|Home | About | Journals | Submit | Contact Us | Français|
Glycogen synthase kinase-3 (GSK3) is a highly conserved protein kinase that is involved in several important cell signaling pathways and is associated with a range of medical conditions. Previous studies indicated a major role of the Dictyostelium homologue of GSK3 (gskA) in cell fate determination during morphogenesis of the fruiting body; however, transcriptomic and proteomic studies have suggested that GSK3 regulates gene expression much earlier during Dictyostelium development. To investigate a potential earlier role of GskA, we examined the effects of loss of gskA on cell aggregation. We find that cells lacking gskA exhibit poor chemotaxis toward cAMP and folate. Mutants fail to activate two important regulatory signaling pathways, mediated by phosphatidylinositol 3,4,5-trisphosphate (PIP3) and target of rapamycin complex 2 (TORC2), which in combination are required for chemotaxis and cAMP signaling. These results indicate that GskA is required during early stages of Dictyostelium development, in which it is necessary for both chemotaxis and cell signaling.
Glycogen synthase kinase-3 (GSK3) is a multifunctional protein kinase that is highly conserved throughout the eukaryotes. In animals, including humans, it is associated with important embryological and physiological signaling processes, such as Wnt and insulin signaling (Cross et al., 1995 ; Papkoff and Aikawa, 1998 ). As a consequence it is linked to major clinical conditions, including cancer, diabetes, and Alzheimer's disease. At the cellular level GSK3 mediates processes, such as regulation of gene expression, control of proteolysis, metabolism, and control of the cytoskeleton (Harwood, 2001 ; Kim et al., 2002 ). The question of how GSK3 exerts its effects on these multiple processes is important for understanding the integration of cell signaling within the cellular context.
The social amoeba Dictyostelium grows and divides in the presence of nutrients in a unicellular form, but when starved it undergoes a developmental program to form a multicellular and differentiated structure, termed the fruiting body (Harwood, 2001 ). Dictyostelium development comprises of two major phases. First, during aggregation phase, cells migrate together by chemotaxis in a process mediated by cAMP pulses. Second, the multicellular aggregate, or mound, undertakes a series of differentiation events and morphogenetic movements to ultimately form a fruiting body, which comprises spore cells within a spherical head, supported by a stalk of vacuolated, cellulose encased stalk cells.
Dictyostelium possesses a single homologue of GSK3, gskA (Harwood et al., 1995 ). It was discovered previously that GskA regulates the processes that control cell fate (Harwood et al., 1995 ). This initially occurs during early stages of multicellular development, when cells first come together to form a multicellular mound. At this point, cells differentiate into three basic cell types; prespore cells that eventually form spores, and prestalk cells (pst) comprising pstA cells that give rise to the stalk and the majority of its associated structures and pstB cells that form the basal disk that anchors the stalk to the substratum (Williams et al., 1989 ). In wild-type cells, 80% of the cells form prespore cells and the remainder form pst cells, of which the majority are pstA. However, in a gskA null mutant, the pstB population expands at the expense of the prespore population, so that the final fruiting body of the basal disk is greatly enlarged and the spore head reduced (Harwood et al., 1995 ). During later stages of fruiting body formation, GskA regulates a second cell fate decision by controlling the differentiation from pstA cells to pstAB cells (Schilde et al., 2004 ), which progress to form the main structural element of the stalk.
In the mound, activation of GskA is mediated by high concentrations of cAMP via the cAMP receptors carC and carD, which control tyrosine phosphorylation of GskA by the kinase ZakA (Kim et al., 1999 ) and an unidentified tyrosine phosphatase (Kim et al., 2002 ). Downstream of GskA lies a β-catenin homologue, Aardvark (Aar; Grimson et al., 2000 ), and a signal transducer and activator of transcription (STAT) transcription factor, STATa (Ginger et al., 2000 ), although the full role of these proteins is currently unclear.
Although the phenotypes described for gskA null mutants occur during multicellular stages, GskA expression and activity is present throughout Dictyostelium development (Plyte et al., 1999 ). A proteomic and microarray analysis of cells at 5 h of development, several hours before the first multicellular developmental stages, revealed ≈150 genes with altered patterns of expression (Strmecki et al., 2007 ). Curiously, of those genes identified that had been characterized previously, all are associated with growth and aggregation phase rather than multicellular development. This raises the question of additional roles of GskA during these earlier stages of Dictyostelium development.
