A conserved, N-terminal tyrosine signal directs Ras for inhibition by Rabex-5

Summary

Dysregulation of the Ras oncogene in development causes developmental disorders, “Rasopathies,” whereas mutational activation or amplification of Ras in differentiated tissues causes cancer. Rabex-5 (also called RabGEF1) inhibits Ras by promoting Ras mono- and di-ubiquitination. We report here that Rabex-5-mediated Ras ubiquitination requires Ras Tyrosine 4 (Y4), a site of known phosphorylation. Ras substitution mutants insensitive to Y4 phosphorylation did not undergo Rabex-5-mediated ubiquitination in cells and exhibited Ras gain-of-function phenotypes in vivo. Ras Y4 phosphomimic substitution increased Rabex-5-mediated ubiquitination in cells. Y4 phosphomimic substitution in oncogenic Ras blocked the morphological phenotypes associated with oncogenic Ras in vivo dependent on the presence of Rabex-5. We developed polyclonal antibodies raised against an N-terminal Ras peptide phosphorylated at Y4. These anti-phospho-Y4 antibodies showed dramatic recognition of recombinant wild-type Ras and Ras^G12V proteins when incubated with JAK2 or SRC kinases but not of Ras^Y4F or Ras^Y4F,G12V recombinant proteins suggesting that JAK2 and SRC could promote phosphorylation of Ras proteins at Y4 in vitro. Anti-phospho-Y4 antibodies also showed recognition of Ras^G12V protein, but not wild-type Ras, when incubated with EGFR. A role for JAK2, SRC, and EGFR (kinases with well-known roles to activate signaling through Ras), to promote Ras Y4 phosphorylation could represent a feedback mechanism to limit Ras activation and thus establish Ras homeostasis. Notably, rare variants of Ras at Y4 have been found in cerebellar glioblastomas. Therefore, our work identifies a physiologically relevant Ras ubiquitination signal and highlights a requirement for Y4 for Ras inhibition by Rabex-5 to maintain Ras pathway homeostasis and to prevent tissue transformation.

Keywords:

Drosophila melanogaster – Phenotypes – RNA interference – Eyes – Phosphorylation – Ubiquitination – Ras signaling

Introduction

In Drosophila, Ras proteins are represented by Ras1/Ras85D and Ras2/Ras64B. Drosophila Ras1/Ras85D corresponds to mammalian H-Ras, N-Ras, and K-Ras and is distinct from the closely related Ras2/Ras64B protein represented by R-Ras in mammals. Using the convention that predominates in the literature, here we refer to Drosophila Ras1/Ras85D as Ras; we refer collectively to mammalian H-Ras, N-Ras, and K-Ras as Ras or individually to specific isoforms as H-Ras, N-Ras, and K-Ras as appropriate; and we refer to Ras2/Ras64B as Ras2 for the Drosophila protein and R-Ras for the mammalian protein. Drosophila Ras and mammalian H-Ras N-Ras and K-Ras share sequence identity in their N-termini but diverge in a C-terminal region called the HyperVariable Region or HVR [1; for review, 2–3] (depicted schematically in S1 Fig). It is well accepted that one of the roles of the HVR is to direct the membrane association of Ras in part by the C-terminal CAAX box (cysteine, aliphatic amino acid, aliphatic amino acid, any amino acid) [1; for review, 2–3].

Ras signaling regulates cell proliferation, growth, differentiation, and cell survival by signaling to a range of downstream effectors including Raf/ERK/MAPK, RalGDS, and PI3K among others [1–12]. Consequently, Ras dysregulation in development alters patterning and causes developmental disorders collectively called “Rasopathies” [for review, 4–9]. Mutational activation and amplification of Ras in differentiated tissues are implicated in cancer [for review, 10–12]. Therefore, mechanisms of attenuating Ras activity are crucial for proper development and to prevent disease.

Rabex-5 (also called RabGEF1), an A20-like E3 ubiquitin ligase, promotes inhibitory mono- and di-ubiquitination of Drosophila Ras and mammalian H-Ras and N-Ras to restrict signaling to downstream effectors [13–16]. Rabex-5 inhibits both wild-type Ras and also the constitutively active oncogenic mutant Ras^G12V (also referred to as RasV12 in the literature) [13–16].

No signal in Ras has been reported to direct its inhibition by Rabex-5, and no ubiquitination targeting motif has been ascribed to Rabex-5 or the A20 family of ubiquitin ligases. We mapped a ubiquitination signal in Drosophila Ras; we report here that Rabex-5 inhibition of Drosophila Ras requires Ras N-terminal tyrosine 4 (Y4). Phenylalanine substitution mutants of Ras at Y4 (to prevent phosphorylation) were insensitive to Rabex-5-mediated ubiquitination in S2 cells and showed Ras gain-of-function phenotypes in vivo. Glutamic acid substitution mutants of Ras at Y4 (to mimic the charge of phosphorylation) showed increased Rabex-5-mediated ubiquitination in S2 cells, and glutamic acid substitution mutants of Ras^G12V at Y4 suppressed oncogenic Ras phenotypes in vivo, dependent on the presence of Rabex-5. JAK2 and SRC kinases are capable of promoting phosphorylation of Ras^WT at Y4, whereas JAK2, SRC, and EGFR can promote phosphorylation of an oncogenic form of Ras, Ras^G12V, at Y4 as measured by recognition by anti-pY4 antibodies.

