The tumor suppressor PP2A Aβ regulates the RalA GTPase

Suppression of PP2A Aβ expression transforms human cells

To assess the role of PP2A Aβ structural subunit in human cell transformation, we suppressed Aβ expression in immortal, non-tumorigenic HEK cells expressing LT, hTERT, and H-RAS (HEK TER) cells using a short hairpin RNA (shRNA) specific for Aβ (shAβ). Introduction of shAβ into HEK TER cells overexpressing FLAG-tagged PP2A Aβ led to suppression of the ectopically expressed Aβ (Figure 1A). We then generated HEK TER cell lines that stably harbored either shAβ (HEK TER-shAβ) or a control shRNA targeting GFP (shGFP) (HEK TER-shGFP). Expression of shAβ in HEK TER cells reduced Aβ mRNA levels by 78±6% compared to those found in control cells (Figure 1B).

In parallel, we generated a HEK TER cell line expressing shRNA targeting the other structural PP2A A subunit (Aα) (Figure 1C). Suppression of the PP2A Aα subunit resulted in a substantial reduction of the levels of the PP2A B55α and B56ε regulatory subunits and the catalytic Cα subunit (Figure 1C, left panels). These observations corroborate prior work that demonstrated that the PP2A regulatory and catalytic subunits exhibit limited stability when not bound to a structural PP2A A subunit (Silverstein et al., 2002). In contrast, suppressing PP2A Aβ expression failed to alter the levels of these regulatory and catalytic PP2A subunits (Figure 1C, right panels). In consonance with these findings, we observed that PP2A-attributable phosphatase activity was decreased by 80±6% or 13±3% in cells expressing shAα or shAβ, respectively, compared to control shGFP cells (Figure 1D). These findings confirm previous observations (Hemmings et al., 1990; Zhou et al., 2003) that concluded that PP2A complexes containing Aβ represent a small fraction of the total PP2A complexes.

We next examined the effects of suppressing either PP2A Aα or Aβ on proliferation, anchorage-independent (AI) growth and tumor formation. Consistent with our prior work (Chen et al, 2005), we found that HEK TER cells expressing very low levels of Aα proliferated poorly and failed to grow in an AI manner or to form tumors (Figure 1E,F, Supplemental Figure 1D). In contrast, suppression of the PP2A Aβ subunit induced a small increase in the rate of proliferation (Figure 1E), conferred the ability to grow in an AI manner, and permitted these cells to form tumors (Figure 1F, Supplemental Figure 1A).

To exclude the possibility that these observations were due to off-target effects of RNAi, we used a second shRNA that targets a different region of the PP2A Aβ mRNA (shAβ-2). This second shRNA also efficiently suppressed Aβ expression in HEK TER cells and permitted these cells to grow in an AI manner (Supplemental Figure 1B,C). Overexpression of a PP2A Aβ allele resistant to suppression by shAβ (Supplemental Figure 1D) reversed the observed transformed phenotype (Figure 1G, 2B). Moreover, we found that suppression of Aβ expression in immortal, human small airway epithelial cells expressing hTERT, LT, and H-RAS (AATER cells) (Lundberg et al., 2002) also permitted AI growth (Figure 1G), indicating that the observed effects of suppressing Aβ are not restricted to a particular cell type. Together, these results suggest that loss of PP2A Aβ expression contributes directly to cell transformation.

Cancer-associated PP2A Aβ mutants are functionally defective

Mutations in PP2A Aβ are found in lung, breast and colon cancers and include missense mutations involving P65S, L101P, V448A, V545A, G8R, K343E and L101P/V448A (Wang et al., 1998). In addition, one PP2A Aβ_. mutation involves the deletion of exon 9, resulting in an in-frame deletion of 45 amino acids (ΔE344-E388). To study the function of these cancer-associated mutants, we generated FLAG-epitope tagged versions of wild-type (WT) PP2A Aβ or cancer-derived Aβ mutants. We overexpressed these Aβ alleles in HEK TER cells and determined whether such Aβ mutants formed functional PP2A complexes. In Aβ immune complexes, we found that the WT PP2A Aβ subunit formed heterotrimers with the catalytic Cα and each of the regulatory B subunits tested (Figure 2A, Supplemental Figure 2A,B). In contrast, each of the cancer-derived Aβ mutants showed clear defects in their ability to form productive PP2A complexes (Figure 2A, lower panel; Supplemental Figure 2A,B,C), corroborating a prior report (Ruediger et al., 2001a).

