The MMAC1 Tumor Suppressor Phosphatase Inhibits Phospholipase C and Integrin-Linked Kinase Activity

Home , Phosphatase, Phospholipase C

Oncogene (2000) 19, 200 ± 209 ã 2000 Macmillan Publishers Ltd All rights reserved 0950 ± 9232/00 $15.00 www.nature.com/onc The MMAC1 tumor suppressor phosphatase inhibits phospholipase C and integrin-linked kinase activity

Alyssa M Morimoto1, Michael G Tomlinson1, Kaname Nakatani2, Joseph B Bolen3, Richard A Roth2 and Ronald Herbst*,1

1Department of Cell Signaling, DNAX Research Institute, 901 California Ave, Palo Alto, California, CA 94304, USA; 2Department of Molecular Pharmacology, Stanford University School of Medicine, Stanford, California, CA 94305, USA; 3Department of Oncology, Hoechst Marion Roussel, Bridgewater, New Jersey, NJ 08807, USA

Loss of the tumor suppressor MMAC1 has been shown Introduction to be involved in breast, prostate and brain cancer. Consistent with its identi®cation as a tumor suppressor, MMAC1 (also known as PTEN or TEP-1) is mutated at expression of MMAC1 has been demonstrated to reduce a high frequency in brain, breast, and prostate tumors cell proliferation, tumorigenicity, and motility as well as as well as in melanomas and endometrial carcinomas, aect cell±cell and cell±matrix interactions of malignant (Guldberg et al., 1997; Kong et al., 1997; Li and Sun, human glioma cells. Subsequently, MMAC1 was shown 1997; Li et al., 1997; Steck et al., 1997). These to have lipid phosphatase activity towards PIP3 and observations suggest that MMAC1 acts as a tumor protein phosphatase activity against focal adhesion suppressor in multiple tissues. Indeed, subsequent kinase (FAK). The lipid phosphatase activity of studies showed that reintroduction of this gene into

MMAC1 results in decreased activation of the PIP3- human glioma cells reduced cell growth, tumorigenicity dependent, anti-apoptotic kinase, AKT. It is thought that in nude mice, and aected motility and cell ± cell this inhibition of AKT culminates with reduced glioma interactions, demonstrating that MMAC1 represents a cell proliferation. In contrast, MMAC1's eects on cell bone ®de tumor suppressor (Furnari et al., 1997; Cheney motility, cell ± cell and cell ± matrix interactions are et al., 1998; Tamura et al., 1998; Morimoto et al., 1999). thought to be due to its protein phosphatase activity Interestingly, germ line mutations in MMAC1 have also towards FAK. However, recent studies suggest that the been linked to the multiple hamartomatous predisposi- lipid phosphatase activity of MMAC1 correlates with its tion syndromes, Cowden's disease and Bannayan ± ability to be a tumor suppressor. The high rate of Zonana. These syndromes are also associated with mutation of MMAC1 in late stage metastatic tumors increased susceptibility to breast and thyroid cancer suggests that eects of MMAC1 on motility, cell ± cell (Liaw et al., 1997; Marsh et al., 1997). and cell ± matrix interactions are due to its tumor Sequence analysis of MMAC1 indicated that this suppressor activity. Therefore the lipid phosphatase gene encodes motifs conserved in dual speci®city activity of MMAC1 may aect PIP3 dependent signaling phosphatases (Li and Sun, 1997; Li et al., 1997; Steck pathways and result in reduced motility and altered cell ± et al., 1997). MMAC1 has been shown to possess cell and cell ± matrix interactions. We demonstrate here protein phosphatase activity towards focal adhesion that expression of MMAC1 in human glioma cells kinase (FAK) (Tamura et al., 1998) and lipid reduced intracellular levels of inositol trisphosphate and phosphatase activity toward phosphatidylinositol 2+ inhibited extracellular Ca in¯ux, suggesting that 3,4,5-trisphosphate (PIP3) (Maehama et al., 1998). MMAC1 aects the phospholipase C signaling pathway. However, it appears that the lipid phosphatase activity In addition, we show that MMAC1 expression inhibits of MMAC1 correlates with tumor suppression (Myers integrin-linked kinase activity. Furthermore, we show et al., 1998). that these eects require the catalytic activity of Many signaling molecules directly bind PIP3 through MMAC1. Our data thus provide a link of MMAC1 to pleckstrin homology (PH) domains (Corvera and

PIP3 dependent signaling pathways that regulate cell ± Czech, 1998) and could thus be aected by reduced matrix and cell ± cell interactions as well as motility. levels of PIP3 due to expression of MMAC1. The

