3099Research Article

IntroductionMotile eukaryotic cells are equipped with machineries that interpretchemoattractant gradients and regulating cell polarization andmotility. The molecular nature of these machineries remains to befully understood. One view suggests that chemotaxis is regulatedby a hypothetical molecular compass that functions upstream ofPI3K and phospholipase A2 (PLA2), which could complement eachother to form surprisingly robust molecular circuits that ensureproper chemotaxis under diverse situations (Chen et al., 2007; vanHaastert et al., 2007). Several studies also showed that PI3K functionis less crucial for efficient chemotaxis under a strong (over 1 μMcAMP) chemoattractant gradient, but is necessary under a weak(below 0.1 μM cAMP) gradient (Loovers et al., 2006; Takeda etal., 2007; Hoeller and Kay, 2007). Inactivation of either the PI3Kor PLA2 pathways significantly perturbs chemotaxis under a weakgradient but not under a strong gradient (van Hastert et al., 2007).When both pathways are compromised, chemotaxis is severelycompromised, even under a strong gradient (van Hastert et al., 2007).Another view, by contrast, suggests that the regulation of pseudopodformation is central in gradient sensing, where PtdIns(3,4,5)P3 isnot the major determinant of the polarization or of the directionalpseudopod formation, but rather controls the frequency ofpseudopod formation (Andrew and Insall, 2007). It is not easy tocompare these different views, considering that each group adoptedvery different chemotaxis assays for their studies, so it seems thatmore investigation into the mechanisms of chemotaxis is necessaryin order to fully understand its nature.

Phosphatidylinositol (3,4,5)-trisphosphate [PtdIns(3,4,5)P3] isstill an important part of the chemotaxis machinery in somesituations, but it is not the molecular compass for gradient sensing,as was suggested initially (Loovers et al., 2006; Hoeller and Kay,

2007; Takeda et al., 2007). Generation of PtdIns(3,4,5)P3 dependsheavily on the regulation of PI3Ks. In Dictyostelium, two PI3Ks,PI3K1 and PI3K2, are mostly responsible for chemoattractantcAMP-dependent PtdIns(3,4,5)P3 generation (Zhou et al., 1995;Huang et al., 2003). Both kinases can localize either at the cytosolor at the plasma membrane in response to chemoattractant signals.The domains for membrane localization and kinase activation havebeen clearly defined by serial deletion analysis. The membrane-targeting domain precedes the central Ras-binding domain, whichmediates activation of the kinase through Ras (Funamoto et al.,2002). Previous studies also established that activation of PI3Kthrough small GTPase Ras facilitates PI3K membrane localizationof PI3K, and thus forms a positive-feedback loop in both achemoattractant signal-dependent and -independent manner (Sasakiet al., 2004; Sasaki et al., 2007). The latter also explains how theRas-PI3K loop is involved in the modulation of basic cell motility(Sasaki et al., 2007).

To identify additional regulators of PtdIns(3,4,5)P3, weperformed restriction enzyme mediated insertion (REMI)mutagenesis on cells expressing GFP-PHcrac protein [a pleckstrinhomology (PH) domain from the cytosolic regulator of adenylylcyclase (CRAC) protein fused to the green fluorescent protein(GFP)]. Mutants that exhibit higher levels of GFP-PHcrac at theplasma membrane in the absence of chemoattractant stimulationwere isolated by monitoring enhanced plasma membranelocalization of GFP-PHcrac proteins. Four of the REMI mutantsexhibited insertions at the same locus, which encodes a GPI-anchored superoxide dismutase (SOD) protein, SodC. A previousstudy using chemical superoxide scavengers suggested that propercontrol of superoxide radicals plays an important role inDictyostelium cell aggregation (Bloomfield and Pears, 2003). The

A genetic screen for Dictyostelium mutant displaying high levelof constitutive phosphatidylinositol (3,4,5)-trisphosphate led tothe finding that the glycosylphosphatidylinositol (GPI)-anchored superoxide dismutase SodC regulates small GTPaseRas. Cells that lack SodC exhibited constitutively high levels ofactive Ras, more membrane localization of GFP-PHcrac, anddefects in chemoattractant sensing, cell polarization andmotility. These defects of SodC-lacking cells were partiallyrestored by expression of wild-type SodC but not by thecatalytically inactive mutant SodC (H245R, H247Q).Furthermore, an inhibition of PI3K activity in SodC-deficientcells by LY294002 only partially restored chemoattractant

sensing and cell polarization, consistent with the fact that SodC-deficient cells have aberrantly high level of active Ras, whichfunctions upstream of PI3K. A higher level of active GFP-RasGwas observed in SodC-deficient cells, which significantlydecreased upon incubation of SodC-deficient cells with thesuperoxide scavenger XTT. Having constitutively high levels ofactive Ras proteins and more membrane localization of GFP-PHcrac, SodC-deficient cells exhibited severe defects inchemoattractant sensing, cell polarization and motility.

Key words: SOD, Ras, PI3K, PtdIns(3,4,5)P3, Chemotaxis, Motility

Summary

The GPI-anchored superoxide dismutase SodC isessential for regulating basal Ras activity and forchemotaxis of Dictyostelium discoideumSudhakar Veeranki, Bohye Kim and Leung Kim* Department of Biological Sciences, Florida International University, Miami, FL 33199, USA*Author for correspondence (e-mail: kiml@fiu.edu)

Accepted 1 July 2008Journal of Cell Science 121, 3099-3108 Published by The Company of Biologists 2008doi:10.1242/jcs.030056

Jour

nal o

f Cel

l Sci

ence

3100

mechanism that underlies radical generation and dismutation, andthe cellular targets that are involved in Dictyostelium aggregationhave yet to be identified. In this report, we will present thecharacterization of SodC and the phenotypes of sodC– cells, andwill discuss potential roles of SodC in chemotaxis through theregulation of Ras activity and of PI3K.

ResultsGeneration of REMI mutants that exhibit higher constitutivelevels of PtdIns(3,4,5)P3To find additional components involved in the regulation ofPtdIns(3,4,5)P3 levels, wild-type cells expressing GFP-PHcracwere randomly mutagenized by the terminator trapping REMImethod (Takeda et al., 2003). GFP-PHcrac has previously beenshown to bind PtdIns(3,4,5)P3 and has been used as a markerfor PtdIns(3,4,5)P3 localization (Huang et al., 2003). Out of~1000 independently isolated insertional mutants, sixindependent remi clones exhibited elevated level of GFP-PHcracat the plasma membrane in the absence of chemoattractantstimulation (Fig. 1A). By contrast, GFP-PHcrac proteins werelocalized uniformly through the cytoplasm and occasionallyenriched at the local membrane ruffles and macropinosomes inwild-type cells. It is expected that the mutants displaying higherPtdIns(3,4,5)P3 levels would be severely defective in chemotaxis.One such example is pten cells, which show severe defect inchemotaxis (Funamoto et al., 2002; Iijima and Devreotes, 2002).remi56 cells, as a representative of REMI mutants, failed to formterritorial streams when tested by under-buffer aggregationassay (Fig. 1B).