Results and discussion

An N-terminal tyrosine-based signal directs Ras for mono- and di-ubiquitination

To elucidate the molecular mechanism of Ras inhibition by Rabex-5 and to advance our understanding of the A20 family of E3 ubiquitin ligases, we mapped a signal in Drosophila Ras responsible for Rabex-5 mediated ubiquitination with a deletion strategy (S1A Fig). In our previous work, we used a double FLAG-His6 tag on full length Ras [13, 16]. Deletion constructs were tagged with a triple tag of GFP-FLAG-His6 so that smaller constructs (corresponding to larger deletions) would be large enough to eliminate concerns of peptide instability. Ubiquitin conjugates of deletion constructs were isolated from Schneider S2 cells using nickel purification (to isolate the His6 tag). Visualization of the conjugates was achieved using the FLAG tag (to visualize Ras) and HA (to visualize ubiquitin which was expressed from an HA-Ub plasmid) as done previously [13, 16]. Ras is also regulated by the E3s Nedd4 [17], βTRCP [18], and LZTR1 [19–20]. Therefore, to map a Rabex-5 ubiquitination signal but without excluding Nedd4, βTRCP, or LZTR1 signals, our initial deletion strategy followed ubiquitination of Ras in Schneider S2 cells without Rabex-5 supplementation (S1A–S1D Fig). As noted, Ras membrane association is directed by a C-terminal CAAX signal which is represented by the amino acids CKML in Drosophila Ras. To properly localize N-terminal constructs, we tagged each deletion construct at its C-terminus with the Drosophila Ras CAAX box CKML (depicted schematically in S1A Fig; sequences listed in the methods section).

Previous work by Jura et al. reported the importance of the HVR for Ras ubiquitination. Inhibitory ubiquitination of H-Ras and N-Ras but not K-Ras was reported in mammalian cells [15]. Notably, replacing the K-Ras HVR with the H-Ras HVR conferred ubiquitination onto K-Ras [15]. This could have reflected the requirement for specific sequences in the H-Ras HVR not present in the K-Ras HVR; alternatively, this could have reflected the importance of the HVR in directing the localization of each Ras isoform to a compartment where the ubiquitination occurs. We report here that the Drosophila Ras HVR was neither sufficient nor required for Ras ubiquitination. C-terminal constructs were not ubiquitinated (S1A and S1B Fig). After narrowing the region sufficient for ubiquitination to the N-terminal 20 amino acids of Ras, we tested the ability of the N-terminal 20 amino acids to serve as a competitive inhibitor. Expressing GFP-Myc-tagged 1-20CKML peptides in excess prevented the formation of Ras-ubiquitin conjugates of full-length FLAG-His6 tagged Ras^WT isolated on nickel beads and detected by anti-HA antibodies (S1E Fig), whereas GFP-myc tagged peptides of a different 20 amino acid region in excess had no effect on Ras^WT ubiquitin conjugates (S1E Fig). We further narrowed the region sufficient to confer Ras ubiquitination in S2 cells to the N-terminal 10 amino acids of Ras (Fig 1A, S1A Fig, S1F and S1G Fig). Co-transfecting cells with Rabex-5 increased the ubiquitination of this region (Fig 1A, S1F and S1G Fig) but not of other small regions of Ras (S1F Fig).

Because the HVR was neither sufficient for ubiquitination nor required for ubiquitination of constructs, our results taken together with the previous work of Jura et al.[15], could be consistent with a requirement for the HVR to direct Ras localization [1; for review, 2–3] to a specific compartment where a sequence shared by H-Ras and K-Ras would then be recognized. In fact, the N-termini of Drosophila Ras and H-Ras, N-Ras, and K-Ras are identical (alignment, S2C Fig). While we cannot rule out a contribution from amino acids 11–189, the first 10 amino acids of Ras were sufficient to confer ubiquitination.

Normally, Ras^WT and Ras^G12V proteins associate with the plasma membrane and are also found in the cytoplasm [1–3, 15] (Fig 1B, S1H Fig S2E–S2E” Fig). Previous work by the Bar-Sagi group has shown that a non-ubiquitinated mutant of mammalian H-Ras (in which all solvent exposed lysines were mutated to arginine) had increased localization to the Golgi from which Ras proteins are known to actively signal [15]. In contrast, an H-Ras ubiquitin fusion protein showed decreased Golgi localization and increased localization in an early endosomal compartment [15]. This suggests that the mechanism of inhibitory ubiquitination is to sequester ubiquitinated Ras proteins away from downstream effectors they would encounter at the membrane or in the Golgi by retaining it in the early endosome. We over-expressed Rabex-5^DPYT (which maintains E3 activity but has impaired Rab5 GEF function) to increase the inhibitory ubiquitination of endogenous Ras proteins without affecting the endosomal compartment in the developing eye using GMR-gal4. Staining for endogenous Ras using anti-pan-Ras antibodies at a specific time window 48 hours after puparium formation revealed a dramatic redistribution of Ras to an intracellular compartment (S1I Fig) not seen in controls (S1H Fig). Taken together with work from the Bar-Sagi group showing redistribution of a Ras ubiquitin fusion construct [15], we speculate that Rabex-5 activity can promote Ras re-localization to prevent its signaling to downstream effectors in a highly conserved fashion.

Most cellular Ras is GDP-loaded [21]. Mutation at or close to Ras codon 12 biases Ras to a GTP-loaded conformation [22]. To define a signal in both GDP- and GTP-loaded Ras conformations, we created alanine substitution mutants of the first ten amino acids in FLAG-His6 tagged full-length Drosophila Ras^WT and FLAG-His6 tagged full-length Drosophila Ras^G12V. Y4A and V7A substitution reproducibly decreased ubiquitination in both Ras^WT and Ras^G12V (S2A and S2B Fig). Curiously, we saw differences in alanine scanning between Ras and Ras^G12V constructs (S2A and S2A’ Fig, summarized in S2B Fig). E3A and K5A substitutions decreased ubiquitination of only Ras^G12V (S2A’ Fig, S2B Fig).