Lastly, we demonstrate that AKT3, an isoform of AKT binding of PIP3 to PH domain-containing proteins such highly expressed in the brain, is also a target for as AKT, phospholipase C (PLC) and integrin linked MMAC1 repression. These data suggest an important kinase (ILK) is thought to facilitate membrane role for AKT3 in glioblastoma multiforme. We therefore targeting and induce conformational changes that propose that repression of multiple PIP3 dependent result directly, or indirectly, in activation (Aoki et al., signaling pathways may be required for MMAC1 to 1998; Delcommenne et al., 1998; Falasca et al., 1998). act as a tumor suppressor. Oncogene (2000) 19, 200 ± Recent studies have shown that PIP3 levels are 209. indeed higher in cells lacking MMAC1 (Haas-Kogan et al., 1998; Stambolic et al., 1998). In addition, lack of Keywords: MMAC1; PTEN; TEP1; ILK; PLC endogenous MMAC1 expression correlated with elevated levels of activated AKT1 (Haas-Kogan et al., 1998; Myers et al., 1998; Stambolic et al., 1998; Suzuki et al., 1998), and ectopic MMAC1 expression resulted in decreased levels of activated AKT and *Correspondence: R Herbst phosphorylated BAD protein (Myers et al., 1997; Wu Received 6 May 1999; revised 30 September 1999; accepted 13 October 1999 et al., 1998a). Therefore, these results are consistent MMAC1 inhibits PLC and ILK AM Morimoto et al 201 with a model in which MMAC1 reduces AKT activity and motility of human glioma cells (Li et al., 1997; and thus increases apoptosis, resulting in tumor Steck et al., 1997; Furnari et al., 1997; Cheney et al., suppression. However, there are a number of observa- 1998; Tamura et al., 1998; Morimoto et al., 1999). tions that suggest that MMAC1 does not inhibit tumor The lipid phosphatase activity of MMAC1 appears to suppression solely through repression of the AKT result in inhibition of AKT1 activity and a growth signaling pathway. suppression phenotype (Haas-Kogan et al., 1998; Expression of MMAC1 in malignant human glioma Stambolic et al., 1998; Suzuki et al., 1998). The cells not only inhibits cell proliferation and tumor- protein phosphatase activity results in FAK depho- igenicity, but also aects motility and cell ± cell and sphorylation which is thought to then cause decreased cell ± matrix interactions (Furnari et al., 1997; Cheney cell motility and cell ± matrix interaction (Tamura et et al., 1998; Tamura et al., 1998; Morimoto et al., al., 1998). However, it appears that the lipid 1999). In addition, loss of MMAC1 correlates with the phosphatase activity of MMAC1 correlates with progression of tumors to a metastatic state, suggesting tumor suppressor activity (Myers et al., 1998). Thus that the tumor suppressor activity of MMAC1 aects far, the AKT signaling pathway has not been linked cell motility and/or cell ± matrix or cell ± cell interac- to either cell ± cell, cell ± matrix interactions or cell tions in vivo. As the AKT signaling pathway has not motility. This suggested that other PIP3 dependent been shown to aect cell ± matrix and cell ± cell signaling pathways may be inhibited by the tumor interactions or cell motility, these observations suggest suppressor activity of MMAC1 resulting in decreased that repression of PIP3 regulated signaling pathways cell motility and/or altered cell ± matrix and cell ± cell distinct from AKT may contribute to the ability of interactions. MMAC1 to suppress tumor formation. We therefore examined whether MMAC1 expres- In contrast to the AKT signaling pathway, both sion aects signaling pathways that are involved in the phospholipase C (PLC) and the integrin linked such processes and are also regulated by PIP3. One kinase (ILK) signaling pathways regulate motility as such candidate is the phospholipase C (PLC) well as cell ± cell and cell ± matrix interactions. The signaling pathway. PLC activation generates inositol

PLC signaling pathway has been linked to the trisphos-phate (IP3) and diacylglycerol (DAG) which motility of human glioma cells (Kyoshmomn et al., result in a ¯ux of intracellular calcium and

1999). Similarly, integrin linked kinase (ILK) aects stimulation of protein kinase C (PKC) in a PIP3 cell motility and cell ± cell and cell ± matrix interac- dependent manner, respectively (Falasca et al., 1998; tions. Overexpression of this PIP3 regulated kinase is Rameh et al., 1998; Rhee et al., 1997). In addition, also sucient to induce tumorigenicity in vivo the PLC signaling pathway has been linked to the (Hannigan et al., 1996; Radeva et al., 1997). As motility of human glioma cells (Khoshyomn et al., both ILK and PLC are PIP3 dependent enzymes 1999). We therefore examined whether expression of

(Hannigan et al., 1996; Falasca et al., 1998), they MMAC1 aected the intracellular levels of IP3 in represent potential downstream targets for MMAC1 human glioma cells. regulation. As such, MMAC1 could aect tumor cell As a control, we ®rst examined whether IP3 levels motility, cell ± cell and cell ± matrix interactions are sensitive to the PI3 kinase inhibitor, LY294002, in through the inhibition of these enzymes and their U373 cells. Asynchronously growing U373 control cells signaling pathways. were treated with vehicle or the PI3 kinase inhibitor,

We show here that reintroduction of MMAC1 into LY294002, and the amount of IP3 present in the lysates human glioblastoma cells reduces extracellular Ca2+ was determined. As shown in Figure 1a, the levels of in¯ux, intracellular inositol trisphosphate (IP3) levels, IP3 were reduced by over 50% when cells were treated and ILK activity. These results thus provide a link with LY294002. To determine whether MMAC1 between MMAC1 and PIP3 dependent signaling similarly aects IP3 levels, we also examined IP3 levels pathways that aect cell motility, cell ± cell and cell ± in cells expressing WT MMAC1 or a catalytically matrix interactions. As the PLC and the ILK signaling inactive form of MMAC1, C124S MMAC1 (Figure pathways are also involved in mitogenesis (Rhee et al., 1b). Expression of wild type MMAC1 consistently

1997; Hannigan et al., 1996), the repression of these reduced the levels of IP3 in U373 cells by 50% or more pathways may also contribute to the ability of in two independently isolated clones of cells stably MMAC1 to inhibit cell proliferation. We also expressing WT MMAC1. In contrast, cells expressing demonstrate that the activity of AKT3, an isoform of the catalytically inactive form of MMAC1 had IP3 AKT highly expressed in the brain, is inhibited by levels comparable to control cells that lack endogenous MMAC1 expression. Therefore, inhibition of both MMAC1. The pretreatment of U373 cells with AKT1 and AKT3 may be required for tumor LY294002, expression of wild type or C124S suppression in the brain. Together our results suggest MMAC1 did not appear to alter the expression level that the tumor suppressor activity of MMAC1 may be of PLCg1 (Figure 1c) the major isoform of PLC the result of pleotropic eects on multiple PIP3 expressed in U373 cells (data not shown), suggesting regulated signaling pathways. that MMAC1 aects the enzymatic activity of PLC.

