Feedback Regulation of Ras Signaling by Rabex-5-mediated Ubiquitination

Rabex-5 Catalyzes Ras Ubiquitination

Ras proteins (HRas, NRas and KRas) are small guanine-nucleotide binding proteins that mediate the transduction of signals from diverse extracellular stimuli to intracellular signaling pathways that control fundamental cellular functions including proliferation, differentiation, and survival. It is now well established that Ras proteins can dynamically partition to different cellular membranes. While all Ras isoforms are found at the plasma membrane, HRas and NRas are also present at the Golgi and endosomes whereas KRas can be found in the endoplasmic reticulum and the outer mitochondrial membrane [6, 7]. Significantly, Ras compartmentalization dictates its signaling capacity and leads to distinct biological outcomes. For example, activation of Ras at the plasma membrane is fast and transient whereas activation of Ras in the Golgi is delayed and sustained [8]. Moreover, in T cells, Ras resides in both the plasma membrane and Golgi, however, it is the activation of the Ras-ERK pathway in the Golgi pool that is essential in specifying the functional T cell population [9]. Lastly, mitochondrial KRas induces apoptosis through its association with Bcl-XL [10].

Several mechanisms have been implicated in the control of Ras intracellular trafficking. KRas which resides predominantly in the plasma membrane is transiently redistributed to intracellular membranes by phosphorylation-dependent alterations of electrostatic interactions with the plasma membrane [10]. HRas and NRas undergo a constitutive depalmitoylation/repalmitoylation cycle, which leads to their continuous shuttling between the plasma membrane and the Golgi [11, 12]. We have shown that in mammalian cells HRas and NRas are subject to mono- and di-ubiquitination and that this modification promotes their endosomal association thereby leading to the attenuation of Ras-ERK signaling [1]. Consistent with these findings, it has been recently reported that in Drosophila, maintaining a threshold of Ras ubiquitination is critical to prevent inappropriate Ras-ERK activation in vivo [13].

To gain insights into the mechanisms that regulate Ras ubiquitination, we sought to determine the identity of the E3 ligase that catalyzes this reaction. We focused on Rabex-5, the mammalian orthologue of yeast Vps9p and a guanine nucleotide exchange factor (GEF) for Rab5 [2, 3]. In addition to its GEF activity, Rabex-5 has been shown to possess an E3 ubiquitin ligase activity mediated by an amino-terminal zinc finger (ZnF) domain which belongs to the A20 ZnF family [14–16]. This catalytic activity along with the observation that Rabex-5 binds to Ras [17] prompted us to investigate the role of Rabex-5 in the regulation of Ras ubiquitination.

We first examined whether Rabex-5 can promote Ras ubiquitination. As shown, the overexpression of Rabex-5 in HEK 293T cells induced the accumulation of mono- and di-ubiquitinated species of ectopically expressed HRas (Figure 1A and Figure S1A) in agreement with the pattern of Ras ubiquitination reported previously [1]. Similar results were obtained using different cell types [such as COS-1 (Figure S1B) and HeLa (data not shown)] as well as an alternative method to isolate and detect ubiquitinated Ras species (Fig. S1C). Consistent with the ectopic expression data, the mono- and di-ubiquitination of endogenous Ras was also stimulated by the overexpression of Rabex-5 (Figure 1B). To test whether this effect was dependent on the ligase activity of Rabex-5, we used a Rabex-5 mutant (Rabex-5 AA) that is impaired in ligase activity due to Tyr to Ala substitutions in residues 25 and 26 which compromise ubiquitin binding [15]. Rabex-5 AA retained a Ras binding activity (Figure S1D) but was ineffective in stimulating Ras ubiquitination (Figure 1B). Thus, the most plausible interpretation of our findings is that Rabex-5 promotes Ras ubiquitination by functioning as its ubiquitin ligase.

To further test this idea, we investigated whether Rabex-5 can directly catalyze the ubiquitination of Ras. To this end, we performed an in vitro ubiquitination assay using recombinant His-HRas and GST-Rabex-5 proteins. It has been reported that when purified as a recombinant protein, full-length Rabex-5 is conformationally unstable [18–20]. Therefore, we employed a recombinant fragment of Rabex-5 (residues 1-76) that encompasses the ubiquitin ligase domain and has been successfully utilized before in in vitro ubiquitination assays [14, 15]. The suitability of this fragment for use in our experiments is further supported by the observations that it contains Ras binding determinants [17] and can interact with Ras directly in vitro (Figure S1E). As shown in Figure 1C, Rabex-5 stimulated the ubiquitination of HRas in vitro whereas Rabex-5 AA was compromised in promoting HRas ubiquitination. Similar results were obtained using a complementary approach (Figure S1F). Notably, Rabex-5 failed to ubiquitinate Rab5, both in vivo (Figure S1G) and in vitro (Figure S1F). Together these findings indicate that HRas is a direct and specific substrate for Rabex-5.

