NatB modulates Rb mutant cell death and tumor growth by regulating EGFR/MAPK signaling through the N-end rule pathways

Despite the prevalence of N-terminal acetylation (Nt-acetylation), little is known of its biological functions. In this study, we show that NatB regulates Rb mutant cell survival, EGFR/MAPK signaling activity, and EGFR signaling-dependent tumor growth. We identify Grb2/Drk, MAPK, and PP2AC as the key NatB targets of EGFR pathway. Surprisingly, NatB activity increases the levels of positive pathway components Grb2/Drk and MAPK while decreases the levels of negative pathway component PP2AC despite these proteins have the same first two amino acids that are recognized by NatB and N-end rule pathways. Mechanistically, we show that NatB regulates Grb2/Drk protein stability through its N-terminal sequences and that Grb2/Drk and MAPK are selectively degraded by the Arg/N-end rule E3 ubiquitin ligase Ubr4, which targets proteins with free N-terminus. In contrast, PP2AC is selectively degraded by the Ac/N-end rule pathway E3 ubiquitin ligase Cnot4 that targets proteins with acetylated N-terminus. These results reveal a novel mechanism by which NatB-mediated Nt-acetylation and N-end rule pathways modulate EGFR/MAPK signaling by inversely regulating the levels of positive and negative components. Since mutation or overexpression that deregulate the EGFR/Ras signaling pathway are common in human cancers and NatB subunits are significant unfavorable prognostic markers, this study can potentially lead to the development of novel therapeutic approaches. Significance Statement Nt-acetylation is often regarded as a constitutive, irreversible, and static modification that is not suited to serve regulatory functions. Our observation that Nt-acetylation by NatB coordinately regulate the levels of positive and negative components of the EGFR/MAPK pathway show that Nt-acetylation and N-end rule pathways can play important roles regulating important signaling pathways. As Acetyl-CoA level, which is influenced by cell metabolism, can be rate limiting for Nt-acetylation, our results also suggest a potentially new mechanism by which cellular metabolic status can regulate growth factor signaling.