Identification and characterization of SodCInsertion loci were identified as described in the Materials andMethods. Remi56, as well as three other mutants had insertionsat the same locus encoding an extracellular superoxide dismutase(SOD) previously identified as SodC (Tsuji et al., 2003). SodCmessage is expressed both in vegetative and aggregative stages,and gradually declined as cells terminally differentiate (VanDriessche et al., 2002; Iranfar et al., 2003). Sequence analysis ofSodC revealed a signal peptide as previously suggested (Tsuji etal., 2003), and a previously undescribed C-terminalglycosylphosphatidylinositol (GPI)-anchoring motif (Fig. 2A).Interestingly, SodC encodes an incomplete SOD domain near theN-terminal end, in addition to a complete SOD near the Cterminus. Three independent sodC– cells were generated byhomologous recombination and confirmed by genomic PCRanalysis (Fig. 2B). sodC– cells were transfected with a GFP-PHcrac expression construct to assess indirectly the level ofPtdIns(3,4,5)P3. sodC– cells exhibited significantly more GFP-PHcrac protein at the plasma membrane than did wild-type cells,similar to remi56 (Fig. 2C). Aberrant plasma membranelocalization of the GFP-PHcrac proteins in sodC– cells was furtherconfirmed by subcellular fractionation. The membrane andcytosolic fractions were prepared from wild-type and sodC– cellsusing membrane filters (Parent et al., 1998), and analyzed bywestern blotting using an anti-GFP antibody. Consistent with theGFP images, significantly more GFP-PHcrac protein was detectedin the membrane fraction of the sodC– cells (Fig. 2D). As a control,the same blot was re-probed with a pan-Ras antibody. As expected,most Ras proteins were detected from membrane fractions. Thesedata suggest that both remi56 and sodC – cells might haveconstitutively higher levels of PtdIns(3,4,5)P3.

Journal of Cell Science 121 (18)

Fig. 1. Generation of REMI mutants exhibiting higher basal levels of GFP-PHcrac at the plasma membrane. (A) remi56 cells displayed more plasmamembrane localization of GFP-PHcrac, which suggests higherPtdIns(3,4,5)P3 levels in the membrane of remi56. (B) Wild type and remi56lines were tested to see whether they are able to form developmentalterritorial streams. remi56 lines failed to aggregate in submerged culture atthe cell density of 2.5�10 cells/cm, whereas wild type cells aggregatednormally.

Fig. 2. Identification of the insertional mutation in the REMI mutants andgeneration of sodC – cells. (A) remi56 displayed an insertion at a locusencoding a superoxide dismutase (SOD) domain-containing protein, SodC.SodC encodes an N-terminal signal peptide with the C-terminal GPI-anchoring sequence (omega domain), and a partial and a full Cu/Zn-superoxide dismutase (SOD) domain. (B) sodC – cells were generated byhomologous recombination from wild-type cells, and confirmed by genomicPCR. (C) sodC – cells showed increased basal membrane localization ofGFP-PHcrac compared with wild-type cells. (D) Western blot analysis ofthe cytosolic and membrane fractions of the wild-type and sodC – cellsshowed more GFP-PHcrac localization at the membrane in sodC than inwild-type cells. Ras proteins are shown as a loading control for membranefractions.

Jour

nal o

f Cel

l Sci

ence

3101Regulation of Dictyostelium chemotaxis by superoxide dismutase

SodC encodes superoxide dismutase with a GPI anchorThe Cu/Zn-SOD domain of SodC was expressed as a GFP fusionprotein under the Actin-15 promoter and detected by anti-GFPantibody (Fig. 3A). When incubated with superoxide radicals, theGFP-SOD domain inhibited superoxide-dependent Formazanformation, whereas the same amount of GFP control showed littleinhibition (Fig. 3B).

To confirm that SodC is actually a GPI-anchored membraneprotein, a Myc-tag was inserted after the signal peptide of the full-length SodC. Myc-SodC proteins were immunopurified and detectedby anti-Myc antibody (Fig. 4A, bottom left). Membrane localizationof Myc-SodC was evident from both western blotting of subcellularfractions (Fig. 4A, bottom right) and indirect immunofluorescencemicroscopy (Fig. 4B). To further test whether there is a GPI anchoredSOD activity, wild-type cells were treated with GPI specificphosphatidylinositol-phospholipase C (PI-PLC) (Kondoh et al.,2005). The media from PI-PLC treated wild-type cells displayed threetimes higher SOD activity than that of the control (Fig. 4C). Bycontrast, the media from sodC– cells displayed no measurableamounts of SOD activity under the same experimental conditions.

Being localized on the outer leaflet of the plasma membranethrough a GPI anchor, SodC may control extracellular superoxidelevel and/or penetration of the radical across the plasma membrane,and thus intracellular superoxide level. When tested by thepreviously published assay method using superoxide-sensitivereagent XTT (Bloomfield and Pears, 2003), wild-type and sodC–

cells displayed no significant difference in the level of extracellularsuperoxide level, as shown in Fig. 4D.

Next, the intracellular superoxide levels were measured usinganother superoxide sensitive reagent: NBT. Contrary to XTT, NBTbecomes an insoluble precipitate upon reduction by superoxide(Choi et al., 2006). The amount of insoluble NBT trapped insidethe sodC– cells was consistently higher (by ~18%) than that in thewild type (Fig. 4E). These data suggest that SodC is a GPI-anchoredsuperoxide dismutase that is involved in the regulation ofintracellular level of superoxide in Dictyostelium cells.

sodC– cells are defective in chemotaxis but not indevelopmentLike remi56 cells, sodC – cells were severely defective inaggregation (Fig. 5A). However, when plated at high cell densities

where chemotaxis is less essential, sodC – cells developedindistinguishably from wild-type cells (Fig. 5B). These data stronglysuggest that SodC is essential for chemotaxis but not fordevelopment.

sodC – cells, after being pulsed for 4 hours, displayed asignificantly compromised chemoattractant sensing during the first20 minutes under both weak and strong cAMP gradients (Fig. 6B,C).However, during the second 20 minutes under both weak and strong

Fig. 3. SodC has SOD activity. (A) The SodC SOD domain was expressed inDictyostelium as a GFP fusion protein under the Actin 15 promoter. GFP andGFP-SOD proteins were purified by immunoprecipitation with anti-GFPantibody, and normalized by western blot analysis using anti-GFP antibody.(B) Relative superoxide levels were compared after incubation with equalamounts of purified GFP or GFP-SOD protein. The relative superoxide levelfrom GFP was set as 1.0. GFP-SOD samples exhibited an average level of 0.3(standard deviation of 0.06) from three independent experiments.

Fig. 4. SodC is a GPI-anchored membrane protein. (A) A Myc epitope tag wasinserted in-frame after the SodC signal peptide. Myc-SodC was expressed inthe wild-type background, and detected by western blotting (bottom left).Membrane and cytosolic fractions were prepared, and Myc-SodC was detectedonly from the membrane fraction (bottom right). (B) Membrane localization ofMyc-SodC was also confirmed by indirect immunofluorescence. (C) Logphase wild-type and sodC– cells were plated at a density of 1�10 cells/cm,and treated with or without 1 unit of GPI-specific PI-PLC for 5 minutes at25°C. Extracellular SOD activities were compared. (D) The relative levels ofextracellular superoxide in wild-type and sodC– cells were measured usingXTT reduction, as described in the Materials and Methods. Virtually identicallevels of the radical were detected from three independent experiments.(E) The relative intracellular superoxide levels were determined using NBT.Cell-trapped NBT levels were ~18±5% (s.d.) higher in sodC than in wild-typecells (data are from three independent experiments).