These data indicated that, similar to a PI3 kinase inhibitor, MMAC1 reduces the intracellular levels of IP and that the catalytic activity of MMAC1 is Results 3 required for this eect. To further examine the potential eect of MMAC1 MMAC1 suppresses the PLC/Ca2+ signaling pathway on the PLC signaling pathway, we addressed whether The presence of MMAC1 in vivo and in vitro MMAC1 aects Ca2+ ¯ux in human glioblastoma cells. correlates with decreased proliferation, tumorigenicity Intracellular Ca2+ ¯ux can occur in response to the

Oncogene MMAC1 inhibits PLC and ILK AM Morimoto et al 202 2+ generation of IP3 by PLC. Therefore, we ®rst examined cells to release Ca from the ER stores in response to the ability of U373 cells to release Ca2+ from the serum (Figure 2a; control+serum). We found that endoreticulum (ER) stores, to ¯ux extracellular calcium U373 control cells release Ca2+ from the ER stores in

into the cells and whether these events require PIP3. response to serum, but that this response is minimal as U373 control cells were suspended in calcium free compared to the extracellular in¯ux of Ca2+ (compare media, incubated with the Ca2+-binding dye indo- Figure 2a, control versus control+serum). The use of 1AM, washed and incubated with EGTA. After 380 s, other stimuli such as EGF or thapsigargin produced cells were exposed to a source of extracellular calcium similar pro®les for Ca2+ eux from the ER stores (data to measure Ca2+ in¯ux (Figure 2a; control). We found not shown) suggesting that the predominant Ca2+ ¯ux that U373 cells exhibit a constitutive extracellular in U373 cells is extracellular in¯ux. We next examined calcium in¯ux. We next examined the ability of U373 whether the eux of Ca2+ from the ER stores and the

Figure 1 Eect of LY294002 and MMAC1 on inositol trisphosphate (IP3) levels in human glioma cells. (a) U373 control cells were preincubated with vehicle or 50 mM LY294002 for 30 min and IP3 was quantitated as described in Materials and methods. (b) The amount of IP3 in U373 control cells and cells stably expressing WT MMAC1 or C124S MMAC1 was determined as above. The results are representative of three independent experiments done with two independently isolated clones of cells stably expressing WT MMAC1 or C124S MMAC1. (c) U373 control cells treated with (lane 2) or without 50 mM LY294002 (lane 1) or two clones of cells stably expressing WT MMAC1 (lanes 3,4) or C124S MMAC1 (lanes 5,6) were examined for PLCg expression by Western blot

Oncogene MMAC1 inhibits PLC and ILK AM Morimoto et al 203 in¯ux of extracellular Ca2+ are aected by the PI3 of MMAC1 exhibit an extracellular Ca2+ in¯ux kinase inhibitor LY294002 (Figure 2a, control; similar to that of control cells (Figure 2b, C124S). LY294002+serum). Whereas there is minimal inhibi- Interestingly, the C124S MMAC1 mutant-expressing tion of the ER store release of Ca2+, the in¯ux of cells consistently exhibited a calcium in¯ux pro®le extracellular calcium was signi®cantly inhibited by slightly dierent than that of control cells (compare treatment with LY294002. These data indicated that Figure 2b control to C124S). We have observed this the predominant ¯ux of Ca2+ in U373 cells is the in¯ux phenotype with multiple independently isolated of extracellular Ca2+, and that this event is PI3 kinase- clones of C124S MMAC1 expressing cells. Similar dependent. enhanced responses have been observed with cells We next examined the eect of expression of wild expressing this mutant form of MMAC1 (Li and type MMAC1 or the catalytically inactive C124S Sun, 1998; Myers et al., 1998; Morimoto et al., mutant of MMAC1 on the ability of U373 cells to 1999) and may be due to stabilization of PIP3 or ¯ux calcium. Similar to LY294002 pretreatment of other components of this signaling pathway by cells (Figure 2a, control; LY294002+serum), expres- MMAC1. Our results demonstrate that expression sion of wild type MMAC1 signi®cantly inhibited the of catalytically active MMAC1 inhibits a PI3 kinase extracellular in¯ux of Ca2+ but did not appear to dependent in¯ux of extracellular Ca2+. Together, our signi®cantly aect the eux of calcium out of the ®ndings are consistent with MMAC1 inhibiting the ER stores (Figure 2b, compare control to WT). In PLC signaling pathway resulting in decreased 2+ contrast, cells expressing a catalytically inactive form amounts of intracellular IP3 and Ca in¯ux.

Figure 2 Extracellular Ca2+ in¯ux in U373 cells is inhibited by LY294002 and MMAC1. The ability of cells to release Ca2+ from 2+ intracellular endoreticulum stores as well as ¯ux extracellular Ca in (CaCl2 arrowhead at 380 s) was examined in response to serum stimulation (arrow at 120 s) or in the absence of any stimuli (no arrow). All cells were suspended in Ca2+ free media 2+ containing 5 mM EGTA. (a) U373 control cells were incubated with media containing Ca at 380 s (control) or treated with 20% 2+ serum at 120 s and then incubated with media containing Ca at 380 s (control+serum) or preincubated with 50 mM LY294002 and then treated with serum and media containing Ca2+ as above (control; LY294002+serum) (b) U373 control cells (control) and cells expressing wild type MMAC1 (WT) and U373 cells expressing C124S MMAC1 (C124S) were stimulated with 20% serum at 120 s (arrow) and incubated with a molar excess of calcium at 380 s (arrowhead). Enhanced eects have previously been observed with cells expressing C124S MMAC1 and may be due to stabilization of PIP3 or other components of this signaling pathway by MMAC1. The results are representative of three independent experiments done with two or more independently isolated clones of cells expressing WT MMAC1 or C124S MMAC1