Introduction
N-terminal acetylation (Nt-acetylation), which is observed in majority of the proteins in eukaryotes, involves the transfer of an acetyl group from acetyl-CoA to the α-amino group of a protein. Nt-acetylation is catalyzed by the N-terminal acetyltransferases (NATs). There are six NATs, NatA-F, in higher eukaryotes with different substrate specificity [1]. Nt-acetylation neutralizes the positive charge and alters chemical properties of the N terminus of the protein.
At biochemical level, Nt-acetylation has been shown to affect localization, interaction, and/or degradation of specific proteins. However, despite the widespread nature of this modification [2], the biological function of NAT and their target Nt-acetylation is still largely unknown.
Studies of engineered β-galactosidases in yeast has established that the Nt-residuals play critical roles in determining protein stability through the N-end rule pathway [3,4]. Two branches of the N-end rule pathway have been identified: the Arg/N-end rule pathway and the Ac/N-end rule pathway [3,5]. In Arg/N-end rule pathway, nonacetylated destabilizing Ntresidues were recognized by the N-recognins, which induce target protein ubiquitination and degradation. N-recognins for the Arg/N-end rule pathway were found to be the UBR-box containing E3 ubiquitin ligases conserved from yeast to mammals. The UBR box is essential for the binding of the N-terminal destabilizing residues and the recognition of the positively charged N-terminal NH3+ group is critical [6]. Interestingly, of the seven UBR box containing E3 ubiquitin ligases found in mammalian genomes, only four (Ubr1, Ubr2, Ubr4, and Ubr5) were found to be N-recognins [3]. On the other hand, cellular proteins with Nt-acetylated residues will be recognized by the Ac/N-recognins, which will target the ubiquitination and degradation of Ntacetylated proteins through the Ac/N-end rule pathway. Two Ac/N-recognins Doa10 and Not4 have been identified in yeast. The Doa10 homolog Teb4 has been shown to function as a mammalian Ac/N-recognins [7], suggesting that the Ac/N-end rule pathway is also conserved.
The recognition by Ac/N-recognins was proposed to be conditional, influenced by protein folding and protein complex formation. The conditional nature of degradation by the Ac/N-end rule pathway provides a mechanism for protein quality control and balance the levels of subunits in a protein complex [5].
Because Nt-acetylation is believed to be largely co-translational and without a corresponding deacetylase, Nt-acetylation is often regarded as a constitutive, irreversible, and static modification that is not suited to serve regulatory functions. In addition, as studies of Nend rule pathways have mostly been carried out in yeast using a reporter protein, very little is known about the regulation of endogenous proteins from multicellular organisms by the two branches of N-end rule pathways. Indeed, it is not known whether Nt-acetylation/N-end rule 4 pathways can regulate the signaling output of major signaling pathways. As there are many potential Nt-acetylation targets in a major signaling pathway, it is not clear whether Ntacetylation/N-end rule pathways will regulate different targets that serve positive or negative functions coordinately, which would be expected if Nt-acetylation/N-end rule pathways were to play an important role in regulating a signaling pathway.
Drosophila provides a genetically tractable model with well-conserved genes and important signaling pathways with humans. In Drosophila developing eye discs, inactivation of the fly Retinoblastoma (Rb) homolog Rbf induces high levels of cell death specifically near the morphogenetic furrow (MF) [8], a region in eye disc where anterior asynchronous progenitor cells arrest in G1 and initiate photoreceptor differentiation. This cell death is mediated by Hid induction and are blocked by mutations that can increase EGFR signaling [9][10][11]. EGFR signaling is activated in the MF and posterior region of the developing eye discs and plays critical roles in preventing cell death in addition to its role in regulating cell cycle and differentiation [12]. Interestingly, reducing EGFR signaling by mutations that function upstream of MAPK preferentially increased apoptosis of rbf mutant cells in posterior eye discs [11,13]. In contrast, mutation of rno, which functions downstream of MAPK and regulates the nuclear output of EGFR signaling [14], did not affect apoptosis of rbf mutant cells. This is consistent with the reported direct regulation of Hid by MAPK [15]. Taken together, these results suggest that rbf mutant cells are more sensitive to reduced level of EGFR/MAPK signaling and that the rbf synthetic lethal screens [13,16,17] can identify additional modulators of EGFR signaling. In this study, we report new alleles of NatB subunits in rbf synthetic lethal screen and identify a novel mechanism by which NatB modulates EGFR signaling by regulating the levels of positive and negative components of the pathway coordinately through the two branches of the N-end rule pathway.

Inactivation of NatB subunits induced synergistic cell death with loss of rbf in developing
Drosophila tissues.
In a genetic screen to identify Rbf synthetic lethal mutations on Drosophila chromosome 3R, we identified four mutants (AB, AE, CJ, and EQ) that fall into the same complementation group. While clones of FRT control or single mutant (paled patches) were readily detected in adult eyes, double mutant clones of rbf and these mutants were not detectable ( Fig. 1A-D, Fig.   S1I-J). To better visualize the loss of double mutant tissues, we generated single or double 5 mutant clones in the Minute background, which allows large mutant clones to be generated due to competitive growth advantage [18]. Indeed, large white patches of AE, EQ, AB, and CJ single mutant clones were observed in Minute background (Fig. S1A, C, E, G). Interestingly, most of the white patches were also lost in conjunction with rbf mutation, leading to the development of much smaller adult eyes (Fig. S1B, D, F, H). These results strongly support the notion that the identified mutants induce synthetic lethality in conjunction with rbf mutation.
Genetic mapping using 3R deficiencies was carried out. We found that the AE complementation group mutants map to the 92B4-C1 cytological region. The following results demonstrate that the AE complementary group mutants are alleles of psid. First, both of the two existing psid alleles, psid-D4 (psid 55D4 ) and psid-D1 (psid 85D1 ) [19,20], failed to complement the AE group mutants; Second, both psid-D4 and psid-D1 showed loss of mutant clones in the presence of rbf mutation, similar to that of AE and EQ mutants ( To demonstrate that psid mutation induced synthetic lethality with rbf, we used anti activated caspase 3 antibody to determine cell death in the developing discs. rbf mutant cells showed significantly increased cell death in the morphogenetic furrow (MF) region but not much in the posterior of the developing eye disc [8,9]  Psid is the regulatory subunit of N-terminal acetyltransferase B (NatB). It binds to the catalytic subunit NAA20 to catalyze the addition of acetyl group to the N terminus of proteins that start with MD, ME, MN, or MQ. Knockdown of NAA20 by RNAi in conjunction with Rbf RNAi also led to significantly decreased adult eye sizes ( Fig. 1G-J), which was correlated with significantly increased cell death in eye/antenna discs (Fig. S1M-O). Taken together, these data suggest that inactivation of NatB induced synergistic cell death with loss of rbf.
NatB regulates MAPK activation at multiple points downstream of EGFR.

