Impairment of ubiquitylation by mutation in Drosophila E1 promotes both cell-autonomous and non-cell-autonomous Ras-ERK activation in vivo

Ras activation can promote cell survival, differentiation and growth. Inappropriate activation of Ras by mutation in components of the pathway can lead to developmental syndromes such as Noonan's syndrome (Schubbert et al., 2007; Gelb and Tartaglia, 2006), and activating mutations in Ras are found in approximately 30% of solid tumors (Duursma and Agami, 2003). Normally, Ras can be activated by upstream receptor tyrosine kinases (RTKs) upon ligand binding when the SH2-domain-containing protein Grb2 (drk in Drosophila) recruits the guanine nucleotide exchange factor, son of sevenless (sos), to catalyze the exchange of GDP bound to Ras for GTP, thereby creating active Ras. Ras (Ras85D in Drosophila, referred to hereafter simply as Ras) promotes activation of Raf (pole hole in Drosophila), MEK (Dsor1 in Drosophila) and eventually activation of the MAP kinase ERK (rolled in Drosophila). In Drosophila, active ERK downregulates specific targets. For example, direct phosphorylation of the transcriptional co-repressor Yan (also known as Anterior open) leads to its export from the nucleus and subsequent ubiquitin-mediated protein degradation (Rebay and Rubin, 1995). ERK can activate transcription, both through inactivation of transcriptional co-repressors such as Yan as well as activation of transcription factors such as the ETS-domain-containing protein Pointed (Pnt) (O'Neill et al., 1994; Brunner et al., 1994).

Ubiquitin is a small protein attached to substrate proteins singly or in a polyubiquitin chain to direct those substrates to a variety of fates including endocytosis or degradation. Initially, ubiquitin is `charged' by the ubiquitin-activating enzyme E1 (or Uba1), and then transferred to ubiquitin-conjugating enzymes (E2s) and ubiquitin ligases (E3s), the specificity factors, which directly attach ubiquitin to substrates (Scheffner et al., 1995; Hershko and Ciechanover, 1998). In signal transduction through Ras, the step at or upstream of Ras activation widely accepted to be regulated by the ubiquitin pathway is RTK endocytosis. For example, Egfr is ubiquitylated by the E3 Cbl causing its endocytosis and downregulation (Levkowitz et al., 1999; Yokouchi et al., 1999; Waterman et al., 2000).

In a previous study (Pfleger et al., 2007), we demonstrated in Drosophila melanogaster that hypomorphic mutations in E1 result in over-representation of mutant tissue in a mosaic eye, extra interommatidial cells, and cell death resistance. By contrast, null mutations in E1 cause cell lethality but stimulate the overgrowth of their neighbors and increase the overall size of the eye. We report here that hypomorphic and null mutations in E1 result in a cell-autonomous increase in Ras activation of ERK. Our findings suggest that Ras activation due to impairment of the ubiquitin pathway could occur downstream of RTK regulation and independent of Grb2/sos. Furthermore, we report our interesting observation that null mutation in E1 also causes the non-autonomous activation of Ras in adjacent tissues. The sensitivity of Ras activation to modulation of E1, a non-specific step in the ubiquitin pathway, suggests that maintaining a threshold of ubiquitylation is crucial to prevent inappropriate Ras-ERK activation in vivo by both cell-autonomous and non-cell-autonomous means.

We showed previously that E1 hypomorphic mutant tissue contains extra interommatidial cells and is resistant to cell death (Pfleger et al., 2007). Extra cells can result from failure of the normal program of cell death and/or from ectopic divisions. We report here that E1 hypomorphic mutant cells also undergo ectopic divisions. In a wild-type larval eye disc, anterior to the morphogenetic furrow (MF), cells undergo active division that synchronizes close to the MF. Immediately anterior to the MF, mitotic cells stain for phosphohistone H3 (pHH3) (Fig. 1A), whereas posterior to the MF most cells are differentiating. An additional round of division occurs in a subset of cells posterior to the MF, referred to as the second mitotic wave (SMW), seen as a thin stripe of anti-pHH3 staining. In eye discs homozygous for E1 hypomorphic alleles Uba1^B1 and Uba1^B2 or Uba1^B1/Uba1^B2 heterozygous eye discs (Uba1^B1/Uba1^B2 shown in Fig. 1B), a significant increase in anti-pHH3 staining anterior to the furrow and in the most posterior region was observed. Overall anti-pHH3 staining, indicated by the number of anti-pHH3-positive spots in the entire disc, increased from 62±5 in FRT42D control discs (n=4) to 132±6 (n=5) in Uba1^B1/Uba1^B2 discs (Fig. 1C) In fact, ectopic proliferation posterior to the MF was significant enough that rather than being confined to a thin stripe of the SMW (on average 13.9±0.5 μm, n=4) in FRT42D control discs, the region of frequent anti-pHH3 staining extended two to three times as far into the region posterior to the MF (on average 33.8±1.9 μm, n=5) in Uba1^B1/Uba1^B2 discs.

