PIPKIγ Regulates Focal Adhesion Dynamics and Colon Cancer Cell Invasion

Focal adhesion assembly and disassembly are essential for cell migration and cancer invasion, but the detailed molecular mechanisms regulating these processes remain to be elucidated. Phosphatidylinositol phosphate kinase type Iγ (PIPKIγ) binds talin and is required for focal adhesion formation in EGF-stimulated cells, but its role in regulating focal adhesion dynamics and cancer invasion is poorly understood. We show here that overexpression of PIPKIγ promoted focal adhesion formation, whereas cells expressing either PIPKIγK188,200R or PIPKIγD316K, two kinase-dead mutants, had much fewer focal adhesions than those expressing WT PIPKIγ in CHO-K1 cells and HCT116 colon cancer cells. Furthermore, overexpression of PIPKIγ, but not PIPKIγK188,200R, resulted in an increase in both focal adhesion assembly and disassembly rates. Depletion of PIPKIγ by using shRNA strongly inhibited formation of focal adhesions in HCT116 cells. Overexpression of PIPKIγK188,200R or depletion of PIPKIγ reduced the strength of HCT116 cell adhesion to fibronection and inhibited the invasive capacities of HCT116 cells. PIPKIγ depletion reduced PIP2 levels to ∼40% of control and PIP3 to undetectable levels, and inhibited vinculin localizing to focal adhesions. Taken together, PIPKIγ positively regulates focal adhesion dynamics and cancer invasion, most probably through PIP2-mediated vinculin activation.


Introduction
Focal adhesions (FAs, also called cell-matrix adhesions) are specific types of large macromolecular assemblies at the ventral surface of cells, functioning as both mechanical machineries and regulatory signaling hubs [1,2]. Temporal and spatial regulation of focal adhesion assembly/disassembly is required for cell migration [3]. During cell migration, nascent focal adhesions (also called focal complexes) are formed to stabilize lamellipodia at the front of cells while focal adhesions are dissolved at the trailing edges of cells [4,5].
Focal adhesion assembly and disassembly are also implicated in cancer invasion, a prerequisite for metastasis. DRR (Down-Regulated in Renal cell carcinoma) associates with actin and microtubules and stimulates glioma invasion by promoting focal adhesion disassembly [6]. The actin cross-linking protein filamin A suppresses focal adhesion disassembly and breast cancer cell invasion [7]. Rho/Rock signaling promotes tumor cell migration and invasion by regulating focal adhesion dynamics through caveolin-1 phosphorylation [8]. FAK also promotes focal adhesion disassembly and cancer invasion [9,10,11]. Focal adhesion dynamics and signaling pathways that regulate this process could be therefore an attractive target for cancer therapy.
Many molecules have been shown to regulate focal adhesion dynamics. FAK regulates focal adhesion turnover [9,12], probably through a dynamin and microtubule-dependent pathway [13]. Paxillin, a focal adhesion adaptor, also modulates focal adhesion dynamics by JNK and PAK-mediated phosphorylation [14,15]. Talin activates integrins and initiates focal adhesion formation [16,17,18], whereas cleavage of talin by calpain mediates focal adhesion disassembly [19]. Calpain also cleaves FAK and paxillin to modulate focal adhesion dynamics [20,21]. We have shown that Smurf1-mediated ubiquitination of the talin head domain, one of the two cleavage products, plays an important role in focal adhesion disassembly and cell migration [22]. ACF7, a microtubule and filamentous actin binding protein, regulates focal adhesion assembly/disassembly through its ATPase activity [23]. However, the molecular mechanisms that control focal adhesion assembly/disassembly are not fully understood.
Phosphatidylinositol phosphate kinase type I c (PIPKIc) is an enzyme that catalyzes ATP-dependent phosphorylation of phosphatidylinositol 4-phosphate (PI(4)P) to generate PI(4,5)P 2 , which regulates a variety of biological processes, including focal adhesion formation [24]. PIPKIc has a well-conserved kinase catalytic domain at the central region [25]. Within the catalytic domain, there is a subdomain called the activation loop, which determines site of phosphorylation on the inositol ring of the substrate. PIPKIc strongly interacts with talin and competes for talin binding with the b integrin tail [26,27]. It is localized at adherens junctions in epithelial cells [28,29]. It has been reported that PIPKIc is required for focal adhesion formation in migratory cells [30]. However, the precise roles of the lipid kinase activities of PIPKIc in focal adhesion dynamics are not defined.
In the present study, we investigated the requirement for the lipid kinase activity of PIPKIc in focal adhesion dynamics and colon cancer invasion. Our results identify an essential role of PIPKIc in focal adhesion assembly/disassembly and cancer invasion.