We next determined whether these Aβ mutants retained phosphatase activity by measuring PP2A Aβ-attributable phosphatase activity in Aβ immune complexes isolated from HEK TER cells expressing WT or mutant PP2A forms of Aβ (Figure 2A). We found that the phosphatase activity associated with mutant Aβ (P65S, V545A, G8R, K343E or D504G) immune complexes was reduced by 48±5%, 70±7%, 68±9%, 52±4%, or 71±9% respectively, compared to the activity levels found with the WT Aβ subunit. We failed to detect any Aβ-attributable phosphatase activity in cells expressing the Aβ mutants L101P/V448A, ΔE344-E388, L101P, and V448A (Figure 2A). Thus, we concluded that cancer-associated mutations of PP2A Aβ mutants fail to form the full compliment of PP2A complexes and demonstrate impaired mutant-specific phosphatase activity.

Loss-of-function of both Aβ alleles is necessary to transform human cells

To determine whether Aβ mutants exhibit gain of function properties, we measured the total PP2A-specific phosphatase activity in cells overexpressing WT Aβ or Aβ mutants. Introduction of WT Aβ or the P65S or G8R Aβ mutants resulted in a slight increase in total phosphatase activity, while no changes in overall activity were found in cells expressing other Aβ mutants (Supplemental Figure 3). Thus, PP2A Aβ mutants did not disrupt endogenous phosphatase activity by acting as dominantly acting negative alleles. Moreover, introduction of the WT allele of PP2A Aβ or any of the mutant PP2A Aβ alleles into immortalized but non-tumorigenic HEK TER cells failed to transform these cells, confirming that these mutations fail to act as dominant oncogenes (data not shown).

In consonance with these observations, Aβ mutant alleles are usually accompanied by the deletion of the second allele in human tumor samples (Tamaki et al., 2004; Wang et al., 1998). To mimic this situation, we generated WT and mutant alleles of Aβ (RAβ) that are resistant to targeting by shAβ and expressed them in HEK TER-shAβ cells (Supplemental Figure 1D, Figure 2B, lower panel). We found that expression of the WT PP2A Aβ (RAβ) in tumorigenic HEK TER-shAβ cells decreased cell proliferation by 36% (Supplemental Figure 4A). In contrast, expression of each of the Aβ mutants failed to alter substantially the proliferation rate of shAβ cells (Supplemental Figure 4A). Introduction of RAβ in HEK TER-shAβ or AATER-shAβ cells also significantly suppressed AI growth (62±8% and 67±9%, respectively) and reduced the number of tumors formed, whereas introduction of other Aβ mutants showed no inhibitory effects on the transformed phenotype (Figures 1G, 2B, Supplemental Figure 4B). These observations confirmed that each of cancer-associated PP2A Aβ mutants is a functionally null allele and that loss-of-function of both Aβ alleles is necessary to induce a fully transformed phenotype.

Restoration of Aβ function in lung carcinoma cells reverses the tumorigenic phenotype

To investigate the role of PP2A Aβ mutations in patient-derived cancer cell lines, we determined the effects of restoring PP2A Aβ function in lung cancer cell lines that harbor PP2A Aβ mutations. The H1450 and H838 lung cancer cell lines harbor PP2A Aβ subunit mutations (Wang et al., 1998). The H1450 cell line carries one Aβ allele containing an in-frame deletion of residues 717-1583 and a second allele with point mutation D504G, while H838 harbors only one mutant Aβ allele (G8R). To examine whether expression of the Aβ subunit reverses the tumorigenic phenotype, we introduced the WT Aβ allele into these lung cancer cell lines harboring mutant PP2A Aβ (Figure 2C). As control, we used the primary AATER cells and two lung carcinoma cell lines - H1437 that harbors two WT Aβ alleles and H2009 that has one WT Aβ allele and a G90D Aβ allele. When overexpressed, the Aβ G90D allele behaved in a manner indistinguishable from WT Aβ with regard to forming active PP2A complexes and suppressing AI growth (data not shown).