Jour

nal o

f Cel

l Sci

ence

3102

cAMP gradients, sodC– cells showed an improvement in cAMPsensing, but only up to ~40 % of the wild-type level. In addition,sodC– cells showed severe motility defects, which were worse underweak cAMP gradient (Table 1). Contrary to the chemotaxis index,the speed of motility did not improve during the whole duration ofthe assay.

sodC – cells displayed less polarization during the first 20minutes under both a weak and strong gradient than did wild-typecells. An improvement in the polarity from 0.6 to 0.3 was observedfrom sodC– cells under a strong gradient during the last 20 minutes,whereas no such improvement was made from sodC– cells undera weak gradient (Fig. 6D). sodC– cells displayed multiple problemsin chemotaxis that cannot easily be overcome by a higherconcentration of or a longer exposure to chemoattractant cAMP.

Re-introduction of wild-type SodC, not the inactive mutantSodC, partially attenuated the chemotaxis defects of sodC–

cellsA previous study showed that substitution of two histidine residuesin the Cu2+-binding motif of the SOD domain with arginine andglutamate led to an effective loss of SOD activity (Wang et al.,2002). Similarly, the catalytically inactive SodC (H245R,H247Q)was generated and expressed in sodC– cells. Levels of SodCtranscripts were compared in wild type, in sodC– cells and in sodC–

cells expressing wild-type SodC or SodC(H245R,H247Q) by RT-PCR (as shown in Fig. 7A). A clear SodC message was detectedfrom wild-type cells, but not from sodC – cells. SodC cellsexpressing wild-type SodC or SodC(H245R,H247Q) displayed acomparable level of SodC message, which is higher than that ofthe endogenous SodC messages.

sodC– cells with this level of re-introduced SodC messages werechallenged with a micropipette filled with 0.1 μM cAMP for 20minutes. The chemotaxis index of sodC– cells expressing full-lengthSodC was near ~90% of the wild-type level, and their speedimproved to ~40% of the wild-type level. By contrast, thechemotaxis index of sodC– cells expressing SodC(H245R, H247Q)was similar to that of the parental sodC– cells (Fig. 7). Althoughthe complementation was less than complete, there was a clear

improvement in chemotaxis and cell polarization by wild-type SodCexpression but not by the mutant SodC in sodC– cells (Table 2;Fig. 7C).

sodC– cells pretreated with PI3K inhibitor LY294002 exhibitedimproved chemotaxisTo determine whether sodC– cells are defective in chemotaxismainly because of the presence of an excessive PtdIns(3,4,5)P3,cells expressing PtdIns(3,4,5)P3 marker GFP-PHcrac were pulsed

Journal of Cell Science 121 (18)

Fig. 5. sodC– cells were defective in aggregation, but not in development.(A) sodC– cells failed to aggregate in submerged culture at low cell density(2.5�10 cells/cm), where wild-type cells aggregate normally. (B) sodC– cellsdisplayed normal development when plated at high cell densities (1.25�10cells/cm) at which chemotaxis can be bypassed.

Fig. 6. sodC– cells were defective in chemotaxis. (A-C) Cells were challengedwith a point source of either 0.1 μM or 10 μM cAMP. Tracing images ofchemotaxing cells were arranged to demonstrate relative directionalmovement, cell shape and distances traveled towards the cAMP point source(circle). Superimposed tracing images were grouped as early (0 to 20 minutes)and late (21-40 minutes) duration as marked. Each tracing image is a 1-minuteinterval. (D) The roundness of chemotaxing cells is summarized as defined inthe Materials and Methods.

Jour

nal o

f Cel

l Sci

ence

3103Regulation of Dictyostelium chemotaxis by superoxide dismutase

for 4 hours with 50 nM cAMP, and either left in DB buffer or treatedwith 15 μM or 50 μM of PI3K inhibitor LY294002 (LY) for 20minutes. sodC– cells expressing GFP-PHcrac showed aberrantplasma membrane localization, which decreased significantly at the

plasma membrane after treatment with 15 μM or 50 μM LY294002(Fig. 8A). However, LY294002-treated sodC– cells still displayedPtdIns(3,4,5)P3 enriched macropinosomes and local ruffles,indicating that PtdIns(3,4,5)P3 levels were attenuated but notcompletely depleted (Fig. 8A).

Then, wild-type and sodC– cells were incubated with 15 μMLY294002 for 20 minutes, and challenged with micropipette filledwith 10 μM cAMP for 20 minutes. Wild-type cells showedcomparable chemotaxis indices and speeds irrespective of 15 μMLY294002 treatment (Fig. 8). Untreated sodC – cells almostcompletely failed to respond to a micropipette filled with 10 μMcAMP (Fig. 8). By contrast, sodC– cells pretreated with LY294002displayed a near wild-type level of gradient sensing, significantlyimproved speed of motility and cell polarization (Table 3; Fig. 8C).By contrast, the restoration of the speed of locomotion by LYtreatment was lesser (Table 3).

sodC– cells displayed aberrant PI3K regulationRegulation of PI3K1 and PI3K2, the two major enzymes largelyresponsible for chemoattractant-induced PtdIns(3,4,5)P3 generationin Dictyostelium, is complex. Through a yet to be identifiedmechanism, cells control membrane and cytoplasmic localizationof PI3K (Funamoto et al., 2003). In addition, activation of Rasproteins leads to an enhanced PI3K activity. Subsequently, morePtdIns(3,4,5)P3 will accumulate, which in turn stimulate membranelocalization of PI3K through a F-Actin-dependent mechanism(Sasaki et al., 2004). In the absence of chemoattractant stimulation,wild-type cells maintain a minor, but detectible, fraction of PI3Kat the plasma membrane (Han et al., 2006; Sasaki et al., 2004).

sodC– cells displayed N-PI3K1-GFP proteins localized uniformlythroughout the plasma membrane, whereas wild-type cells displayeda strong enrichment at the leading front (Fig. 9A). This was furtherconfirmed by western blot analysis on the membranous andcytosolic fractions prepared from cells expressing equivalent amountof N-PI3K1-GFP proteins (Fig. 9B). Localization of the other majorregulator of PtdIns(3,4,5)P3, PTEN, was also examined. Nosignificant difference in the localization of GFP-fused PTENproteins was observed from wild-type and sodC– cells (Fig. 9A).The aberrancy in the localization of N-PI3K1-GFP in sodC– cellswas further highlighted when examined upon global stimulationwith cAMP. Instead of a transient membrane localization observedin wild type, N-PI3K1-GFP showed membrane localization beforethe stimulation, which almost failed to change in response to globalcAMP stimulation (Fig. 9C). sodC– cells consistently displayed lesspolarized cell shapes.