Close examination of Uba1^B1 homozygous adults revealed occasional overgrowths, such as outgrowths on the leg, proboscis or humeral regions (humeral outgrowth, Fig. 1D) consistent with ectopic divisions. Although these outgrowths were rare (1-2%), they were never observed in homozygous populations of the wild-type parent chromosome, FRT42D (n>1000).

Within the developing Drosophila eye, Ras signaling can promote growth, proliferation, cell fate specification and cell survival. Normally, Ras is activated after ligand binding by RTKs. The RTK Egfr promotes growth and proliferation in the early eye and is later required for survival of postmitotic cells. Absence of Egfr results in failure of proper eye differentiation. By contrast, mutation in the RTK sevenless leads to failure of differentiation of only R7 photoreceptors (Kurada and White, 1999). The ability of Ras to promote different outcomes appears to be dependent on different thresholds of ERK activity (Halfar et al., 2001).

Inappropriate Ras activation could explain the cell death resistance and overgrowth of E1 mutant tissue. Multiple components in the pathway can be evaluated in Drosophila using antibodies and genetic tools (Fig. 2A). In particular, Ras activation leads to dual phosphorylation of ERK (dpERK) (activated ERK), which leads to the ubiquitin-mediated degradation of the transcriptional co-repressor Yan, activation of the ETS transcription factor pointed (pnt), and expression of transcriptional targets such as pnt itself and the high-threshold target argos (aos).

Staining larval eye discs containing clones homozygous for FRT42D (the wild-type FRT42D chromosome without pigment or fluorescent markers) and clones homozygous for FRT42D P[W+, UbiGFP] (a wild-type FRT42D chromosome containing a pigment marker, W+, and a fluorescent marker, UbiGFP) with dpERK antibodies to assess ERK activity reveals clustering of ERK activation in the MF, with diffuse staining in the posterior region of the disc (Gabay et al., 1997; Spencer et al., 1998) (Fig. 2B-D). In eye discs containing clones homozygous for the E1 hypomorphic mutant Uba1^B1 and clones homozygous for FRT42D P[W+, UbiGFP], there was an increase in the size of dpERK clusters in the Uba1^B1 mutant clones (GFP-negative) (Fig. 2E-K). To quantify this increase, we measured the area of dpERK clusters in Uba1^B1 clones (n=13) and FRT42D P[W+, UbiGFP] clones (n=7) from yweyFLP; FRT42D P[W+, UbiGFP]/FRT42D Uba1^B1 mosaic eye discs and in FRT42D clones (n=8) and FRT42D P[W+, UbiGFP] clones (n=14) from yweyFLP; FRT42D P[W+, UbiGFP]/FRT42D discs. dpERK clusters in Uba1^B1 mutant clones were 2.9 times the size of FRT42D control clones (Fig. 2K). No dramatic increase in dpERK staining was seen in Uba1^B1 homozygous clones in the posterior region where staining was diffuse. However, there may be subtle increases (arrowheads in Fig. 2E-G) that are difficult to demonstrate convincingly with this antibody by immunofluorescence.

ERK activation promotes transcription of target genes including pnt (Gabay et al., 1996; Frankfort and Mardon, 2004) and the high threshold target aos (Brodu et al., 2004), which are then expressed at high levels in the posterior region of the eye disc. pnt and aos expression, as assessed by pnt-lacZ and aos-lacZ reporters, occurred in Uba1^B1 homozygous clones (GFP-negative) anterior to the MF, where pnt and aos are not normally observed (Fig. 3A-F, aos; supplementary material Fig. S1, pnt). In yweyFLP; FRT42D P[W+, UbiGFP]/FRT42D Uba1^B1; aos-lacZ/+ mosaic eye discs, no strong aos expression was seen in 21 wild-type clones just anterior to the MF. By contrast, nine of 18 Uba1^B1 clones in this region showed groups of cells with high aos expression. Similarly, in yweyFLP; FRT42D P[W+, UbiGFP]/FRT42D Uba1^B1; pnt-lacZ/+ discs, no strong pnt expression was seen in 14 wild-type clones just anterior to the MF, but three of 11 Uba1^B1 clones in this region showed groups of cells with high pnt expression. Changes in aos posterior to the MF are difficult to detect given the high aos expression. However, confocal scanning at non-saturating intensities revealed increased aos in some mutant clones posterior to the MF compared with wild-type clones (Fig. 3G-I). These data suggest that Ras signaling is increased in mutant clones.

Because the changes in dpERK, pnt and aos are subtle and do not affect every mutant cell, we compared the overall level of dpERK between control FRT42D larvae and larvae homozygous for Uba1^B1 or Uba1^B2. Western blotting for E1 protein (Fig. 2L) showed the expected doublet corresponding to two E1 isoforms for FRT42D and Uba1^B2 larvae (lanes 1 and 3), and only the longer isoform for Uba1^B1 (lane 2) that contains a mutation in the start methionine of the shorter isoform (Pfleger et al., 2007). Uba1^B2 is a temperature-sensitive allele that shows decreased thioester formation with ubiquitin at room temperature, consistent with a partial loss of function, and which adopts the phenotype of complete loss-of-function alleles at 30°C (Pfleger et al., 2007). The dpERK signal was strongly elevated in Uba1^B1 and Uba1^B2 homozygous larvae raised at room temperature compared with the FRT42D larva (Fig. 2M, upper panel). Stripping and reprobing for total ERK (Fig. 2M, lower panel) demonstrated that the increase in dpERK reflects an increase in the proportion of active ERK in mutant larvae.