Oncogene MMAC1 inhibits PLC and ILK AM Morimoto et al 204 immune precipitates (Figure 3c). Together, these MMAC1 regulates integrin linked kinase activity results indicate that expression of MMAC1 in human Integrin linked kinase (ILK) has been implicated in glioblastoma cells down-modulates ILK activity. the regulation of cell ± cell and cell ± matrix interactions (Hannigan et al., 1996; Radeva et al., 1997; Wu All AKT isoforms are expressed in normal brain tissue et al., 1998b). This serine/threonine kinase is also a

PIP3-dependent signaling protein (Delcommenne et al., MMAC1 has been shown to repress the activity of 1998), making it a candidate target for regulation by AKT1, however it is not known whether MMAC1 MMAC1. To examine a potential link between MMAC1 and ILK, we ®rst asked whether the expression patterns of ILK and MMAC1 are similar. Normal adult mouse organs were homogenized and proteins were immune precipitated with cross-linked anti-MMAC1 antibody. The immune precipitates were then Western blotted with anti-MMAC1 antibody. MMAC1 protein was detected in all tissues tested, with the highest levels in brain and barely detectable levels in the heart. Interestingly, a doublet of MMAC1 protein was consistently detected in mouse liver tissue (Figure 3a). Protein lysates were immune precipitated and then Western blotted with anti-ILK antibody (Figure 3b). High levels of ILK protein were detected in every tissue examined. Importantly, ILK protein was easily detected in the brain (Figure 3b) where MMAC1 is expressed (Figure 3a) and is thought to act as a tumor suppressor (Li et al., 1997; Steck et al., 1997). To examine whether MMAC1 aects ILK activity, lysates from U373 cells expressing wild type MMAC1, C124S MMAC1 or control cells were immune precipitated for ILK protein and kinase assays were performed using myelin basic protein as a substrate (Figure 3c). Kinase assays were quantitated using a phosphoimager. Expression of wild type MMAC1 suppressed ILK activity by 40%, and the PI3 kinase inhibitor LY294002 inhibited ILK activity by 28%. Inhibition of ILK kinase activity (by approximately 40%) was consistently observed in two independently isolated clones of cells expressing wild type MMAC1. In contrast, cells expressing C124S MMAC1 consistently exhibited ILK kinase activity slightly above Figure 3 MMAC1 represses the activity of the PIP3-regulated kinase, ILK. Protein from mouse tissues was immune precipitated (18%) that of control cells (Figure 3c). These results and then Western blotted with (a) anti-MMAC1 antibody (b) are consistent with the enhanced Ca2+ ¯ux observed in anti-ILK antibody. (c) U373 cells expressing wild type MMAC1 C124S MMAC1-expressing cells (Figure 2b, C124S). In or C124S MMAC1 or control cells were pretreated with 50 mM addition, treatment of cells expressing C124S MMAC1 PD98059 or 50 mM LY294002 for 30 min at 378C. Lysates were immune precipitated with anti-ILK antibody and kinase assays with LY294002 reduced ILK activity, demonstrating were performed using myelin basic protein (MBP) as a substrate. that ILK activity in C124S MMAC1-expressing cells is The top panel shows MBP phosphorylation and the bottom panel PI3 kinase-dependent (Figure 3c). As a control, U373 shows an anti-ILK Western blot of corresponding ILK immune cells were also treated with the MEK inhibitor, PD precipitates. Kinase assays were quantitated using a phospho- 98059. No signi®cant eect on precipitable ILK imager and ImageQuant software, Molecular Dynamics. The results are representative of two independent experiments using activity was detected indicating that MAPK activity two independently isolated clones of wild type MMAC1 and was not present at detectable levels in anti-ILK C124S MMAC1-expressing cells

Figure 4 Expression pattern of AKT isoforms in normal mouse tissue. Mouse organs (indicated at the top of the ®gure) were homogenized and lysates were immune precipitated and then Western blotted with AKT isoform-speci®c antibodies. AKT1 (lanes 1, 4, 7, 10, 13, 16), AKT2 (lanes 2, 5, 8, 11, 14, 17), AKT3 (lanes 3, 6, 9, 12, 15, 18). The migration of AKT and of the immunoglobulin heavy chain are indicated to the right of the ®gure

Oncogene MMAC1 inhibits PLC and ILK AM Morimoto et al 205 aects all three AKT isoforms. To address whether 1997; Steck et al., 1997), U373 and A172 cells were MMAC1 aects all AKT isoforms, we ®rst examined grown in serum free media overnight and then whether the isoforms exhibit expression patterns stimulated with serum. Lysates were immune pre- similar to MMAC1. Protein lysates were immune cipitated with AKT isoform-speci®c antibodies and precipitated and then Western blotted with AKT kinase assays were performed using histone H2B as a isoform-speci®c antibodies (Figure 4). AKT2 was substrate (Figure 5a,b). AKT1 exhibited very low expressed at the highest levels in the liver, testes and basal activity which is stimulated by serum twofold in spleen whereas AKT1 expression was high in the U373 cells and ®vefold in A172 cells (Figure 5b), in brain, heart, lung, spleen and testes and AKT3 agreement with recently published studies (Haas- expression was highest in brain, testes, spleen and Kogan et al., 1998; Myers et al., 1998; Stambolic et lung. All isoforms were also detected at low levels in al., 1998). In both cell lines AKT2 activity was very the remaining tissues tested. Importantly, AKT low and not altered signi®cantly by serum stimulation. isoforms were detected in all tissues in which In U373 cells, basal AKT3 activity was high and not MMAC1 protein was detected (compare Figure 4 to signi®cantly serum inducible. Similarly, AKT3 basal Figure 3a). As AKT1 and AKT3 expression were high activity was high and only increased by 18% upon in brain tissue and the loss of MMAC1 appears to be serum stimulation of A172 cells. Consistent with involved in approximately 40% of glioblastoma previously published observations, serum stimulated multiforme tumors (Li et al., 1997; Steck et al., AKT1 activity and basal AKT3 activity are inhibited 1997), these results suggested that both AKT1 and by the PI3 kinase inhibitor wortmannin in U373 cells AKT3 represent potential targets for MMAC1 in the (Downward 1998 and references within; Nakatani et brain. al., 1999 and data not shown). Therefore, all three AKT isoforms are active in GBM cells that lack endogenous expression of MMAC1. However, the All AKT isoforms are expressed and active in GBM cells isoforms exhibit distinct activation pro®les with AKT3 To determine whether all AKT isoforms are expressed possessing high basal activity and little response to and active in glioblastoma multiforme (GBM) cells serum stimulation, whereas AKT1 activity is serum that lack endogenous MMAC1 expression (Li et al., inducible.

