates was counted. Values are means ± SEM for three independent experiments (*, P < 0.05 [t test]; one-way ANOVA with Bonferroni's posttest indicates that the differences between cnrN− cells with PDE treatment and wild-type or cnrN−/cnrNOE controls are not significant).
Cells with chemotaxis defects, such as torA− or pten− cells, form small aggregation territories or do not aggregate (36, 65). To determine if cnrN− cells have chemotaxis defects which may contribute to the small-territory formation phenotype, we examined the chemotactic responses of cnrN− cells. In both small-droplet assays and Boyden chamber assays, cnrN− cells appeared to show chemotactic responses similar to those of wild-type cells (data not shown).
cnrN − cells form normal streams when starved at low densities.Cells starved at different densities may form fruiting bodies of different sizes (3). When wild-type cells were starved at 2.5 × 105 cells/cm2, they formed normal streams both on agar plates and in chamber slides, whereas when they were starved at 1 × 105 cells/cm2 or 0.4 × 105 cells/cm2, they formed small or almost no streams (Fig. 7A and B). In contrast, when cnrN− cells were starved at 2.5 × 105 cells/cm2, they formed many small territories with few streams, whereas when they were starved at low densities, they formed much larger territories with normal streams (Fig. 7A and B). Assuming that diluting cells lowers their overall extracellular cAMP level, the observation that diluting cnrN− cells increases territory size and allows streams to form supports the hypothesis that cnrN− cells form small territories due to excessively large cAMP pulses.
cnrN − cells can affect wild-type group size, and cnrN− cells form normal streams when starved at low densities. Wild-type (WT) and cnrN− cells were starved at the indicated densities either on agar plates (A) (bar, 1 mm) or in chamber slides (B) (bar, 0.1 mm). Images were taken when stream formation occurred. Wild-type cells, a mixed population of 90% wild-type and 10% cnrN− cells, and a mixed population of 80% wild-type and 20% cnrN− cells (C) or cnrN− cells, a mixed population of 90% cnrN− and 10% wild-type cells, and a mixed population of 80% cnrN− and 20% wild-type cells (D) were starved on filter pads. Aggregates were photographed 12 h after starvation. Bar, 0.2 mm. (E) The number of aggregates was counted. Values are means ± SEM for three independent experiments (*, P < 0.001; one-way ANOVA with Bonferroni's posttest).
Supplying starved wild-type cells with large cAMP pulses induces cells to form small groups (62). gα9− cells accumulate high levels of cAMP and can induce wild-type Ax3 cells to form small aggregates when a mixed population of 10% gα9− cells and 90% Ax3 cells is starved (10). Similarly, cnrN− cells can induce wild-type cells to form small groups when the percentage of cnrN− cells in the mixed population reaches 20% (Fig. 7C and E). This further supports the hypothesis that high levels of cAMP cause cnrN− cells to form small groups. A total of 20% wild-type cells in the mixed population could not induce cnrN− cells to form normally sized aggregates (Fig. 7D and E), suggesting that the abnormality of cnrN-mediated size regulation cannot be rescued by extracellular factors secreted by wild-type cells.
cnrN affects cAMP production but not cAMP degradation.cAMP is produced by AC and degraded by a family of PDEs in Dictyostelium (2, 41). To determine whether cnrN affects cAMP degradation and/or cAMP production, we measured PDE activities and GTPγS-induced AC activity. After cells were developed for 6 h, secreted PDE activities in CM and PDE activities in total cell lysates of three different cell lines were similar (Fig. 8A and B). In addition to G protein-mediated AC activation in response to cAMP stimulation, AC can also be activated by GTPγS, an analog of GTP which completely activates all GTP binding proteins (41). When developed cells were lysed in the absence of GTPγS, AC activities in all three cell lines were similar, whereas when cells were lysed in the presence of GTPγS, the AC activity of cnrN− cells was significantly higher that that of wild-type or cnrN−/cnrNOE cells (Fig. 8C). The results suggest that cnrN− cells accumulate high levels of cAMP because of increased cAMP production rather than decreased cAMP degradation.