Expression of WT Aβ in the H1450 and H838 cell lines suppressed their doubling rate by 68±8% and 58±9%, respectively (Figure 2C,DTable 1) and reduced the number of AI colonies formed by these cell lines by 77±8% and 42±5%, respectively (Figure 2E). Moreover, expression of WT Aβ into H1450 significantly suppressed tumor growth (Figure 2F). In contrast, introduction of WT Aβ into the H1437, H2009 or AATER cell lines, each of which carries a WT Aβ allele, failed to inhibit cell proliferation, AI growth or tumor growth, suggesting that the decreased cell proliferation and AI growth observed in H1450 and H838 cells was not an artifact induced by overexpression of Aβ (Figure 2C-F, Table 1).

Cell cycle analysis by double 5-bromo,3′-deoxyuridine/propidium iodide (BrdU/PI) staining showed that the introduction of WT Aβ into H838 and H1450 cells increased the percentage of cells present in the G1 phase of the cell cycle (Table 1). In contrast, expression of WT Aβ failed to affect the cell cycle distribution of AATER, H1437 and H2009 cells. In all of these cells, we did not detect any evidence of increased numbers of apoptotic cells after overexpression of WT Aβ (Table 1), suggesting that the decreased proliferative rate induced by WT Aβ in lung carcinoma cell lines harboring mutant Aβ was due to the lengthening of cell cycle rather than apoptosis.

Suppression of PP2A Aα or Aβ induces transformation by distinctive mechanisms

We previously reported that partial suppression of PP2A Aα imparts a tumorigenic phenotype (Figure 3A) by selectively eliminating PP2A B56γ containing complexes (Chen et al., 2004). Although the two distinct PP2A A isoforms, Aα and Aβ, share 86% sequence identity (Hendrix et al., 1993), it remained unclear whether these isoforms share overlapping functions. To determine if Aα and Aβ were interchangeable, we overexpressed Aα in HEK TER cells in which we suppressed Aβ expression (HEK TER-shAβ). Overexpression of PP2A Aα in these cells failed to revert the transformed phenotype induced by Aβ suppression, indicating that the Aα isoform does not functionally substitute for loss of Aβ expression (Figure 3A). Consistent with these observations, we confirmed that only the PP2A Aα isoform forms complexes with ST (Zhou et al., 2003) (Figure 3B, Supplemental Figure 5). These observations indicate that suppression of Aβ induces transformation by a mechanism distinct from that caused by partial depletion of PP2A Aα or by expression of ST.

The small GTPase RalA interacts with PP2A Aβ

To study the mechanism by which PP2A Aβ suppression contributes to transformation, we purified PP2A Aα and Aβ complexes to identify proteins that bind uniquely to Aβ. Specifically, we overexpressed FLAG-epitope tagged forms of Aα and Aβ in 293T cells, isolated the immune complexes formed with these proteins (Figure 3C) and identified proteins that associate with the Aα and Aβ subunits by mass spectrometry (Supplemental Figure 6). We found that the PP2A catalytic Cα and several regulatory B subunits associated with both the Aα- and Aβ-specific complexes, confirming that these conditions permitted the isolation of functional heterotrimers (Figure 3C, Supplemental Figure 6A). Although most proteins identified by this approach interacted with both Aα and Aβ, we found a small number of proteins that bound only to PP2A Aβ. When we attempted to verify these putative Aβ interacting proteins, we found that the small GTPase protein RalA was the only protein that specifically formed complexes with Aβ subunit (Figure 3D, Supplemental Figure 6B).