Figure 6 MMAC1 expression represses both AKT1 and AKT3 kinase activity in human glioma cells. (a) U373 control cells, cells stably expressing wild type MMAC1 or C124S MMAC1 were immune precipitated with cross linked anti-MMAC1 antibody and Western blotted with anti-MMAC1 antibody. The migration of MMAC1 protein is indicated to the right of the ®gure. (b) Figure 5 All AKT isoforms are expressed and active in human U373 control cells (lanes 1, 3, 5) cells stably expressing wild type glioblastoma cells. (a) U373 cells were treated with (+) or MMAC1 (lanes 2, 4, 6) or C124S MMAC1 (lane 7) were immune without (7) 20% serum for 30 min at 378C. Lysates were precipitated with AKT isoform-speci®c antibodies and in vitro immune precipitated using AKT isoform-speci®c antibodies and kinase assays were performed using histone H2B as a substrate. in vitro kinase assays were performed using histone H2B as a The top panel shows H2B phosphorylation and the bottom panel substrate. (b) A172 cells were treated as in (a). Each top panel of shows AKT expression levels. The migration of AKT and of a and b shows histone H2B phosphorylation and each bottom immunoglobulin heavy chain are shown to the right of the panel shows AKT expression levels. The migration of AKT and bottom panel. The kinase assays were quantitated using a of immunoglobulin heavy chain are indicated to the right of the phosphoimager and ImageQuant software, Molecular Dynamics. bottom panels. Kinase assays were quantitated using a The results are representative of three independent experiments. phosphoimager and ImageQuant software, Molecular Dynamics. Similar results were found with two independently isolated clones The results are representative of two independent experiments of wild type and C124S MMAC1-expressing cells

Oncogene MMAC1 inhibits PLC and ILK AM Morimoto et al 206 activity. The PLCg signaling pathway has previously MMAC1 suppresses AKT3 activity in human glioma been linked to the migration of endothelial cells in cells response to PDGF (Ronnstrand et al., 1999). In To examine whether MMAC1 has an eect on AKT1 addition, inhibition of the PLCg signaling pathway and AKT3, we examined the activity of each AKT has been shown to decrease tumor cell motility and isoform in U373 cells ectopically expressing wild type invasiveness (Turner et al., 1997; Khoshyomn et al., MMAC1 (Figure 6a). U373 cells were immune 1999). Therefore the inhibition of the PLC signaling precipitated with AKT isoform-speci®c antibodies and pathway by MMAC1 may aect the motility of kinase assays were performed using histone H2B as a malignant human glioma cells. substrate. Kinase assays were quantitated using a Repression of the PLC signaling pathway may also phosphoimager. Expression of wild type MMAC1 contribute to the anti-proliferative phenotype induced consistently reduced AKT3 kinase activity by 30%, by MMAC1. Decreased intracellular levels of the whereas expression of catalytically inactive MMAC1 PLC product, DAG, could reduce PKC activity. PKC did not signi®cantly alter AKT3 activity (Figure 6b). activation occurs in response to multiple mitogenic The basal activity of AKT2 was extremely low and signals (Rhee et al., 1997), and ampli®cation of expression of MMAC1 did not considerably alter its multiple isoforms of PKC have been found in activity (Figure 6b). Expression of MMAC1 also malignant gliomas (Baltuch et al., 1996). We are suppressed AKT1 activity by 41% (Figure 6b), in currently examining whether expression of MMAC1 agreement with previous studies (Haas-Kogan et al., represses activation of various PKC isoforms in GBM 1998; Myers et al., 1998; Stambolic et al., 1998; Suzuki cells. In addition, PLC is commonly activated in et al., 1998). These results demonstrate that the activity response to stimulation of growth factor receptor of the two AKT isoforms that are highly expressed in tyrosine kinases (Rhee et al., 1997). Therefore, the brain, AKT3 and AKT1, is suppressed by repression of the PLC pathway by MMAC1 may MMAC1. also down-modulate signaling from tyrosine kinase pathways constitutively activated in gliomas (Vass- botn et al., 1994). Interestingly, calcium has also been Discussion shown to regulate Ca2+/calmodulin-dependent protein kinase kinase (CaM-KK) which represses AKT in a