6
Our previous studies suggest that mutations such as TSC2 and axin induce synergistic cell death with rbf most strongly in anterior eye discs due to induction of excessive cellular stress [16,17,21]. In contrast, mutations that cause deficiency in EGFR/MAPK signaling induce synergistic cell death with rbf mainly in posterior eye discs [13]. The observed psid, rbf synergistic cell death in posterior eye discs prompted us to determine whether psid mutation affect EGFR/MAPK signaling. EGFR signaling is upregulated in posterior eye discs, which play important roles regulating cell cycle, cell survival, and differentiation [12,22]. Indeed, psid mutation significantly decreased EGFR signaling as shown by reduced Aos-lacZ reporter expression and reduced level of active Diphosphorylated ERK (pERK) in psid mutant clones 2E-E"). As the Psid S678D mutant is defective in binding to NAA20 while the Psid S687A mutant retains the ability to bind NAA20 [20], our results suggest that the reduced MAPK activation in psid mutant clones is mediated by reduced NatB activity. In support of this, knockdown of NAA20 in eye discs also significantly reduced EGFR signaling as shown by reduced Aos-lacZ  3L). To demonstrate that the observed inhibition of activated EGFR-induced pERK level is mediated by the loss of NatB activity, the ability of expressing WT or mutant Psid proteins to rescue EGFR-induced pERK was determined. The WT and S687A mutant form of Psid that can bind NAA20 were able to restore EGFR-induced pERK levels in psid mutant clones ( Fig. 3G-H') while the NAA20 binding defective Psid S678D mutant failed to rescue ( Fig. 3I-I'). Furthermore, knockdown of NAA20 by RNAi also strongly inhibited activated EGFR-induced pERK (Fig. 3J-K'). Taken together, these results suggest that NatB modulates the activities of multiple targets that function between EGFR and Raf to regulate MAPK activation.
To determine whether inhibition of EGFR signaling induced MAPK activation contributed to the synergistic cell death of rbf, psid double mutant cells, we tested the ability of activated EGFR, Ras, or Raf to rescue rbf, psid double mutant cell death. While activated EGFR failed to rescue rbf, psid cell death, activated Ras induced partial rescue and activated Raf rescued cell death to background levels ( Fig. 3M-Q). These results support the notion that reduced MAPK activity in psid mutant clones contributed to the synergistic cell death with rbf.

psid mutation inhibited activated EGFR-induced tumor growth
Activation of EGFR signaling in conjunction with scrib mutation induces cephalic and male gonadal tumors when Ey-FLP was used to induce MARCM clones [24][25][26]. Interestingly, S3H-H'). Furthermore, expressing Psid WT or Psid SA but not the NAA20 binding defective Psid SD mutant restored EGFR CA scrib tumor growth in psid mutant background (Fig. S3F). These results suggest Psid-regulated NatB activity is require for activated EGFR-induce tumor growth.

NatB stabilizes Drk, which promotes EGFR induced MAPK activation
Since NatB catalyze the addition of acetyl group to the N terminal proteins that starts with MD, ME, MN, or MQ, we analyzed components of the EGFR/MAPK pathway and identified three potential NatB targets, Drk (fly Grb2 homolog), PP2AC (catalytic subunit of PP2A), and 8 Sprouty (Spry), which function between EGFR and Raf. Two of these three proteins, Drk and PP2AC, showed high sequence conservation in the first two amino acid, the N-terminal region, as well as the overall protein. The high level of sequence conservation suggest that the regulation of these two proteins by NatB/N-end rule pathways and the overall function of these proteins are likely conserved. Therefore, we initially focused on Drk and PP2AC.
We first characterized an antibody against Drk [27]. We found Drk RNAi clones