cnrN negatively regulates cAMP production. Wild-type (WT), cnrN−, or cnrN−/cnrNOE cells were developed for 6 h. (A) CM was collected, and secreted PDE activities in CM were measured. (B) Cells were lysed, and intracellular plus membrane-associated PDE activities were measured. Values in panels A and B are means ± SEM for three independent experiments. In both panels A and B, the differences are not significant (one-way ANOVA with Bonferroni's posttest; P > 0.05). (C) Cells were lysed in the presence of 50 μM GTPγS. The GTPγS-induced AC activity was determined by measuring cAMP levels. Values are means ± SEM for three independent experiments. The GTPγS-induced AC activity of cnrN− cells was significantly higher than that of wild-type or cnrN−/cnrNOE cells (*, P < 0.05; one-way ANOVA with Bonferroni's posttest, comparing the three GTPγS-induced values).
cnrN negatively regulates PIP3 accumulation and Akt activation upon cAMP stimulation.PTEN negatively regulates PI3K-dependent pathways by dephosphorylating PIP3, and pten− cells overaccumulate PIP3 in response to cAMP stimulation (35, 36). cnrN shares 41% identity with DdPTEN, and this prompted us to examine if cnrN has similar effects on PI3K-dependent pathways. Five seconds after cAMP stimulation, cnrN− cells showed a higher and prolonged peak level of PIP3 compared to that of wild-type cells (Fig. 9A and B). This indicates that cnrN negatively regulates PIP3 accumulation.
cnrN negatively regulates PI3K-dependent pathways. (A) Wild-type (WT) and cnrN− cells were developed in phosphate-free buffer for 6 h and then incubated with 32PO4 for 1 h. After cAMP stimulation, cells were lysed at the indicated time points, and lipids were extracted and analyzed by TLC. The black arrows indicate PIP3, and the open arrows indicate PIP2. (B) Levels of PIP3 were determined by densitometry. The 0-hour time point represents the value in cells prior to stimulation. Values were normalized to the value of the wild type at 0 s. Values are means ± SEM for three independent experiments. The levels of PIP3 in cnrN− cells at 5, 20, and 60 s were significantly higher than those in wild-type cells (t tests; P < 0.05). (C) Cells were developed for 6 h and then lysed at the indicated time points after cAMP stimulation. The membrane fraction was collected by centrifugation and analyzed by Western blotting using anti-Akt antibodies. (D) Levels of membrane-bound Akt were determined by densitometry. The 0-hour time point represents the value in cells prior to stimulation. Values were normalized to the value of the wild type at 0 s. Values are means ± SEM for three independent experiments. The Akt translocation level in cnrN− cells at 5 s was significantly higher than that in wild-type or cnrN−/cnrNOE cells (one-way ANOVA with Bonferroni's posttest; P < 0.05). (E) Cells were lysed with SDS sample buffer at the indicated time points. Akt phosphorylation was analyzed by Western blotting using anti-phospho-threonine antibodies, shown in the upper panel. The black arrow indicates phosphorylated Akt, which is the top band in the blot. The same blot was stripped and restained with anti-Akt antibodies to indicate the total levels of Akt and the position of Akt on the blot, as shown in the lower panel. The open arrow indicates Akt. Data are representative of three independent experiments. (F) Levels of phosphorylated Akt were determined by densitometry. The 0-hour time point represents the value in cells prior to stimulation. Values were normalized to the value of the wild type at 0 s. Values are means ± SEM for three independent experiments. The Akt phosphorylation level in cnrN− cells at 5 s was significantly higher than that in wild-type or cnrN−/cnrNOE cells (one-way ANOVA with Bonferroni's posttest; P < 0.05).
Akt is a downstream effector of PI3K signaling pathways and affects a wide variety of cellular events (45, 48). To confirm that cnrN affects PI3K pathways, we examined if cnrN affects Akt activation. Translocation to the plasma membrane and phosphorylation at the two threonine residues are two processes that result in the activation of Akt (48). As previously observed (31), affinity-purified anti-Akt antibodies stained a band at 53 kDa, approximately the predicted size of Akt (Fig. 9C). As shown in Fig. 9C and D, compared to wild-type and cnrN−/cnrNOE cells, cnrN− cells showed a higher peak level of Akt membrane translocation 5 s after cAMP stimulation, which is consistent with the above result that the level of PIP3 is higher in cnrN− cells upon cAMP stimulation. To detect Akt phosphorylation, we used a standard assay in which Western blots are stained with anti-phospho-threonine antibodies (43, 48). The upper panel of Fig. 9E shows proteins detected by anti-phospho-threonine antibodies, and the lower panel shows the proteins on the same blot detected by anti-Akt antibodies. As previously described (43), the phosphorylated Akt band (Fig. 9E, solid arrow) overlapped the Akt band (Fig. 9E, open arrow). Although total Akt levels appeared to vary slightly, there was no consistent difference from experiment to experiment. As shown in Fig. 9E and F, the level of transient Akt phosphorylation in cnrN− cells was higher than that in wild-type and cnrN−/cnrNOE cells, which is also consistent with the observation that the level of Akt membrane translocation was higher in cnrN− cells. These results indicate that cnrN negatively regulates PI3K-dependent pathways.