The tyrosine and valine important for ubiquitination of Drosophila Ras^WT and Ras^G12V and the lysine important for ubiquitination in Ras^G12V are entirely conserved in human H-Ras, N-Ras, and K-Ras as well as in Drosophila Ras2 (S2C Fig). To establish if Rabex-5 can also promote ubiquitination of Ras2, we expressed FLAG-His6 tagged Ras2 in Schneider S2 cells. Rabex-5 promoted Ras2 ubiquitination (S2D Fig) to a similar extent as Ras^WT. The crucial tyrosine is not conserved in R-Ras (alignment, S2C Fig), so it is unclear if the ability of Rabex-5 to promote Ras2 ubiquitination would be conserved for human R-Ras.

In transfected Schneider S2 cells, most cells show Ras associated with the membrane and also in intracellular puncta in the cytoplasm. In a population of transfected cells, some individual cells showed enrichment of Ras signal at the membrane or in intracellular puncta (S2E–S2E” Fig). Because work by the Bar-Sagi group indicated the importance of localization for Ras ubiquitination [15], we assessed localization of the alanine scanning constructs. Localization of each alanine mutant disrupting ubiquitination confirmed that a lack of ubiquitination did not result from a failure to localize to the plasma membrane and these cytoplasmic puncta (S2F–S2M Fig).

Previous proteomic studies have identified phosphorylation in mammalian Ras at Y4 [23] and also at other tyrosines [24–30]. Ras^WT purified from S2 cells was recognized by anti-phosphotyrosine (p-tyr) antibodies (S2N–S2P Fig). Ras^Y4F (phenylalanine substitution to preserve structure but without the hydroxyl group) showed decreased recognition by anti-p-tyr antibodies compared to Ras^WT (S2O and S2P Fig) consistent with phosphorylation at Y4 and also at other tyrosines in Drosophila Ras as seen in mammalian Ras.

Negative charge at Y4 increases Rabex-5 mediated Ras ubiquitination

We created glutamic acid substitution mutant Ras^Y4E to mimic the charge of phosphorylation in order to test the effects of negative charge at Y4 in vitro and in vivo. Ras^Y4E and Ras^Y4F associated with the plasma membrane and intracellular puncta as seen with Ras^WT (Fig 1B–1D). The basal level of Ras^Y4E ubiquitination increased compared to Ras^WT, and Rabex-5-mediated Ras^Y4E ubiquitination increased compared to Rabex-5-mediated ubiquitination of Ras^WT (Fig 1E, quantified in Fig 1E’ and 1E”, raw data in S1 File). In contrast, Ras^Y4F showed lower basal ubiquitination and no Rabex-5-mediated ubiquitination (Fig 1E–1E”).

Ras^Y4F shows Ras gain-of-function phenotypes in vivo

To investigate a biological role for Drosophila Ras Y4, we created inducible transgenic lines for Ras^WT, Ras^Y4E, Ras^Y4F, Ras^G12V, and double mutants Ras^Y4E,G12V and Ras^Y4F,G12V. Each transgene was tagged with the N-terminal FLAG-His6 tag used in our in vitro studies and was inserted at the same attp40 genomic site to rule out position insertion effects. Transgenes expressed at similar levels (S3B Fig for larvae). Ras^Y4E and Ras^Y4F expressed with Act5C-gal4 rescued the early lethality of Ras loss-of-function alleles to the same extent as Ras^WT transgenes (S3I Fig), suggesting that genetically Y4E and Y4F mutations are capable of carrying out Ras function and do not inactivate Ras inherently.

When Ras signaling is kept within an appropriate range during development, growth, proliferation, and cell fate decisions occur normally. When Ras signaling is in excess, a variety of growth and patterning phenotypes result depending on the developmental timing and tissue context. Expression of Ras^WT in many contexts such as in the eye (using GMR-gal4 and ey-gal4) and the wing (using c765-gal4) does not disrupt growth or disrupt patterning and cell fate decisions (summarized in S3A Fig), presumably because endogenous mechanisms of regulating Ras buffer the increased Ras expression to within the normal range. However, expressing Ras^WT with constitutive driver Tub-gal4 or dorsal wing driver MS1096-gal4 resulted in the Ras gain-of-function phenotype of ectopic wing veins (Fig 2B, 2B’ and 2E; S3D and S3D’ Fig, S3G Fig) compared to control wings (Fig 2A, 2A’ and 2D; S3C and S3C’ Fig, S3F Fig). Ras^Y4F increased wing vein disruption (Fig 2C, 2C’ and 2F; S3E and S3E’ Fig, S3H Fig) compared to Ras^WT and statistically significantly decreased wing size compared to controls (Fig 2G, raw data in S2 File). The increase in severity of phenotype suggests that Y4F substitution increases Ras activity.

Y4F substitution enhances oncogenic Ras phenotypes in vivo

Expressing Ras^G12V in the wing causes wing vein abnormalities (Fig 3B, 3E and 3H; S4B Fig, S4E Fig, S4H Fig) compared to controls (Fig 3A, 3D and 3G; S4A Fig, S4D Fig, S4G Fig). Y4F mutation in oncogenic Ras, Ras^Y4F,G12V, enhanced wing vein abnormalities and further reduced wing size and caused some wings to appear crumpled (Fig 3C, 3F and 3I; S4C Fig, S4F Fig, S4I Fig). Increasing Rabex-5 E3 activity by co-expressing Rabex-5^DPYT [13, 31] to a level with no wing phenotype (Fig 3D’; S4D’ Fig) suppressed Ras^G12V phenotypes (Fig 3E’; S4E’ Fig) but did not suppress Ras^Y4F,G12V phenotypes (Fig 3F’; S4F’ Fig). The enhanced phenotypes of Ras^Y4F,G12V compared to Ras^G12V and the suppression of Ras^G12V but not Ras^Y4F,G12V by Rabex-5 E3 activity are consistent with a model that Y4F mutation allows Ras^G12V to evade Rabex-5-mediated inhibition and emphasize the importance of Ras Y4 for targeting Ras for inhibition.