To determine if Ras activation contributes functionally to the cell-autonomous overgrowth of E1 hypomorphic mutant tissue, we removed one copy of Ras in eyes containing clones homozygous for the mutations Uba1^B1 or Uba1^B2. The over-representation of mutant tissue in these mosaic eyes was notably suppressed for both alleles (shown for Uba1^B2) (Fig. 4A-C). Ras can signal through multiple downstream effectors, including ERK and phosphoinositide-3-kinase (PI3K)/AKT. Western blot analysis revealed no obvious increase in AKT activation in Uba1^B1 or Uba1^B2 mutant larvae compared to FRT42D control larvae (not shown). In contrast to suppression by Ras, removing one copy of AKT had no effect on over-representation of mutant tissue in a mosaic eye (not shown).

E1 hypomorphic mutations result in resistance to cell death caused by overexpression of pro-apoptotic activities such as eiger, grim and hid (Pfleger et al., 2007). Ras signaling promotes cell survival, and the normal cell death observed in the posterior region of the larval eye can be suppressed by Ras-ERK activation (Yang and Baker, 2003). Thus, increased Ras activity in these mutants may explain their cell death resistance. Uba1^B1/Uba1^B2 eyes with only one functional copy of Ras showed decreased suppression of eiger-induced cell death (Fig. 4F) compared with Uba1^B1/Uba1^B2 eyes (Fig. 4E), suggesting that Ras plays a role in resistance to eiger-induced cell death. However, removing one copy of Ras did not eliminate cell death resistance of Uba1^B1/Uba1^B2. Additional factors may contribute, including stabilization of the caspase inhibitor DIAP1, which accumulates in E1 mutant tissue (Pfleger et al., 2007; Lee et al., 2008).

The increase in anti-pHH3 staining of homozygous Uba1^B1 eye discs (not shown) or Uba1^B1/Uba1^B2 eye discs (Fig. 4H) was dominantly suppressed by Ras mutation (Fig. 4I,L). Quantifying overall proliferation revealed an average decrease from 132±6 anti-pHH3 spots for Uba1^B1/Uba1^B2 eye discs (n=5) to 89±7 (n=4) when one copy of Ras was removed (Fig. 4L). Importantly, ectopic proliferation posterior to the MF was decreased and again restricted to a thin stripe (15.9±0.6 μm, n=4) (Fig. 4I), resembling anti-pHH3 staining of FRT42D control discs (Fig. 4G). This suggests that the ectopic proliferation in mutant tissue is largely promoted by Ras.

Ubiquitylation downregulates RTK signaling (Levkowitz et al., 1999; Yokouchi et al., 1999; Waterman et al., 2000). To investigate if the E1 phenotypes result from failure to ubiquitylate upstream RTKs, we examined anti-pHH3 staining in Uba1^B1/Uba1^B2 larval eye discs heterozygous for mutation in the RTK Egfr, the Grb2 homolog drk and the guanine nucleotide exchange factor sos. Surprisingly, removing one copy of Egfr, drk or sos showed no suppression of ectopic divisions (Fig. 4J-L). Increased anti-pHH3-positive cells were still observed (150±18 for removing Egfr, n=6; 139±10 for removing drk, n=7; and 132±10 for removing sos, n=7) and substantial ectopic proliferation still extended posterior to the MF more than twice as far as in a wild-type disc. Furthermore, removing one copy of sos (the component just upstream of Ras, therefore likely to mediate signaling from Egfr as well as other RTKs) did not reduce the over-representation of mutant tissue in a mosaic eye (not shown).

Only a small percentage of Uba1^B1 homozygotes reach adulthood. Mutation in Ras dramatically suppressed this lethality, whereas mutation in Egfr, drk and sos showed little or no effect (effects of Ras, Egfr and drk on Uba1^B1/Uba1^B1 lethality are shown in supplementary material Fig. S2). Together, these data support the argument that Ras is the limiting component in activation of the pathway in these mutants. In other studies, it has been established that phenotypes caused by activating mutations in upstream RTKs including Egfr and sevenless or expression of oncogenic Her2/Neu, are sensitive to the gene dosage of both drk and sos. Because those phenotypes can be dominantly suppressed by mutation in drk (including suppression by the same allele we have used) or sos (Read et al., 2005; Doyle and Bishop, 1993; Olivier et al., 1993; Simon et al., 1991; Settle et al., 2003), Grb2 and sos are limiting components when the pathway is activated at upstream steps. Because the phenotypes we examined were not dominantly suppressed by mutation in drk or sos, these components are not limiting in the E1 mutants. The simplest model to explain our findings, therefore, is that there is a step in Ras signaling affected by mutation in E1 downstream of RTKs and possibly independent of Grb2/sos, potentially Ras itself.