Here we have shown that expression of the tumor PIP3-independent manner (Yano et al., 1998), suppressor MMAC1 in human glioblastoma cells indicating that PLC signaling may overlap with the

reduces the intracellular levels of IP3, inhibits extra- AKT pathway (Figure 7). Therefore, by repressing the cellular Ca2+ in¯ux, and inhibits ILK activity. In PLC signaling pathway, MMAC1 could eectively addition, we have identi®ed AKT3 as an additional alter cell motility as well as proliferation signals in target for MMAC1 regulation. We have also demon- AKT-dependent as well as AKT-independent path- strated that the catalytic activity of MMAC1 is ways. required for these eects. We and others have ILK may similarly aect cell motility as well as previously shown that expression of catalytically cell ± cell and cell ± matrix interactions of glioma cells. active MMAC1 is required to inhibit not only the Overexpression of ILK induces phosphorylation and proliferation of glioma cells, but also to reduce inactivation of glycogen synthase kinase-3 (GSK3) and saturation density, motility, and anchorage-indepen- may thus aect levels of the cell ± cell interaction dent growth (Tamura et al., 1998; Li and Sun, 1998; protein, b-catenin (Delcommenne et al., 1998). Over- Morimoto et al., 1999). Although the inhibition of expression of ILK also reduces expression of E- AKT1 by MMAC1 (Haas-Kogan et al., 1998; Stambolic et al., 1998; Suzuki et al., 1998) may contribute to MMAC1's anti-proliferative phenotype, the AKT signaling pathway has not been linked to cell ± matrix and cell ± cell interactions or motility. Dephosphorylation of FAK by MMAC1 is thought to aect cell motility and cell ± matrix interactions (Tamura et al., 1998). However, it appears that the lipid phosphatase activity of MMAC1 correlates with tumor suppressor activity (Myers et al., 1998). The loss of MMAC1 correlates with the progression of tumors to a metastatic state (Li et al., 1997; Steck et al., 1997). This suggests that the tumor suppressor activity of MMAC1 aects cell motility and/or cell ± matrix or cell ± cell interactions in vivo. Therefore, inhibition of

PIP3 dependent signaling pathways involved in cell motility and cell ± matrix interactions may be required for MMAC1's function as a tumor suppressor. The studies presented here identify a connection

between MMAC1 and PIP3 dependent signaling molecules that aect cell ± cell and cell ± matrix interactions and cellular motility. MMAC1 aects the levels Figure 7 MMAC1 aects multiple PIP3-dependent signaling pathways. MMAC1 represses the PIP -dependent enzymes PLC, 2+ 3 of intracellular IP3 and the in¯ux of extracellular Ca . ILK and AKT and may thus aect glioma cell proliferation, cell ± These results suggest that MMAC1 inhibits PLC cell interactions and motility. Details are given in the text

Oncogene MMAC1 inhibits PLC and ILK AM Morimoto et al 207 cadherin (Wu et al., 1998) and decreases adhesion of were maintained in Dulbecco's modi®ed eagle's media cells to the extracellular matrix (Hannigan et al., 1996; (DMEM) with 10% fetal calf serum, penicillin (250 U/ml), Wu et al., 1998). In this manner, MMAC1 inhibition streptomycin (25 mg/ml) and L-glutamine (5 mg/ml). For of ILK activity may aect cell ± cell contacts as well as serum stimulation, cells were maintained in serum free cellular motility. ILK has also been linked to cellular DMEM (minimal media) for 16 h, washed three times with PBS and maintained in minimal media for an additional 3 h. proliferation through its eect on the G -S transition of 1 Cells were then treated with DMEM containing 20% serum the cell cycle (Radeva et al., 1997). This eect has been for 30 min at 378C. For inhibition studies, cells were attributed to phosphorylation of AKT1 on serine 473 pretreated with 100 nM wortmannin (Calbiochem) or 50 mM by ILK (Delcommenne et al., 1998). Human AKT3 has LY294002 (Calbiochem) or 50 mM PD 98059 (Calbiochem) recently been cloned and appears to have the PDK2 for 30 min at 378C. serine phosphorylation site, similar to AKT1 (Brod- beck et al., 1999). If ILK indeed has PDK2 activity, then the inhibition of AKT1 and AKT3 by MMAC1 Antibodies expression could be due to combined eects on Cross linked anti-MMAC1 antibodies were generated as membrane recruitment of AKT and/or downregula- described previously (Morimoto et al., 1999). AKT isoform tion of ILK. We are currently generating a dominant- speci®c antibodies were purchased from UBI (AKT1, negative mutant of ILK to examine whether ILK AKT2, AKT3) and a rabbit polyclonal antibody generated repression aects both cellular proliferation and cell ± against the N-terminus of AKT3 (Nakatani et al., 1999). cell interactions. Anti-ILK antibody and anti-ERK1 and ERK2 MAP kinase We also show that MMAC1 expression represses antibody was purchased from UBI. AKT3 activity but does not appear to signi®cantly aect AKT2 activity. Basal AKT2 activity was extremely low in both cell lines examined, so it is Immune precipitations and Westerns dicult to address whether we would be able to detect Cells were lysed in 1% Nonidet P-40, 50 mM MOPS pH 7.0, subtle eects on AKT2 by MMAC1. Thus far, few 150 mM NaCl, 5% Glycerol, 0.4 mM EDTA pH 8 (lysis dierences between the AKT isoforms have been found buer). Lysates were immune precipitated with and washed (Konishi et al., 1995; Brodbeck et al., 1999). three times in lysis buer. Protein was eluted from the beads Importantly, our results show that the AKT3 isoform with SDS sample buer (Novex) containing 10% b- mercaptoethanol at 378C. Proteins were separated by SDS ± exhibits high basal activity in two human glioma cell PAGE and Western blotted in 5% blocking solution lines that lack endogenous MMAC1 expression. As (BioRad) in 50 mM Tris pH 8, 150 mM NaCl, 0.05% MMAC1 has been implicated in the etiology of both Tween-20. HRP-linked Protein A was used with ECL glioblastoma multiforme and prostate cancer, it is (Amersham) as a detection agent. interesting to note that AKT3 is also constitutively active in prostate cancer cells lines (Nakatani et al., 1999). In contrast to AKT3, the majority of AKT1 Mouse tissues activity is serum inducible, suggesting that these Organs were harvested from normal mice, Dounce homo- isoforms are dierentially regulated. Lastly, our genized in lysis buer, and centrifuged for 10 min at 14 000 ®nding that the AKT isoforms have distinct tissue r.p.m. at 48C three times. Lysates were then incubated with speci®c patterns of expression suggests that the formalin ®xed S. aureus (Calbiochem) for 1 h at 48C and isoforms have unique functions that result in cell type centrifuged for 10 min at 14 000 r.p.m. at 48C three times. speci®c eects downstream of MMAC1. Equivalent amounts of total protein (2 mg) from each tissue Results from in vivo experiments indicate a role for was used for the anti-ILK, anti-AKT and anti-MMAC1 MMAC1 in cell ± cell and cell ± matrix interactions and immune precipitates. motility. Mouse embryos lacking MMAC1 exhibit severely disorganized blastocysts (Stambolic et al., 1998), suggesting that expression of MMAC1 is AKT kinase assays required for proper cell ± cell signaling. Furthermore, Protein lysates were incubated with the appropriate AKT mutations in MMAC1 predominate in advanced antibody for at least 2 h at 48C. Protein A-agarose beads invasive tumors (Li et al., 1997; Steck et al., 1997). (UBI) were added for an additional 2 h incubation. Immune These ®ndings suggest a role for MMAC1 in metastatic precipitates were washed three times with 25 mM HEPES progression, consistent with its putative role in pH 7, 1 M Nacl, 0.1% BSA, 10% Glycerol, 1% Triton X-100 and once with 20 m HEPES pH 7, 10 mM MgCl ,10mM adhesion and motility. Interestingly, ILK overexpres- M 2 MnCl2, 0.2 mM EGTA (kinase buer). Beads were resus- sion resulted not only in cell transformation, but also pended in kinase buer containing 1 mM DTT, 5 mM ATP, in increased metastasis in in vitro models (Hannigan et 10 mCi[g-32P]ATP and 500 ng histone H2B (Boehringer al., 1996; Radeva et al., 1997; Delcommenne et al., Mannheim). Reactions were incubated at 308C for 30 min 1998; Wu et al., 1998), consistent with the possibility and terminated by the addition of 26SDS sample buer that ILK may mediate many of the tumor suppressor containing 10% b-MeOH. Kinase assay reactions were functions of MMAC1. quantitated using a phosphoimager and ImageQuant software, Molecular Dynamics.