cnrN − cells exhibit an increased level of actin polymerization.When Dictyostelium cells sense a cAMP gradient and move up the gradient, free actin monomers are polymerized to form F-actin, most of which is accumulated at the leading edge of the polarized cells (25). cAMP-dependent PI3K activation induces actin polymerization (55). cnrN negatively regulates PI3K pathways, so cnrN may also negatively regulate actin polymerization. To measure cAMP-stimulated actin polymerization, cells were lysed by detergent and the detergent-insoluble part containing F-actin was analyzed by SDS-polyacrylamide gel electrophoresis (SDS-PAGE). The levels of F-actin 5 to 20 s after cAMP stimulation in cnrN− cells were higher than those in wild-type and cnrN−/cnrNOE cells (Fig. 10A), which suggested that cnrN inhibits actin polymerization. In agreement with this result, we also found that cnrN− cells show a higher level of cell motility (Fig. 10B and C). To determine the effect of cnrN on the intracellular distribution of F-actin, we observed F-actin by staining cells with phalloidin. As shown in Fig. 10D, F-actin was located mainly at the periphery of resting cells or at the leading edge of polarized cells in all three cell lines. The results suggested that whereas cnrN inhibits the extent of actin polymerization, cnrN may not affect F-actin distribution.
cnrN negatively regulates actin polymerization but does not affect localization of F-actin. (A) Wild-type (WT), cnrN−, or cnrN−/cnrNOE cells were developed for 6 h, treated with caffeine for 20 min, and stimulated with cAMP. Cells were lysed in buffer containing 0.05% Triton X-100 at the indicated time points. The detergent-insoluble part containing F-actin was analyzed by SDS-PAGE. Levels of F-actin were determined by densitometry. The 0-hour time point represents the value in cells prior to stimulation. Values were normalized to the value of the wild type at 0 s. Values are means ± SEM for seven independent experiments. The levels of actin polymerization in cnrN− cells at 5, 10, 15, and 20 s were significantly higher than those in wild-type or cnrN−/cnrNOE cells (one-way ANOVA with Bonferroni's posttest; P < 0.05). (B) The motility of cnrN− cells was measured. Values are means ± SEM for four independent experiments (one-way ANOVA with Bonferroni's posttest; * P < 0.001). (C) Motility histogram of more than 100 individual cells for each cell line. (D) Cells were starved for 6 h and stained with Alexa fluor 488-labeled phalloidin. In all three cell lines, F-actin was localized at the periphery of rounded cells, shown in the upper panel, and the leading edge of all polarized cells, shown in the lower panel. Bars, 10 μm.
DISCUSSION
Much remains to be understood about how organisms regulate the size of a group of cells. In Dictyostelium, two key mechanisms of size regulation are the cAMP-mediated breakup of a field of cells into territories in which cells move toward the aggregation center and form dendritic aggregation streams and the CF-mediated breakup of aggregation streams, which allows the formation of individual groups that eventually develop into fruiting bodies (6, 39). We describe here that CnrN, a protein phosphatase with sequence similarity to PTENs, regulates aggregation territory size, and thus group size, by decreasing the cAMP pulse size during development. The small-group phenotype of cnrN− cells can be rescued by treating cells with PDE or by simply starving cells at a low density to decrease the cAMP pulse size during development.
The REMI mutagenesis method by which the cnrN gene was identified was performed with smlA− cells, which form small groups, and thus REMI mutants forming large fruiting bodies were selected. However, when we deleted the PTEN phosphatase domain of CnrN in wild-type and smlA− cells, we found that the resulting cnrN− and smlA−/cnrN− cells formed small fruiting bodies, which indicates that the loss of CnrN causes the small-fruiting-body phenotype. One possible reason for the phenotypic difference between smlA−/cnrN REMI cells and smlA−/cnrN− cells is that the REMI insertion, which is localized between the PTEN phosphatase domain and the lipid binding domain, might cause altered localization and/or activity of CnrN and thus the large fruiting bodies. Another possible reason is that more than one insertion occurred during mutagenesis, and other insertions may cause the large-fruiting-body phenotype of the cnrN REMI mutant.