PIPKIc promotes focal adhesion formation
To examine a possible role for PIPKIc in focal adhesion formation, CHO-K1 cells were transfected with EGFP-PIPKIc or EGFP vector, respectively. The cells were re-plated on fibronectin, fixed with paraformaldehyde, incubated with an anti-paxillin monoclonal antibody, and then stained with Dylight 549 conjugated goat anti-mouse IgG. Focal adhesions were lined up around the edges of the EGFP vector-transfected cells, with few focal adhesions in the centers of the cells, whereas PIPKIc expression dramatically stimulated focal adhesion formation in the centers of the cells (Fig. 1A). Over-expression of PIPKIc stimulates an increase in focal adhesion numbers (Fig. 1B) and the increase was mainly contributed by small focal adhesions (,3 mm 2 ) (Fig. 1C). The stimulation of focal adhesion formation by PIPKIc relies on its interaction with talin, because PIPKIc W647F , a mutant that is deficient in the interaction with talin, was not able to promote focal adhesion formation (Fig. 1A PIPKIc kinase activity is required for stimulation of focal adhesion formation PI(4,5)P 2 binds talin and strengthens the interaction between talin and the b integrin tail, stimulating integrin clustering [31]. PI(4,5)P 2 also binds vinculin and unmasks the actin and talin binding sites on vinculin, promoting focal adhesion formation [32]. Therefore, PIPKIc-stimulated PI(4,5)P 2 synthesis could be essential for promoting focal adhesion formation. To test this hypothesis, we mutated two lysine residues, K188 and K200, to arginine residues within the ATP-binding site of PIPKIc. The mutant PIPKIc K188,200R and the WT were purified from CHO-K1 cells and the kinase activities were examined in vitro by using mass spectrometry to quantitate production of PI(4,5)P 2 from PI(4)P. As shown in Fig. 2A, mutation at K188 and K200 reduced kinase activity by 95%. EGFP-tagged mutant PIPKIc K188,200R and WT PIPKIc were stably expressed in CHO-K1 cells, and their effects on focal adhesion formation were examined after paxillin staining. Cells that stably express WT PIPKIc formed focal adhesions around the edges and in the centers of cells, whereas cells that express PIPKIc K188,200R , similar to parental cells, possessed small focal adhesions, most of which were around the edges of the cells, and had a defect in spreading (Fig. 2B). Quantitative analysis indicated that PIPKIc increased focal adhesions by more than 2 fold, whereas PIPKIc K188,200R did not significantly promote focal adhesion formation (Fig. 2C&D). These results indicate that PIPKIc activity is essential for focal adhesion formation in CHO-K1 cells.
To further verify the requirement of PIPKIc activity for focal adhesion formation, we tested the capacity of another kinase dead mutant, PIPKIc D316K [33], to promote focal adhesion formation as described above. As shown in Fig. S1, the average focal adhesion number (per cell) in cells expressing PIPKIc D316K was approximately 45, which is approximately the same as that in cells expressing EGFP vector (Fig. 1). The focal adhesions in cells expressing PIPKIc D316K accumulated at the edges of the cells, with very few focal adhesions in the center of the cells. This result confirms the essential role of PIPKIc activity in focal adhesion formation.