If impaired ubiquitylation of Ras promotes overgrowth in these mutants, we should be able to isolate ubiquitylated Ras from normal cells. Unfortunately, existing antibodies that recognize Drosophila Ras in western analysis do not work well in immunopurification or immunofluorescence experiments. Therefore, we expressed in S2 cells Ras tagged at its N-terminus by a FLAG epitope followed by six histidines (6His) together with ubiquitin tagged at its N-terminus with an HA epitope. Pulldown of Ras on Ni-NTA beads isolated species of Ras recognized in western analysis by both the tag on Ras (FLAG) and the tag on ubiquitin (HA) (Fig. 5). By size, these conjugates represent primarily di-ubiquitylated Ras. We also observed a less frequent population of mono-ubiquitylated Ras (supplementary material Fig. S3). Unavailability of antibodies for immunofluorescence or to immunopurify Drosophila Ras currently precludes analysis of endogenous Ras ubiquitylation in vivo. Our ongoing efforts are developing tagged Ras transgenic lines to address this in the future.

Whereas poly-ubiquitylation branching from lysine 48 of ubiquitin is associated with degradation by the proteasome, mono- and di-ubiquitylation are not associated with ubiquitin-mediated protein degradation (Thrower et al., 2000). Consistent with regulation of Ras by mono- and di-ubiquitylation, we did not see accumulation of Ras in larvae homozygous for Uba1^B1 or Uba1^B2 compared to control larvae (supplementary material Fig. S3).

We isolated a number of E1 mutations that completely lacked biochemical activity when engineered and tested in vitro in our previous study (Pfleger et al., 2007). In particular, Uba1^A1 can encode only a truncated form lacking one ThiF domain and the active site required for formation of a thioester with ubiquitin and therefore can be considered a true null allele. In eyes containing clones homozygous for the null allele Uba1^A1, the homozygous Uba1^A1 cells die but surprisingly stimulate the dramatic overgrowth of neighboring cells. This indicates that null mutation in E1 has non-autonomous effects on growth and/or proliferation (Pfleger et al., 2007). Our findings of non-autonomous overgrowth phenotypes for E1 mutations have now been confirmed by a later report (Lee et al., 2008).

Because null mutation in E1 is cell lethal, very few E1 null clones remain by the third instar stage when created early in eye development. To analyze E1 null clones, we inducibly created clones with a heatshock using hsFLP so they could be examined before they disappeared. Therefore, unlike the mosaic discs presented where the larval eye had grown from a disc containing only homozygous wild-type and homozygous mutant tissue, the following immunohistochemical analysis was performed where only small wild-type and mutant clones were present in otherwise heterozygous discs.

The dpERK pattern was altered in mosaic eye discs containing Uba1^A1 clones. Normally, there is a clear, even column of dpERK clusters in the MF. By contrast, in all ywhsFLP; FRT42D P[W+, UbiGFP]/FRT42D Uba1^A1 discs, the column of dpERK clusters was uneven. dpERK clusters were observed in Uba1^A1 clones (GFP-negative), indicating that Ras can be activated in E1 null tissue. No dpERK clusters were observed anterior to the MF in or adjacent to 48 control FRT42D homozygous clones in ywhsFLP; FRT42D P[W+, UbiGFP]/FRT42D discs. By contrast, of 104 Uba1^A1 homozygous clones anterior to the MF in ywhsFLP; FRT42D P[W+, UbiGFP]/FRT42D Uba1^A1 discs, dpERK clusters were found in seven Uba1^A1 clones or adjacent to 14 Uba1^A1 clones (Fig. 6A-F). Consistent with increased dpERK activation, we saw an increase in transcription of pnt using a pnt-lacZ reporter (supplementary material Fig. S4). These data are consistent with a cell-autonomous increase in Ras signaling through ERK in E1 null clones, as seen in E1 hypomorphic mutants.

As noted, dpERK staining of mosaic eye discs containing Uba1^A1 clones showed strong clusters of dpERK adjacent to 14 of 104 Uba1^A1null clones. We also sometimes saw a ring of dpERK staining surrounding Uba1^A1 clones in the MF (Fig. 6A-C, upper arrow; and data not shown). Typically, high levels of Yan are observed in the posterior region of the eye disc. Activation of ERK promotes Yan degradation. We observed decreased Yan staining in tissue surrounding larger Uba1^A1 clones (Fig. 6G-L) posterior to the MF. Measuring the intensity of Yan staining in tissue adjacent to three distinct Uba1^A1 clones revealed an average decrease by 50%. Correspondingly, in the region just anterior to the MF where aos is not expressed normally, we saw a striking ring of aos expression in tissue surrounding large Uba1^A1 clones (Fig. 6M-R). In ywhsFLP; FRT42D P[W+, UbiGFP]/FRT42D Uba1^A1 discs, no aos expression was seen anterior to the MF in tissue distant from Uba1^A1 clones. By contrast, for 14 of 22 large Uba1^A1 clones in the region of the MF, clear aos expression was seen in adjacent wild-type tissue anterior to the MF. Increased aos was also observed surrounding some mutant clones posterior to the MF (supplementary material Fig. S4). The decrease in Yan levels and increase in dpERK and aos expression surrounding Uba1^A1 clones are consistent with the non-autonomous activation of Ras-ERK in adjacent tissue.