Negative charge at Y4 suppresses oncogenic Ras phenotypes

If preventing phosphorylation at Y4 allowed Ras to evade inhibition by Rabex-5, then mimicking Y4 phosphorylation would be predicted to increase targeting by Rabex-5, thus inhibiting Ras activity. Notably, Ras^Y4E,G12V exhibited suppressed phenotypes compared to Ras^G12V in all contexts tested. Ras^Y4E,G12V expression in the early eye with ey-gal4 (Fig 4C, 4E and 4F; S5C Fig, S5E and S5F Fig) resembled a control eye (Fig 4A, 4D and 4F; S5A Fig, S5D Fig, S5F Fig) lacking the Ras^G12V overgrowth and tissue outgrowths (Fig 4B, 4D and 4E; S5B Fig, S5D and S5E Fig). Ras^Y4E,G12V expression later in eye development with GMR-gal4 (Fig 4I, S5I and S5J Fig) resembled a control eye (Fig 4G, S5G Fig), not a rough Ras^G12V eye (Fig 4H, S5H–S5J Fig). Ras^Y4E,G12V expressed in hemocytes with He-gal4 (Fig 4L, S5M Fig) resembled controls (Fig 4J, S5K Fig) not Ras^G12V-induced hemocyte over-proliferation (Fig 4K, S5L Fig). Ras^Y4E,G12V expressed in the wing with c765-gal4 (Fig 4O; S5P Fig) resembled a control wing (Fig 4M; S5N Fig) not the Ras^G12V wing vein phenotype (Fig 4N; S5O Fig). Ras^Y4E,G12V expressed in the dorsal wing with MS1096-gal4 (Fig 4Q; S5R Fig) suppressed the lethality of Ras^G12V and caused disrupted wings compared to control wings (Fig 4P; S5Q Fig). These results demonstrate that negative charge at Ras Y4 is a mechanism for inhibiting Ras in vivo; this inhibition is strong enough to block oncogenic Ras visible phenotypes and to prevent oncogenic Ras^G12V–induced lethality.

Ras^Y4E expressed with Act5C-gal4 rescued the early lethality of Ras loss-of-function alleles to the same extent as Ras^WT transgenes (S3I Fig), and Ras^Y4E,G12V expressed in the dorsal wing promoted Ras phenotypes (Fig 4Q; S5R Fig). These data suggest that Y4E mutation does not inherently inactivate Ras, for example by causing misfolding. The ability of negative charge at Y4 to potently block oncogenic Ras activity is consistent with a model that Y4 phosphorylation leads to Ras inhibition by Rabex-5.

Suppression of oncogenic Ras phenotypes by Y4 phosphomimic substitution requires Rabex-5

If Y4E substitution increased Ras targeting by Rabex-5 in vivo as seen in vitro (Fig 1E–1E”), then reducing Rabex-5 gene dosage should elicit oncogenic Ras phenotypes. Increased expression of Ras^Y4E,G12V resulted in obvious oncogenic Ras phenotypes (Fig 4S; S5T Fig) presumably due to overwhelming endogenous Rabex-5. Low-level Rabex-5 RNAi that on its own results in no obvious phenotype (Fig 4T; S5U Fig) and resembles a control wing (Fig 4R; S5S Fig) together with Ras^Y4E,G12V expression resulted in obvious ectopic wing veins (Fig 4V; S5W Fig). The lack of phenotype upon Ras^Y4E,G12V expression (Fig 4O and 4U; S5P Fig, S5V Fig) but extra wing vein phenotype upon concurrent Rabex-5 reduction (Fig 4V; S5W Fig) is consistent with a model that endogenous Rabex-5 strongly restricts Ras^Y4E,G12V activity.

JAK2 and SRC kinases can promote phosphorylation of recombinant Ras^WT and Ras^G12V at Y4