The activity of PIPKIc is essential for its promoting focal adhesion dynamics To examine whether PIPKIc lipid kinase activity is required for focal adhesion dynamics, CHO-K1 cells that stably express EGFP-PIPKIc or -PIPKIc K188,200R were transfected with mDsRedpaxillin and then plated on MatTek dishes (with a glass coverslip at the bottom) precoated with fibronectin (5 mg/ml) and grown for 3 hr. TIRF images of mDsRed-paxillin were taken using a Nikon TIRF microscope and the temperature was maintained at 37uC using an INU-TIZ-F1 microscope incubation system (Tokai Hit). Images were recorded at 1-min intervals for a 120 min period. Focal adhesion assembly and disassembly rate constants were calculated as described previously [22]. The FA assembly and disassembly rate constants in cells expressing PIPKIc K188,200R were 0.08360.014 and 0.07260.010 min 21 , respectively, which are similar to those in normal CHO-K1 cells that we reported previously [22], whereas FA assembly and disassembly rate constants were 0.19060.019 and 0.13660.014 min 21 , respectively, in cells expressing WT PIPKIc (Fig. 3). This result indicates that the inositol lipid kinase activity of PIPKIc is required for its stimulation of focal adhesion assembly and disassembly.
An essential role for PIPKIc in focal adhesion formation in colon cancer cells Focal adhesions have been implicated in regulating cancer invasion, while the role of focal adhesions in colon cancer cells has not been well defined. To determine if the activity of PIPKIc influences focal adhesion formation in colon cancer cells, HCT116 cells that stably express EGFP-PIPKIc or -PIPKIc K188,200R were plated on fibronectin and stained for paxillin. As shown in Fig. 4A, most of the focal adhesions were located to the peripheral region, and HCT116 cells expressing PIPKIc K188,200R had much fewer focal adhesions than those expressing the WT enzyme. The average focal adhesion number in cells expressing WT PIPKIc was 22.5/cell, whereas that in cells expressing PIPKIc K188,200R was 11.5/cell, both of which were fewer than those observed in CHO-K1 cells (Fig. 4B). In addition, PIPKIc K188,200R had more effect on the smaller focal adhesions than the larger ones (Fig. 4C).
To test whether PIPKIc is essential for focal adhesion formation in HCT116 cells, HCT116 cells were infected with recombinant lentiviruses that express PIPKIc shRNA or shRNA control and were selected with puromycin. As shown in Fig. 5A, expression of PIPKIc shRNA resulted in dramatic reduction in the endogenous PIPKIc level of HCT116 cells. The cells were then plated on fibronectin and stained for paxillin. Surprisingly, PIPKIc knockdown almost abolished focal adhesion formation in HCT116 cells (Fig. 5B). PIPKIc depletion reduced focal adhesion number by about 74% (Fig. 5C). Different from PIPKIc K188,200R , PIPKIc shRNA had more effect on the larger focal adhesions than the smaller ones (Fig. 5D). These results indicate an essential role of PIPKIc in focal adhesion formation in HCT116 colon cancer cells. This dramatic effect of PIPKIc knockdown did not occur in MDA-MB-231 human breast cancer cells, where PIPKIc knockdown only partially inhibited focal adhesion formation (data not shown).

