PTEN regulates motility but not directionality during leukocyte chemotaxis

The localization at opposite cell poles of phosphatidylinositol-3 kinases and PTEN (phosphatase and tensin homolog on chromosome 10) governs Dictyostelium chemotaxis. To study this model in mammalian cells, we analyzed the dynamic redistribution of green fluorescent protein (GFP)-tagged PTEN chimeras during chemotaxis. N- or C-terminus GFP-tagged PTEN was distributed homogenously in the cytoplasm of chemotaxing PTEN-negative Jurkat cells and PTEN-positive HL60 cells. Moreover, we did not detect uropod accumulation of endogenous PTEN in chemoattractant-stimulated HL60 cells. Cell fractionation indicated that both endogenous and ectopically expressed PTEN were confined largely to the cytosol, and that chemoattractant stimulation did not alter this location. PTEN re-expression in Jurkat cells or PTEN depletion by specific siRNA in HL60 cells did not affect cell gradient sensing; PTEN nonetheless modulated chemoattractant-induced actin polymerization and the speed of cell movement. The results suggest a role for PTEN in regulating actin polymerization, but not directionality during mammalian cell chemotaxis.

PI3K activity (Maehama and Dixon, 1998). Several protein substrates of PTEN have also been proposed, including focal adhesion kinase (FAK) and PTEN itself Raftopoulou et al., 2004). In addition to its ability to suppress cell growth and survival (Borlado et al., 2000;Stambolic et al., 1998), evidence suggests that PTEN controls mammalian cell movement. Pten -/fibroblasts show an increased migration rate, which is reduced after expression of wild-type PTEN in these cells (Liliental et al., 2000;Tamura et al., 1998). Although this phenotype is common to cells that have lost PTEN expression, there are also apparently contradictory findings. Some studies suggest that loss of PTEN increases Bcell chemotaxis to CXCL12 (Fox et al., 2002;Suzuki et al., 2003), whereas another group reported reduced chemotaxis in pten -/-B cells (Anzelon et al., 2003).
The mechanism by which PTEN regulates cell migration is unclear. In some cases, the increased cell migration has been attributed to augmented PtdIns(3,4,5)P 3 levels (Liliental et al., 2000;Tamura et al., 1998). Nonetheless, the PTEN effect on migration was also ascribed to its protein phosphatase activity. Indeed, expression of the naturally occurring PTEN mutant G129E, which has lost lipid phosphatase activity but retains protein phosphatase activity, inhibited migration of some tumor cells ; this effect might be mediated by PTEN-induced dephosphorylation of FAK. More recent evidence suggests a role for PTEN C2 domain phosphorylation in the inhibition of PTEN-deficient human glioma cell migration, independent of its lipid phosphatase activity

6207
The localization at opposite cell poles of phosphatidylinositol-3 kinases and PTEN (phosphatase and tensin homolog on chromosome 10) governs Dictyostelium chemotaxis. To study this model in mammalian cells, we analyzed the dynamic redistribution of green fluorescent protein (GFP)-tagged PTEN chimeras during chemotaxis. N-or C-terminus GFP-tagged PTEN was distributed homogenously in the cytoplasm of chemotaxing PTEN-negative Jurkat cells and PTENpositive HL60 cells. Moreover, we did not detect uropod accumulation of endogenous PTEN in chemoattractantstimulated HL60 cells. Cell fractionation indicated that both endogenous and ectopically expressed PTEN were confined largely to the cytosol, and that chemoattractant stimulation did not alter this location. PTEN re-expression in Jurkat cells or PTEN depletion by specific siRNA in HL60 cells did not affect cell gradient sensing; PTEN nonetheless modulated chemoattractant-induced actin polymerization and the speed of cell movement. The results suggest a role for PTEN in regulating actin polymerization, but not directionality during mammalian cell chemotaxis. (Raftopoulou et al., 2004). The mechanism by which PTEN affects cell migration may be closely linked to its location in moving cells. For example, it is difficult to reconcile the model of FAK dephosphorylation by PTEN with the location of PTEN at the cell rear, as FAK is enriched at the leading edge of fibroblast and lymphocytes (Friedl et al., 1998).
Here we analyzed the dynamic redistribution of GFP-tagged PTEN chimeras in chemotaxing cells. In PTEN-deficient Jurkat T cells (Xu et al., 2002) and in HL60 cells (which express endogenous PTEN), both N-and C-terminal GFPtagged PTEN (GFP-PTEN or PTEN-GFP, respectively) showed cytoplasmic distribution almost identical to that of ds-Red2 fluorescent protein (RFP) cotransfected into these cells. The cytosolic restriction of PTEN, confirmed by cell fractionation, was independent of its catalytic activity. PTEN re-expression in Jurkat cells or PTEN depletion with specific siRNA in HL60 cells did not affect chemotaxis toward specific attractants. By contrast, catalytically active PTEN diminished chemoattractant-induced actin polymerization, and diminished cell speed. Whereas these findings indicate a role for PTEN in leukocyte motility, the data suggest that some chemoattractant signal amplification mechanisms are not conserved between Dictyostelium and mammalian cells.