No motif or recognition signal responsible for targeting Ras for inhibition by Rabex-5 has previously been identified in Ras. Our findings demonstrate that Rabex-5-mediated inhibition of Drosophila Ras requires conserved Ras residue Tyrosine 4 in vitro and in vivo to regulate Ras. We cannot rule out a requirement for the tyrosine hydroxyl group in protein interactions contributing to this inhibition separate from its role in phosphorylation. However, to more specifically address phosphorylation of Drosophila Ras at Y4, we generated anti-phospho-Y4 (anti-pY4) rabbit polyclonal antibodies (described further in the methods). Anti-pY4 polyclonal antibodies recognized a number of proteins in cell lysates with presumably similar epitopes (S6A Fig), and they also recognized unphosphorylated Ras proteins expressed in and purified from bacteria at a baseline level (Fig 5A and 5B, S6B–S6H Fig). Baseline recognition of unmodified Ras proteins makes it difficult to assess subtle changes in phosphorylation of Y4 from cellular samples. Alanine scanning (S2 Fig) identified a role for Y4 and V7 in ubiquitination of both Ras^WT and Ras^G12V; these residues represent a YXXV motif known to be recognized by tyrosine kinases JAK2 [32] and SRC [33–34]. Our alanine scanning also showed a role for E3 in Ras^G12V but not in Ras^WT; EY is a motif recognized by EGFR [35]. Recombinant His6-tagged Ras^WT and Ras^G12V proteins purified from bacteria were incubated in the presence or absence of purchased recombinant active JAK2, SRC, and EGFR kinases. Reproducibly, anti-p-Y4 antibodies showed a dramatic increase in recognition of Ras^WT and Ras^G12V proteins compared to the baseline recognition of non-phosphorylated recombinant protein after incubation with JAK2 and SRC kinases (Fig 5A and 5B; S6B–S6F Fig). Anti-pY4 antibodies did not show a difference in recognition of Ras^WT protein versus Ras^WT protein incubated with EGFR kinase (Fig 5A and 5B, S6C Fig, S6E Fig) but showed dramatically increased recognition of Ras^G12V incubated with specific preparations of EGFR protein (Fig 5A, S6D Fig) but not with all EGFR preparations (Fig 5B, S6F Fig) compared to its recognition of Ras^G12V protein alone. To confirm that recognition depended on Y4, we repeated these assays with recombinant His6-tagged Ras^Y4F and Ras^Y4F,G12V proteins. Anti-pY4 antibodies did not show a difference in recognition between Ras^Y4F or Ras^Y4F,G12V proteins versus Ras^Y4F or Ras^Y4F,G12V proteins incubated with JAK2, SRC, or EGFR kinases (Fig 5B, S6G and S6H Fig). These findings are consistent with a model that JAK2 and SRC kinases are capable of promoting phosphorylation of Ras^WT and Ras^G12V at Y4 in vitro. Our results could also be consistent with a model that EGFR can promote phosphorylation of Ras^G12V at Y4 in vitro. However, given that this activity varied between preparations of commercial EGFR purified from HEK293 cells, it is possible that this activity relies on a co-purifying factor and is not intrinsic to EGFR. Alternatively, EGFR activity for Ras Y4 may require specific modifications of EGFR or co-factors not consistent between preparations. Anti-pY4 antibodies also recognized JAK2 and SRC kinases to varying degrees depending on specific preparations (S6C–S6H Fig), presumably due to the similar epitopes of their auto-phosphorylation sites with their site in the Ras N-terminus used to generate the anti-pY4 antibodies. Taken together, our in vivo phenotypic data (S3–S5 Figs) and in vitro kinase assay data (Fig 5, S6 Fig) are consistent with a model that the Y4 requirement involves phosphorylation potentially by JAK2, SRC, and/or EGFR tyrosine kinases (summarized schematically in Fig 5J). Serine and threonine phosphorylation have been reported to be required for recognition by the SCF family of cullin ring ligases [36; for review, 37–38]. Reports of tyrosine phosphorylation directing recognition by an E3 are relatively rare. A role for Y4 phosphorylation in inhibition by Rabex-5 could reflect an increased affinity of Rabex-5 for Ras phosphorylated at Y4 as with the SCF and its phosphorylated substrates. Alternatively, a phosphorylated tyrosine could serve another role such as recruitment of an SH2 domain containing adaptor protein that facilitates Ras interaction with Rabex-5.

Reducing hop levels enhances Ras^G12V but not Ras^Y4F,G12V

The Y4 kinase would be expected to increase inhibitory ubiquitination of Ras proteins, thus would serve to inhibit Ras biological outputs. JAK2, SRC, and EGFR have all been described to activate Ras signaling in various contexts. If one or more of these kinases also acts as the Y4 kinase to increase Rabex-5-mediated inhibitory Ras ubiquitination, this would serve as a feedback mechanism to ensure precise pathway outputs. Unfortunately, this also means that testing a role for these kinases to inhibit Ras in vivo using genetic interactions is challenging. Expressing EGFR (S7D Fig), Ras^G12V (Fig 5D and 5D’, S7B Fig), or Ras^Y4F,G12V (Fig 5E and 5E’, S7C Fig) individually in the eye using GMR-gal4 leads to a rough eye, some loss of eye pigment, and in some eyes, the appearance of black tissue around the periphery of the eye compared to a control eye (Fig 5C and 5C’, S7A Fig). Indeed, co-expressing EGFR and Ras^G12V with GMR-gal4 (S7E Fig) led to a dramatic enhancement of both the EGFR and the Ras^G12V rough eye phenotypes. Co-expression led to more significant loss of red eye pigment and an increased prevalence of black tissue at the periphery of the eye (S7E Fig). This differs from the phenotype of co-expressing EGFR and Ras^Y4F,G12V which also led to significant loss of eye pigment, but the black tissue phenotype appeared not only at the periphery of the eye but also in other regions of the eye (S7F Fig). The spread of black tissue from the periphery of the eye to other regions of the eye could reflect a greater enhancement of phenotype consistent with a role for EGFR to both restrict Ras^G12V activity and also to promote signaling through endogenous Ras but an inability to restrict Ras^Y4F,G12V, or could reflect the enhanced phenotype of Ras^Y4F,G12V generally. Thus, these genetic interactions are difficult to interpret.

Low level RNAi of hopscotch (hop, an ortholog of JAK2) in the eye using GMR-gal4 leads to no phenotype on its own (Fig 5F and 5F’) and resembles a control eye (Fig 5C and 5C’). RNAi to hop concurrent to expressing Ras^G12V (Fig 5G and 5G’, quantified in 5I, raw data in S3 File) consistently enhanced the phenotype of appearance of black tissue in the eye, a quantifiable phenotype we have used previously to reflect severity of phenotype [13] (Fig 5I). Importantly, this enhancement required Y4; we did not see enhancement of this black tissue phenotype upon hop RNAi concurrent to expressing Ras^Y4F,G12V (Fig 5H and 5H’, quantified in 5I). These in vivo interactions taken together with the in vitro kinase assays would be consistent with Hop/JAK2 activity normally acting to promote Y4 phosphorylation to restrict Ras activity in this context.