PIPKIc positively modulates adhesion strength of colon cancer cells to fibronectin
To examine whether the activity of PIPKIc regulates cell adhesion strength on fibronectin, HCT116 cells that stably express EGFP-PIPKIc or -PIPKIc K188,200R were stained with calcein-AM and seeded on a 96-well plate that was pre-coated with different concentrations of fibronectin. The calcein fluorescence was read, and then the plate was inverted and centrifuged at 1506g for 5 min. The calcein fluorescence was read again. As shown in Fig. 6A, the cell adhesion strength difference between the cells expressing PIPKIc K188,200R and the WT was not significant at high concentrations of fibronection. However, at low concentrations of fibronectin, the cells expressing PIPKIc K188,200R had significantly lower adhesion strength as compared to those expressing the WT. PIPKIc knockdown also caused a significant decrease in cell adhesion strength in HCT116 cells (Fig. 6B).
These results indicate that PIPKIc activity regulates cell adhesion strength in colon cancer cells.

PIPKIc is essential for invasion but not the migration of colon cancer cells
It has been reported that PIPKIc plays a role in cell migration. To test whether PIPKIc regulates the migration of HCT116 colon cancer cells, we employed time-lapsed wound-healing assays to examine the migration of HCT116 cells expressing EGFP-PIPKIc and -PIPKIc K188,200R , respectively, in the presence of HGF. As shown in Fig. S2 A&B, cells expressing PIPKIc K188,200R migrated slightly faster than those expressing the WT enzyme as measured by time-lapse wound-healing assays. Also, depletion of PIPKIc had no significant effect on the migration of HCT116 cells in transwell assays (Fig. S2 C&D). These results suggest that PIPKIc activity is not essential for the migration of HCT116 colon cancer cells.
To test the role of PIPKIc in cancer invasion, the invasion of HCT116 cells that stably express PIPKIc shRNA or a shRNA control in the absence and presence of HGF was examined by using Matrigel invasion assays. As shown in Fig. 7 (A&B), PIPKIc knockdown resulted in significant reduction in the invasion of HCT116 cells either in the absence or in the presence of HGF. The basal and HGF-stimulated invasive capacities of HCT116 cells stably expressing PIPKIc K188,200R are also significantly lower than those of the cells expressing the WT enzyme (Fig. 7C&D). These results indicate that PIPKIc positively regulates the invasion of HCT116 colon cancer cells. PIPKIc regulates focal adhesion formation through activating vinculin by PI(4,5)P 2 To dissect the mechanism by which PIPKIc regulates focal adhesion formation, we set out to analyze the levels of polyphosphoinositides in HCT116 cells that express PIPKIc shRNA or a shRNA control by using mass spectrometry. Depletion of PIPKIc in HCT116 cells had no significant effect on the level of PIP, but caused significant reduction in PIP 2 level (note that these methods cannot distinguish positional enantiomers of these lipids so it is possible that PI(3,4)P 2 may contribute significantly to residual levels of PIP 2 in these cells). Interestingly, in support of this idea knockdown of PIPKIc reduced PIP 3 levels below the detection limit of our assay (Fig. 8). These results indicate that PIPKIc is a key enzyme responsible for the production of PI(4,5)P 2 and PI(3,4,5)P 3 in HCT116 cells.
To see whether PI(3,4,5)P 3 is essential for focal adhesion formation in colon cancer cells, HCT116 cells were treated with LY294002, a specific inhibitor of phosphatidylinositol 3-kinase, and focal adhesion formation in these cells was examined after paxillin staining. LY294002 (20 mM) had no significant effect on focal adhesion formation (data not shown). Taken together with the above observations, this finding indicates that PI(4,5)P 2 , but not PI(3,4,5)P 3 , is required for focal adhesion formation. PI(4,5)P 2 binds and activates vinculin and is implicated in regulating focal adhesion formation [32]. If PIPKIc-mediated production of PI(4,5)P 2 is essential for focal adhesion formation in HCT116 cells, depletion of PIPKIc would reduce PI(4,5)P 2 levels and prevent vinculin from localizing to focal adhesions. To test this hypothesis, endogenous PIPKIc-depleted cells were infected with retroviruses expressing a codon-modified PIPKIc (rescue) (Fig. 9A), and the cells were stained for vinculin. Vinculin was rarely localized to focal adhesions in PIPKIc-depleted HCT116 cells, whereas re-expression of PIPKIc in PIPKIc-depleted cells resulted in a dramatic increase in focal adhesion-localized vinculin (Fig. 9B). Quantitative analysis indicated that PIPKIc rescue restored focal adhesion formation in PIPKIc-depleted cells (Fig. 9C&D), as compared to the data in Fig. 5. These data suggest that PIPKIc may regulate focal adhesion formation through PI(4,5)P 2 -mediated vinculin activation.