To examine the F-Actin organization in camp-pulsed sodC– cells,cells were stained with TRITC-phalloidin (Fig. 9D). Althoughpolarized wild-type cells showed a dominant leading front enrichedwith F-Actin, sodC– cells exhibited numerous filopodial extensionsand membrane ruffles around their more round cell bodies. Whenstimulated with 10 μM cAMP, wild-type cells displayed a twofoldincrease in F-Actin level after 5 seconds of stimulation, which

Table 1. Summary of chemotactic indices andmotilities of sodC– cells

cAMP Response 0-20 minutes 21-40 minutes

sodC– 0.1 μM Chemotaxis index 0.16±0.36 0.31±0.34Speed (μm/minute) 1.6±0.3 2.2±1.2

10 μM Chemotaxis index 0.00±0.34 0.29±0.36Speed (μm/minute) 5.2±1.3 5.7±2.5

Wt 0.1 μM Chemotaxis index 0.82±0.10Speed (μm/minute) 9.2±2.6

10 μM Chemotaxis index 0.79±0.12Speed (μm/minute) 9.2±4.7

Mean values with standard deviations from three independent experimentsare shown.

Table 2. Summary of chemotactic indices and motilities ofsodC– cells expressing either wild-type SodC or the

SodC(H245R,H247Q) double-point mutant

SodC(H245R,0.1 μM cAMP SodC::sodC– H247Q)::sodC–

Chemotaxis index (mean±s.d.) 0.66±0.31 0.29±0.27Speed (μm/minute) (mean±s.d.) 3.5±1.4 1.0±0.2

Fig. 7. Defects in sodC– cells were partially rescued by SodC but not by aSodC(H245R,H247Q) double-point mutation. (A) RT-PCR experiment with aprimer set specific for SodC was used to detect the level of SodC transcript. Aspecific RT-PCR product was obtained from wild type, but not from sodC–

cells as expected. Ig7 transcripts were used as a control. Similar levels of thetranscripts were observed in sodC– cells expressing wild-type SodC and theSodC(H245R,H247Q) mutant under Actin-15 promoter. (B) sodC– cellsexpressing wild-type SodC and SodC(H245R,H247Q) mutant were challengedwith a micropipette filled with 0.1 μM cAMP for 20 minutes. Twenty stacks oftracing images are shown with a 100 μm scale bar. Summary of the roundness(C) is shown.

Jour

nal o

f Cel

l Sci

ence

3104

rapidly decreased close to the basal level. By contrast, sodC– cellsexhibited ~20% higher basal F-Actin level compared with that ofthe wild type, but showed no further increase in response to cAMPstimulation (Fig. 9E).

SodC regulates Ras, an upstream regulator of PI3KRas is an upstream regulator of PI3K, and subject to regulation bysuperoxide in vitro (Sasaki et al., 2004; Cox and Der, 2003; Heoand Campbell, 2005). We tested whether Ras is aberrantly regulatedin sodC– cells by using GFP-RBD (Ras-Binding Domain) protein,which was previously used as an active Ras marker in Dictyosteliumcells (Sasaki et al., 2004). Active Ras proteins in cells were indirectlyvisualized by monitoring GFP-RBD proteins after 4 hours ofpulsing. As shown in Fig. 10A, wild-type cells almost alwaysshowed polarized GFP-RBD localization after 4 hours of cAMPpulsing. Cells with sodC background, by contrast, displayedstrikingly disorganized pattern of GFP-RBD protein (Fig. 10A).Furthermore, more active Ras proteins were detected from sodC–

cells than wild-type cells by GST-RBD pull-down assay (Fig. 10B)(Sasaki et al., 2004). In addition, when stimulated globally with 10μM cAMP, sodC– cells displayed higher basal Ras proteins thanwild-type cells with no further increase in the active Ras level. Bycontrast, wild-type cells displayed the transient Ras activation asexpected (Fig. 10C).

To further determine whether the loss of SodC directly affectRas activity or indirectly through regulating superoxide level, cellswere stimulated with conditioned medium with or without asuperoxide scavenger XTT. A previous study has shown thatconditioned medium prepared by cAMP pulsing stimulates cellularsuperoxide generation (Bloomfield and Pears, 2003). sodC– cellsdisplayed a higher basal level of active Ras (as expected), butshowed a significant decrease in active Ras levels upon depletionof superoxide by incubation with radical scavenger XTT (4 mM)for 10 minutes at room temperature (Fig. 10D). A much lower levelof active endogenous Ras was observed from wild-type cells, whichwas also susceptible to XTT treatment (Fig. 10D).

Next, the identity of aberrantly activated Ras species in sodC–

cells was determined. A previous study identified that RasG is oneof the major Ras proteins that regulates chemotaxis and is capableof binding to the human Raf1-RBD when activated (Sasaki et al.,2004). Consistent with this, basal activity of GFP-RasG wasidentified to be higher in sodC– cells than in wild-type cells (Fig.10E). In addition, stimulation of wild-type cells with the CM showeda modest increase in the level of active GFP-RasG, which wassusceptible to XTT treatment (Fig. 10F). By contrast, sodC– cellsdisplayed higher basal level of active GFP-RasG, which was notresponsive to the stimulation with the CM but was susceptible toXTT treatment (Fig. 10F).

DiscussionThe majority of superoxide radicals are generated either during theoxidative respiration in the mitochondria, or as products of the

Journal of Cell Science 121 (18)

Fig. 8. LY294002 treatment attenuated chemotaxis defects of sodC– cells.(A) Cells expressing the PtdIns(3,4,5)P3 marker GFP-PHcrac were pulsed for4 hours with 50 nM cAMP, and either left in DB buffer or treated with 15 μMor 50 μM LY294002 (LY) for 20 minutes. GFP-PHcrac aberrantly localized atthe plasma membrane of sodC– cells but not of wild type. Membranelocalization of GFP-PHcrac in sodC– cells largely disappeared after LYtreatment. GFP signals at the membrane of sodC– cells were reminiscent offine filopodial extensions, which also disappeared after LY treatment.(B) Wild-type and sodC– cells were pulsed for 4 hours with 50 nM cAMP, andtreated with and without 15 μM LY294002 (LY) for 20 minutes. Cells werethen challenged with micropipettes filled with 10 μM cAMP for 20 minutes.Twenty stacks of cell tracing images 1 minute apart, are shown with a 100 μmscale bar. (C) Roundness values of the analyzed cells are shown.

Table 3. Summary of chemotactic indices and motilities ofwild-type and sodC– cells treated with LY294002

Response Wild type sodC–

–LY Chemotaxis index (mean±s.d.) 0.79±0.12 0.00±0.33–LY Speed (μm/minute) (mean±s.d.) 9.2±4.7 5.4±1.3+15 μM LY Chemotaxis index (mean±s.d.) 0.70±0.19 0.80±0.17+15 μM LY Speed (μm/minute) (mean±s.d.) 9.8±1.6 6.0±3.1

Jour

nal o

f Cel

l Sci

ence

3105Regulation of Dictyostelium chemotaxis by superoxide dismutase

NADPH oxidase complex at the plasma membrane (Hancock et al.,2001; Droge, 2002). Currently, the mechanism of superoxidegeneration in Dictyostelium is not clear, but a previous report hasshown that Dictyostelium cells produce superoxide radicals in acAMP-independent mechanism (Bloomfield and Pears, 2004). Wefound that sodC– cells maintained extracellular superoxide, similarto wild-type cells, but the total intracellular superoxide level wasincreased by up to ~18% compared with wild-type cells. Consideringthat this modest increase was the average value within the cell, thelocal level at the near vicinity of the plasma membrane could beconsiderably to be higher than locations deeper inside, where a number

of intracellular SOD proteins exist. Thus, SodC, being localized onthe outer leaflet of the plasma membrane, may regulate the level ofextracellular superoxide radicals and/or the flux of the radical intothe cell. A previous study has shown that superoxide radicals couldrapidly become neutralized by protonation and permeable to theplasma membrane (Korshunov and Imlay, 2002).