To determine whether Rabex-5 is essential for Ras ubiquitination in vivo, we investigated whether Rabex-5 silencing suppresses Ras ubiquitination. The capacity to detect a significant level of Ras ubiquitination under conditions of endogenous Rabex-5 expression is a prerequisite for this assay. Because the ubiquitination signal of wild-type HRas is barely detectable under basal conditions (Figure 1A and Figure S1B), we resorted to the use of activated HRas (HRas G12V) which, for reasons delineated below (see Figure 4F), displays a robust ubiquitination even in the absence of Rabex-5 overexpression. Transfection of shRNA against Rabex-5 was accompanied by suppression of HRas ubiquitination and this effect could be rescued by the overexpression of shRNA-resistant Rabex-5 WT but not Rabex-5 AA constructs (Figure 1D). We conclude that Ras ubiquitination in vivo is predominantly controlled by Rabex-5.

Rabex-5-mediated Ubiquitination Controls the Endosomal Partitioning of Ras

Mono- and di-ubiquitination of plasma membrane proteins serves as a signal for endosomal sorting [21, 22]. Accordingly, we have shown that the ubiquitination of HRas promotes its endosomal association [1]. Since Rabex-5 regulates the ubiquitination of HRas, we predicted that the partitioning of Ras to the endocytic compartment would be augmented in the presence of Rabex-5. To test this prediction, we analyzed the effect of Rabex-5 overexpression on HRas localization by fluorescence microscopy. Under basal conditions, GFP-HRas displayed a characteristic distribution pattern with prominent presence on the plasma membrane and Golgi (Figure 2A). Upon Rabex-5 overexpression, a significant enhancement of endosome-associated GFP-HRas was detected (Figures 2B and 2D). Conversely, depletion of Rabex-5 by shRNA resulted in reduced levels of endosomal GFP-HRas (Figure S2A). The endosomal identity of the GFP-HRas-labeled vesicular structures was ascertained using the early endosomal marker EEA1 (Figure 2E). The majority of the GFP-HRas-labeled endosomes were also Rabex-5 positive (Figures 2B and 2E) consistent with the documented physical interactions between the two proteins [17] and the notion that Ras is a substrate of Rabex-5. Moreover, since Rabex-5 resides mostly in the early endosomes (Figure 2E) [23], it is possible that the ubiquitination of Ras takes place at the endosomal membrane, and, if so, would function as an endosomal retention rather than endosomal targeting signal.

As a GEF for Rab5, Rabex-5 promotes endosomal fusion [2, 18] as reflected by the accumulation of enlarged endosomes in Rabex-5 expressing cells (Figures 2A, 2B, 2E and Figure S2B). To rule out the possibility that the Rabex-5-dependent accumulation of HRas in the endosomes is due to a general stimulatory effect of Rabex-5 on endosomal trafficking, we examined the localization of GFP-HRas in the presence of a Rabex-5 mutant that was rendered defective in GEF activity by mutation in the Vps9 domain [3, 23]. As previously demonstrated [23], this mutant (Rabex-5 D313A) was localized to endosomes albeit of smaller size relative to Rabex-5 WT-containing endosomes (Figure 2A). Significantly, this mutant, which possesses Ras ubiquitination activity (Figure S2C), retained the ability to enhance Ras endosomal association (Figures 2C and 2D). Hence, by acting as a ubiquitin ligase for Ras, Rabex-5 tightly controls the pool size of endosomal Ras.

Rabex-5-mediated Ubiquitination Attenuates Ras-induced ERK Activation

We have previously demonstrated that HRas ubiquitination and the resulting endosomal retention negatively regulates Ras-dependent recruitment of Raf to the plasma membrane and ERK activation [1]. Given the role of Rabex-5 in promoting HRas ubiquitination and endosomal association, we surmised that ERK activation by Ras would be attenuated by overexpressing Rabex-5. As illustrated in Figures 3A and S3A, the overexpression of Rabex-5 WT or Rabex-5 D313A led to the suppression of HRas-induced ERK activation indicating that this effect is independent of the GEF activity of Rabex-5. In contrast, the ligase activity of Rabex-5 is essential for its attenuating effect on HRas-induced ERK activation as indicated by the observation that this effect was not exhibited by Rabex-5 AA or Rabex-5 D313A/AA mutants (Figure 3A and Figure S3A).

To substantiate the link between Rabex-5-mediated Ras ubiquitination and downstream ERK signaling, we took advantage of a previous finding that Ras ubiquitination displays isoform specificity. In particular, we have shown that KRas is not modified by ubiquitination [1]. Consistent with this finding, the overexpression of Rabex-5 did not induce KRas ubiquitination (Figure 3B). Moreover, KRas-mediated ERK activation was not affected by Rabex-5 overexpression (Figure 3A) confirming that the negative effect exerted by Rabex-5 on HRas signaling is mediated by its HRas ubiquitination activity. It should be noted that NRas, which has been previously shown to be ubiquitinated [1], is also a target for Rabex-5-mediated ubiquitination (Figure S3B) and thus, in principle, could be subjected to a similar mode of regulation.