Materials and methods ILK kinase assays Cells Cells were resuspended in lysis buer and incubated with A172 and U373 cells were from ATCC. U373 cells stably 4 mg of ILK antibody and protein A agarose for 2 h at 48C. expressing wild type MMAC1 or C124S MMAC1 were Beads were washed three times with lysis buer and once isolated as described previously (Morimoto et al., 1999). Cells with 50 mM HEPES pH 7, 10 mM MnCl2,10mM MgCl2,

Oncogene MMAC1 inhibits PLC and ILK AM Morimoto et al 208

2mM NaF, 1 mM Na3VO4 (ILK kinase buer). Beads were temperature in calcium free HBSS. After labeling, cells resuspended in ILK kinase buer containing 5 mM cold ATP, were washed and resuspended in calcium free HBSS 10 mCi[g-32P]ATP and 30 mg myelin basic protein (UBI). The supplemented with 20 mM HEPES buer. Cells were reactions were incubated at 308C for 20 min and terminated preincubated with 5 mM EDTA for 5 min and/or 50 mM by the addition of 26SDS sample buer. The kinase assay LY294002 (Calbiochem) for 30 min and then stimulated reactions were quantitated as described above. with media containing 20% serum or PBS and subsequently incubated with a molar excess of calcium chloride. Measurement of calcium ¯ux was performed using a FACSVantage (Becton Dickinson, Mountain View, CA, IP3 binding assays USA). Cells were incubated in lysis buer for 10 min on ice and 0.2 volumes of ice cold 20% perchloric acid were added. Lysates were incubated an additional 20 min on ice and centrifuged at 14 000 r.p.m. for 15 min at 48C. Supernatants were neutralized with ice cold KOH to a pH of 7.5 and centrifuged for 15 min at 14 000 r.p.m. at 48C. Supernatants were

equalized for protein concentration and IP3 levels were Acknowledgments

quantitated using an IP3 binding assay (Amersham We are grateful to Madeline Fort for providing mouse TRK1000) as per manufacturer's instructions. tissues and thank the FACS facility for help with the calcium ¯ux assays. We also thank Emma Lees and Jing Wang for helpful comments on the manuscript and Calcium ¯ux assays Maribel Andonian and Gary Burget for graphics support. U373 cells (2.56106) were incubated with 1 mM indo-1AM DNAX Research Institute is fully supported by Schering- (Molecular Probes, Eugene, OR, USA) for 30 min at room Plough Corporation.