Furthermore, the finding that chemotactic defects of sodC– cellswere alleviated by expression of wild-type SodC but not with thecatalytically inactive mutant SodC strongly suggests that thedismutation of the radical by SodC is indeed important in theregulation of Dictyostelium chemotaxis.

The previous study of pten cell chemotaxis showed thatalleviation of excessive PtdIns(3,4,5)P3 by LY294002 treatment

Fig. 9. Aberrant localization of PI3K in sodC– cells. (A) The membranelocalization domain of PI3K1 (N-PI3K1) fused with GFP was expressed inwild-type and sodC– cells. Aggregation-competent polarized wild-type cellsclearly demonstrated localized PI3K membrane translocation at the leadingedge, whereas sodC– cells showed no PI3K polarization around themembrane. By contrast, GFP-PTEN localization was indistinguishablebetween wild-type and sodC– cells. (B) 0.01% Triton X100 fraction showedthat more N-PI3K1-GFP proteins were aberrantly enriched in the membranefraction of sodC– cells than of wild-type cells. (C) Cells expressing PI3K1-LD-GFP proteins were pulsed with 50 nM cAMP for 4 hours, and stimulatedwith 10 μM cAMP. Membrane translocation of PI3K1-LD protein wasrecorded at 10-second intervals. Clear membrane localization of PI3K1-LDwas observed in wild-type cells, but no such changes were seen in sodC– cells.(D) Cells were pulsed for 4 hours, fixed and stained with TRITC-phalloidin, asdescribed in the Materials and Methods. Two representative images are shownfor each wild-type and sodC– cell. Wild-type cells displayed more polarizedcell bodies with a leading edge enriched with F-Actin. By contrast, sodC– cellswere much more round than the wild type and showed numerous filopodia-likestructures instead of a dominant pseudopodium. (E) Pulsed cells werestimulated with 10 μM cAMP as indicated, and lysed with a F-Actin buffercontaining 0.2 % of Triton X-100 and TRITC-phalloidin, and the F-Actinlevels were measured as described in the Materials and Methods. sodC– cellsdisplayed a higher basal level of F-Actin compared with wild type, but nowild-type-like response was observed after cAMP stimulation.

Fig. 10. Ras proteins were not properly regulated in sodC– cells. (A) ActiveRas proteins are visualized by GFP-RBD signals on the plasma membrane.Wild-type cells often displayed well-organized RBD signal on one side of acell, whereas sodC exhibited broad GFP-RBD signal around cellularperipheries. Images were captured after 4 hours of cAMP pulsing. (B) Asignificantly higher basal level of active Ras was detected in extracts fromsodC– cells compared with wild-type cells after 4 hours of pulsing. (C) Cellswere pulsed with 50nM cAMP for 4 hours, and then stimulated with 10 μMcAMP as indicated. Total Ras protein levels were first normalized by westernblot using anti-Pan-Ras antibody. Active Ras proteins were determined byGST-RBD assay (Sasaki et al., 2004). (D) A higher basal level of active Ras insodC– cells was significantly decreased upon depletion of superoxide byincubation with radical scavenger XTT (4 mM) for 10 minutes at roomtemperature. As in C, wild-type cells showed a lower level of active Ras.(E) sodC– cells displayed higher basal level of active GFP-RasG than did wild-type cells. (F) GFP-RasG was modestly activated by conditioned medium(CM) and was susceptible to XTT in wild-type cells. In sodC– cells, GFP-RasG showed higher basal activity and was susceptible to XTT, but no furtherstimulation of GFP-RasG was observed with CM.

Jour

nal o

f Cel

l Sci

ence

3106

could significantly improve chemoattractant sensing (Chen et al.,2003). Consistent with the previous study, LY294002-treated sodC–

cells displayed near wild-type-like efficiency in chemoattractantsensing. Under the same experimental conditions, wild-type cellsdisplayed no significant change in gradient sensing and only a minordecrease in motility, which were comparable with previous reports(Loovers et al., 2006; Takeda et al., 2007). Contrary to thechemotaxis index, the speed of locomotion was only partiallyrestored from LY294002-treated sodC– cells. Considering that 15μM of LY294002 treatment did not completely depletePtdIns(3,4,5)P3 (Fig. 8A), an excessive PtdIns(3,4,5)P3 depletionis unlikely to be the reason for the incomplete restoration ofchemotaxis of sodC– cells by LY294002 treatment. The presenceof high level of active Ras proteins in sodC– cells would haveprevented more complete rescue after LY294002 treatment.

Previous studies have uncovered that Ras proteins, probablyRasG, activate PI3K, which in turn controls a PtdIns(3,4,5)P3-dependent cascade that involves extracellular cAMP productionthrough adenylyl cyclase activation (Sasaki et al., 2004). RasG isalso known to control cytoskeletal remodeling independently ofcAMP production (Zhang et al., 1999). A later study also suggeststhat Ras proteins, most likely RasG, control actin cytoskeletalremodeling through the TORC2 complex, in addition to thePtdIns(3,4,5)P3 pathway (Lee et al., 2005). In addition, cells witha constitutively high level of active RasG(G12T) displayednumerous fine filopodial extensions, membrane ruffles anddecreased cell motility (Khosla et al., 1996; Zhang et al., 1999),which are also prominent in sodC– cells. Furthermore, both Rasand F-Actin levels are constitutively higher, and no further activationwas observed in response to cAMP stimulation in sodC– cells.Under chronic activation of Ras proteins, sodC– cells seem to losetheir ability to regulate PI3K (Fig. 9C) and F-Actin synthesiscorrectly (Fig. 9E) in response to cAMP.

A previous study has indicated that the majority of the Ras-GTPbound to the human Raf1-RBD in aggregation-competent cells wasRasG (Sasaki et al., 2004). Consistent with this, we discovered thatGFP-RasG was aberrantly activated in sodC– cells when assayedwith the human Raf1-RBD (Fig. 10E). Furthermore, incubation ofsodC– cells with superoxide scavenger XTT lowered the level ofactive endogenous Ras proteins and of GFP-RasG (Fig. 10D,F).Wild-type cells also displayed increases in endogenous Ras andGFP-RasG levels in a superoxide scavenger XTT-sensitive manner(Fig. 10D,F). We therefore propose that SodC indirectly suppressesthe level of active Ras proteins by lowering superoxide level.Considering that RasG functions upstream of not only PI3K butalso of other pathways, such as TORC2, an indirect regulation ofRasG by SodC-mediated regulation of superoxide thus seems to beessential for proper chemoattractant sensing, cell polarization andmotility.