CF regulates stream breakup, and thus group size, partially by increasing cell motility (63). To determine whether CnrN is a part of CF signaling, we tested the sensitivity of cnrN− cells to CF. cnrN− group size is insensitive to CF, suggesting at first glance that CnrN is necessary for CF signal transduction. In contrast, the motility of cnrN− cells is sensitive to CF, suggesting the second possibility that CnrN is not a key component of the CF signaling pathway that regulates cell motility. To address the discrepancy between the two possibilities, we overexpressed or disrupted cnrN in smlA− cells, which accumulate high levels of CF, and found that neither high CnrN activity nor the absence of CnrN activity altered the smlA− group size. This implies that it is more likely that CnrN affects some, but not all, parts of the CF signaling pathway.
In their small territories, cnrN− cells appear to mostly aggregate directly into groups with relatively little stream formation. Computer simulations predicted that even though CF does affect motility in cnrN− cells, with very short streams there will be correspondingly less stream breakup, and with the cnrN− groups tending to have a larger diameter than the wild-type stream thickness, the cnrN− groups appear to resist being broken apart by CF regulating motility and adhesion (data not shown). Thus, the loss of CnrN appears to block the effect of CF on group size without actually being part of the CF signal transduction pathway. Together with the previous observation that CF does not affect territory size (5, 6), while a loss of CnrN reduces territory size, these results suggest that CnrN may regulate group size by using a CF-independent mechanism.
A number of previous findings suggested that territory size is negatively affected by excessively high levels of cAMP (4, 10, 24). For example, cells developed on agar plates with 0.6 mg/ml cAMP form 20 times more aggregation territories, with a smaller size, than those for cells developed with 0.15 mg/ml cAMP (4). In agreement with this, we found that the level of accumulated cAMP in cnrN− cells is much higher than that in wild-type or cnrN−/cnrNOE cells. Treating cells with exogenous PDE at the territory formation stage or simply diluting cnrN− cells rescued the small-territory phenotype of cnrN− cells, which supports our hypothesis that the small territories and small fruiting bodies formed by cnrN− cells are caused by the abnormally large cAMP pulses rather than by chemotaxis defects, which cause small or no territories in torA− cells or pten− cells (36, 65).
cnrN − cells have elevated and extended responses to cAMP stimulation and form small territories. Similarly, cells lacking Gα9 (gα9− cells) also show elevated and extended responses to cAMP stimulation, including higher levels of PIP3 and cAMP accumulation, and form a large number of small aggregation territories and small fruiting bodies (10, 11). It is unclear yet if cnrN and Gα9 are part of the same pathway regulating territory size. A total of 10% gα9− cells can induce wild-type cells in a mixed population to form small groups (10). Interestingly, 10% cnrN− cells cannot do this, but 20% cnrN− cells have a similar effect. One possible reason is that the cAMP produced by 10% cnrN− cells is not enough to affect the wild-type cells in the mixture. This is also consistent with the result that the level of cAMP accumulation in gα9− cells is more than twice that in cnrN− cells (11).
We envision that a break between two aggregation territories will begin when cells are sufficiently far apart that a cAMP signal cannot cross the gap and thus cannot attract and entrain cells on the other side of the gap. This could occur when either the cAMP signal is low or when cAMP receptors are desensitized by an oversaturated cAMP signal. Exposing cells to abnormally high concentrations of cAMP may cause the rapid desensitization of cAMP receptors and even the degradation of receptors (68). Therefore, it is likely that high levels of cAMP in cnrN− and gα9− cells may desensitize cAMP receptors and disturb the relay and propagation of cAMP, thus causing the streaming defect and the formation of small territories. Alternatively, low levels of cAMP may not be propagated far from the aggregation center, so the distal cells may not sense the signal and move toward the cAMP gradient, which may also cause the formation of small territories. For example, cells lacking phosphodiesterase inhibitor (PDI), the inhibitor of the extracellular PDE, do not generate spiral cAMP waves and form territories about 50 times smaller than those of wild-type cells (52). This suggests that cnrN, Gα9, and PDI maintain an optimal cAMP pulse size to allow the formation of territories of the proper size.
