The Ras association domain family (Rassf) is one of the Ras effectors, which can bind to several GTP-charged Ras-like GTPases. The Rassf proteins are widely conserved beyond species from nematode to human. To explore the novel functions of Rassf proteins, we took advantage of nematode C. elegans as a model animal with only one Rassf homolog, T24F1.3 (rasf-1). The rasf-1-mutant as well as rasf-1-knockdown animals were found to be more sensitive to oxidative stress of arsenite than in wild type, indicating that rasf-1 is involved in oxidative stress response. We next screened for proteins that interact with RASF-1 by the yeast two-hybrid system and identified RAB-39 Rab GTPase as an interacting partner of RASF-1. We not only confirmed specific binding between these molecules but also demonstrated that RASF-1 binds to GTP-bound form but not GDP-bound form of RAB-39. Importantly, rab-39 mutant animals were also sensitive to oxidative stress, which was dependent on rasf-1 according to the epistasis analysis. Moreover, Rassf1 and Rab39, mammalian homologs of rasf-1 and rab-39, respectively, were shown to interact with each other in vitro. These results indicate that the RASF-1 functionally interacts with RAB-39 and that the interaction between their homologs is conserved in mammals.
The Ras association domain family (Rassf) comprises ten members from Rassf1 to Rassf10, some of which have various isoforms in mammals. One characteristic feature of this family is the presence of the Ras association domain (RA), which can be found either C-terminal (Rassf1–Rassf6) or N-terminal (Rassf7–Rassf10) (Avruch et al. 2009; Underhill-Day et al. 2011). Thus, the Rassf proteins are divided into two groups by the position of the RA domain. This domain mediates binding to GTP-charged Ras or Ras-like GTPases, at least in vitro when overexpressed (Avruch et al. 2009). Another important protein–protein interaction domain, the Sav-Rassf-Hpo (SARAH) domain, is only found in Rassf1-6. Rassf1, the most studied Rassf, is involved in a variety of biological functions. It has been reported that Rassf1 activates Mst1 (mammalian sterile 20-like kinase-1) through its SARAH domain, mediates the activation of Bax by K-Ras (Vos et al. 2006) and promotes apoptosis (Khokhlatchev et al. 2002; Praskova et al. 2004). Moreover, hypermethylation of the Rassf1 promoter with silencing of Rassf1 gene has been implicated in a number of human tumors (Dammann et al. 2005). There are a number of reports on the functions of Rassf5 (Nore1A) as well. However, other Rassf members have been rather poorly characterized, and the whole picture of Rassf has not been presented. C. elegans has only one Rassf homolog gene, T24F1.3, which we name as rasf-1 hereafter. RASF-1 possesses a RA domain in the C-terminal, and a SARAH domain, hence, belongs to the classical Rassf family. It has been reported that RASF-1 interacts with Mst1 in vitro (Khokhlatchev et al. 2002). This is one of the reasons why we considered RASF-1 as a prototype of Rassf and used C. elegans as a small animal model in search for novel functions of conserved classical Rassf genes. In this study, we found that animals harboring the deletion allele for rasf-1 are more sensitive to oxidative stress than wild type. We next identified Rab GTPases RAB-39 as a binding partner of RASF-1 by yeast two-hybrid screening. Rassf proteins were originally designated on the basis of sequence homology to domains that associate with Ras-like small GTP-binding proteins. Various GTP-bound Ras-like proteins bind to effector proteins to mediate distinct biological responses. There are 150 Ras-like proteins encoded in the human genome which can be grouped by homology or function, as being similar to Ras, Rho, Rab, Arf (ADP-ribosylation factor) or Ran. Little is known about the GTP-binding proteins that may interact with the majority of the RASSF family or how they function. Rab GTPases are conserved regulators of multiple aspects of intracellular membrane trafficking and dynamics (Zerial & McBride 2001). RAB-39 is one of the Rab GTPases that is most closely related to the mammalian Rab39 GTPases. Rab39 functions as a membrane-associated GTPase required for intracellular vesicular trafficking and for regulation of endo- and exocytosis in mammals (Chen et al. 2003). Here, we show that rab-39 mutant and RNAi-treated animals were more sensitive to oxidative stress than wild type and that RASF-1 was shown to be genetically linked to RAB-39 in oxidative stress response by epistasis analysis. Additionally, Rassf1 and Rab39, mammalian homologs of rasf-1 and rab-39, respectively, have been shown to interact with each other in vitro, suggesting that this interaction is conserved also in mammals.