Materials and MethodsGeneration of restriction enzyme mediated insertion (REMI)mutants and sodC– cellsRemi clones were generated following the terminator trapping REMI protocol (Takedaet al., 2003) from Ax3 cells expressing the GFP-PHcrac protein. Individual Remiclones were identified using the 3� Race Core Kit (TAKARA): the cDNAs weregenerated from RNA by RT-PCR with the oligo dT-3 sites adaptor primer from the3� Race kit, and then amplified by PCR using a Blasticidin-specific primer (5�-CGAGTGGTAAGTCCTTGT GG-3�) and the oligo dT-3 adaptor primer. TheBlasticidin cassette was identified at the 1004th nucleotide of the SodC open readingframe in remi56. Three other independent Remi clones also contained the Blasticidincassette at the identical position of SodC. sodC– cells were generated by homologousrecombination from a wild-type cell (JH10), and confirmed by genomic PCR using

a primer set (5�-GGTGGTGTTGAAGGTATT-3� and 5�-TTCAACATTACCAC-CATTTGC-3�). The absence of SodC transcript was confirmed by RT-PCR using thefollowing primer set: 5�-ATGAGACTTTTATCTGTATTAG-3� and 5�-TTAAAG-CAAAGCAAAGATAATTG-3�.

Generation of the full-length SodC, GFP-SodC and myc-SodCconstructsThe full-length SodC was cloned by RT-PCR using the primer set 5�-ATGAGACTTTTATCTGTATTAG-3� and 5�-TTAAAGCAAAGCAAAGATAATTG-3�, and confirmed by sequencing. GFP-SOD was constructed by excising the SODdomain from full-length SodC in the TOPO-TA vector by HinCII digestion andsubcloned into the PGEX-4T-2 vector (Pharmacia). The SOD domain was then excisedusing EcoRI and AccI, and subcloned in-frame into the Dictyostelium GFP expressionvector pDEXH-GFP (Faix et al., 1992). The construct was confirmed by sequencing,and expression was verified by western blotting with anti-GFP antibody (1:1000,Covance).

Myc-SodC expression plasmid was constructed as follows. The full-length SodCcDNA in TOPO vector was amplified with M13 primer and a primer encoding SodCSignal Peptide (SP)/Myc sequence and NdeI site (CCATATGTTAAATCTTCTTC -TGAAATTAATTTTTGTTCAAAAGCATATTGGTAATCGGCTTTTGCAATGGAA ATAC3�), subcloned into a TOPO vector (pTOTP-SP-Myc-NdeI), and confirmed bysequencing. Both pTOPO-SodC and pTOPO-SP-Myc-NdeI were digested with XhoI(TOPO vector) and NdeI (at 211th bp of SodC) and gel purified. The SP-Myc-Nde1(153bp) insert was ligated to the digested vector to create Topo-SP-Myc-SodC, inwhich a Myc epitope replaced 124 bp (76~210 nucleotides) of SodC. After sequenceconfirmation, the SP-Myc-SodC was released with EcoRI digestion and ligated toEcoRI-digested pExp4 vector to create SP-Myc-SodC.

A SodC mutant with a disrupted copper-binding site (SodC(H245R, H247Q)) (Wanget al., 2002) was generated with a mutant primer (ccggtttatcttatcaagctcatggtttcAGAgttcAACaatttggtgatgtttcatcgg, where capital letters denotes mutations) and theQuickchange site-directed mutagenesis kit (Stratagene).

SOD activity assaySOD activity was measured using the SOD assay kit (Dojindo) according to themanufacturer’s instructions. WST-1 solution (200 μl) was mixed with the xanthineoxidase-containing enzyme mix (20 μl) and the SOD-containing samples (20 μl) andwere incubated at 25°C for 15 minutes. The relative superoxide levels weredetermined by measuring the OD450 of the reaction mix after 20 minutes at 25°C.WST-1 formazan has a molar absorption of 3.7�10 at 450 nm. Mean values fromthree independent experiments are shown with error bars representing standarddeviations.

For testing GPI cleavage of SodC, cells were treated with phosphatidylinositol-specific phospholipase C (PI-PLC, Molecular Probes) prior to the SOD assay. Forthese, 1�10 log phase cells were washed and resuspended with 200 μl of 1�PBS.PI-PLC (1.0 U, 10 μl) was added to each sample, and the reaction mixtures wereincubated at 25°C for 5 minutes. The cell-free media were saved and their SODactivities were measured as described above. Mean values from three independentexperiments are shown with error bars representing standard deviations.

Superoxide quantification: XTT and NBT assaysExtracellular superoxide levels were measured by using XTT as described previously(Bloomfield and Pears, 2003). Ten cells were pulsed with 50 nM cAMP for 4 hoursat 20�10 cells/ml. Equivalent amount of cells (~1.5�10) were harvested andresuspended with 0.15 ml of DB containing 0.5 mM XTT for 10 minutes at 22°C.Amount of reduced XTT was measured spectrophotometrically at 470 nm.

Levels of intracellular superoxide were measured by using NBT (Nitro bluetetrazolium salt) as previously described (Choi et al., 2006). Cells (2.5 ml) at a densityof 2�10 cells/ml were pulsed with 50 nM cAMP for 4 hours, and resuspended with1 ml of DB containing 0.2 mM NBT for 30 minutes at 22°C. Cells were then washedtwice with DB, once with methanol and air-dried. Dried pellets were solubilized with0.24 ml of 2 M KOH and 0.28 ml of DMSO (dimethyl sulfoxide). The intracellularNBT extracted from cells were measured spectrophotometrically at 620 nm.

Submerged aggregation and cAMP chemotaxis assaysFor submerged aggregation experiments, log phase cells were harvested, washed andplaced under DB at the cell densities of 2.5�10 cells/cm. After 10 hours at 22°C,cell migration, streaming and aggregation were observed.

For chemotaxis assays, log phase cells were differentiated with 50 nM pulses ofcAMP for 4 hours. Aggregation-competent cells were plated at a density of 6�10cells/cm. A micromanipulator with a glass capillary needle (Eppendorf Femtotip)was filled with either 100 nM or 10 μM cAMP solution. The responses of the cellswere followed by time-lapse video recording using a CoolSNAP digital camera. Theroundness of a chemotaxing cell, which represents polarity of cells, is defined as theratio of ellipsoidal short radius divided by its long radius was calculated as describedelsewhere (Loovers et al., 2006). The chemotactic index, which is defined as thedistance moved in the direction of the pipette divided by the total distance moved,was computed from the centroid positions (Loovers et al., 2006).