The loss of cnrN leads to higher levels of cAMP accumulation, PIP3 accumulation, Akt activation, and actin polymerization in response to cAMP stimulation, which suggests that cnrN plays an important role in negatively regulating PI3K-dependent pathways. Although we have not been able to determine whether cnrN directly dephosphorylates PIP3 in vitro, the similarity of cnrN to PTEN suggests that the high level of PIP3 accumulation in cnrN− cells, like that in pten− cells, is caused by decreased dephosphorylation of PIP3 rather than by increased PIP3 production. Our hypothesis is thus that cnrN inhibits cAMP production by negatively regulating PI3K-dependent pathways to allow the formation of normally sized territories. After stream formation within these territories, if a stream is excessively large, CF then increases cell motility, in part by potentiating PI3K pathways, and decreases cell-cell adhesion (31, 54, 63), which leads to the breakup of large streams into groups (Fig. 11).
Model of group size regulation in Dictyostelium. At the aggregation territory formation stage, cnrN downregulates AC activity to prevent the overaccumulation of cAMP, possibly by negatively regulating PI3K pathways, which in turn allows the formation of normal territories and streams. After the stream formation stage, CF increases cell motility by positively regulating cytoskeleton remodeling, which results in the breakdown of large streams to small aggregates. In cells lacking cnrN, although chemotaxis is not impaired, cAMP levels and/or signaling is largely altered, which causes the formation of small territories with few streams. The resulting streams are too small for CF-mediated stream breakup to occur, and thus crnN− group size is insensitive to addition or depletion of CF.
PTEN has been studied extensively in a variety of organisms and is a tumor suppressor in mammalian systems due to its inhibitory effect on PI3K-dependent pathways which regulate cell motility, growth, and survival (22, 59). Loss of PTEN causes NIH 3T3 cells to migrate faster, possibly due to the increased activity of Rac/Cdc42, which potentiates actin polymerization (61). In Dictyostelium, PTEN is generally acknowledged to play an important role in sensing the cAMP gradient and establishing cell polarity, but its role in chemotaxis is somewhat unclear (28, 37). Aggregation of the original pten− cells is severely impaired, likely due to the generation of multiple pseudopods during movement (12, 36, 37, 67). In contrast, Hoeller and Kay reported that their pten− cells, or even sextuple mutants in which PTEN and all five known PI3Ks are disrupted, showed essentially normal chemotaxis (34). Unlike the original pten− cells, but similar to gα9− cells, cnrN− cells are able to aggregate, especially at a low cell density. This suggests that cnrN does not play a major role in chemotaxis. The recent finding that the phospholipase A2 pathway is necessary for chemotaxis adds to the complexity of how chemotaxis is regulated (12, 67).
cnrN − cells move faster, show elevated PI3K-dependent responses to cAMP, and form small aggregation territories compared to wild-type cells. gα9− cells also have a higher cell motility and form small territories (10, 11), but they show stronger PI3K-dependent responses than cnrN− cells do, as indicated by the higher levels of cAMP accumulation and prolonged PIP3 accumulation (11). This suggests that Gα9 plays a more critical role in regulating chemotactic responses and behaviors. pten− cells also show stronger cAMP-induced responses than those of cnrN− cells; for example, in pten− cells, cAMP-stimulated PIP3 accumulation is higher and the high PIP3 level persists longer than observed in cnrN− cells, and actin polymerization remains at a high level even 5 min after cAMP stimulation (35, 36). These results indicate that cnrN and PTEN, regardless of their similarities in amino acid sequence as well as inhibitory activity on PI3K pathways, play different roles in Dictyostelium cells, with cnrN playing a relatively subtle role to regulate cAMP production in order to optimize aggregation territory size.
ACKNOWLEDGMENTS
We thank Bill Loomis for the kind gift of anti-gp24 antibodies, Robin Williams for advice on inverse PCR, Peter Devreotes and Miho Iijima for the protocol for and helpful discussion of the in vivo PIP3 assay, Diane Hatton for assistance with sequence analysis, Debra Brock and Tong Gao for helpful suggestions on filter pad assays and adhesion assays, and Darrell Pilling for critically reading the manuscript.
Richard Gomer was an investigator of the Howard Hughes Medical Institute. This work was supported by grant C-1555 from the Robert A. Welch Foundation and by NIH grant GM074990.
FOOTNOTES
- Received 25 June 2008.
- Accepted 24 July 2008.
- Copyright © 2008 American Society for Microbiology
REFERENCES