RalA has been implicated in the regulation of exocytosis, transcription, secretion, cell transformation and tumor progression [reviewed in (Feig, 2003; Feinstein, 2005)]. When we overexpressed HA-tagged-RalA and FLAG-tagged Aβ in 293T cells, we found that immune complexes isolated using either HA- or FLAG-specific antibodies contained both RalA and Aβ (Figure 3E). We also found that endogenous RalA interacts with FLAG-tagged Aβ in 293T cells (Figure 3D) and that both RalA and Aβ were found in membrane fraction of HEK TER-Aβ cells (Supplemental Figure 7A). In contrast, PP2A Aα failed to form complexes with RalA (Figure 3D,E), confirming that RalA binds specifically PP2A Aβ-containing complexes. We note that RalA-Aβ complexes were comprised of 16±5% of the total RalA and 9±2% of the total Aβ proteins (Supplemental Figure 7B). These observations indicate that the interaction between the PP2A Aβ complex and RalA involves a small fraction of the total PP2A Aβ complexes and RalA and suggests that these interactions are likely short-lived as would be expected for interactions between phosphatases and their substrates.

We then determined whether cancer-associated Aβ mutants form complexes with RalA by purifying PP2A Aβ immune complexes from HEK TER cells expressing WT Aβ or cancer-associated Aβ mutants (G8R, V454A, and V448A) representing each of the categories defined in Supplemental Figure 2C. WT Aβ_. but not Aβ point mutants bound endogenous RalA (Figure 3F). This observation makes it unlikely that interaction between WT Aβ and RalA is non-specific and suggests that loss of Aβ function also impairs RalA complex formation.

RalA is necessary for transformation induced by PP2A Aβ suppression

To assess whether RalA expression was required for cell transformation induced by loss of Aβ expression, we suppressed RalA expression using two different RalA-specific shRNAs (shRalA-1 and shRalA-2) (Figure 5A, upper panel). Immunoblotting with a RalA-specific antibody revealed that both shRalA-1 and shRalA-2 efficiently suppressed RalA expression. Suppression of RalA inhibited the proliferation of HEK TER cells expressing shGFP (Figure 4A). In contrast, suppression of RalA failed to affect the proliferation rate of the HEK TER-shAβ, HEK TER-LowAα, and HEK TER-ST cell lines. These findings are consistent with prior observations that showed that RalA is not limiting for serum-dependent proliferation of tumorigenic cells under standard culture conditions (Chien and White, 2003).

To examine the contribution of endogenous RalA to oncogenic transformation, we then tested the consequences of suppressing RalA expression on AI growth. Depletion of RalA inhibited the ability of HEK TER cells expressing shAβ to form AI colonies by 94±5% (Figure 4B). In contrast, suppression of RalA in HEK TER cells expressing ST or in which Aα expression was decreased by 50% (LowAα) only partially inhibited the ability of these cells to form AI colonies (37±6% and 61±6%, respectively) (Figure 4B), consistent with prior observations that showed that oncogenic alleles of H-RAS also activate RalGEFs (Lim et al., 2005). These findings indicate that RalA expression is necessary for the transformed phenotype induced by suppression of Aβ.

PP2A Aβ negatively regulates RalA activity

To determine whether the interaction of Aβ with RalA perturbs RalA function, we examined RalA activity in HEK TER cells in which Aβ expression was suppressed. RalA activity was analyzed by determining the amount of GTP-bound, and therefore active, RalA bound to recombinant Ral-BP1 protein. We found a 4.5-fold increase in GTP-bound RalA levels in cells lacking Aβ compared to control cells expressing shGFP (Figure 4C). Expression of ST or partial suppression of Aα had no effect on the steady state amount of GTP-bound RalA (Figure 4C), consistent with our prior observations that suggested that Aβ induces transformation by a mechanism distinct from that caused by partial inhibition of Aα or ST overexpression.

In contrast, introduction of exogenous PP2A Aβ inhibited RalA activity in both HEK TER and HEK TER-shAβ cells as measured by decreased levels of GTP-bound RalA (Figure 4D). To test whether cancer-derived Aβ_. mutants affected RalA activity, we overexpressed the shRNA-resistant G8R Aβ_. mutant in HEK TER cells lacking Aβ expression. We found that the G8R Aβ mutant that cannot form complexes with RalA (Figure 3F) also failed to affect RalA activity (Figure 4D). Consistent with these findings, the H838 lung cancer cell line that harbors a single mutant Aβ allele exhibited approximately 4-fold higher levels of RalA activity compared to the H2009 carcinoma cell line that harbors WT Aβ_. (Figure 4D). Overexpression of WT Aβ in both H838 and H2009 cell lines decreased the amount of GTP-bound RalA identified using this Ral-BP1 binding assay. Moreover, introduction of WT PP2A Aβ into H838 cells decreased the amount of activated RalA to the level found in H2009 cells (Figure 4D). Together, these experiments indicate that the interaction of Aβ with RalA inhibits RalA activity.