Psid regulates Drk through its N-terminal sequences.
Since NatB proteins modify Protein N-terminus, which may affect protein stability [3], we directly tested this idea by expressing a GFP protein tagged with Drk N-terminal 10 amino acids (N-Drk-GFP) together with the β-gal control using the MARCM system. The ratio of GFP to βgal levels were determined in WT or psid mutant background and compared. In comparison to the WT control, mutation of either psid D1 or psid D4 significantly decreased levels of N-Drk-GFP  [28,29]. These observations suggest that PP2A has an overall negative effect on RTK/Ras induced MAPK activation in vivo.
We first used PP2AC RNAi to identify an antibody that can recognize the endogenous fly protein. As shown in Fig. 6E, PP2AC RNAi significantly decreased PP2AC protein signal, indicating that this PP2AC antibody can specifically detect the endogenous protein.
Interestingly, MARCM clones with either psid D1 or psid D4 mutation showed increased PP2AC protein levels ( Fig. 6A-B'). The increased PP2AC protein was dependent on the psid mutation since expression of WT Psid blocked the increased PP2AC levels (Fig. 6C-C'). Furthermore, knockdown of NAA20 also significantly increased PP2AC levels (Fig. 6D-D'). These results showed that inactivation of NatB activity increased the levels of PP2AC, the catalytic subunit of PP2A that negatively regulates EGFR/Ras-induced MAPK activation.
Overexpression of PP2AC in mouse heart was shown to increase PP2AC level, reduce the phosphorylation level of its targets, and impair cardiac function [30]. To determine whether increased PP2AC levels contribute to reduced MAPK activation in psid mutant clones, we tested effects of knockdown PP2AC in psid mutant clones. Since PP2AC knockdown induced significantly levels of cell death, baculovirus p35 protein, which does not affect pERK levels when expressed alone (Fig. 6J), was expressed in conjunction with PP2AC knockdown.
PP2AC-RNAi cells, which were observed mostly in the basal region of the eye disc even in the presence of p35 expression, showed increased pERK levels (Fig. 6M-M"'). This result is consistent with the previous finding that PP2AC has an overall negative effect on RTK/Ras induced MAPK activation [28]. Importantly, psid mutant cells with PP2AC knockdown showed increased pERK levels, in contrast to the reduced pERK levels observed in psid mutant clones

PP2AC and Drk are regulated by distinct branches of the N-end rule pathways
Since PP2AC levels are increased when NatB activity was inactivated, we hypothesized that the N-terminally acetylated PP2AC but not the N-terminally unmodified PP2AC is preferentially degraded. N-terminally acetylated proteins are degraded by the Ac/N-end rule pathway mediated by Doa10 or Not4 E3 ubiquitin ligases in yeast and the Doa10 homolog Teb4 in mammalian system [5]. The Doa10/Teb4 and Not4 homologs in Drosophila are CG1317 and Cnot4, respectively. We generated an allele of fly Teb4 that deleted a significant portion of the open reading frame including the initiation ATG (Fig. S5A). Clones of cells with teb4/CG1317 mutation were quite small and did not affect PP2AC levels ( Fig. S5B-B'). These results suggest that Teb4/CG1317 does not significantly affect PP2AC degradation. On the other hand, inactivating the fly Not4 E3 ubiquitin ligases homolog Cnot4 with RNAi significantly increased levels of PP2AC protein (Fig. 6F-F') but not PP2AC mRNA (Fig. S5C). Additionally, overexpression of Cnot4 significantly decreased the basal levels of PP2AC (Fig. 6G-G'). These results suggest that Cnot4 but not Teb4/CG1317 contribute to the degradation of PP2AC.
The above results showed that PP2AC is degraded by the Ac/N-end rule E3 ubiquitin ligase Cnot4 while Drk is degraded by the Arg/N-end rule pathway E3 ubiquitin ligase Ubr4. We further determined whether PP2AC could also be significantly affected by the Arg/N-end rule pathway and whether Drk significantly affected by the Ac/N-end pathway. Ubr4 knockdown did not significantly affect PP2AC levels ( Fig. 5J-J') despite significantly increased the levels of Drk ( Fig. 5F-F'). In addition, knockdown of Ubr1 and Ubr5, the other two fly N-recognins of the Arg/N-end rule pathway, did not significantly affect PP2AC levels either (Fig. S5D-E'). These results suggest that PP2AC is not significantly affected by the Arg/N-end rule pathway E3 ubiquitin ligases. On the other hand, overexpression of Cnot4 did not significantly affect Drk levels despite significantly reduced PP2AC levels ( Fig. 6G-H'). In addition, mutation of Teb4/CG1317 did not significantly affect Drk levels either (Fig. S5F). These results show that Drk protein level was not significantly affected by the Ac/N-end rule E3 ubiquitin ligases.
Taken together, our results show that PP2AC is preferentially targeted for degradation by the Ac/N-end rule pathway E3 ubiquitin ligase Cnot4 while Drk is preferentially targeted for degradation by the Arg/N-end rule pathway E3 ubiquitin ligase Ubr4.