Here, we report an in-depth investigation of aggregation and chemotaxis of gskA null mutants. We find that gskA null mutant cells have defective chemotaxis toward cAMP and folate, a growth phase chemoattractant. The chemotaxis defect is accompanied by suppression of phosphatidylinositol 3,4,5-trisphosphate (PIP3) levels. However, we observe that loss of gskA causes a much greater effect on cAMP-mediated chemotaxis than can be accounted for by loss of PIP3 signaling alone (Lee et al., 2005 ; Hoeller and Kay, 2007 ; Takeda et al., 2007 ). We find that protein kinase B (PKB)R1 is not phosphorylated in gskA mutants, a protein kinase related to PKB but regulated in a PIP3-independent manner. Loss of both PIP3 signaling and PKBR1 phosphorylation has a strong effect on chemotaxis. Finally, we show that gskA null mutant cells do not generate cAMP pulses during aggregation, demonstrating that GskA plays a major role in the regulation of chemotaxis and early Dictyostelium development.
The gskA null strain (ID Strain: DBS0236114) was obtained from the Dicty Stock Centre (http://dictybase.org/) and was originally generated in the Dictyostelium discoideum strain AX2 background. The wild-type strain refers to AX2 and is the same strain denoted in Bloomfield et al. (2008) . Cells were cultured axenically on plates in HL5 with glucose medium (Formedium HLG0102) before experiment. gskA null mutants were grown in media supplemented with 10 μg/ml blasticidin, and gskA null transformants carrying GskA-green fluorescent protein (GFP) (gskA-GSKA) and immobilized metal assay of phosphorylation (IMPA)-GFP (gskA-IMPA) were grown in medium supplemented with 40 μg/ml G418 (catalog no. 11811–023; Invitrogen). All cells were maintained at 22°C.
The GskA-GFP2 rescue protein for gskA was constructed by inserting the coding regions of GskA into pDXA GFP2 at the Kpn1 and Nsi1 restriction sites, such that the fusion protein was expressed under the control of a constitutive Actin15 promoter. By introducing point mutations into the coding region of the pDXA GskA-GFP2, the kinase-dead construct pDXA GskA-GFP2 K85R was made with the QuikChange site-directed mutagenesis kit (Strategene, La Jolla, CA) as per the manufacturer's direction. The coding sequence of the IMPA gene was amplified from the cDNA clone FCBP15, obtained form the Japanese D. discoideum cDNA project (Morio et al., 1998 ). The insert was subsequently ligated into the pTX-GFP vector.
cAMP was measured as described previously (Snaar-Jagalska and Van Haastert, 1994 ) by using isotope dilution assay kits (GE Healthcare, Little Chalfont, Buckinghamnshire, United Kingdom). Cells were starved in KK2 buffer (6.5 mM KH2PO4, 3.8 mM K2HPO4, pH 6.2) for 5 h in shaking culture at a concentration of 1 × 107 cells/ml. To determine cAMP levels, cells were resuspended at 5 × 107 cells/ml, and 100 μl of cells was stimulated with 5 μM 2′-deoxy-cAMP in the presence of 5 mM dithiothreitol. Reactions were terminated by the addition of 100 μl of 3.5% (vol/vol) perchloric acid on ice. Samples were neutralized with 50 μl of KHCO3 (50% saturated), followed by centrifugation for 2 min at 14,000 × g at 4°C. Finally, 50 μl of each sample was added to the cyclic AMP 3H assay system (GE Healthcare).
Wild-type cells transformed with the plasmid WF38 (PHCRAC-GFP; Parent et al., 1998 ) were pulsed with cAMP for 5 h with an end concentration of 100 nM. Cells were subsequently stimulated by the addition of 1 μM cAMP, and protein translocation was recorded by fluorescence videomicroscope with a 60× objective.