PP2A Aβ dephosphorylates RalA

We next determined whether PP2A Aβ complexes dephosphorylate RalA. Using a phospho-serine specific antibody, we found the accumulation of a serine-phosphorylated form of RalA in cell lines in which Aβ expression was suppressed (Figure 5A). This observation suggested that RalA was a direct substrate for Aβ PP2A complexes.

A prior report indicated that RalA can be phosphorylated by Aurora A kinase at Ser194 (Wu et al., 2005). To identify other potential RalA phosphorylation site(s), we isolated RalA from HEK TER-shAβ cells and performed a mass spectrometric analysis. Although we obtained highly redundant coverage of most of the RalA protein, peptides corresponding to the carboxyl-terminal portion of the protein were underrepresented. This portion of RalA is highly basic and contains a prenylation site. Of the portion of RalA that we were able to assess, we found evidence of RalA phosphorylation at Ser11 and Ser183 (data not shown) but no other serine or threonine residues.

To determine whether any of these serine residues were the target of Aβ-containing PP2A complexes, we replaced each of these three serine residues with alanine and generated HA-tagged S11A, S183A and S194A RalA mutants. We co-expressed these RalA alanine substitution mutants and FLAG-tagged Aβ in 293T cells and determined whether these RalA mutants retained the ability to bind Aβ (Figure 5B). We found that Aβ bound to both WT RalA and the S11A RalA mutant. In contrast, the S183A and S194A RalA mutants showed diminished ability to form complexes with Aβ (Figure 5B). These observations suggested that PP2A Aβ binds to and dephosphorylates RalA at Ser183 and Ser194.

To confirm these observations, we created antibodies that recognize RalA phosphorylated at Ser183/Ser194 by immunizing mice with two KLH-conjugated phospho-peptides (phospho-Ser183 and phospho-Ser194). We found that the sera from one of these immunized mice and several hybridoma supernatants derived from this immunized mouse contained antibodies specific for RalA phosphorylated at Ser183 and Ser194 (Figure 5C, Supplemental Figure 8A-C). Specifically, sera from one of the immunized mice harbored antibodies that recognized both phosphorylated and non-phosphorylated forms of RalA. One hybridoma line derived from this mouse (XL13), bound only phosphorylated RalA.

We then used these phospho-specific antibodies to determine whether Ser183 and Ser194 in RalA were targets of PP2A Aβ. In cells lacking the expression of Aβ, a condition where we previously found evidence for increased RalA activity, these antibodies recognized higher levels of phosphorylated RalA, compared to cells expressing Aβ. To verify the specificity of these antibodies, we treated RalA immune complexes with calf intestinal phosphatase (CIP), which eliminated the RalA species detected by these phosphospecific antibodies without affecting total RalA levels (Supplemental Figure 8A-C). Moreover, when we overexpressed WT RalA and S183A, S194A and S183/194A RalA mutants in HEK TER cells lacking Aβ expression (HEK TER-shAβ), we found that the S183A and S194A RalA mutants showed decreased levels of phospho-RalA compared to WT RalA. Importantly, we failed to detect phosphorylated RalA in cells expressing the double S183/194A RalA mutant with these phosphospecific antibodies (Figure 5C). These observations confirmed that Ser183 and Ser184 in RalA are the targets of PP2A Aβ-containing complexes.

We then analyzed phospho-RalA levels in lung cancer cell lines harboring Aβ mutations. Phosphorylated forms of RalA were undetectable in the H1437 and H2009 cell lines that harbor WT Aβ. In contrast, we found that RalA was phosphorylated in H838 and H1450 cell lines that lack functional Aβ alleles (Figure 5D). Together, these studies strongly implicated Ser183 and Ser194 in RalA as substrates of PP2A complexes containing Aβ.