Materials and Methods
Cloning and expression constructs cDNA encoding human PTEN was a gift of R. Pulido (CIC, Valencia, Spain) and the PTEN-C/S (C124S) mutants, generated with the Quick Change site-directed mutagenesis kit (Stratagene) were subcloned in pEGFP-N3 and pEGFP-C2 vectors (Clontech) to generate the GFPchimeras. PTEN and PTEN-C/S cDNA were subcloned in the bicistronic plasmid pRV-GFP (Genetrix, Spain). pRFP-GPI was obtained as described (Gómez-Moutón et al., 2004). Jurkat and HL60 cells differentiated in dimethyl sulfoxide (2ϫ10 7 cells) were transfected by electroporation (BioRad) or by using the T-cell Nucleofector Kit (Amaxa Biosystem) with GFP-PTEN, PTEN-GFP or PTEN-C/S-GFP (C124S), or with PTEN or PTEN-C/S cDNA subcloned in the bicistronic plasmid pRV-GFP. Expression was analyzed 6-14 hours post-transfection by FACS (Coulter) and live cells were isolated on Ficoll gradients. In vitro PTEN phosphatase activity was measured in extracts of Jurkat cells transfected with the different chimeras or untagged PTEN in a Malachite-Green-based assay (Upstate Biochemicals, #17-351), using PtdIns(3,4,5)P 3 (Echelon) as a substrate, according to the manufacturer's instructions. An affinity purified, constitutive active PTEN form supplied with the kit was used as positive control.
siRNA experiments HL60 cells were transfected with a pool of PTEN-specific siRNA or with pooled non-specific siRNA duplexes (100 nM; SMARTpool, Dharmacon) using Jet-PEI (Polyplus transfection). Control and PTEN-specific siRNA were co-transfected with a Cy3-labeled nonspecific duplex control (Dharmacon), which were used to determine transfection efficiency (98-100%) by FACS. Silencing was confirmed 24-48 hours post-transfection by western blot.

Time-lapse confocal videomicroscopy and chemotaxis assays
Real-time cell chemotaxis was studied by time-lapse confocal microscopy, as described (Gómez-Moutón et al., 2004). Starved cells were plated (1 hour, 37°C) on fibronectin (Fn)-coated chamber coverslips (Nunc). Chemotaxis studies were performed at 37°C using a heating plate and a micromanipulation system (Narishige) adapted to a Leica confocal microscope. Stimulus was supplied in micropipettes (1-2 µm) prepared in a Kopf pipette puller using thinwall glass capillaries with an inner filament (Clark Electromedical Instruments), filled with CXCL12 (100 nM; Peprotech) or fMLP (100 nM; Sigma) in serum-free RPMI and sealed at the back. Fluorescence and phase contrast images were recorded at established time intervals and videos were processed with NIH-Image J software. Fluorescence scanning was performed with MicroImage software (Olympus Optical). Cell motility speed and movement paths were analyzed with MetaMorph Imaging System (Universal Imaging).
For transwell assays, 1-2ϫ10 5 PTEN-expressing Jurkat cells, or Jurkat cells pretreated with LY294002 or siRNA-transfected HL60 cells in serum-free medium were seeded in the upper chamber; lower chambers were filled with serum-free medium alone or containing CXCL12 (25 nM) or fMLP (100 nM). After incubation (2.5 hours, 37°C), cell number in the lower chamber was estimated by FACS. The percentage of migrating cells was calculated as the quotient between the number of migrating fluorescent cells in the lower chamber and the fluorescent cells in the initial input (ϫ100).