RIN1 Is Required for Rabex-5-mediated HRas Ubiquitination

Rabex-5 has been shown to bind to the Rab5 effector Rabaptin-5 [2]. This interaction serves to recruit Rabex-5 to early endosomes in a Rab5 dependent manner via the interaction of Rabaptin-5 with Rab5-GTP [18]. In mammalian cells, the levels of Rab5-GTP are positively regulated not only by Rabex-5, but also by the Rab5 GEF proteins of the RIN family, the best characterized of which is RIN1 [24]. Therefore, we sought to determine whether RIN1 participates in the regulation of Ras ubiquitination. RIN1 overexpression stimulated the ubiquitination of HRas (Figure 4A), and conversely, HRas ubiquitination was reduced to basal levels upon siRNA-mediated suppression of RIN1 expression (Figure 4B). Moreover, RIN1-mediated HRas ubiquitination is Rabex-5 dependent as indicated by the failure of RIN1 to enhance HRas ubiquitination upon suppression of Rabex-5 (Figure 4C). Thus, RIN1 appears to be an essential component of the regulatory cascade that controls Ras ubiquitination. This newly identified function for RIN1 might account, at least in part, for its reported activity as a negative regulator of Ras signaling [4].

To further investigate the mechanism by which RIN1 participates in the regulation of HRas ubiquitination we tested the effect of a RIN1 mutant that is defective in Rab5 GEF activity, RIN1 E574A [25], on HRas ubiquitination. As illustrated in Figure 4D, the Rab5 GEF activity of RIN1 was required for its promoting effect on HRas ubiquitination. Hence, it appears that RIN1-mediated activation of Rab5 serves as a prerequisite for Rabex-5-dependent HRas ubiquitination. This idea is supported by the finding that the dependence of HRas ubiquitination on RIN1 could be obviated by the expression of a constitutively active form of Rab5, Rab5 Q79L (Figure S4A). Moreover, Rab5 Q79L stimulated HRas ubiquitination in a Rabex-5-dependent manner (Figure S4B). Together this data is consistent with a model in which RIN1, through its function as a Rab5 GEF, promotes the Rab5-GTP-dependent recruitment of Rabaptin-5/Rabex-5 complex to endosomal sites where HRas ubiquitination takes place.

In addition to its function as a GEF for Rab5, RIN1 is a Ras effector [4, 5]. It directly interacts with GTP-bound Ras through a canonical Ras association domain, and this interaction potentiates RIN1 GEF activity [26]. The role of RIN1 in promoting Ras ubiquitination along with its Ras effector function raises the intriguing possibility that Ras ubiquitination might be subject to feedback regulation through the activation of the RIN1/Rab5/Rabex-5 axis. To test this idea, HRas WT and HRas G12V were differentially tagged with T7 and HA epitopes, respectively, and co-expressed in COS-1 cells. As illustrated in Figure 4E, the level of ubiquitination of HRas WT was significantly enhanced in the presence of HRas G12V indicating that HRas activation can promote HRas ubiquitination. The link between HRas activation and the status of HRas ubiquitination also predicts that HRas G12V would be preferentially ubiquitinated relative to HRas WT. Indeed, at basal conditions the level of HRas G12V ubiquitination is appreciably higher in comparison to HRas WT (Figure 4F). A similar difference has been observed in HEK 293T and HeLa cells (data not shown). HRas WT and HRas G12V displayed similar in vivo interactions with Rabex-5 (Figure S4C) ruling out differential Rabex-5 binding as the mechanism underlying the preferential ubiquitination of HRas G12V. Of note, this result is at variance with our previous demonstration of no significant differences between the level of ubiquitination of HRas WT and HRas G12V in CHO cells [1]. As we have proposed previously, it is conceivable that in these cells the differences might be obscured because of altered relative stoichiometries of one or more of the regulatory components that control Ras ubiquitination.

In summary, our study identifies Ras as a target of the E3 ligase Rabex-5. By promoting Ras ubiquitination and endosomal association, Rabex-5 can function to dynamically shift the partitioning of Ras between different membrane compartments thereby contributing to the spatial control of Ras-dependent cellular responses. The Rab5 GEF activity of Rabex-5 has been also implicated in the control of receptor-mediated Ras signaling through the modulation of receptor endocytosis [23]. Hence, the dual function of Rabex-5 as an E3 ligase for Ras and as a GEF for Rab5 provides a molecular platform for the integration of endocytic trafficking and signaling.