NatB inactivation decreased the level of MAPK, another positive component of the EGFR/MAPK pathway.
The above results showed that NatB inactivation inhibited EGFR/MAPK pathway by decreasing the levels of the conserved positive regulator Drk and increasing the levels of the conserved negative regulator PP2AC. To gain better understanding of how NatB/N-end rule pathway modulate EGFR/MAPK signaling, we characterized the effect of NatB on two additional potential targets, Spry and MAPK. MAPK has high level of overall sequence conservation except the N-terminal region. Interestingly, a previous study showed that MAPK is degraded by the Poe/Ubr4 E3 ubiquitin ligase, suggesting that MAPK is regulated by the Arg/N end rule pathway [31]. Therefore, it is very interesting to determine whether NatB affects MAPK levels.
We found that inactivation of NatB with psid mutation moderately reduced MAPK levels in wing disc (Fig. 7C), using an anti-MAPK antibody that can detect the endogenous MAPK protein (Fig. 7F). In addition, significant reduced MAPK protein levels were also observed in both psid D1 or psid D4 mutant clones in fat body (Fig. 7A-B'). Therefore, NatB also regulates MAPK levels. In support of the previous report that MAPK is degraded by Ubr4, knockdown of Ubr4 with p35 expression significantly increased MAPK levels (Fig. 7D-D'), while p35 expression alone or knockdown of Ubr1 or Ubr5 had no effect (Fig. 7G-I'). In addition, knockdown of Ubr4 also blocked psid mutation induced-decrease in MAPK levels ( Fig. 7C-E').
Taken together, NatB activity also increase the level of MAPK, another positive component of the EGFR signaling pathway.
In contrast to the highly conserved sequences of Drk and PP2AC, Spry has low overall sequence conservation except in a cysteine-rich region near the C-terminus [32]. Staining with an antibody against Spry [32] detected elevated signals in posterior eye discs that was significantly reduced by Spry RNAi (Fig. S6A-A', white arrowheads) and background signals in other parts of eye discs that was not affected by Spry-RNAi (Fig. S6A-A', yellow arrowheads). This is consistent with the observations that Spry expression is dependent on RTK signaling [32,33]. Decreased levels of Spry was detected in psid mutant clones (Fig. S6B). Since psid mutation reduces EGFR/MAPK signaling, which in turn regulates Spry expression, the observed decreased Spry level in posterior eye disc could be due to reduced EGFR signaling in psid clones. As we could not detect basal levels of Spry in other parts of discs with no RTK signaling, we could not exclude the possibility that NatB may affect Spry protein stability in addition to the role on Spry expression.