Characterization of PTEN chimeras
To analyze PTEN dynamics in living cells, we constructed PTEN dynamics in chemotaxis chimeric proteins by tagging GFP to the PTEN N-or Cterminus and characterized the in vitro activity of these chimeras expressed in Jurkat cells in a Malachite-Green-based phosphatase assay using PtdIns(3,4,5)P 3 as a substrate. We found that both PTEN-GFP (N-terminal tag) and GFP-PTEN (C-terminal tag) effectively dephosphorylated this substrate, whereas the catalytically inactive PTEN-C/S-GFP showed only minimal phosphatase activity compared with mock Jurkat cells (Fig. 1A). Notably, active GFP chimeras showed lipid phosphatase activity similar to that of untagged PTEN overexpressed at comparable levels ( Fig. 1A).
To study the phosphatase activity of the chimeras in living cells, we took advantage of the fact that protein kinase B (PKB/AKT) phosphorylation and activation require PI3K products (Xu et al., 2002). Expression of PTEN-GFP, but not of a catalytically inactive mutant (PTEN-C/S-GFP), reduced basal phospho-AKT levels in Jurkat cells (Fig. 1B). As in other reports (Xu et al., 2002), PTEN-GFP did not eliminate AKT activation after CXCL12 stimulation (Fig. 1B), although it reduced the peak of CXCL12-induced AKT phosphorylation. In contrast to the PTEN-GFP chimera, basal phospho-AKT levels in Jurkat cells were not affected by GFP-PTEN chimera overexpression (Fig. 1C), although this chimera dephosphorylated PtdIns(3,4,5)P 3 in the in vitro lipid phosphatase assay (Fig. 1A).
Western blot analysis of Jurkat cells expressing PTEN chimeras with anti-PTEN and anti-GFP antibodies always developed a single band corresponding to the predicted molecular weight for the fusion between PTEN and GFP. This suggests that the chimeras were expressed as fusion proteins. To confirm this point, we performed immunofluorescence staining of unstimulated and chemoattractant-stimulated HL60 cells expressing these chimeras, using an anti-PTEN antibody. We found coincident cytoplasmic staining of PTEN-GFP (green label) and of an anti-PTEN antibody (red label) in both unstimulated and fMLP-stimulated cells (see supplementary material, Fig. S1). Similar results were obtained with the GFP-PTEN chimera (not shown).

GFP-PTEN and PTEN-GFP do not polarize during chemotaxis
We initially analyzed PTEN dynamics in chemotaxing human cells by expressing a C-terminal GFP-tagged PTEN fusion protein, similar to that used in the Dictyostelium studies. HL60 cells co-expressing PTEN-GFP and the inert cytosolic RFP probe were exposed to an fMLP gradient delivered by micropipette. PTEN-GFP levels were approximately 1.5-fold higher than those of endogenous PTEN (see supplementary material, Fig. S2A). Most PTEN-GFP was found in cytoplasm, with no redistribution during chemotaxis ( Fig. 2A; see also supplementary material, Movie 1); some nuclear staining was also seen in several cells. In some frames, PTEN-GFP accumulated in the cytosol area facing the attractant source; in other frames, weaker cytosolic labeling was seen at the rear. Fluorescence intensity scanning showed that PTEN-GFP accumulation coincided with increased RFP labeling in the same cell areas (Fig. 2B); PTEN-GFP accumulation at the sides was not observed (  the raft marker RFP-GPI (glycosylphosphatidylinositol). Whereas the bulk of PTEN-GFP was distributed homogeneously, RFP-GPI polarized to the front and rear of moving, PTEN-GFP-expressing HL60 cells (Fig. 2D,E). These results indicate that specific markers are polarized in PTEN-GFP-expressing cells.
The PTEN C-terminal region has multiple phosphorylation sites for casein kinase II and a PDZ-binding domain that can interact with membrane-associated guanylate kinase inverted (MAGI) proteins (Georgescu et al., 1999;Torres and Pulido, 2001;Wu et al., 2000). As these interactions could be impeded in the PTEN-GFP chimera, we studied GFP-PTEN dynamics in fMLP-stimulated HL60 cells. GFP-PTEN levels were approximately twofold those of endogenous PTEN (see supplementary material, Fig. S2A). During fMLP-induced chemotaxis of HL60 cells coexpressing GFP-PTEN and RFP, most GFP-PTEN was evenly distributed in cytosol (Fig. 3A,B; see also supplementary material, Movie 3).