For chemotaxis to folate, cells were grown axenically to a density of 1 × 106, and 4 × 105/cm2 cells were washed twice in 20% media/KK2 and replated in a LabTek coverglass chambered well (Nunc 155379; Nalge Nunc International, Rochester, NY) with 2 ml of 20% media/KK2. Folate (25 mM) was delivered via a Femtotip (5242 952.008; Eppendorf, Hamburg, Germany), and images were recorded for every 20 s under 10× phase objective. For aggregation, 5 × 106 cells were washed and plated on KK2 agar. Images were taken every 30 s for cAMP pulse movies under 4× objective for at least 24 h. Postprocessing was carried out with ImageJ (National Institutes of Health, Bethesda, MD). For chemotaxis, 5 × 107 cells were shaken in KK2 buffer (Formedium KK29907) for 5 h while being pulsed every 6 min with cAMP to an end concentration of 10−7 M. The chemotaxis-competent cells were placed in a Zigmond chamber (Z02; Neuro Probe, Gaithersburg, MD) in a cAMP gradient (source at 10−6 M cAMP, sink with no cAMP). Differential interference contrast images of cells were captured with a 20× objective at 6-s intervals for 15 min. Cell movement was analyzed using Dynamic Image Analysis System (DIAS), version 3.4.1 (Soll Technologies, Iowa City, IA; Curreli et al., 2001 ). Statistical analysis was carried out using the nonparametric Kruskal–Wallis test, with a post hoc Dunn's multiple comparison test using Prism 4 (GraphPad Software, San Diego, CA).
Cells (1 × 108) were starved and pulsed for 4 h with 100 nM cAMP. At the fourth hour, cells were pelleted and resuspended in 600 μl of KK2. Radioactive [γ-32P]ATP was added (1.2 MBq), and the cells were pulsed with cAMP for a further hour (100 nM/6 min).
Phospholipids were extracted as described in Konig et al. (2008) : cells were added to 750 μl of acidified A (chloroform/methanol/HCl (1M) ratio: 40:80:1) and sonicated. A further 250 μl of chloroform and 450 μl of 1 M HCl were added to the cells, and then the cells were vortexed and spun, and the lower phase was collected. The process was repeated two more times, once with chloroform/methanol/HCl (1M) (ratio: 10:10:4) and the second time with methanol/EDTA (100 mM) (ratio: 1:0.9) and subsequently dried. Samples were then resuspended in chloroform/methanol (ratio: 2:1) and spotted onto a TLC plate (Cole-Parmer Instrument, Vernon Hills, IL). Plates were developed with a chloroform/acetone/methanol/acetic acid/double-distilled H2O solvent system (ratio 46:17:15:14:8), and phospholipids were visualized by spraying with molybdenum Blue spray reagent (Sigma Chemical, Poole, Dorset, United Kingdom). Radiolabeled lipid spots were imaged by exposing the plate to Hyperfilm (GE Healthcare) for 3 d. PIP3 measurements were carried out according to manufacturer's instructions (K-2500s PIP3 Mass ELISA kit; Echelon Biosciences, Salt Lake City, UT). Anti-phospho protein kinase C (PKC; pan) antibody used for detection of PKBA and PKBR1 was purchased from Cell Signaling Technology (190D10; Danvers, MA) and analyzed as described previously (King et al., 2009 ).