This report focuses on the importance of a specific tyrosine in directing Rabex-5 mediated ubiquitination of Drosophila Ras (model, Fig 5J). The site of inhibitory ubiquitination in Drosophila Ras and mammalian H-Ras and N-Ras has not been identified. There are a number of solvent-exposed lysines in mammalian Ras proteins reported to be sites of ubiquitination [20, 39–47], many of which are conserved in Drosophila Ras (summarized in S1 Table). K117 mono-ubiquitination has been reported to increase intrinsic nucleotide dissociation which facilitates GDP-GTP exchange, thus activating Ras. K147 mono-ubiquitination has been shown to decrease the rate of GTP hydrolysis mediated by GTPase Activating Proteins [41], which also activates Ras. K170 has been shown to serve as a site of LZTR1-mediated inhibitory ubiquitination [20]. We speculate that one of these other lysines could serve as the biologically relevant site of Rabex-5 mediated inhibitory mono- and di-ubiquitination of Ras proteins. In addition to the potential for inhibitory ubiquitination to promote re-localization of Ras (SS1I Fig and [15]), it is possible that Rabex-5 mediated inhibitory Ras ubiquitination decreases GDP-GTP exchange or increases the rate of GTP hydrolysis in opposition to the activating ubiquitination events at K117 and K147.

Importantly, Y4 mutations in H-Ras (HRas^Y4H) have been reported in cerebellar glioblastomas [48]. H-Ras^Y4H tumor variants taken together with our work showing gain-of-function phenotypes upon mutation at Y4 suggest that Y4 phosphorylation is important for maintaining appropriate restriction of Ras activity and that mutation at Y4 evades such inhibition to promote tissue transformation.

Materials and methods

Reproducibility

The reported work represents reproducible experiments that reflect a minimum of three well-controlled, independent trials. For phenotypes that are subjective (not quantifiable), independent lab members scored progeny blind to avoid bias.

Tissue culture

S2 cells were cultured using standard methods at 25°C in Schneider’s Drosophila Medium (1X) (Gibco, 21720–024). Cells were transfected using Effectene Transfection Reagent (Qiagen, Cat # 301427) according to manufacturer instructions with plasmids pUAST-HA-Ub, Act-gal4, UAS myc Rabex-5, pIE^1-4 Flag-His6-GFP Ras WT, pIE^1-4 Flag-His6-GFP Ras 1-100CKML, pIE^1-4 Flag-His6-GFP Ras 81–189, pIE^1-4 Flag-His6-GFP Ras HVR, pIE^1-4 Flag-His6-GFP Ras 180–189, pIE^1-4 Flag-His6-GFP Ras 61-80CKML, pIE^1-4 Flag-His6-GFP Ras 41-60CKML, pIE^1-4 Flag-His6-GFP Ras 21-40CKML, pIE^1-4 Flag-His6-GFP Ras 1-20CKML, pIE^1-4 Flag-His6-GFP Ras 1–10 CKML, pIE^1-4 Flag-His6-GFP Ras 1–60 CKML, pIE^1-4 Flag-His6 Ras WT, pIE^1-4 Flag-His6 Ras M1A, pIE^1-4 Flag-His6 Ras T2A, pIE^1-4 Flag-His6 Ras E3A, pIE^1-4 Flag-His6 Ras Y4A, pIE^1-4 Flag-His6 Ras Y4E, pIE^1-4 Flag-His6 Ras Y4F, pIE^1-4 Flag-His6 Ras Y4F, pIE^1-4 Flag-His6 Ras E5A, pIE^1-4 Flag-His6 Ras L6A, pIE^1-4 Flag-His6 Ras V7A, pIE^1-4 Flag-His6 Ras V8A, pIE^1-4 Flag-His6 Ras V9A, pIE^1-4 Flag-His6 Ras G10A, pIE^1-4 Flag-His6 Ras M1A,G12V, pIE^1-4 Flag-His6 Ras E3A,G12V, pIE^1-4 Flag-His6 Ras Y4A,G12V, Flag-His6 Ras E5A,G12V, Flag-His6 Ras L6A,G12V, Flag-His6 Ras V7A,G12V, Flag-His6 Ras V8A,G12V, Flag-His6 Ras V9A,G12V, Flag-His6 Ras G10A,G12V as indicated and harvested after 48–72 hours. Importantly, amounts of pUAST-HA-Ub and Act-gal4 (to direct ubiquitin over-expression) were kept constant across experiments. Constructs are summarized in Table 1, and protein sequences of Ras constructs are detailed in Table 2.

Schneider S2 cell extract preparation

Schneider S2 cell extracts were prepared immediately from harvested cells or from frozen cell pellets. Lysates were prepared in lysis buffer, 1X NP40 buffer (50 mM HEPES pH 7.4, 1% NP40, 1mM EDTA, 150 mM NaCl), 8M Urea, 5 mM beta-mercaptoethanol, 10 mM imidazole supplemented with Roche protease inhibitor cocktail and 1mM each of PMSF and Sodium Orthovanadate) and used in pull down assays described above or analyzed by Western.

Nickel pull down of tagged Ras from Schneider S2 cells (for Western analysis)

Ni-NTA agarose beads (Qiagen) were equilibrated as per manufacturer’s instructions and then washed twice with binding buffer, pH 8.0 (1X PBS, 8 M Urea, 0.5% NP40, 10 mM Imidazole, 360 mM NaCl) followed by incubation with binding buffer, pH 8.0 supplemented with 1g/ml BSA. Lysates were then incubated with beads on a nutator for 2h at 4°C. Beads were then transferred to mini-columns on a vacuum assembly and washed 3X with wash buffer pH 6.3 (1X PBS, 8 M Urea, 0.5% NP40, 30 mM Imidazole, 360 mM NaCl). Purified His-tagged proteins were eluted from the beads by washing with elution buffer pH 6.3 (1X PBS, 8 M Urea, 0.5% NP40, 500 mM Imidazole, 360 mM NaCl). Eluate was boiled in 1X loading buffer for 10 minutes before storage at -20°C.