To determine whether the phosphorylation state of RalA contributes to transformation induced by loss of Aβ function, we created S183A and S194A RalA mutants resistant to the RalA-specific shRNA by introducing silent substitutions into the RalA coding sequence (Supplemental Figure 8D). We overexpressed these shRNA-resistant RalA mutants in HEK TER cells in which both RalA and Aβ were suppressed (HEK TER-shAβ-shRalA). The overexpression of WT RalA or the S11A RalA mutant restored RalA activity while expression of the S194A or S183A RalA mutants was unable to complement the suppression of endogenous RalA (Figure 5E). These observations suggest that PP2A Aβ-mediated dephosphorylation of RalA at Ser183 and Ser194 down-regulates its activity. Consistent with these observations, suppression of RalA in HEK TER-shAβ cells inhibited AI growth and tumor formation, and expression of a shRalA-resistant form of WT RalA or the S11A RalA mutant restored the tumorigenic phenotype caused by Aβ suppression (Figure 5F). In contrast, overexpression of shRalA-resistant S183A or S194A RalA mutants failed to reverse the transforming phenotype induced by loss of Aβ expression. Together, these observations implicate the control of phosphorylation of RalA at Ser183 and Ser194 by PP2A Aβ as a critical regulatory step in RalA function.

PP2A Aα and Aβ: Distinct tumor suppressor complexes

Several PP2A complexes have been implicated in cancer development [reviewed in (Schonthal, 2001)]. The observation that mutations of the PP2A Aβ structural subunit occur in many human tumors suggested that this subunit participates in malignant transformation. Here we show that Aβ loss-of-function cooperates with the expression of LT, hTERT and H-RAS to convert human cells to a transformed phenotype. Moreover, that the introduction of a WT Aβ allele into lung cancer cell lines carrying loss-of-function Aβ mutations partially reversed the tumorigenic phenotype of these cells. These findings, together with prior observations in human tumor specimens (Zhou et al., 2003), implicate PP2A Aβ as a tumor suppressor gene.

Somatic mutations of the PP2A structural Aα subunit have also been reported in human cancers (Calin et al., 2000; Ruediger et al., 2001b). Although these distinct PP2A A isoforms, α and β, are 86% identical (Hendrix et al., 1993), it remained unclear whether these two isoforms are functionally interchangeable (Gotz et al., 2000). We previously showed that near complete suppression of PP2A Aα subunit expression induced cell death by apoptosis while cancer-associated PP2A Aα mutations contributed to human cell transformation by creating a state of haploinsufficiency (Chen et al, 2005). In contrast, loss-of-function of both Aβ alleles leads to human cell transformation. These observations are consistent with prior studies, which showed that mutations of the Aα gene occur on one allele while alterations of the Aβ gene are found in both alleles (Wang et al., 1998). Here we also found that the Aα isoform cannot functionally substitute for Aβ. Consistent with prior observations, we also found that only the Aα isoform interacts SV40 ST, suggesting that the transformed phenotype induced by suppression of Aβ occurs by distinct mechanism from that induced by expression of ST. Collectively, these observations indicate that PP2A complexes containing Aα or Aβ are functionally distinct and contribute to cell transformation by discrete mechanisms (Figure 6).

PP2A Aβ binds and regulates the small GTPase RalA

Here we show that PP2A Aβ interacts with and modulates the activity of the small GTPase RalA. RalA is a multifunctional protein, regulating membrane transport, apoptosis, transcription, cell migration and cell proliferation [reviewed in (Camonis and White, 2005; Feig, 2003)]. RalA activation results in activation of several signaling pathways, including the activation of phospholipase D1 and Src kinase (Goi et al., 2000; Jiang et al., 1995), vesicle transport (Moskalenko et al., 2002), filopodia formation (Ohta et al., 1999), epidermal growth factor-induced cell motility (Gildea et al., 2002) and AI growth (Chien and White, 2003; Lim et al., 2005; Panner et al., 2006). The activation of known RalA effector proteins such as RalBP1 or the exocyst has not previously been directly linked to tumorigenesis. However, other signaling molecules regulated by RalA, such as phospholipase D1, Src, JNK, NF-κB and cyclin D, may contribute to cancer progression (Feig, 2003; Lim et al., 2005).