Dictyostelium cells were starved and shaken with pulses of 100 nM cAMP for 5 h. Cells (2 × 106) were then harvested and pellets were snap-frozen. RNA was isolated using an illustra RNA spin mini kit (25-0500-72; GE Healthcare), and its integrity was checked on a 1% formaldehyde agarose gel. Genomic DNA contamination was checked by PCR using existing primers for a known gene. cDNA was prepared using the First-Strand cDNA kit (11483188001; Roche Diagnostics, Mannheim, Germany) using 1 μg of RNA and random primers. A minus-reverse transcriptase control was also set up. qRT-PCR was carried out on a Chromo4 machine (Bio-Rad laboratories, Hercules, CA) with the Opticon2 detection software. The fluorophore used for detection was SYBR Green. iq SYBR Green master mix (170-8884; Bio-Rad Laboratories) was used for the reactions with 100 μg of cDNA. Housekeeping genes incorporated for quantification were Ig7 (www.dictybase.com; ID Strain: DDB_G0294034), impk1, and impk2. The genes of interest were carA-1 (DDB_G0273397), acaA (DDB_G0281545), and pdsA (DDB_G0285995). Data analysis was carried out using the qbase software Jo VandeSomple and Jan Hellemans (Hellemans et al., 2007 ). Standard curves were created for all genes by using a 3-point 10-fold dilution series of Ax2 0 h cDNA, and all samples also were normalized to the Ax2 0 h cDNA. Primer sequences are as follows: ig7 forward, TCCAAGAGGAAGAGGAGAACTGC and ig7 reverse, TGGGGAGGTCGTTACACCATTC; ipmk1 forward, GCAGGTTCAACACCATTCAAAAAATC and ipmk1 reverse, TCCAACACTATCCATTCCTTACCATC; ipmk2 forward, TGGTAGTTTTTTGAGTGTCAGCCC and ipmk2 reverse, TGATGATGTTGTTGTTGTTG TTGTAGTG; carA-1 forward, ATGTTGGGTTGTATGGCAGTG and carA-1 reverse, AGGGAAACCACCATTGACAG; acaA forward, CATTCTAGAGGCGGTATTGGC and acaA reverse, GGAGAAAATGTCTGATTTCGCTT; and pdsA forward, CCATTGGGTACAACTGGTGGA and pdsA reverse, AACTGCCCATGATGGATAGGT.
We observed previously that gskA null mutant cells had unusual aggregation, forming small mounds ~2 h before wild-type strains. This was independent of parental background, being seen in mutants created from both DH1 and AX2 parental wild-type strains (Harwood et al., 1995 ; Schilde et al., 2004 ). Small mounds also were seen in cells treated with lithium, an inhibitor of GSK3 (Williams et al., 1999 ). This prompted a more in depth investigation of gskA null mutants.
To investigate the role of GskA during aggregation, wild-type and gskA null mutant cells were plated for development and recorded using time-lapse videomicroscopy. As seen previously, gskA null mutant cells formed small mounds ≈2 h earlier than wild-type cells (Supplemental Movie 1). These small mounds formed without the multicellular streams seen in wild-type cells (Figure 1A). Furthermore, the small mounds were unstable, and they often disaggregated before reaggregating in different positions on the substratum (Figure 1B). Restoring GskA activity via expression of the wild-type gskA cDNA in gskA null mutants, rescued aggregation to that seen in the wild type (Figure 1C). However, no phenotypic rescue was seen after expression of a mutant gene that lacks kinase activity (GskAK85R), indicating a requirement for kinase activity to mediate GskA function (Figure 1C).
We tested the ability of gskA null cells to undergo chemotaxis to cAMP. Cells were pulsed with cAMP for 5 h and then placed in a gradient formed from a 1 μM cAMP source. gskA null mutant cells had a complete loss of chemotaxis. Cell chemotaxis can be measured as the chemotactic index (CI), where +1 represents maximally accurate chemotaxis toward the cAMP source, −1 represents chemotaxis away from the cAMP source, and 0 is random movement neither toward nor away from the cAMP source. We also have calculated the percentage of cells that respond positively toward cAMP (% cells respond), and this represents the proportion of cells that have CI values of ≥0.7, with cos−1 (0.7) = 45° and cos−1 (1.0) = 0°. Hence, cells with CI = 1.0, chemotax in a straight line to cAMP, whereas any cell that deviates 45° on either side of this would exhibit a value of 0.7.
Wild-type cells possessed a mean CI of 0.9, with 88.7% of wild-type cells exhibiting chemotaxis toward cAMP compared with 31.5% of gskA mutant cells. This is a slightly higher percentage (25%) than expected from cells when they randomly move. The gskA null mutant cells have a strongly decreased mean CI value of 0.18 and in addition exhibited an approximate 50% decrease in cell speed and directionality, a measure of cell turning as they migrate toward the cAMP source. Re-expression of GskA in the gskA null mutant restores CI, cell speed, and directionality to that of the wild-type cells. These results indicate a strong chemotaxis defect toward cAMP in the gskA null mutant (Figure 2A).