Larval extract preparation

Extracts were prepared from larvae of the specified genotypes in Phosphosafe Extraction Reagent (Novagen) supplemented with protease inhibitors (Complete Phosphatase Inhibitor Cocktail, 1mM PMSF and 1mM Sodium Orthovanadate). Lysates of individual larvae were loaded per lane to demonstrate transgene protein levels.

Development of anti-phospho-Y4 antibodies

A project was initiated with Genscript to produce MTE{pY}KLVVVGC peptides to immunize rabbits. Genscript immunized rabbits and purified phospho-specific polyclonal antisera which recognized MTE{pY}KLVVVGC peptides and delivered purified antibodies to us following purification. Antibodies were tested against cell lysates, larval lysates, and purified recombinant proteins.

Western Blot analysis

Western blots used Immobilon-FL Transfer Membrane (Millipore, Cat # IPFL00010) and were visualized using the Li-Cor Odyssey System. Primary antibodies were anti-pY4 (rabbit polyclonal, this study; 1:2000); anti-FLAG (rabbit, Sigma Catalog # F7425-.2MG, 1:1000), anti-Pan Ras (mouse, Millipore Sigma Catalog # OP40100UG, 1:1000), anti-FLAG M2 (mouse, Sigma Catalog # F1804-5MG, 1:1000), anti-HA (mouse, Roche Catalog # 12CA5, 1:1000), anti-alpha tubulin (mouse, Sigma T9026.-.2ML, 1:8000), anti-phospho-tyrosine 4G10 (mouse, EMD Millipore Catalog # 05–321, 1:1000); secondary antibodies were Alexa Fluor goat anti-mouse 488 (Invitrogen, Catalog # A11001, 1:10,000), Alexa Fluor goat anti-rabbit 680 (Invitrogen, Catalog # A21076, 1:10,000), Alexa Fluor goat anti-mouse 680 (Invitrogen, Catalog # A21057, 1:10,000), DyLight goat anti-mouse 800 (Thermo Scientific, Catalog # 35521, 1:10,000), and DyLight goat anti-rabbit 800 (Thermo Scientific, Catalog # 35571, 1:10,000).

Recombinant proteins

Ras^WT and Ras mutants (Ras^Y4F, Ras^G12V, and Ras^Y4F,G12V) were cloned into pet-28 vectors for bacterial expression and purification of His6-tagged proteins. Proteins were expressed from these plasmids in Rosetta-gami B(DE3)PLysS strains (Novagen, 711374) following induction by IPTG. Proteins were purified on nickel beads. Protein sequences of Ras mutants are detailed in Table 2.

Kinase assays

Active JAK2 (SRP0171), EGFR (SRP6404), and SRC (S1076) recombinant proteins were purchased from Sigma. Recombinant Ras^WT or Ras mutants on beads were incubated in the presence or absence of 100 micrograms (S6B Fig) or 20 micrograms (Fig 5A and 5B, S6C–S6H Fig) recombinant kinase proteins in kinase assay buffer (25mM TrisHCl pH7.5, 10mM MgCl₂) for thirty minutes, quenched with sample buffer, run on a gel, and analyzed by Western.

Western quantification and adjustment

Raw tiff files from the Li-Cor Odyssey were split from full color into specific channels corresponding to single antibodies that were then converted to grayscale using Adobe Photoshop. Images were cropped to fit figure panels; adjustments to brightness and contrast were applied uniformly to the entire images not to portions of an image. Quantification of percent ubiquitination utilized gels in which signals were not over-saturated. To quantify percent ubiquitination, gel lanes were analyzed using Image J; we summed the signal of unconjugated and ubiquitinated Ras bands in a single lane to define total Ras signal in that lane and then calculated the proportion of each band per total to indicate percent conjugated to ubiquitin out of total Ras. In this manner, we could make comparisons of percent conjugation to ubiquitin in one lane to the percent conjugation to ubiquitin in another lane from different lanes on the same gel within the same experiment.

Drosophila experiments

Ras constructs were cloned into pUAST-attB with the FLAG and His6 sequences of MDYKDDDDKRGSHHHHHHALE preceding the Ras coding sequence. UAS Flag-His6-Ras^WT, UAS Flag-His6-Ras^Y4E, UAS Flag-His6-Ras^Y4F, UAS Flag-His6-Ras^G12V, UAS Flag-His6-Ras^Y4E,G12V, and UAS Flag-His6-Ras^Y4F,G12V plasmids were sent to BestGene for injection and generation of transgenic lines at the attp40 locus. Lines were balanced over CyO or SM6-TM6B balancers and then maintained as true-breeding homozygous stocks. Genomic DNA was sequenced to confirm each insert. Gal4 drivers were obtained from the Bloomington Drosophila Stock center or other labs in the Drosophila community. UAS hop^IR (P{TRiP.JF01267}), UAS EGFR were from the Bloomington Stock center. UAS Rabex-5^IR (P{GD14133}v46329 CG9139GD14133) was obtained from the VDRC, VDRCID dna14133 and was characterized in our previous study [13]. Crosses were performed at the indicated temperatures on standard Drosophila medium. Raw wing images were converted to grayscale using Adobe Photoshop. Brightness and contrast of eye and wing images were adjusted using Adobe Photoshop to maximize clarity; adjustments were applied to the entire images. Genotypes are summarized below, and identifiers are annotated in Table 1. Protein sequences for Ras transgenes are detailed in Table 2.