We observed a decrease or absence of aos in many larger Uba1^A1 clones. Given the ERK activation in Uba1^A1 clones, decreased aos expression is perplexing. As noted, the global defect in ubiquitin-mediated protein degradation most likely leads to the failure to degrade transcriptional co-repressors such as Yan within Uba1^A1 clones, as evident by strong Yan staining within these clones (Fig. 6G-L). Thus, high threshold transcriptional targets such as aos may not be expressed despite high ERK activity in these circumstances.

Previously, we saw accumulation of Wingless and activation of JNK in Uba1^A1 clones; however, mutation in wingless or introduction of dominant negative basket, the Drosophila JNK homolog, failed to suppress the overgrowth (Pfleger et al., 2007). We also reported increased Notch and Delta staining in Uba1^A1 clones immediately posterior to the MF, but decreased Notch and Delta in clones in the MF (Pfleger et al., 2007). A later study also reports changes in Notch (Lee et al., 2008). In the non-autonomous overgrowth mutants erupted and vps25, Notch accumulation increases the secretion of unpaired, which then promotes the growth of surrounding tissues via the Jak/Stat pathway and can be suppressed by removing one copy of Stat92E (Moberg et al., 2005; Vaccari and Bilder, 2005; Herz et al., 2006). The role of Notch accumulation in E1 null clones posterior to the MF to the non-autonomous overgrowth is unclear, given (1) the failure of mutation in Stat92E to dominantly suppress the non-autonomous overgrowth in mosaic eyes containing clones homozygous for Uba1^A1 (Pfleger et al., 2007) and (2) the absence of an increase in the production of unpaired (Pfleger et al., 2007). The role of decreased Notch in the MF in E1 null clones to the overgrowth phenotype is also unclear. However, removing one copy of E1 by introducing null mutation Uba1^A1 dominantly enhanced the dominant wing-notching phenotype of two Notch alleles (N^fa-1 and N^55e11, data not shown) indicating that decreased E1 can lead to biologically relevant decreases in Notch signaling. Despite the lack of suppression by Stat92E on overall organ size caused by null mutation of E1, another group suggests that reducing Notch or Stat92E gene dosage affects aspects of the phenotype of a temperature-sensitive allele of Uba1 (Lee et al., 2008), indicating a possible role for Notch accumulation in some E1 phenotypes. Indeed, Notch and Ras can act both antagonistically as well as cooperatively (Sundaram, 2005; Doroquez and Rebay, 2006).

The increased activation of Ras in tissue adjacent to E1 null cells we report here could play an important role in the non-autonomous overgrowth either in conjunction with or independent of other signaling pathways. As we reported previously (Pfleger et al., 2007), homozygous Uba1^A1 clones created using eyFLP die by adulthood, so the mosaic larval eyes result in overgrown adult eyes composed entirely of wild-type tissue. For simplicity, overgrown eyes resulting from mosaic larval eyes containing clones homozygous for Uba1^A1 will be referred to as `Uba1^A1 mosaic eyes' despite the lack of mutant tissue at adult stages. To address the functional contribution of Ras to the non-autonomous overgrowth, we removed one copy of Ras in Uba1^A1 mosaic eyes (Fig. 7A; supplementary material Fig. S5). The non-autonomous overgrowth was dramatically suppressed. The resulting adult eyes were similar in size and shape to wild-type eyes. In mosaic eye discs, tissue adjacent to Uba1^A1 clones showed no obvious decrease in Yan staining when one copy of Ras was removed (supplementary material Fig. S6). This suggests that inappropriate Ras activation no longer occurred in tissue adjacent to Uba1^A1 clones. As noted, Ras can signal through multiple effectors. Although we cannot rule out a role for Ras-dependent activation of AKT in the non-autonomous overgrowth, removing one copy of AKT (Fig. 7B) showed no obvious effect on the overgrowth of Uba1^A1 mosaic eyes.

The sensitivity to Ras gene dosage could have been within E1 null cells, in responding tissue, or both. To address if Ras-ERK activation in responding cells was growth-relevant, we investigated how cells mutant for another component of the pathway would respond to non-autonomous signals. PTP-ER, protein tyrosine phosphatase-enhancer of Ras1, negatively regulates ERK (Karim and Rubin, 1999). Mosaic eyes in which cells homozygous for Uba1^A1 signal to cells homozygous for the mutation PTP-ER^f02707 (created by generating flies of the genotype yweyFLP; FRT42D PTP-ER^f02707/FRT42D Uba1^A1) were dramatically overgrown and exhibited additional outgrowths in PTP-ER^f02707 mutant tissue (Fig. 7C, right eye) compared with mosaic eyes in which Uba1^A1 cells signal to wild-type cells. These outgrowths contained cuticle-like tissue as well as recognizable ommatidia with red pigment, confirming they contained the PTP-ER^f02707 mutation (a pW+ insertion). Eyes entirely mutant for PTP-ER^f02707 were of normal size (Fig. 7C, left eye). Because PTP-ER mutant tissue does not overgrow on its own, the enhanced outgrowths in PTP-ER mutant tissue support a model in which null mutation in E1 causes growth-relevant activation of Ras-ERK signaling in adjacent cells. These added outgrowths are not seen in eyes mosaic for PTP-ER^f02707 and Uba1^B1 or PTP-ER^f02707 and Uba1^B2 (not shown), consistent with our findings that E1 hypomorphic mutants have cell-autonomous effects on Ras-ERK activation, whereas null mutations in E1 have both cell-autonomous and non-autonomous effects on Ras-ERK.