Discussion
Besides serving as the precursor of other second messengers, PI(4,5)P 2 itself binds many cytoskeletal and focal adhesion proteins and is believed to be a key regulator of focal adhesion dynamics [34]. PI(4,5)P 2 binds vinculin to unmask the talin-binding sites on vinculin [32]; it also binds talin thus stabilizing talin-integrin interactions [35]. PIPKIc is thought to be the enzyme that generates PI(4,5)P 2 spatially and temporally for focal adhesion formation during cell migration [27,34]. On the other hand, PI(4,5)P 2 has not been detected at focal adhesions and the role of PIPKIc in regulating focal adhesion dynamics is controversial [36]. PIPKIc has been shown to be required for focal adhesion formation during EGF-stimulated cell migration [30], whereas it has also been reported that expression of PIPKIc caused cell rounding and focal adhesion disassembly [26].
We show here that expression of PIPKIc at low levels in CHO-K1 cells stimulated focal adhesion formation (Fig. 1), whereas kinase-dead mutants, PIPKIc K188,200R and PIPKIc D316K failed to promote focal adhesion formation (Fig. 2, Fig. 4 and Fig. S1). Furthermore, PIPKIc knockdown almost completely abolished focal adhesion formation in HCT116 colon cancer cells (Fig. 5). In addition, expression of PIPKIc promoted focal adhesion assembly and also disassembly rates, while PIPKIc K188,200R was unable to do so (Fig. 3). These results identify an essential role of PIPKIc in regulating focal adhesion dynamics.
Although direct evidence is lacking, PI(4,5)P 2 has been well implicated in regulating integrin activation. PI(4,5)P 2 binds talin and blocks its self-inhibition (head-tail interaction) thus promoting its interaction with the b integrin tail [35]. It also stimulates integrin clustering [31]. The increase in both talin-integrin interaction and integrin clustering should stimulate integrin activation. We found that HCT116 cells expressing PIP-KIc K188,200R have significantly lower adhesion strength as compared to those expressing the WT, and depletion of PIPKIc by shRNA also caused a significant decrease in cell adhesion strength in HCT116 cells (Fig. 6), suggesting that PIPKIc may modulate integrin activation in colon cancer cells.
It has been reported that PIPKIc is required for EGFstimulated migration of MDA-MB-231 human breast cancer cells and HeLa human ovarian cancer cells [30,37]. However, our results here show that neither expressing PIPKIc K188,200R , a kinase-dead mutant, nor depletion of PIPKIc inhibit the migration of HCT116 cells (Fig. S2). The MDA-MB-231 and HeLa cells migrate much faster than HCT116 cells, suggesting that PIPKIc is essential for fast-moving cells, but not for slow-migrating cells like HCT116 cells. This is supported by our unpublished result that PIPKIc K188,200R dramatically inhibits the migration of Clone A cells, a fast-moving colon cancer cell line.
On the other hand, PIPKIc seems to be required for the invasion of both fast-(such as MDA-MB-231) and slow-invading (such as HCT116) cancer cells. Depletion of PIPKIc has been shown to inhibit EGF-stimulated invasion of MDA-MB-231 cells [37], and either expressing PIPKIc K188,200R or the depletion of PIPKIc inhibits the invasion of HCT116 cells (Fig. 7). These results indicate an essential role of PIPKIc in controlling cancer invasion.
Previous studies have focused on the role of focal adhesion disassembly in regulating cancer invasion. DRR promotes glioma invasion by promoting focal adhesion disassembly [6]. Filamin A suppresses breast cancer cell invasion by inhibiting focal adhesion disassembly [7]. FAK also promotes focal adhesion disassembly and cancer invasion [9]. However, our results show that both expression of PIPKIc K188,200R , the kinase-dead mutant, and depletion of PIPKIc impair focal adhesion formation, accompanying the inhibition of the invasion of HCT116 (Fig. 4, 5 and 7). In addition, PIPKIc stimulates both focal adhesion assembly and disassembly (Fig. 3), suggesting that focal adhesion turnover, which requires focal adhesion assembly and disassembly spatially and temporarily, regulates the invasion of cancer cells.
PIPKIc is a major enzyme that controls polyphosphoinositide metabolism in HCT116 cells. PIPKIc knockdown results in significant reduction in the level of PI(4,5)P 2 and decreases PI(3,4,5)P 3 levels even more substantially (Fig. 8). PI(3,4,5)P 3 plays an important role in tumorigenesis and cancer metastasis. However, PIPKIc knockdown does not affect the activation of Akt, a major target of PI(3,4,5)P 3 , in HCT116 cells (data not shown), probably because Akt can be activated by PI(3,4)P 2 , which is not directly affected by PIPKIc.
Inhibition of PI 3-kinase using LY294002 does not influence focal adhesion formation in HCT116 cells, suggesting that PI(4,5)P 2 instead of PI(3,4,5)P 3 is responsible for PIPKIc-mediated focal adhesion formation. PI(4,5)P 2 promotes vinculin binding to talin and actin and has been shown to be essential for focal adhesion formation [32]. Our result shows that PIPKIc regulates vinculin localizing to focal adhesions in HCT116 cells (Fig. 9). Taken together, it is most likely that PIPKIc regulates focal adhesion formation through PI(4,5)P 2 -mediated vinculin activation.
Cancer invasion is a complicated process, requiring spatial and temporary regulation of cell-matrix adhesions, cell protrusions and matrix-degradation [38]. PI(4,5)P 2 generated by PIPKIc regulates cancer invasion through modulating cell adhesion dynamics. Although PI(3,4,5)P 3 is not essential for focal adhesion formation, it may also play a role in other aspects of cancer invasion, probably through activating Rac [39].