Consistent with prior reports, we observed that RalA suppression diminished the ability of transformed cells to form AI colonies (Chien and White, 2003; Lim et al., 2005). Strikingly, RalA depletion in cells lacking Aβ completely abolishes AI growth, indicating that RalA is necessary for transformation induced by suppression of Aβ. Furthermore, we found that cancer-derived Aβ mutants fail to form complexes with RalA protein. Since Aβ mutations occur throughout its coding sequence, we speculate that these cancer-associated mutations fail to bind RalA due to disruption of functional heterotrimeric PP2A Aβ complexes rather than through specific interactions with these mutant residues. This model is consistent with recent structural determinations of the PP2A Aα complex (Cho and Xu, 2007; Xu et al., 2006). Functionally, these Aβ mutants thus fail to down-regulate RalA resulting in the increased RalA activity observed in cells lacking functional Aβ_. complexes.

We also found accumulation of RalA phosphorylated at Ser183/Ser194 in cells lacking Aβ expression, suggesting that RalA activity can be modulated by (de)phosphorylation. Prior studies have identified two pathways to activate RalA. The cycling of RalA between GDP and GTP bound forms is regulated by guanine nucleotide exchange factors (GEFs), which promote GTP binding, and GTPase activating proteins, which promote GTP hydrolysis (Feig, 2003; Quilliam et al., 2002). At least four RalGEFs are known which are stimulated by activated RAS. Moreover, several GAPs down-regulate RalA (Emkey et al., 1991). In addition, the elevation of intracellular Ca²⁺ levels and interaction of RalA with calmodulin also enhances the binding of GTP to RalA in vitro [reviewed in (Feig, 2003; van Dam and Robinson, 2006)]. Together, these observations identify phosphorylation of RalA as another mechanism that regulates RalA activity and cell transformation.

A prior report demonstrated that the phosphorylation of the RAS family members Rap1a and Rap1b proteins occurs at C-terminal residues Ser180 or Ser179, respectively. This change in phosphorylation modulates the ability of Rap1a to associate and regulate its downstream targets (Altschuler and Lapetina, 1993; Hu et al., 1999; Quilliam et al., 1991). RalA has also been shown to be phosphorylated at Ser194 by Aurora kinase A (Wu et al., 2005). Here we have identified two additional RalA phosphorylation sites - Ser11 and Ser183. Using alanine substitution Aβ mutants, we identified S183 and S194 as residues dephosphoryalted by PP2A Aβ and required for the interaction of RalA and PP2A Aβ. Although PP2A Aβ is likely to have other substrates in addition to RalA, the regulation of RalA phosphorylation appears to play a critical role in transformation associated with loss of functional Aβ in human cancers. In agreement with these findings, overexpression of Aurora A kinase was shown to increase the amount of GTP-bound RalA, thereby promoting collagen I-induced migration and AI growth in MDCK cells (Wu et al., 2005). Thus, the phosphorylation of RalA at Ser183 and Ser194 may represent a RAS-independent mechanism of RalA regulation; however, further studies are required to identify the kinase(s) that are responsible for RalA phosphorylation at these sites.

PP2A Aβ, RalA and human cell transformation

Several recent studies implicate the activation of RalA in cell transformation. In HEK cells expressing LT, ST and hTERT, substitution of an oncogenic allele of H-RAS with a Ras effector mutant that binds selectively to RalGEFs (RAS V12, G37) led to AI growth. However, in these experiments, the additional expression of Ras mutants that bind either Raf (RAS V12, S35) or PI3K (RAS V12, C40) was necessary to achieve tumorigenicity (Hamad et al., 2002; Rangarajan et al., 2004), indicating that the activation of RalGEFs contributes to but does not suffice to substitute for RAS. Furthermore, overexpression of constitutively-activated forms of RalA (G23V or Q72L) induces AI growth of immortalized HEK cells and human fibroblasts expressing LT, ST and hTERT (Lim et al., 2005). These observations implicate the activation of RalGEFs as an important Ras effector pathway in human cells.