Adult Uba1^A1 mosaic eyes were created by generating homozygous mutant and homozygous wild-type cells early in eye development across the entire eye. All wild-type cells would probably receive non-autonomous signals from Uba1^A1 cells at the earliest stages before and during their proliferation to populate the entire eye field (schematic in supplementary material Fig. S5). Thus, if activation of Ras-ERK in these responding cells causes the adult eye phenotype, it should resemble the expression of activated Ras across the entire eye early in eye development. Expressing RasV12 or RasV12S35 (an effector loop mutant that signals primarily through Raf/ERK) (Karim and Rubin, 1998; Uhlirova et al., 2005; Prober and Edgar, 2002; Rodriguez-Viciana, 1997; Kinashi et al., 2000; Pacold et al., 2000) early in the entire eye results in dramatic overgrowth and an increase in organ size (Karim and Rubin, 1998) (Fig. 7D; supplementary material Fig. S5). Intriguingly, the extent of overgrowth resembles the overgrowth of wild-type tissue in Uba1^A1 mosaic eyes (Fig. 7D; supplementary material Fig. S5). By contrast, expressing RasV12G37 or RasV12C40 (effector loop mutants that signal primarily through RalGDS or PI3K, respectively) (Karim and Rubin, 1998; Uhlirova et al., 2005; Prober and Edgar, 2002; Rodriguez-Viciana, 1997; Kinashi et al., 2000; Pacold et al., 2000) early across the entire eye caused no obvious overgrowth (supplementary material Fig. S5). Therefore, activation of Ras-ERK in tissue adjacent to Uba1^A1 null clones would be sufficient to drive overgrowth and increase overall eye size in Uba1^A1 mosaic eyes.

Does the cell-autonomous increase in Ras activation in E1 null clones promote the non-autonomous activation of Ras? It is difficult to test this experimentally. As noted, experiments testing the functional relevance of Ras adjusted the Ras gene dosage in both E1 null cells and in responding, overgrowing cells. However, we can address whether cell-autonomous Ras signaling through ERK is sufficient to promote non-autonomous Ras activation. Using mosaic analysis with a repressible call marker (MARCM) strategies (Lee and Luo, 2001; Wu and Luo, 2006), we created clones of cells expressing RasV12S35 labeled with GFP. As with E1 null clones (Fig. 6G-L), Yan staining decreased in wild-type tissue (in this experiment, wild-type tissue was GFP-negative) adjacent to RasV12S35-expressing clones (GFP-positive) in mosaic eye discs (Fig. 7E-J). Because Yan is degraded upon Ras activation, this decrease indicates that the increased Ras-ERK signaling in RasV12S35 clones is sufficient to cause the non-autonomous activation of Ras in nearby cells. A similar effect was seen in tissue surrounding RasV12-expressing clones but was not observed in tissue surrounding FRT42D control clones (supplementary material Fig. S6).

We have presented data that in vivo, impaired ubiquitin pathway function due to mutation in E1 results in a growth-relevant, cell-autonomous increase in Ras-ERK activity. It is widely accepted that RTK endocytosis is regulated by ubiquitylation and that a failure of RTK ubiquitylation promotes increased signaling through Ras (Levkowitz et al., 1999; Yokouchi et al., 1999; Waterman et al., 2000). We cannot rule out contributions from upstream regulators of Ras to the phenotypes of E1 mutants in our system; however, mutation in Ras dominantly suppressed the increased proliferation and pupal lethality of E1 hypomorphs strongly, whereas mutations in Egfr, drk and sos did not. One possible explanation is that multiple upstream steps that converge on Ras are regulated by ubiquitylation. Alternatively, it is possible that an as-yet-unidentified regulator of Ras is regulated by ubiquitylation. However, the simplest model to explain our findings is that the cell-autonomous increase in Ras activity may be independent of Egfr and Grb2/sos and occurs at the step of Ras. Indeed, we have demonstrated that Drosophila Ras is ubiquitylated. Our findings suggest the exciting model that decreased ubiquitylation of Ras itself causes increased activation of ERK. This may be a mechanism highly conserved between Drosophila and mammals, because a recent study reports di-ubiquitylation of H-Ras and N-Ras in vitro (Jura et al., 2006). Whereas Jura et al. established ubiquitylation of H-Ras and N-Ras in a tissue-culture context, the physiological relevance of Ras ubiquitylation has not been investigated. Our Drosophila studies for the first time demonstrate that in vivo, the activation of Ras is highly sensitive to ubiquitylation. We demonstrated that impairing ubiquitylation leads to increased Ras-ERK activation that promotes ectopic cell proliferation and confers increased resistance to cell death in vivo in a developmental context.