To exclude the possibility that the cells could not sense cAMP, we examined expression of genes involved in chemotaxis and cell signaling. The earlier microarray was carried out in cells that had been starved for 5 h and matched the developmental state of those used in our chemotaxis experiments (Strmecki et al., 2007 ). We reexamined the results of the microarray but failed to detect differences in gene expression of the cAMP receptor genes. carA is the major receptor expressed during aggregation (Sun and Devreotes, 1991 ), and we used qRT-PCR to confirm that there was no difference in carA gene expression between wild-type and gskA null mutant cells (Figure 2B). The microarray showed altered expression of pdsA and pdiA genes, all which play a role in aggregation and we confirmed that induction of pdsA, which encodes a cAMP phosphodiesterase, is indeed substantially reduced in gskA null cells (Figure 2B). Finally, a survey of major genes expressed during aggregation showed that the expression of the adenylyl cyclase acaA gene was induced to only 30% of the wild type. This may have been below the threshold to detect on the microarray (Figure 2B).
To confirm that gskA null cells can sense and respond to cAMP, we examined the response of phosphatase and tensin homologue (PTEN) to cAMP stimulation. In wild-type cells, an accumulation of PTEN protein was seen on the cell membrane but was lost to the cytosol within seconds of cAMP accumulation. This relocalization was still present in gskA null cells (Figure 2C), demonstrating that the loss of gskA did not block cAMP sensing and some downstream events.
To examine chemotaxis in a context independent of cAMP signaling, we investigated the ability of the gskA null mutant to chemotax toward folate. This occurs during growth of Dictyostelium where cells are attracted to folic acid and pterins (Pan et al., 1972 , 1975 ), substances that are secreted by bacteria, the Dictyostelium food source. We recorded the chemotaxis response of growth phase cells to 25 mM folic acid, delivered via a micropipette. We again observed a substantial chemotaxis defect in gskA null cells compared with wild type (Figure 2D). Together, these results indicate a requirement of GskA in Dictyostelium chemotaxis.
To investigate the chemotaxis deficit in gskA null mutant cells during early development, we examined PIP3, a downstream effector of cAMP and mediator of the chemotaxis response. To do this, we monitored phosphorylation of PKBA, the Dictyostelium homologue of PKB. As in its animal homologues, PKBA contains a pleckstrin homo0logy (PH)-domain that mediates translocation to the plasma membrane via PIP3 binding, where the protein is subsequently phosphorylated (Kamimura et al., 2008 ). It therefore is a good monitor for PIP3 synthesis after cAMP stimulation. In wild-type cells, PKBA phosphorylation occurred within seconds of cAMP stimulation, with a peak between 10 and 20 s, and then declined to basal levels by 60 s. No cAMP stimulated elevation of PKBA occurred in gskA null mutant cells (Figure 3A). We can exclude a failure to express the PKBA gene because no difference in expression between mutant and wild-type cells was observed either in the microarray (Strmecki et al., 2007 ) or by qRT-PCR (data not shown).
To pursue this further, we monitored PIP3 synthesis in living cells by expression of the PIP3-specific binding protein PHCRAC-GFP (Parent et al., 1998 ). When cells are stimulated with a global cAMP signal, PIP3 is generated on the entire plasma membrane, causing translocation of PHCRAC-GFP from the cytosol to the membrane. In wild-type cells, PHCRAC-GFP was recruited to the membrane within 10 s of cAMP addition and is subsequently lost after 30 s. No PHCRAC-GFP membrane translocation was observed in the gskA null cells (Figure 3B) and when levels of fluorescence were quantified, PHCRAC-GFP was suppressed to only 10% of that seen in wild-type cells (Figure 3C).
We previously showed that overexpression of the inositol monophosphatase gene IMPA in wild-type cell elevates PIP3 by increased synthesis of the phosphatidylinositol 3-kinase (PI3-kinase) substrate phosphatidylinositol 4,5-bisphosphate (PIP2; King et al., 2009 ). We used IMPA overexpression to elevate PIP3 in gskA null cells and examined its effects. Overexpression of IMPA in gskA null mutant cells not only rescued chemotaxis, restoring mean cell speed, directionality, and CI values of gskA null mutant cells to wild-type control values (Figure 4A), it also restored PKBA phosphorylation (Figure 4B). This not only indicated that GskA plays a major role in PIP3 signaling, it again demonstrated that gskA null mutant cells could sense sufficient cAMP to restore chemotaxis when this downstream effector is restored.