Genotypes of fly images (Figures and Supplemental Figures):

w; Tub-gal4/+ (Fig 2A and 2A’; S3C Fig, S3C’ Fig)

w; UAS Ras^WT/+; Tub-gal4/+ (Fig 2B and 2B’; S3D Fig, S3D’ Fig)

w; UAS Ras^Y4F/+; Tub-gal4/+ (Fig 2C and 2C’; S3E Fig, S3E’ Fig)

MS1096-gal4 (Fig 2D; S2F Fig; Fig 4P, S5Q Fig)

MS1096-gal4; UAS Ras^WT (Fig 2E; S3G Fig)

MS1096-gal4; UAS Ras^Y4F (Fig 2F; S3H Fig)

w; c765-gal4/+ (Fig 3A, 3D and 3G, S3A Fig, S3D Fig, S3G Fig; Fig 4M and 4R, S5N Fig, S5S Fig)

w; UAS Ras^G12V/+; c765-gal4/+ (Fig 3B, 3E and 3H, S4B Fig, S4E Fig, S4H Fig; Fig 4N, S5O Fig)

w; UAS Rabex-5^DPYT/+; c765gal4/+ (Fig 3D’, S4D’ Fig)

w; UAS Ras^G12V/Rabex-5^DPYT; c765gal4/+ (Fig 3E’, S4E’ Fig)

w; UAS Ras^Y4F,G12V/+; c765gal4/+ (Fig 3C, 3F and 3I, S4C Fig, S4F Fig, S4I Fig)

w; UAS Ras^Y4F,G12V/Rabex-5^DPYT; c765-gal4/+ (Fig 3F’, S4F’ Fig)

w; ey-gal4/+ (Fig 4A, left eye in 4D and 4F, S5A Fig, left eye in S5D Fig, S5F Fig)

w; ey-gal4/UAS Ras^G12V (Fig 4B, right eye in 4D and 4E, S5B Fig, right eye in S5D Fig, S5E Fig)

w; ey-gal4/UAS Ras^Y4E,G12V (Fig 4C, left in in 4E, right eye in 4F, S5C Fig, left eye in S5E Fig, right eye in S5F Fig)

w; GMR-gal4/+ (Fig 4G, Fig 5C and 5C’, S1H–S1H” Fig, S5G Fig, S7A Fig)

w; GMR-gal4/UAS Ras^G12V (Fig 4H, Fig 5D and 5D’, S5H Fig, right eye in S5J Fig, S7B Fig)

w; GMRgal4/ UAS Ras^Y4E,G12V (Fig 4I, S5I Fig, left eye in S5J Fig)

w; He-gal4, UAS GFP.nls/+ (Fig 4J, S5K Fig)

w; UAS Ras^G12V/+UAS GFP.nls/+ (Fig 4K, S5L Fig)

w; UAS Ras^Y4E,G12V/+; UAS GFP.nls/+ (Fig 4L, S5M Fig)

MS1096-gal4; UAS Ras^Y4E,G12V (Fig 4Q, S5R Fig)

w; UAS Ras^Y4E,G12V/+; c765-gal4/+ (Fig 4O and 4U, S5P Fig, S5V Fig)

w; UAS Ras^Y4E,G12V; c765-gal4 (Fig 4S, S5T Fig)

w; UAS Rabex-5^IR/+; c765-gal4/+ (Fig 4T, S5U Fig)

w; UAS Rabex-5^IR/ UAS Ras^Y4E,G12V; c765gal4/+ (Fig 4V, S5W Fig)

w; GMR-gal4/+; UAS hop^IR/+ (Fig 5F and 5F’)

w; GMR-gal4/UAS Ras^G12V; UAS hop^IR/+ (Fig 5G and 5G’)

w; GMR-gal4/UAS Ras^Y4F,G12V; UAS hop^IR/+ (Fig 5H and 5H’)

w; GMR-gal4/+; UAS Rabex-5^DPYT/+ (S1I–S1I” Fig)

w; GMR-gal4/UAS Ras^Y4F,G12V (Fig 5E and 5E’, S7C Fig)

w; GMR-gal4/UAS Ras^G12V; UAS EGFR/+ (S7C Fig)

w; GMR-gal4/UAS Ras^G12V; UAS EGFR (S7C Fig)

Genotypes of flies in Western blots:

w; UAS Ras^WT/+; Act5C-gal4/+ (S3B Fig lane 1)

w; UAS Ras^Y4E/+; Act5C-gal4/+ (S3B Fig lane 2)

w; UAS Ras^Y4F/+; Act5C-gal4/+ (S3B Fig lane 3)

Statistical analysis

Wings were measured using Image J software. Wing size comparisons were analyzed using GraphPad Prism software. Unpaired T-tests were used to compare two groups (e.g. controls versus Ras^Y4F over-expressing wings), and one way ANOVA analysis was used for experiments considering three groups (e.g. control versus Ras^WT versus Ras^Y4F). Chi square analysis (using the CHITEST function in Microsoft Excel was used to compare percentages in Fig 1E–1E” for percent of Ras constructs conjugated to ubiquitin and in Fig 5I to compare the percentage of eyes with black tissue. The expected values for each comparison were calculated based on control values applied to the N for other samples.

Supporting information

S1 Fig [a]
An N-terminal Tyrosine-based signal directs Ras for mono- and di-ubiquitination.

S2 Fig [a]
Ras Tyrosine 4 is important for Ras ubiquitination.

S3 Fig [a]
Non-phosphorylatable Ras shows Ras gain-of-function phenotypes .

S4 Fig [a]
Non-phosphorylatable oncogenic Ras enhances oncogenic Ras phenotypes).

S5 Fig [g]
Ras Y4 phosphomimic suppresses the phenotypes of oncogenic Ras dependent on the presence of Rabex-5.

S6 Fig [a]
JAK2, SRC, and EGFR promote Ras phosphorylation at Y4.

S7 Fig [a]

S1 File [1]
Raw Data for .

S2 File [xlsx]
Raw data for .

S3 File [xlsx]
Raw data for .

S1 Table [k]
K-Ras, N-Ras, and H- Ras ubiquitination sites.