Reagents
Anti-paxillin antibody (clone 5H11) was from Millipore; Anti-PIPKIc polyclonal antibody was from Cell Signaling Technology; Anti-tubulin antibody and pLKO1 lentivirus PIPKIc shRNA (sequence: CCG GGC AGT CCT ACA GGT TCA TCA ACT CGA GTT GAT GAA CCT GTA GGA CTG CTT TTT G) were from Sigma; DyLight 549 conjugated goat anti-mouse IgG (H+L) was from Thermo Scientific; Fibronectin and recombinant HGF were from Akron Biotech; Growth factor reduced Matrigel was from BD Bioscience; The plasmid pEGFP-PIPKIc was a gift from Dr. Mark Ginsberg (University of California-San Diego); Pfu Ultra was from Agilent Technologies; DNA primers were synthesized by Integrated DNA Technologies.

Plasmid construction
The plasmid pEGFP-PIPKIc W647F was generated by pfu Ultrabased PCR using pEGFP-PIPKIc as template

Cell culture, transfections and infections
CHO-K1 Chinese hamster ovary cells, HCT116 human colon cancer cells and 293T human embryonic kidney cells were from the American Type Culture Collection and were maintained in DMEM medium (Mediatech, Inc.) containing 10% fetal bovine serum (FBS), penicillin (100 U ml 21 ) and streptomycin (100 mg ml 21 ). CHO-K1 cells were transfected with Safectine RU50 (Syd Labs) according to the manufacturer's protocol. HCT116 cells that stably express EGFP-PIPKIc WT, or -PIPKIc K188, 200R , were obtained by transfecting the cells with TurboFect (Fermentas) and sorting EGFP-positive cells after G418 selection in the University of Kentucky Flow Cytometry Facility, or by infecting with pBabe retrovirus and selecting with puromycin. HCT116 cells stably expressing shRNA control or PIPKIc shRNA were obtained by infecting with pLKO1 lentivirus and selected with puromycin.

Preparation of viruses and cell infection
The 293T cells were transfected with pBabe retrovirus or pLKO1 lentivirus system using Safectine RU50 transfection reagent according to the manufacturer's protocol. The medium of transfectants was collected at 48 and 72 h, filtered through 0.45-mm filter and concentrated by ultracentrifugation. The virus particles were applied to overnight cultures of HCT116 cells for infection. Cells stably expressing recombinant proteins were obtained by growing infected cells in the presence of 1 mg/ml puromycin for 10 days.

Immunofluorescence staining and TIRF imaging
Cells were plated on glass coverslips that were precoated with 5 mg/ ml fibronectin. For immunofluorescence staining, the cells were fixed with 4% paraformaldehyde permeabilized with 0.5% Triton X-100,   blocked with 5% BSA in PBS, and then incubated with anti-paxillin mAb. Paxillin was then visualized by incubating with DyLight 549 Conjugated goat Anti-mouse IgG (H+L). Paxillin staining and EGFP fluorescence were viewed by using a Nikon Eclipse Ti TIRF microscope equipped with a 606, 1.45 NA objective, CoolSNAP HQ2 CCD camera (Roper Scientific). Images were acquired and analyzed by using NIS-Elements (Nikon). To quantify the number and area of focal adhesions, paxillin immunofluorescence images were thresholded to include only focal adhesions and the number and area were calculated by using the software.