References

Aoki M, Batista O, Bellacosa A, Tsichlis P and Vogt LiawD,MarshDJ,LiJ,DahiaPLM,WangSI,ZhengZ, PK. (1998). Proc. Natl. Acad. Sci. USA., 95, 14950 ± BoseS,CalKM,TsouHC,PeacockeM,EngCand 14955. Parsons R. (1997). Nature Genet., 16, 64 ± 67. Baltuch GH and Yong VW. (1996). Brain Res., 71, 143 ± 149. MarshDJ,DahiaPLM,ZhengZ,LiawD,ParsonsR,Gorlin Brodbeck D, Cron P and Hemmings BA. (1999). J. Biol. RJ and Eng C. (1997). Nat. Genet., 16, 333 ± 334. Chem., 274, 9133 ± 9136. Maehama T and Dixon J. (1998). J. Biol. Chem., 273, 13375 ± Cheney IW, Johnson DE, Vaillancourt M-T, Avanzini J, 13378. Morimoto A, Demers GW, Wills KN, Shabram PW, Bolen Myers MP, Stolarov JP, Eng C, Li J, Wang SI, Wigler MH, JB, Tavtigian SV and Bookstein R. (1998). Cancer Res., 58, Parsons R and Tonks NK. (1997). Proc. Natl. Acad. Sci. 2331 ± 2334. USA, 94, 9052 ± 9057. Corvera S and Czech MP. (1998). Trends in Cell Biol., 8, MyersM,PassI,BattyIH,VanderKaayJ,StolarovJP, 442 ± 446. Hemmings BA, Wigler MH, Downes CP and Tonks NK. Delcommenne M, Tan C, Gary V, Rue L, Woodgett J and (1998). Proc. Natl. Acad. Sci USA, 95, 13513 ± 13518. Dedhar S. (1998). Proc. Natl. Acad. Sci. USA., 95, 11211 ± Morimoto AMM, Berson AE, Fujii GH, Steck PA, Tavtigian 11216. SV, Bookstein R and Bolen JB. (1999). Oncogene, 18, Downward J. (1998). Science, 279, 673 ± 674. 1261 ± 1266. FalascaM,LoganSK,LehtoVP,BaccanteG,LemmonMA Nakatani K, Thompson DA, Barthel A, Sakaue H, Liu W, and Schlessinger J. (1998). EMBO J., 17, 414 ± 422. Weigel RJ and Roth RA. (1999). J. Biol. Chem. 274, Furnari FB, Lin H, Su Huang H-J and Cavenee WK. (1997). 21528 ± 21532. Proc. Natl. Acad. Sci., 94, 12479 ± 12484. Radeva G, Petrocells T, Behrend E, Leung-Hagesteijn C, Guldberg P, Straten P-t, Birck A, Ahrenkiel V, Kirkin AF Films J, Slingerland J and Dedhar S. (1997). J. Biol. Chem., and Zeuthen J. (1997). Cancer Res., 57, 3660 ± 3663. 272, 13937 ± 13944. Haas-Kogan D, Shalev N, Wong M, Mills G, Yount G and RamehLE,RheeSG,SpokesK,KazlauskasA,CantleyLC Stokoe D. (1998). Curr. Biol., 8, 1195 ± 1198. and Cantley LG. (1998). J. Biol. Chem., 273, 23750 ± Hannigan GE, Leung-Hagesteijn C, Fitz-Gibbon L, Coppo- 23757. lino MG, Radeva G, Filmus J, Bell JC and Dedhar S. Rhee SG and Bae YS. (1997). J. Biol. Chem., 272, 15045 ± (1996). Nature, 379, 91 ± 96. 15048. Kong D, Suzuki A, Zou T-T, Sakurada A, Kemp LW, Ronnstrand L, Siegbahn A, Rorsman C, Johnell M, Hansen Wakatsuki S, Yokoyama T, Yamakawa H, Furukawa T, K and Heldin CH. (1999). J. Biol. Chem., 274, 22089 ± Sato M, Ohuchi N, Sato S, Yin J, Wang S, Abraham JM, 22094. Souza RF, Smolinski KN, Meltzer SJ and Horii A. (1997). Steck PA, Pershouse MA, Jasser SA, Yung WKA, Lin H, Nat. Genet., 17, 143 ± 144. Ligon AH, Langford LA, Baumgard ML, Hattier T, Davis KonishiH,KurodaS,TanakaM,MatsuzakiH,OnoY, T, Frye C, Hu R, Swelund B, Teng DH-F and Tavtigain Kameyama K, Haga T and Kikkawa U. (1995). Biochem. SV. (1997). Nature Genet., 15, 356 ± 362. Biophys. Res. Comm., 216, 526 ± 534. Suzuki A, de la Pompa JL, Stambolic V, Elia AJ, Sasaki T, Khoshyomn S, Penar PL, Rossi J, Wells A, Abramson DL del Barco Barrantes I, Ho A, Wakeman A, Itie A, Khoo W, and Bhushan A. (1999). Neurosurg., 44, 568 ± 577. Fukumoto M and Mak TW. (1998). Curr. Biol., 8, 1169 ± Li D-M and Sun H. (1997). Cancer Res., 57, 2124 ± 2129. 1178. Li D-M and Sun H. (1998). Proc. Natl. Acad. Sci. USA., 95, Stambolic V, Suzuki A, de la Pompa JL, Brothers GM, 15406 ± 15411. MirtsosC,SasakiT,RulandJ,PenningerJM,Siderovski Li J, Yen C, Liaw D, Podsypanina K, Bose S, Wang SI, Puc J, DP and Mak TM. (1998). Cell, 95, 29 ± 39. Miliaresis C, Rodgers L, McCombie R, Bigner SH, Tamura M, Gu J, Matsumoto K, Aota S, Parsons R and Giovanela BC, Ittman M, Tycko B, Hibshoosh H, Wigler Yamada KM. (1998). Science, 280, 1614 ± 1617. MH and Parsons R. (1997). Science, 275, 1943 ± 1947.

Oncogene MMAC1 inhibits PLC and ILK AM Morimoto et al 209 Turner T, Epps-Fung MV, Kassis J and Wells A. (1997). Wu C, Keightley SY, Leung-Hagesteijn C, Radeva G, Clin. Cancer Res., 3, 2275 ± 2282. Coppolino M, Goicoechea S, McDonald JA and Dedhar Vassbotn FS, Ostman A, Langeland N, Holmsen H, S. (1998b). J. Biol. Chem., 273, 528 ± 536. Westermark B, Heldin C-H and Nister M. (1994). J. Cell Yano S, Tokumitsu H and Soderling TR. (1998). Nature, Physiol., 158, 381 ± 389. 396, 584 ± 587. Wu X, Senechal K, Neshat M, Whang YE and Sawyers CL. (1998a). Proc. Natl. Acad. Sci. USA, 95, 15587 ± 15591.

Oncogene