Journal of Cell Science 121 (18)

Jour

nal o

f Cel

l Sci

ence

3107Regulation of Dictyostelium chemotaxis by superoxide dismutase

Fractionation of the membrane and cytoplasmFractionations of cells expressing GFP-PHCRAC proteins have been describedelsewhere (Parent et al., 1998). Ten vegetative cells were washed and resuspendedwith 150 μl of membrane lysis buffer [20 mM TrisCl (pH 7.7), 2 mM MgSO4] andfilter-lysed into 1 ml of cold PM buffer (5 mM KH2PO4, 5 mM Na2HPO4, 2 mMMgSO4) by filtration through a Nucleopore filter (0.2 μm). Cell lysates wereimmediately centrifuged at 12,000 g for 1 minute at 4°C to separate the membrane-containing pellets from the cytosolic supernatant fractions. The supernatant was mixedwith 4�SDS loading dye and the pellets were solubilized with 50 μl of 1�SDSloading dye. The membranous fractions (10 μl) and 20 μl of the cytosolic fractionswere separated on a 4-20% gradient SDS-PAGE. GFP-PHCRAC localization wasanalyzed by western blotting using an anti-GFP antibody (1:1000, Covance).

The PI3K-containing membranous fractions were prepared according to thepublished procedures (Han et al., 2006; Sasaki et al., 2004). Cells (2.5�10) wereresuspended with 200 μl of 1�PBS and mixed with an equal volume of 0.02% ofTriton X-100 solution and incubated on ice for 5 minutes. All solutions containedprotease inhibitors (Roche, Complete Mini). Mixtures were then centrifuged at 12,000g for 5 minutes at 4°C. The cytosolic supernatant fractions were separated from themembranous pellet fractions. The supernatants were mixed with 4�SDS loading dyeand the pellets were solubilized with 100 μl of 1� SDS loading dye. Membranousfractions (1 μl) and 40 μl of the cytosolic fractions were analyzed by western blotusing anti-GFP antibodies described above.

GFP-fusion proteins and immunofluorescence microscopyThe N-PI3K1-GFP and GFP-RBD constructs have been described previously (Sasakiet al., 2004). All fluorescent images were obtained using a 100� oil-immersion lenson a Leica DM IRB inverted epifluorescence microscope. For indirect fluorescencemicroscopy, cells were permeabilized with 0.01% Triton X-100 in 1�PBS for 10minutes, and fixed with 3.7% formaldehyde for an additional 20 minutes at 22°C.Fixed cells were washed twice with 1�PBS, then incubated with an anti-Myc antibody(1:100 dilution, Santa Cruz Biotech) for an additional 2 hours at 22°C, and washedthree times with 1�PBST (1�PBS, 0.3% Triton X-100) for 30 minutes at 22°C.Rhodamine-conjugated anti-Rabbit goat IgG (1:200 dilution, Molecular Probes) wasused as a secondary antibody.

AntibodiesAnti-GFP antibodies were from Covance for western blot analysis (1:1000 dilution)and from eBioscience for immunoprecipitation (5 μl per each IP). Anti-Myc and anti-Pan-Ras antibodies were from Calbiochem (Ab-3).

Ras-binding assayRas assays were as described by Sasaki et al. (Sasaki et al., 2004). Cells were pulsedwith cAMP for 4 hours and then lysed with cell lysis buffer [20 mM TrisCl (pH 7.7),150 mM NaCl, 1% Triton X-100, 5% glycerol, 1 mM EDTA, 0.1% β-mercaptoethanoland 1� Roche Protease Inhibitor mix]. The whole cell lysates were then mixed with5 μg of purified GST-RBD (Ras Binding Domain) on Glutathione-sepharose beadsfor 2 hours at 4°C, and the GST-RBD/active Ras complexes were washed three timeswith cell lysis buffer. The active Ras proteins bound with GST-RBD were visualizedby western blotting with anti-Pan-Ras antibody (Calbiochem, Ab-3). Wild-type andsodC– cells expressing GFP-RBD proteins were pulsed with 50 nM cAMP in DBfor 4 hours and monitored under epifluorescent microscope.

Ras activation with conditioned medium (CM)A previous study has shown cellular superoxide production in response to stimulationwith conditioned medium prepared after cAMP pulsing (Bloomfield and Pears, 2003).One hundred million cells in 5 ml DB were pulsed with 50 nM cAMP pulses for 4hours, and the supernatant fraction was saved as the conditioned medium (CM) afterseparation of cells by centrifugation. Ras activation in response to superoxidegeneration was determined using the GST-RBD assay as described earlier. Superoxideradicals were depleted by incubation with the scavenger XTT (4 mM). Ras activationwas measured by GST-RBD assay described earlier.

F-Actin assayCells were cAMP pulsed at 2�10 cells/ml for 4 hours, briefly centrifuged andresuspended with PM buffer (5 mM KH2PO4/K2HPO4, 2 mM MgCl2) at 3�10cells/ml. Cells were stimulated with 10 μM cAMP for times as indicated at the figurelegends. At each time point, 100 μl of cells were taken and mixed with 1 ml of actinbuffer [3.7% formaldehyde, 10 mM PIPES, 0.1% Triton X-100, 20 mMK2HPO4/KH2PO4, 5 mM EGTA, 2 mM MgCl2, 250 nM TRITC-phalloidin (pH 6.8)].Cells were fixed and stained for 1 hour at 21°C on an orbital shaker. Cytoskeletalfractions were pelleted by centrifugation and washed with 1 ml of methanol andshaken overnight. Methanol-extracted TRITC-phalloidin was quantified by fluorimetry(540 nm excitation, 575 nm emission).

F-Actin staining with TRITC-phalloidinBoth wild-type and sodC– cells were pulsed at 2�10 cells/ml in DB for 4 hours.Cells were plated on eight-well chambers at a density of 100�10 per cm and incubatedfor 5 minutes at 22°C. Then cells were washed twice gently with 1�PBS. Fixation

was carried out by adding 3.7% formaldehyde in 1�PBS for 10 minutes at 22°C.Subsequently, cells were permeablized with 0.01% Triton X in PBS for 5 minutes at22°C. The cells were washed three times with 1�PBS before the addition of 1�PBScontaining 0.5 μM TRITC-phalloidin and 0.5 % BSA. After 30 minutes of incubationat 22°C, cells were washed three times with PBS and examined under a fluorescentmicroscope.

We thank Peter Devreotes, Richard Firtel, Carole Parent and theDictyostelium stock center for providing various GFP-constructs. Wethank Tong Sun and Osmin Anis for technical assistance. S.V. is a FIUpresidential fellow.

ReferencesAndrew, N. and Insall, R. H. (2007). Chemotaxis in shallow gradients is mediated

independently of PI3K by biased choices between random protrusion. Nat. Cell Biol. 9,193-200.

Bloomfield, G. and Pears, C. (2003). Superoxide signalling required for multicellulardevelopment of Dictyostelium. J. Cell Sci. 116, 3387-3397.

Chen, L., Iijima, M., Tang, M., Landree, M. A., Huang, Y. E., Xiong, Y., Iglesias, P. A.and Devreotes, P. N. (2007). PLA2 and PI3K/PTEN pathways act in parallel to mediatechemotaxis. Dev. Cell 12, 603-614.

Choi, H. S., Cha, Y. N. and Kim, C. K. (2006). Taurine chloramines inhibits PMA-stimulatedsuperoxide production in human neutrophils perhaps by inhibiting phosphorylation andtranslocation of p47. Int. Immunopharmacol. 6, 1431-1440.

Cox, A. D. and Der C. J. (2003). The dark side of Ras: regulation of apoptosis. Oncogene22, 8999-9006.