Cell location of PTEN is independent of its catalytic activity
We studied the dynamics of the inactive PTEN-C/S-GFP during chemotaxis of HL60 cells co-expressing the RFP-GPI raft marker. PTEN-C/S-GFP levels were approximately twofold those of endogenous PTEN (see supplementary material, Fig. S2A). RFP-GPI polarized during cell movement, whereas PTEN-C/S-GFP was evenly distributed in cytosol (Fig. 4A,B; see also supplementary material, Movie 4). Cytosolic PTEN-C/S-GFP distribution was similar to that of RFP in moving Jurkat cells (see supplementary material, Fig.  S4). Taken together, the results suggest that PTEN-C/S-GFP localization remains homogeneous in chemotaxing cells.

PTEN chimeras and endogenous PTEN remain evenly distributed in cytoplasm after stimulation
To analyze the amount of PTEN chimera bound to the membrane before and after stimulation, we subfractionated PTEN-GFP-or PTEN-C/S-GFP-expressing Jurkat cells and found that most of these chimeras were confined to the cytosol of unstimulated or CXCL12-stimulated cells (Fig. 5). In contrast to PTEN chimeras, CXCL12 stimulation induced recruitment of the class IA PI3K p85 regulatory subunit to the membrane shortly after stimulation (Fig. 5). CXCL12 stimulation induced p85 membrane recruitment even in mock cells (Fig. 5), which have a high basal PI3K activity level, as these cells do not express PTEN. The results concur with those in Fig. 1B, showing that CXCL12 induced AKT phosphorylation even in PTEN null cells.
We also studied the location of endogenous PTEN in chemoattractant-stimulated HL60 cells, and found homogeneous distribution of endogenous PTEN in fMLPstimulated PTEN-expressing HL60 cells stained with an anti-PTEN antibody (Fig. 6A). The specificity of this antibody was tested in CEM cells (Fig. 6B), a leukemia lymphoblastic cell line with homozygous deletion of the PTEN gene (Sakai et al., 1998). Cell fractionation showed that endogenous PTEN was confined largely to cytosol in unstimulated or fMLP-stimulated cells (Fig. 6C), whereas the p85 regulatory subunit translocated to the plasma membrane shortly after fMLP stimulation.

PTEN regulates actin polymerization but not cell orientation
We induced PTEN depletion with specific siRNA in HL60 cells. PTEN siRNA-transfected HL60 cells showed a 50% reduction in PTEN protein levels compared with non-specific siRNA-transfected or untransfected cells (Fig. 7A). Despite this reduction, chemotaxis toward fMLP was comparable between cells with normal or reduced PTEN levels (Fig. 7B). As we could not abrogate PTEN expression completely in HL60 cells, we performed the opposite experiment, reexpressing PTEN in a true PTEN-null cell line. We compared chemotaxis between GFP-expressing Jurkat cells and cells transfected with a bicistronic plasmid (not a chimeric protein) containing PTEN or PTEN-C/S and GFP (Fig. 7C); again, chemotaxis was independent of PTEN protein levels or PTEN activity (Fig. 7D). Nonetheless, Jurkat chemotaxis was sensitive to the PI3K inhibitor LY294002 (see supplementary material, Fig. S5), confirming previous observations (Sotsios et al., 1999). LY294002 treatment also inhibited basal migration in the absence of stimulus, suggesting that this drug could affect the general motility of Jurkat cells.  Single cell assays showed that orientation and persistence of movement toward the chemotactic source was PTENindependent (Fig. 7E), although PTEN-expressing cells moved more slowly than PTEN-null cells (Fig. 7E). PTEN expression influenced CXCL12-induced actin polymerization. CXCL12 stimulation of Jurkat cells induced a rapid increase in actin polymerization that peaked in seconds, followed by a slow phase in which polymerization was sustained for minutes (Fig.  7F). Expression of active PTEN, but not the catalytically inactive mutant, attenuated both actin polymerization phases, and F-actin returned to basal levels more rapidly in PTENexpressing Jurkat cells than in controls (Fig. 7F). Collectively, the results suggest that PTEN has no major role in cell orientation but participates in actin cytoskeleton remodeling and cell speed.