We used IMPA overexpression to probe how loss of gskA affects PIP3 signaling. One possibility is that GskA regulates IMPA activity or other targets on the inositol–PIP2 pathway. A lack of PIP2 would then lead to a decrease in PIP3 synthesis, in a similar mechanism to that seen with lithium treatment (King et al., 2009 ). In contrast, elevation of PIP3 may bypass a block in an alternative pathway, such as activation of PI3-kinase. We measured synthesis of phosphatidyl inositol phosphates (PIPs) in wild-type, gskA null mutant, and gskA null mutant IMPA-overexpressing cells. We observed no difference in phosphatidylinositol (PI), phosphatidylinositol phosphate (PIP), and PIP2 synthesis between wild-type and gskA null mutant cells (Figure 4C). This argued against a requirement for GskA for PIP2 synthesis. In contrast, we saw an increase in PIP2 in gskA null mutant IMPA-overexpressing cells (Figure 4C). This suggested that IMPA overexpression bypasses a deficit in PIP3 signaling by elevating PIP2 and hence PIP3 synthesis.
Finally, we measured PIP3 directly and found that in response to cAMP, gskA null mutant cells did not synthesize PIP3 but seemed to lose it. This response is the opposite to that observed in either wild-type or gskA null mutant IMPA-overexpressing cells (Figure 4D), which rapidly synthesize new PIP3 under these conditions. Overall, the data showed that very little PIP3 is formed in gskA null mutant cells.
We noted that the cAMP chemotaxis phenotype seen in the gskA null mutant was much stronger than that observed for loss of PIP3 signaling alone (Hoeller and Kay, 2007 ; Takeda et al., 2007 ; King et al., 2009 ), and sought evidence for a further molecular defect in the gskA null mutant. Previously, it has been found that signaling via the kinase target of rapamycin (TOR) acting in the TORC2 complex was capable of mediating cAMP chemotaxis when PIP3 synthesis was inhibited (Kamimura et al., 2008 ). This alternative chemotaxis mechanism acts via a second PKB homologue, known as PkgB, or PKBR1 (Meili et al., 2000 ). This kinase protein is associated with the plasma membrane via myristoylation and not via a PH domain–PIP3 interaction (Meili et al., 1999 ) and is activated by phosphorylation from TORC2 (Lee et al., 2005 ). We examined PKBR1 phosphorylation in gskA null mutant cells after cAMP stimulation, and found that as seen for PKBA, PKBR1 phosphorylation was also lost (Figure 4B). We could detect PKBR1 expression (data not shown), suggesting a signaling failure upstream of PKBR1. In contrast to PKBA phosphorylation, PKBR1 phosphorylation was not restored by overexpression of IMPA (Figure 4B), indicating that this second molecular mechanism is independent of PIP3 signaling. These results indicated that GskA is required for two intracellular signaling processes, explaining why it possesses a very strong chemotaxis defect.
Although it restored chemotaxis, IMPA overexpression did not rescue aggregation or any other aspect of morphogenesis (Figure 5A and Supplemental Movie 2). This indicated that the aggregation defect was not a simple consequence of loss of chemotaxis, and may arise from an additional signaling deficiency. To test whether these cells can generate cAMP signals, cells were developed for 5 h in suspension and stimulated with the cAMP analogue 2-deoxyadenosine 3,5-monophosphate (dcAMP), and cAMP synthesis was measured. There was almost undetectable cAMP synthesis in gskA null- and gskA null-expressing IMPA cells (Figure 5B).
Dark-field optics was used to examine the cell signaling activity during aggregation. When viewed under these conditions, changes in cell shape in response to cAMP pulses can be seen as a change in light scattering (Tomchik and Devreotes, 1981 ). In wild-type cells, waves of light scattering propagate from the center of the aggregation field pass through the developing cell field as spiral waves (Figure 5C and Supplemental Movie 3A). In contrast to wild-type cells, no light scattering waves were observed in gskA null mutant cells or gskA mutants overexpressing IMPA (Figure 5C and Supplemental Movie 3B). These observations suggested that although PIP3 signaling is sufficient to rescue chemotaxis, it was unable to restore cAMP signaling and fully rescue aggregation.