Here we demonstrate that suppression of PP2A Aβ leads to RalA activation, which substitutes for ST expression to induce both AI growth and tumorigenicity in human cells expressing LT, hTERT and H-RAS. Together with our prior observations that implicated PI3K signaling as a target of B56γ suppression or ST expression (Chen et al., 2004; Zhao et al., 2003), these findings confirm that the pathways activated by ST and H-RAS are highly overlapping. Consistent with these observations, constitutively active forms of RalA cooperate with H-RAS or Raf to induce cell transformation (Urano et al., 1996). Indeed, we and others have found that expression of a degradation resistant form of c-Myc (Yeh et al., 2004) or expression of an activated allele of PI3K (Zhao et al., 2003) also substitute for ST in human cell transformation. Further studies are necessary to determine whether these effector pathways induce particular transformation phenotypes that are not adequately assessed in standard in vitro transformation assays or whether the activation of multiple RAS effector pathways is necessary to bypass regulatory feedback loops normally induced by physiological Ras signaling.

Although the specific targets of RalA critical for cell transformation are not yet known, elevated levels of GTP-bound RalA have been reported in pancreatic cancer cell lines (Lim et al., 2005; Lim et al., 2006), and we found that lung cancer cell lines lacking functional PP2A Aβ exhibit elevated levels of activated RalA. Thus, the activation of RalA may represent one critical downstream effector of RAS signaling, one that is regulated by the antagonistic actions of proto-oncogenes such as RAS and tumor suppressor genes such as PP2A Aβ.

These observations establish PP2A Aβ as a serine threonine phosphatase tumor suppressor gene that contributes to cell transformation by regulating the phosphorylation status and activity of RalA. Moreover, since PP2A complexes composed of PP2A Aα and Aβ regulate different signaling pathways, these observations challenge the view that phosphatases act promiscuously and constitutively. Instead these findings suggest that phosphatases such as PP2A regulate specific phosphorylation events necessary for the initiation and/or maintenance of tumors. Thus, elucidating the role of PP2A and other phosphatases in these processes is likely to provide further insights into cell transformation and identify additional cancer targets.

Protein phosphatase activity

The protein phosphatase activity in immune complexes was determined as described (Chen et al., 2004). For measuring phosphatase activity in whole cell lysates, and free phosphate was first removed from cell lysates by passage through a Sephadex G-25 column (Amersham, Piscataway, NJ). Serine/threonine phosphatase activity in these cell lysates was then measured by the capacity to dephosphorylate a synthetic phosphothreonine peptide RRA(pT)VA specific for PP2A. Specificity was confirmed using purified PP2A core enzyme in the presence or absence of 5 nM OA. After 30 min incubation at RT, free phosphate was measured colorimetrically.

Proliferation, AI growth and tumor formation

1×10⁴ cells were plated in triplicate and counted every 4 d. For population doublings, a seeding density of 1×10⁴ cells in a 10-cm plate was used for all cultures. Soft agar and tumorigenicity assays were performed as described (Hahn et al., 1999). AI colonies were counted 4 wk after seeding at 10× magnification. The number of tumors formed was determined 40 d after injection.

RalA Activation Assay

The RalA activation assay was performed using RalA activation kit (Upstate Biotechnology, Lake Placid, NY). Cell lysates (2 mg) were pre-cleared with glutathione agarose (Amersham Biosciences, Piscataway, NJ) and then incubated with 20 μl of RalBP1 agarose for 1 h. The beads were washed, and the samples were subjected to 12% SDS-PAGE, transferred to nitrocellulose membranes and immunoblotted with an anti-RalA antibody (BD Biosciences, San Jose, CA).

Generation of phosphospecific RalA antibodies

To generate phosphospecific antibodies, 6 mice were immunized with two KLH-conjugated phospho-peptides (phospho-Ser183 and phospho-Ser194). After 3 immunizations, murine sera from 5 mice were tested by immunoblotting BSA-conjugated phosphorylated RalA peptides. Hybridomas were generated from one of these mice; 1,000 supernatants were tested by ELISA; and the 13 positive supernatants were then analyzed by immunoblotting RalA phospho-peptides. We used the sera from the positive immunized mice and the supernatant from one positive parental line, XL13.