Discussion
PI3K and PTEN are proposed as components of an excitationinhibition system, to explain how a cell senses and responds to shallow chemoattractant gradients. In this model, PTEN is a global inhibitor that concentrates at the sides and rear of cells to confine PI3K activation to the front of cells (Funamoto et al., 2002;Iijima and Devreotes, 2002). Whereas PTEN shows this pattern in Dictyostelium, two apparently contradictory results were reported in mammalian cells (Li et al., 2003;Xu et al., 2003). We performed a systematic analysis of real-time distribution of N-and C-terminustagged PTEN chimeras, with or without catalytic activity, in chemotaxing PTEN-expressing or PTEN-null cells. We found no evidence of endogenous or chimeric PTEN accumulation or localization in directionally-or homogenously-stimulated Journal of Cell Science 117 (25)  cells. PTEN was evenly distributed throughout the chemotaxing cell cytosol, with a pattern similar to that of the inert cytosolic RFP probe.
We found that both N-and C-terminus GFP-tagged chimeric PTEN are homogeneously dispersed in the cytosol of fMLP or CXCL12-stimulated cells. This contrasts with Dictyostelium cells, in which around 10% of PTEN associates the cell perimeter after chemoattractant stimulation (Iijima and Devreotes, 2002). Our data are nonetheless consistent with other studies in mammalian cells, which indicate that the bulk of PTEN is in the cytoplasm and/or nucleus (Deichmann et al., 2002;Gimm et al., 2000;Wu et al., 2000). The mechanisms regulating PTEN activity and subcellular distribution are not yet fully known. As its substrate is at the plasma membrane, a PTEN fraction should be found at this site; indeed, a GFPtagged PTEN mutant bearing the phosphatase and C2 domains persists at the membrane (Das et al., 2003). The PTEN Nterminus contains a putative PtdIns(4,5)P 2 binding motif, important for PTEN membrane association in Dictyostelium (Iijima and Devreotes, 2002). Indeed, PTEN-GFP, but not GFP-PTEN, reduced phospho-AKT levels in Jurkat cells, which suggests a role for the PTEN N-terminal region in the regulation of phosphatase activity. C-terminal phosphorylation is suggested to regulate PTEN membrane binding, activation and stability, as it serves as an electrostatic switch for PTEN membrane interaction (Das et al., 2003;Torres and Pulido, 2001). Although the PTEN-GFP reduced basal phospho-AKT levels, an indirect readout of PtdIns(3,4,5)P 3 levels, we cannot rule out that the presence of GFP at the C-terminus could prevent specific phosphorylation/dephosphorylation events required for PTEN localization.
We therefore analyzed in detail the location of endogenous PTEN in HL60 cells. Immunofluorescence analysis showed that endogenous PTEN was largely confined to the cytosol of unstimulated or fMLP-stimulated HL60 cells, and this was again confirmed by cell fractionation. Moreover, we did not detect preferential accumulation of the endogenous protein at the cell rear or sides in polarized cells. This result is consistent with a previous report in which no PTEN accumulation was found in differentiated HL60 cells (Xu et al., 2003), in apparent contradiction of preferential PTEN accumulation at the uropod in primary neutrophils (Li et al., 2003). This raised the intriguing possibility that PTEN location in chemotaxing cells may be governed by cell type-specific factors. It should be noted, however, that we observed AKT PH domain recruitment to the leading edge of differentiated HL60 cells (Gómez-Moutón et al., 2004), indicating that PI3K signaling is restricted to the cell front in these cells.