Droge W. (2002). Free radicals in the physiological control of cell function. Physiol. Rev.82, 47-95.

Faix, J., Gerisch, G. and Noegel, A. (1992). Overexpression of the csA cell adhesionmolecule under its own cAMP-regulated promoter impairs morphogenesis inDictyostelium. J. Cell Sci. 102, 203-214.

Funamoto, S., Meili, R., Lee, S., Parry, L. and Firtel, R. A. (2002). Spatial and temporalregulation of 3-phosphoinositides by PI 3-kinase and PTEN mediates chemotaxis. Cell109, 611-623.

Han, J. W., Leeper, L., Rivero, F. and Chung, C. Y. (2006). Role of RacC for the regulationof WASP and phosphatidylinositol 3-kinase during chemotaxis of Dictyostelium. J. Biol.Chem. 281, 35224-35234.

Hancock, J. T, Desikan, R. and Neill, S. J. (2001). Role of reactive oxygen species incell signalling pathways. Biochem. Soc. Trans. 29, 345-350.

Heo, J. and Campbell, S. L. (2005). Superoxide anion radical modulates the activity ofRas and Ras-related GTPase by a radical-based mechanism similar to that of nitric oxide.J. Biol. Chem. 280, 12438-12445.

Hoeller, O. and Kay, R. (2007). Chemotaxis in the absence of PIP3 gradients. Curr. Biol.17, 813-817.

Huang, Y. E., Iijima, M., Parent, C. A., Funamoto, S., Firtel, R. A. and Devreotes, P.(2003). Receptor-mediated regulation of PI3Ks confines PI(3,4,5)P3 to the leading edgeof chemotaxing cells. Mol. Biol. Cell 14, 1913-1922.

Iijima, M. and Devreotes, P. (2002). Tumor suppressor PTEN mediates sensing ofchemoattractant gradients. Cell 109, 599-610.

Iranfar, N., Fuller, D. and Loomis, W. F. (2003). Genome-wide expression analyses ofgene regulation during early development of Dictyostelium discoideum. Eukaryotic Cell2, 664-670.

Khosla, M., Spiegelman, G. B. and Weeks, G. (1996). Overexpression of an activatedrasG gene during growth blocks the initiation of Dictyostelium development. Mol. Cell.Biol. 16, 4156-4162.

Kondoh, G., Tojo, H., Nakatani, Y., Komazawa, N., Murata, C., Yamagata, K., Maeda,Y., Kinoshita, T., Okabe, M., Taguchi, R. et al. (2005). Angiotensin-converting enzymeis a GPI-anchored protein releasing factor crucial for fertilization. Nat. Med. 11, 160-166.

Korshunov, S. S. and Imlay, J. A. (2002). A potential role for periplasmic superoxidedismutase in blocking the penetration of external superoxide into the cytosol of Gram-negative bacteria. Mol. Microbiol. 43, 95-106.

Lee, S., Comer, F. I., Sasaki, A., McLeod, I. X., Duong, Y., Okumura, K., Yates, Jr,3rd, Parent, C. A. and Firtel, R. A. (2005). TOR complex 2 integrates cell movementduring chemotaxis and signal relay in Dictyostelium. Mol. Biol. Cell 16, 4572-4583.

Loovers, H. M., Postma, M., Keizer-Gunnink, I., Huang, Y. E., Devreotes, P. N. andvan Haastert, P. J. (2006). Distinct roles of PI(3,4,5)P3 during chemoattractant signalingin Dictyostelium: a quantitative in vivo analysis by inhibition of PI3-kinase. Mol. Biol.Cell 17, 1503-1513.

Parent, C. A., Blacklock, B. J., Froehlich, W. M., Murphy, D. B. and Devreotes, P. N.(1998). G protein signaling events are activated at the leading edge of chemotactic cells.Cell 95, 81-91.

Sasaki, A. T., Chung, C., Takeda, K. and Firtel, R. A. (2004). Localized Ras signalingat the leading edge regulates PI3K, cell polarity, and directional cell movement. J. CellBiol. 167, 505-518.

Sasaki, A. T., Janetopoulos, C., Lee, S., Charest, P. G., Takeda, K., Sundheimer, L. W.,Meili, R., Devreotes, P. N., Firtel, R. A. (2007). G protein-independent Ras/PI3K/F-actin circuit regulates basic cell motility. J. Cell Biol. 178, 185-191.

Takeda, K., Saito, T., Tanaka, T., Morio, T., Maeda, M., Tanaka, Y. and Ochiai, H.(2003). A novel gene trap method using terminator-REMI and 3� rapid amplification ofcDNA ends (RACE) in Dictyostelium. Gene 312, 321-333.

Jour

nal o

f Cel

l Sci

ence

3108

Takeda, K., Sasaki, A. T., Ha, H., Seung, H. A. and Firtel, R. A. (2007). Role of

phosphatidylinositol 3-kinases in chemotaxis in Dictyostelium. J. Biol. Chem. 282, 11874-

11884.

Tsuji, A., Akaza, Y., Nakamura, S., Kodaira, K. and Yasukawa, H. (2003). Multinucleation

of the sodC-deficient Dictyostelium discoideum. Biol. Pharm. Bull. 26, 1174-1177.

Turner, B. J., Atkin, J. D., Farg, M. A., Zang, D. W., Rembach, A., Lopes, E. C., Patch,

J. D., Hill, A. F. and Cheema, S. S. (2005). Impaired extracellular secretion of mutant

superoxide dismutase 1 associates with neurotoxicity in familial amyotrophic lateral

sclerosis. J. Neurosci. 25, 108-117.

Van Driessche, N., Shaw, C., Katoh, M., Morio, T., Sucgang, R., Ibarra, M., Kuwayama,

H., Saito, T., Urushihara, H., Maeda, M. et al. (2002). A transcriptional profile of

multicellular development in Dictyostelium discoideum. Development 129, 1543-1552.

van Haastert, P. J., Keizer-Gunnink, I. and Kortholt, A. (2007). Essential role of PI3-kinase and phospholipase A2 in Dictyostelium discoideum chemotaxis. J. Cell Biol. 177,809-816.

Wang, J., Xu, G., Gonzales, V., Coonfield, M., Fromholt, D., Copeland, N. G., Jenkins,N. A. and Borchelt, D. R. (2002). Fibrillar inclusions and motor neuron degenerationin transgenic mice expressing superoxide dismutase 1 with a disrupted copper-bindingsite. Neurobiol. Dis. 10, 128-138.

Zhang, T., Rebstein, P. J., Khosla, M., Cardelli, J., Buczynski, G., Bush, J., Spiegelman,G. B. and Weeks, G. (1999). A mutation that separates the RasG signals that regulatedevelopment and cytoskeletal function in Dictyostelium. Exp. Cell Res. 247, 356-366.

Zhou, K., Takegawa, K., Emr, S. D. and Firtel, R. A. (1995). A phosphatidylinositol (PI)kinase gene family in Dictyostelium discoideum: biological roles of putative mammalianp110 and yeast Vps34p PI 3-kinase homologs during growth and development. Mol. Cell.Biol. 15, 5645-5656.

Journal of Cell Science 121 (18)

Jour

nal o

f Cel

l Sci

ence