Discussion
Our results suggest that NatB activity modulates EGFR signaling by altering the levels of multiple positive and negative components of the pathway. The N-terminus of protein have been shown to influence protein stability through the N-end rule pathway with branches that selectively degrade N-terminally acetylated protein or N-terminally unmodified protein [5].
Interestingly, we found that NatB targets that are highly conserved positive components of EGFR signaling pathway are selectively targeted for degradation by the Arg/N end rule E3 ubiquitin ligase Ubr4 (Fig. 7J). In contrast, NatB target that is a highly conserved negative component of the pathway, PP2AC, is selectively targeted for degradation by the Ac/N end rule 13 E3 ubiquitin ligase Cnot4 (Fig. 7J). As E3 ubiquitin ligases for the two branches of the N-end rule pathway selectively recognize the positively charged non-acetylated N-terminus or the acetylated N-terminus that lacks the positive charge, respectively, NatB activity, which shifts the level of protein Nt-acetylation, will lead to coordinated changes in the levels of positive and negative components of the pathway (Fig. 7J, S7A-B). Therefore, inhibition of NatB activity will results in the loss of N-terminal acetylated Drk/Grb2 and MAPK and the degradation of the Nterminal non-acetylated proteins, leading to decreased levels of positive components (Fig. S7A).
Furthermore, inhibition of NatB activity will also results in accumulation of N-terminal nonacetylated PP2AC and prevents its degradation by Ac/N end rule E3 ubiquitin ligase Cnot4, leading to increased levels of the negative component of the pathway (Fig. S7A). The inverse change in the levels of positive and negative components of the pathway underlie NatB activitymediated regulation of EGFR signaling (Fig. 7J, S7A-B).
Although N-terminal acetylation was generally viewed as a constitutive, co-translational modification that does not have regulatory function, our results suggest N-terminal acetylation by NatB may provide important regulatory function. First, our results show that NatB activity coordinately regulates the levels of positive and negative components of the EGFR signaling pathway. Second, we found that knockdown of Ubr4 in WT background induced significant elevated levels of both Drk/Grb2 and MAPK. As Ubr box proteins specifically recognizes proteins with non-acetylated proteins, these results suggest that significant levels of Drk/Grb2 and MAPK are synthesized without N-terminal acetylation in normal growth conditions and these proteins are targeted for degradation by Ubr4. It is likely that these proteins are partially Nt-acetylation under normal growth conditions, which could potentially be regulated by increasing or decreasing the level of NatB activity or Acetyl-CoA, a key substrate of N-terminal acetylation.
Acetyl-CoA has been suggested as a central metabolite and second messenger of the cellular energetic state [34]. The levels of Acetyl-CoA, which is influenced by the availability of nutrients and growth factors, can vary significantly in cells and influence the levels of protein acetylation by various acetyl transferases [34,35]. Indeed, Acetyl-CoA levels was shown to regulate the N-terminal acetylation of several NatA targets, which modulates the sensitivity of cells to apoptosis [36]. As the Km for NatB from Candida is quite high (around 50 µM) [37], it is likely that N-terminal acetylation by NatB will also be influenced by Acetyl-CoA levels, which will link the regulation of EGFR signaling with cellular energetic status. Therefore, our study suggest a novel mechanism by which the activity of EGFR signaling is regulated by the cellular energetic status.
In addition to Acetyl-CoA level, NatB activity may also be regulated. It was shown that a conserved serine in Psid just downstream of the NAA20 interaction domain regulated NatB complex formation in a phosphorylation-dependent manner (Stephan et al., 2012). Similarly, we found that expressing the phosphorylation resistant but not the phosphorylation mimic form of Psid could rescue the effect of psid mutation on EGFR signaling. These results suggest NatB activity is regulated by cellular signaling pathways through phosphorylation. Taken together, our study suggest that the level of N-terminal acetylation is potentially regulated by cellular signaling and nutrient status, which can in turn regulate the strength of EGFR signaling output by coordinately regulated the levels of multiple positive and negative regulators of the pathway ( Fig. S7A-B). The high sequence conservation of these EGFR signaling components suggest that the mechanisms we uncovered in Drosophila will likely be conserved in mammalian systems. Indeed, a recent study found that inhibition of NatB significantly decreased EGFinduced ERK activation in human liver cancer cells [38]. As mutation or overexpression of EGFR or Ras that deregulate EGFR signaling are quite common in human cancers and NatB subunits are significant unfavorable prognostic markers for human cancers [39][40][41], our results could potentially provide a new strategy to develop therapeutic interventions for these cancers.
The regulation of MAPK activity by NatB described in this study also provides a mechanism by which Nt-acetylation by NatB is required to ensure olfactory receptor neuron (ORN) survival as described previously [20]. It was shown that the phospho-resistant Psid SA mutant rescued the psid cell-loss phenotype while the phosphomimetic Psid SD mutant failed to rescue [20]. In this study, we showed that the phospho-resistant Psid SA but not the phosphomimetic Psid SD was able to rescue psid mutation-mediated MAPK inhibition and rbf psid synergistic cell death.
We showed that psid mutation significantly inhibited dpERK level induced by activated EGFR and Ras but not activated Raf. At first glance, these results do not seem to be consistent with the result that MAPK is also regulated by NatB. The most likely explanation is that the effect of reduced MAPK levels is compensated by the effect of increased levels of PP2AC. It was shown that activation of Raf by Ras-GTP involve activating phosphorylation as well as inhibitory phosphorylation by ERK feedback regulation [42]. It is possible that in experiments involving activated Raf-induced MAPK activation, the main function of PP2A is to remove the inhibitory phosphorylation on Raf. In contrast, in experiments involving activated Ras-induced 15 MAPK activation, the dominant function of PP2A is to remove the activating phosphorylation.
Consistent with this, reducing the PP2AC gene dosage was found to impair signaling from activated Raf but stimulate signaling from activated Ras [28]. Therefore, psid mutation induced increased PP2AC potentially compensated decreased MAPK levels, resulting no obvious inhibition of activated Raf-induced dpERK level.