Our inability to detect membrane-associated PTEN after cell subfractionation may be caused by rapid PTEN shuttling between the cytosol and the membrane. Dephosphorylation of the C-terminus is crucial for PTEN recruitment to the membrane but enhances its proteolytic turnover (Das et al., 2003;Torres and Pulido, 2001). This turnover would preclude the observation of PTEN accumulation at the membrane in cell fractionation or immunofluorescence analyses. Despite this turnover, one would predict certain PTEN accumulation in the cytosolic areas at which PTEN is actively recruited in chemotaxing cells. Moreover, if only a very small PTEN fraction is functionally active in moving cells, it is questionable whether this amount of PTEN is sufficient to function as a global PI3K inhibitor during migration, as proposed for Dictyostelium. This is especially relevant, as chemoattractants induced robust recruitment of PI3K to the cell membrane (Fig.  5, Fig. 6C) and persistent accumulation of active PI3K at the leading edge of moving HL60 cells (Gómez-Moutón et al., 2004).
We therefore analyzed the effect of PTEN in HL60 and Jurkat cell migration. First, we tried to inhibit PTEN expression in HL60 cells using siRNA. Although PTEN expression was reduced around 50%, HL60 chemotaxis toward fMLP was unaffected. Since small amounts of PTEN were detected in these cells, it can be argued that this is sufficient to provide the small fraction of functional PTEN needed for chemotaxis. To rule out this possibility, we used Jurkat cells, a PTEN-deficient cell line. We found that PTEN re-expression reduced the speed of moving cells, as reported for other cell types (Liliental et al., 2000;Tamura et al., 1998); however, PTEN-expressing Jurkat cells sensed and responded to chemoattractant gradients as did mock-transfected Jurkat cells. These results contrast with those reported for Dictyostelium, in which PTEN expression increases the speed and controls the directionality of moving cells (Funamoto et al., 2002;Iijima and Devreotes, 2002).
Evidence suggests that the PI3K pathway might be important for the chemotaxis of specific cell types. In lymphocytes, contradictory results have been obtained using PI3K inhibitors (Sotsios et al., 1999;Cronshaw et al., 2004). Moreover, it was recently shown that lymphocyte homing is controlled mainly by DOCK-2 activity, whereas PI3Kγ activation has a minor role in this process (Nombela-Arrieta et al., 2004;Ward, 2004). In HL60 cells and neutrophils, it was proposed that PI3K is required for directionality (Li et al., 2003;Xu et al., 2003). Our results suggesting that PTEN has a minor influence on cell orientation do not contradict this view; indeed, we found that chemoattractant stimulation induced further PI3K activation in PTEN-null cells, despite the high basal PI3K activity in these cells. PTEN overexpression did not prevent chemoattractant-induced PI3K activation, as determined by AKT phosphorylation, although it reduced the basal PI3K activity of PTEN-null cells. These results concur with previous observations indicating that PTEN does not abolish PI3K activation following stimulation, but has a major role in regulating basal PtdIns(3,4,5)P 3 levels (Myers et al., 1998;Stambolic et al., 1998;Xu et al., 2002). This may explain the sustained actin polymerization phase observed in mock (pten null) and PTEN-C/S-expressing compared to PTEN-expressing Jurkat cells.
We propose that, at least in some mammalian cells, PTEN is not essential for a cell to sense and respond to shallow chemoattractant gradients. Accumulation of PI3K products at the leading edge of these moving cells may be a consequence of preferential activation of specific receptors at this cell site (Gómez-Moutón et al., 2004;van Buul et al., 2003). PTEN may nonetheless have other functions in mammalian cell chemotaxis by controlling actin cytoskeleton remodeling, either through its lipid or protein phosphatase activities (Raftopoulou et al., 2004).