How does Ras ubiquitylation restrict signaling through downstream effectors? It is possible that ubiquitylated Ras adopts a conformation that no longer interacts with Raf. Alternatively, ubiquitylation may alter Ras localization, thus isolating it from downstream effectors. Indeed, Jura et al. showed that a construct of H-Ras fused to ubiquitin (to mimic constitutively ubiquitylated Ras) preferentially localizes to the endosomes (Jura et al., 2006).

It is generally assumed that E1 activity is not limiting; decreasing E1 activity so it becomes limiting could amplify substrate specificities (Salvat et al., 2000) such that some ubiquitin-mediated processes are affected early and dramatically whereas others are affected to a lesser extent or at a later time. The extreme sensitivity of E1 phenotypes to Ras gene dosage strongly supports the argument that Ras regulation is affected early and/or dramatically upon a decrease in E1 function and implies that maintaining sufficient activity of the ubiquitin pathway is crucial to prevent inappropriate Ras-ERK activation in vivo.

We have also presented our intriguing observation that that there is growth-relevant Ras-ERK activation in cells adjacent to E1 null clones, and we have shown that this Ras activation mimics oncogenic Ras. What is the mechanism underlying non-autonomous Ras activation? Given the pleiotropic effects caused by the global loss of ubiquitylation, elucidating this experimentally is difficult. Previous work in mammalian systems reports that Ras activation increases the release of EGF-like ligands (Normanno and Ciardiello, 1997; Peles et al., 1992; Yarden and Peles, 1991), and we have demonstrated here that a cell-autonomous increase in Ras signaling through ERK is sufficient to promote activation of Ras in neighboring cells. Thus, it is possible that the cell-autonomous increase in Ras activation in E1 null cells promotes the non-autonomous Ras activation. Investigating the role of ubiquitylation in preventing non-autonomous Ras activation in the future will be exciting and may elucidate the ability of stromal cells to promote growth and invasiveness of adjacent tumor cells.

By demonstrating that maintaining sufficient ubiquitin pathway activity is crucial for Ras regulation both cell-autonomously and non-autonomously, we provide further support for our previous suggestion that E1 may be a tumor suppressor gene (Pfleger et al., 2007). In fact, one study using comparative genomic hybridization reports a loss in DNA copy number of the human E1 chromosomal region in breast cancer lines and tumors (Larramendy et al., 2000). Microarray (Ross et al., 2004) and/or serial analysis of gene expression (SAGE) (Boon et al., 2002; Silva et al., 2004; Allinen et al., 2004) methods reveal significantly decreased E1 RNA levels in many cancer cell lines and tumors. Previous SAGE studies have shown that E1 levels drop dramatically in the leukocytes and luminal epithelial cells of invasive ductal carcinomas compared to those of normal breast tissue and ductal carcinomas in situ (Allinen et al., 2004), potentially implicating E1 loss in breast cancer progression. Given these reports and our findings of Ras-dependent overgrowth due to mutation of E1 in vivo, we propose that downregulating E1, either by mutation or other means, could be a mechanism employed by tumor cells to achieve cell death resistance and Ras activation. Identification of the ubiquitin ligase or ligases targeting Ras is of high importance, as such ligase(s) may play a crucial role in normal proliferation and may be dysregulated in developmental disorders and in cancer.

Genotypes of adult eyes: w; FRT42D Uba1^B1/FRT42D Uba1^B1 (Fig. 1D); yweyFLP/w; FRT42D P[W+, UbiGFP]/FRT42D (Fig. 4A); yweyFLP/w; FRT42D P[W+, UbiGFP]/FRT42D Uba1^B2 (Fig. 4B); yweyFLP/w; FRT42D P[W+, UbiGFP]/FRT42D Uba1^B2; Ras85D^e1B/+ (Fig. 4C); w; FRT42D Uba1^B1;GMR-gal4, UAS-eiger/SM6-TM6B (Fig. 4D); w; FRT42D Uba1^B1/FRT42D Uba1^B2; GMR-gal4, UAS-eiger/+ (Fig. 4E); w; FRT42D Uba1^B1/FRT42D Uba1^B2; Ras85D^e1B/GMR-gal4, UAS-eiger (Fig. 4F); yweyFLP/w; FRT42D P[W+, UbiGFP]/FRT42D Uba1^A1 (Fig. 7A-B, D, left head); yweyFLP; FRT42D P[W+, UbiGFP]/FRT42D Uba1^A1; Ras85D^e1B/+ (Fig. 7A right head); yweyFLP; FRT42D P[W+, UbiGFP]/FRT42D Uba1^A1; AKT/+ (Fig. 7B, right head); yweyFLP; FRT42D PTP-ER ^f02707/FRT42DP[W⁺] 47A l(2)cl-R11¹ (Fig. 7C, left head); yweyFLP; FRT42D PTP-ER^f02707/FRT42D Uba1^A1 (Fig. 7C, right head); w; ey gal4/UAS RasV12 (Fig. 7D, right head).