Time-lapse live fluorescence imaging
CHO-K1 cells that stably express EGFP-PIPKIc WT, or -PIPKIc K188, 200R were transfected with mDsRed-paxillin. At 24 h post-transfection, the cells were trypsinized and plated on MatTek dishes (with a glass coverslip at the bottom) that had been precoated with fibronectin (5 mg ml 21 ). The cells were cultured for 3 hours and TIRF images were taken using the Nikon Eclipse Ti TIRF microscope stated earlier and the temperature and humidity were maintained by using a INU-TIZ-F1 microscope incubation system (Tokai Hit). Images were recorded at 1-min intervals for a 120 min period. Focal adhesion assembly and disassembly rate constants were analyzed as described previously [22].

Centrifugation assays
HCT116 cells growing on 60-mm dishes were incubated in 2 ml of 2 mg/ml calcein-AM in opti-MEM for 20 min at 37uC. The Cells were then trypsinized, washed and re-suspended in normal growth media containing 10%FBS. The cell suspensions (100 ml, 40,000 cells/well) were added to a 96-well plate coated with different concentrations of fibronectin, centrifuged at 180g in a Beckman Coulter Allegra X-15R centrifuge (SX4750A rotor) and allowed to attach for 1h in 37uC CO2 incubator. Each well was then carefully aspirated to remove floating cells and refilled with fresh PBS-dextrose. An initial fluorescence (F 0 ) was read to determine the density of cells before detachment using a Promega Glomax multi+ Detection system (490 nm excitation, 510-570 nm emission). The lid was then removed, and the plate was covered with sealing tape and centrifuged upside down at 150 g for 5 min to detach the cells. The wells were carefully aspirated and refilled with fresh PBS-dextrose. The fluorescence after centrifuging (F a ) was read to determine the density of cells that remain attached.

Invasion Assays
One hundred microliters of Matrigel (1:30 dilution in serumfree DMEM medium) was added to each Transwell polycarbonate filter (6 mm diameter, 8 mm pore size, Costar) and incubated with the filters at 37uC for 4 h. HCT116 cells were trypsinized and washed 3 times with DMEM containing 1% FBS. The cells were resuspended in DMEM containing 1% FBS at a density of 1610 6 cells/ml. The cell suspensions (100 ml) were seeded into the upper chambers, and 600 ml of DMEM medium containing 1% FBS and 5 mg/ml Fibronectin with or without 50 ng/ml HGF were added to the lower chambers. The cells were allowed to invade for 36 h in a CO2 incubator. The invaded cells were fixed for 15 min with 3.7% formaldehyde and stained using 0.1% crystal violet in 10% ethanol for 30 min. The number of invaded cells per field was counted under a light microscope at 6200. MatTek glass bottom dishes coated with 5 mg/ml fibronectin, and grown to 90% confluency. The medium was then changed to DMEM containing 1%FBS and 10 ng/ml HGF for 6 h. A wound was made on the confluent monolayer, and time-lapse cell migration was recorded using a Nikon Biostation IMQ. The pictures were extracted from time-lapse movies. (B) Quantification of the migration speed of HCT116 cells that stably express PIPKIc or PIPKIc K188,200R using NIS-Elements AR 3.2. (n = 4, *P,0.05). (C) Depletion of PIPKIc by using shRNA had no significant effect on the migration of HCT116 cells. The migration of cells stably expressing shRNA control or PIPKIc shRNA were examined in the absence and presence of HGF (50 ng/ml) by Transwell migration assays. In brief, transwell polycarbonate filters (6 mm diameter, 8 mm pore size, Costar) were coated with fibronectin (5 mg/ml) over night. HCT116 cells were trypsinized and washed 3 times with DMEM containing 1% FBS. The cells were resuspended in DMEM containing 1% FBS at a density of 1610 6 cells/ml. The cell suspensions (100 ml) were seeded into the upper chambers, and 600 ml of DMEM medium containing 1% FBS and 5 mg/ml Fibronectin with or without 50 ng/ml HGF were added to the lower chambers. The cells were allowed to migrate for 24 h in a CO2 incubator. The migrated cells were fixed for 15 min with 3.7% formaldehyde and stained using 0.1% crystal violet in 10% ethanol for 30 min. The number of migrated cells per membrane was counted under a light microscope at 6200. (D) Quantification of the migration of HCT116 cells that stably express shRNA control or PIPKIc shRNA. n = 3, error bar = mean 6 s.e.m, P.0.05. (TIF) Materials S1 Quantitation of phosphoinositides by HPLC ESI tandem Mass Spectrometry. (DOCX)