Here, we report that loss of GskA has substantial effects on Dictyostelium aggregation, causing a major defect in chemotaxis and cAMP signaling. This phenotype correlates with the presence of GskA activity during early stages of development and the results of previous microarray and proteomic analysis (Plyte et al., 1999 ; Strmecki et al., 2007 ). These results reveal an unexpectedly strong chemotaxis phenotype, which has been previously missed, and indicates that the terminal developmental phenotype of Dictyostelium mutants may not be a good indicator of defective chemotaxis.
Loss of gskA has a substantial effect on chemotaxis toward both cAMP and folate. These two chemotactic responses are mediated by different receptors and G proteins (Kumagai et al., 1989 ; Hadwiger et al., 1994 ; Kim et al., 1998 ). Together with the expression of the cAMP receptor carA and the presence of the PTEN response, these observations argue that the requirement for GskA lies at the level of effector activation rather than with signal sensing. We do not know exactly how loss of GskA inhibits folate chemotaxis, and our further conclusions are based on our investigation of the better characterized cAMP chemotaxis response.
We found that loss of gskA caused a major decrease in PIP3 signaling, which could be bypassed by elevation of PIP3 after overexpression of IMPA. Previous observations have shown that PIP3 signaling is able to elicit the chemotaxis response (Kortholt et al., 2007 ), explaining how this restores chemotaxis to gskA null mutant cells. Although PIP3 signaling is sufficient for chemotaxis, it is not essential and a simple loss of PIP3 cannot totally explain the severe chemotaxis phenotype observed for the gskA null mutant cells. Loss or inhibition of PI3K or PIP3 synthesis reduces speed and directionality in chemotaxing cells; however, it has only a minor effect on the CI value (Hoeller and Kay, 2007 ; King et al., 2009 ). In contrast, although we also saw reduced speed and directionality in the gskA null mutant, we also measured a large drop in the CI value. We found that loss of gskA blocks phosphorylation of PKBR1, a PIP3-independent component of a parallel signaling pathway that also mediates chemotaxis to cAMP (Kolsch et al., 2008 ). We conclude that GskA acts upstream of both PIP3- and PKBR1-mediated signaling (Figure 6).
In our experiments, we monitored PKBA and PKBR1 activation through antibodies specific to motifs phosphorylated by the protein kinase PDK-1 (Kamimura and Devreotes, 2010 ). This in turn is dependent on phosphorylation by TOR kinase as part of the TORC2 complex after cAMP stimulation. These events occur at the plasma membrane so that PKBR1 is dependent solely on TORC2 activation, whereas PKBA phosphorylation requires simultaneous TORC2 activation and PIP3 synthesis. This presents a paradox: if TORC2 activation is required for both PKBA and PKBR1, how can overexpression of IMPA restore PKBA phosphorylation without an effect on PKBR1? We noted however that a small amount of PKBA phosphorylation occurs in the absence of TORC2. This is evident in the report from (Kamimura et al., 2008 ; Liao et al., 2010 ) where PKBA phosphorylation is observed in pia mutant cells, which lacks the Dictyostelium homologue of TORC2 component Rictor. We also observed PKBA phosphorylation in a rip3 null mutant, lacking a second component of TORC2 (Supplemental Figure 1, top), but an absence of PKBR1 phosphorylation. We propose that elevation of PIP3, as a consequence of IMPA overexpression, brings more PKBA to the membrane where it is phosphorylated through a TORC2-independent kinase. The identity of this kinase is not known, but a possible candidate could be the PDK1 homologue PdkA (Kamimura and Devreotes, 2010 , Liao et al., 2010 ). Although both PdkA and its paralogue can function in the cytosol, PdkA can translocate to the plasma membrane in response to cAMP through binding to PIP3 (Kamimura and Devreotes, 2010 ), and so IMPA overexpression could enhance PKBA phosphorylation by increasing the concentration of both PKBA and PdkA on the membrane. Furthermore, PKBR1 null cells undergo normal chemotaxis (Meili et al., 2000 ), indicating that PKBA phosphorylation alone is sufficient for chemotaxis.