For immunofluorescence, genotypes of larval eye discs: w; FRT42D (Fig. 1A); w; FRT42D Uba1^B1/FRT42D Uba1^B2 (Fig. 1B); yweyFLP; FRT42D P[W+, UbiGFP]/FRT42D (Fig. 2B-D); yweyFLP; FRT42D P[W+, UbiGFP]/FRT42D Uba1^B1 (Fig. 2E-J); yweyFLP; FRT42D P[W+, UbiGFP]/FRT42D Uba1^B1; argos-lacZ/+ (Fig. 3A-I); w; FRT42D (Fig. 4G); w; FRT42D Uba1^B1/FRT42D Uba1^B2 (Fig. 4H): w; FRT42D Uba1^B/FRT42D Uba1^B2; Ras85D^e1B/+ (Fig. 4I); w; FRT42D Uba1^B1, Egfr^k05115 /FRT42D Uba1^B2 (Fig. 4J); w; FRT42D Uba1^B1, drk^k02401/FRT42D Uba1^B2 (Fig. 4K); w; sos^e4g, FRT42D Uba1^B1/FRT42D Uba1^B2 (Fig. 4L); ywhsFLP; FRT42D P[W+, UbiGFP]/FRT42D Uba1^A1 (Fig. 6A-L); ywhsFLP; FRT42D P[W+, UbiGFP]/FRT42D Uba1^A1; argos-lacZ/+ (Fig. 6M-R); yweyFLP; FRT42D tubPGal80/UAS-RasV12S35, FRT42D; tubPGal4, UAS-mCD8-GFP (Fig. 7 E-J).

Larvae were dissected and stained using standard protocols with the exception of dpERK. For dpERK, larvae were dissected in PBS containing 8% paraformaldehyde and phosphatase inhibitors (phosphatase inhibitor cocktail III, Calbiochem). Eye discs were imaged on a Leica TSC-SP confocal microscope. Primary antibodies were dpERK M8159 (1:100, Sigma); anti-Yan 8B12H9 (1:100, DSHB), anti-β-gal 40-1a (1:10 DSHB); anti-pHH3 (1:200 Upstate). Secondary antibodies were Alexa Fluor 647 goat anti-rabbit, Alexa Fluor 647 goat anti-mouse (Molecular Probes/Invitrogen).

Individual spots positive for anti-pHH3 staining were counted across entire eye discs for a series of eye discs of each genotype, and average and s.e.m. were calculated. The width of the region of high proliferation just posterior to the MF was measured in several places per disc and averaged for individual discs for multiple discs for each genotype, and average and s.e.m. were calculated. dpERK clusters contained entirely within clones were traced in Adobe Photoshop to determine area in pixels. A factor to convert area from pixels to μM² was determined from the scale bar. Regions of decreased Yan staining in tissue adjacent to Uba1^A1 clones were traced and intensity determined using the histogram function in Adobe Photoshop. The same trace area was used to determine baseline Yan intensity by measuring intensity for three distinct regions of tissue in the same disc far from Uba1^A1 clones.

Larval extracts were prepared using PhosphoSafe Extraction Reagent (Novagen) supplemented with protease inhibitors. Parallel western analyses were performed: (1) multiple larvae of each genotype were homogenized together and equivalent micrograms were loaded per lane; (2) lysates from individual larvae were loaded entirely per lane. In each case, the same trend was observed. Westerns were visualized using the Li-Cor Odyssey. Primary antibodies were dpERK M8159 (1:2000) and total ERK M5670 (1:2000), Sigma; anti-E1 (1:500, EMD Biosciences); anti-pAKT (1:1000) and anti-total AKT (1:1000), Cell Signaling Technology. Secondary antibodies were Alexa Fluor goat anti-rabbit 680 (1:20,000) and Alexa Fluor goat anti-mouse 680 (1:20,000) (Molecular Probes/Invitrogen).

S2 cells were cultured using standard methods at 25°C. Cells were transfected with Actin-Gal4 and UAS-FLAG-His6-Ras to express tagged Ras, with Actin-Gal4 and UAS-HA-Ubiquitin to express tagged ubiquitin, and with Actin-Gal4, UAS-FLAG-6His-Ras and UAS-HA-Ubiquitin to co-express tagged Ras and ubiquitin. Transfections used Cellfectin (Invitrogen) according to manufacturer protocols. Cells were washed in 1 × PBS and lysed in 20 mM HEPES pH 7.4, 10% glycerol, 1% NP40, 1 mM EDTA, 150 mM NaCl, 8 M urea, protease inhibitors. Lysates were incubated with Ni-NTA resin, washed and eluted with imidazole. Primary antibodies were anti-HA 12CA5 (1:1000, Roche) and anti-FLAGM2 (1:1000, Sigma). Secondary antibodies were Alexa Fluor goat anti-mouse 680 (1:10,000) (Molecular Probes/Invitrogen).

w; UAS-RasV12/CyO, w; UAS-RasV12S35/CyO, w; UAS-RasV12G37/CyO and w; UASRasV12C40 were crossed to eygal4/SM6-TM6B. To positively mark RasV12S35-expressing cells, UAS-RasV12S35 was recombined onto the FRT42D chromosome and crossed to the MARCM stock yweyFLP; FRT42D tubPGal80; tubPGal4, UAS-mCD8-GFP/SM6-TM6B.