Drosophila stocks and genetics
The following fly stocks were used in this study: rbf 15aΔ [17] Main genetic technologies used in this study include: FLP/FRT system to generate regular loss of function mosaic clones [43]; MARCM system to generate mosaic clones with both mutation and ectopic expression [44]; UAS/Gal4 and Flp-out or CoinFLP system to induce ectopic expression of RNAi in clones or in whole eyes [45][46][47].

Genetic screen for rbf synthetic lethal mutations and generation of N-Drk-GFP transgenic fly and CG1317/Teb4 deletion allele.
Ethyl methanesulfonate (EMS)-induced mutant screen was carried out as described [17].

16
The CG1317/Teb4 deletion allele was generated by crossing the P-element insertion line (BL20646) with the transposase line ∆2-3. Independent excision lines that have lost eye color were established. Deletions were identified by PCR using P element primers and primers flanking the P element. The breakpoints were determined by sequencing of the PCR products.

Immunostaining, BrdU incorporation, and antibodies
Immunostaining was performed as previously described [13]. For dpErk staining, dissected discs were fixed with 8% formaldehyde in PBS for 1 hr. For PP2AC staining, dissected discs were fixed with 4% formaldehyde in 100mM Lysine for 1 hr on ice, and saponin

Quantification of cell death levels and N-Drk-GFP levels in developing imaginal discs
Cell death level was determined by the percentage of clone area (pixels) that have above background levels of caspase 3 (C3) signal using the Histogram function in Photoshop as described previously [21]. Background level of C3 signal was determined from the adjacent WT tissues that have no apoptosis. The average and standard deviation of percent cell death for each genotype discs was then determined from at least six imaginal discs and then compared.
Two-way student T test was used to determine the significance of statistical differences between different genotypes.
To investigate the effect of psidin mutation on N-Drk-GFP levels, MARCM clones were generated and marked by LacZ expression. LacZ expression levels were determined by anti-βgal staining and used as an internal control. Exposure time of imaging was optimized with the brightest samples and used for all samples. GFP and β-gal signal brightness were calculated using the Histogram function in Photoshop. Normalized GFP levels were showed as relative ratios of signal brightness of GFP to that of β-gal with the normalized GFP level in WT control set as 1. The average and standard deviation of relative GFP levels for each genotype discs was then determined using at least six imaginal discs and compared. Two-way student T test was used to determine the significance of statistical differences between different genotypes.

RNA isolation and quantitative real-time PCR
Total RNA was extracted using TRI reagent (Invitrogen) from about 30 eye/antenna discs dissected from 3rd instar larvae cultured at 25 °C with expression of RNAi constructs driven by eyFLP, Act>CD2>Gal4. cDNA synthesis and qRT-PCR reactions were performed as previously described [48]. Ribosomal protein gene rp49 was used as an internal control in qPCR analysis.
The averages and standard deviations of at least three independent replicates were shown.