Further apparent complexity relates to earlier results concerning the effects of lithium on Dictyostelium chemotaxis. Lithium inhibits both GSK3 and IMPA (Van Lookeren Campagne et al., 1988 ; Van Dijken et al., 1996 ; Ryves et al., 1998 ; Ryves and Harwood, 2001 ) and therefore would be expected to produce the strong chemotaxis effect seen in the gskA null mutant. However, previously we found that lithium treatment of aggregation competent wild-type cells suppressed PIP3 signaling, but left PKBR1 phosphorylation and cAMP synthesis unaffected (King et al., 2009 ). We demonstrated that this result arose from inhibition of IMPA, leading to a decrease in PIP2 and subsequent PIP3 synthesis. Arguing against the possibility that GskA becomes insensitive to lithium, we found that nuclear export of Dd-STATa (Ginger et al., 2000 ), which is mediated by GskA phosphorylation, is suppressed by lithium in aggregation competent cells (Supplemental Figure 1, bottom left). We therefore propose that these differences arise through the timing of GskA action. Acute inhibition of GskA in aggregation competent cells, practically defined as cells pulsed with cAMP for 4 h before addition of lithium for 1 h, has no effect on chemotaxis or cAMP synthesis. In contrast, loss of GskA activity at the beginning of development or in growth phase cells causes the block to chemotaxis and signaling. In support of this hypothesis, we found that long-term treatment with lithium present from the initiation of cell development caused a substantial suppression of cAMP synthesis (Supplemental Figure 1, bottom right). This suggests that GskA activity acts as a permissive signal to initiate or set up chemotaxis but is not required continuously either as a direct requirement within the chemotaxis response or for maintenance of a chemotactic competent state. How GskA may achieve this is unclear; however, possibilities could range from constitutive phosphorylation of a protein that acts upstream of PIP3 and TORC2 signaling to regulation of gene expression of a component, or components, required for chemotaxis and signaling.
What is clear is that GskA is required in the regulation of some aggregation-specific gene expression. Here, we have shown decreased expression in both pdsA, a cAMP phosphodiesterase, and acaA, the adenylyl cyclase required for generation of cAMP during aggregation. Loss of expression of these genes could account for loss of cAMP signaling during aggregation, but not the GskA requirement for chemotaxis. The earlier proteomic and microarray analysis of gskA null mutant cells (Strmecki et al., 2007 ) indicates several genes that may contribute to the chemotaxis phenotype. Loss of gskA reduces expression of sodC, a superoxide dismutase, which has been shown to be required for chemotaxis toward cAMP (Veeranki et al., 2008 ). The sodC mutant causes an increase in RasG activity even in the absence of cAMP stimulation, a known upstream activator of PI3-kinase (Sasaki et al., 2004 ), and could contribute to the higher basal levels of PKBA phosphorylation before cAMP stimulation (Figure 3). The microarray also showed that expression of the gene alrA is reduced in gskA null mutant cells. The AlrA, protein encodes an aldehyde reductase, which when disrupted also suppresses cell motility, although the mechanism is unclear (Ehrenman et al., 2004 ). Curiously, both GskA and AlrA have the capacity to alter glucose metabolism within the cell, but the significance of this is not known. In contrast, loss of gskA elevates expression of ampA. The product of this gene is antiadhesion protein that accumulates during aggregation (Varney et al., 2002 ), and elevated levels of AmpA could explain the instability of the small mounds observed during the aggregation of the gskA null.
Together, this study indicates that GskA is an important regulator of the aggregation process and is essential for cell chemotaxis. Our observations however also indicate that the role of GskA is probably complex, acting through a combination of mechanisms that are required at a minimum for chemotaxis and signaling but that also may include metabolism and cell adhesion. These may integrate to during early development to allow aggregation proceed to the multicellular stage, in which GskA mediates cell fate determination.
We thank Peter D. Watson for invaluable help in creating the dark-field movies and advice related to microscopy. We also thank Gerry Weeks and Parvin Boulourani for advice regarding the folate experiments and Tsuyoshi Araki and Jeff Williams for the Dictyostelium monoclonal antibodies against STATa (D4). This work was funded by a Wellcome Trust Programme grant to A.J.H.
This article was published online ahead of print in MBoC in Press (http://www.molbiolcell.org/cgi/doi/10.1091/mbc.E09-10-0891) on June 9, 2010.