C. elegans rasf-1 is the ortholog of mammalian classical Rassfs
Searching of C. elegans database revealed only one worm homolog of Rassfs, T24F1.3, which we named rasf-1. rasf-1 is located on the second chromosome, and it has two variants, rasf-1a and rasf-1b. rasf-1a is composed of 10 exons (615 aa), while rasf-1b is of 8 exons (554 aa). rasf-1 encodes a RASSF family protein. tm5002 allele of rasf-1 lacks 308 nucleotides with an addition of CTAG in the genomic rasf-1 locus (Fig. 1A). Analysis of the amino acid sequence of the RASF-1 protein revealed significant homology to the mouse Rassf1A (mRassf1A). RASF-1 and mRassf1A share 41% amino acid identity in the RA domain, 44% identity in the SARAH domain and 53% identity in the N-terminus DAG (diacylglycerol/phorbol ester) domain (Fig. 1B). tm5002 allele produces mutated RASF-1 with loss of DAG domain. To visualize the expression pattern of rasf-1 in animals, we constructed rasf-1a cDNA with gfp fusion gene with 3 kb promoter region and injected it into wild-type animals. The expression of RASF-1::GFP was observed in pharynx, various neurons and vulva (Fig. 1C).
rasf-1 mutant animals are more sensitive to oxidative stress than wild type
rasf-1 mutant animals were viable and showed no obvious morphological aberrations. To determine the function of rasf-1, rasf-1 mutant animals were treated with an oxidative stressor, sodium arsenite. rasf-1 mutant animals were more sensitive to oxidative stress compared with wild type (WT) (Fig. 2A). To confirm the phenotypes observed in the rasf-1 mutant animals, we performed feeding RNAi of rasf-1 gene. Quantitative RT-PCR showed that the expression of rasf-1 gene was reduced by RNAi (Fig. 2B, upper left). Furthermore, rasf-1 knock down was verified at the protein level using Prasf-1::rasf-1::gfp transgene animals (Fig. 2B, upper right). Similar to rasf-1 mutant animals, rasf-1 knockdown animals were more sensitive to oxidative stress compared with those of control RNAi (Fig. 2B, bottom). These observations suggest that rasf-1 is involved in oxidative stress response in C. elegans.
RASF-1 proteins interact with Rab GTPase, RAB-39
To define the signaling mechanism of rasf-1, we screened a C. elegans cDNA library prepared from whole worms for genes whose protein products associate with RASF-1 by yeast two-hybrid system with a ‘bait’ plasmid containing sequences encoding full length of rasf-1 gene. Then, we obtained rab-39 Rab GTPase, a Ras superfamily small G-protein, as a candidate from transformants. To confirm the interaction between RASF-1 and RAB-39, each gene was subcloned into mammalian expression vectors to generate a FLAG epitope-tagged RASF-1 and HA epitope-tagged RAB-39, respectively. Co-immunoprecipitation experiments using HEK293T cells transiently transfected with these plasmids revealed that HA-RAB-39 is associated with FLAG-RASF-1 (Fig. 3A). It is known that Ras effector proteins specifically interact with GTP-formed Ras or Ras-like GTPases (van der Weyden & Adams 2007). Comparison of the amino acid sequence of the RAB-39 protein with mRab39A and mRab39B indicated that they are highly conserved entire sequences including a GXXXXGK(S/T) motif, involving switching from GDP to GTP form by GEF and DXXGQ motif, involving switching from GTP to GDP form by GAP (Bourne et al. 1991; Vetter & Wittinghofer 2001) (Fig. 3B). These regions were exploited for the construction of S- to N-dominant negative ‘GDP-locked’ and Q to L constitutively active ‘GTP-locked’ forms of the proteins for overexpression studies (Pal et al. 2003). To address whether the RASF-1 interacts with GTP-bound form of RAB-39 as other Ras-like GTPases, we generated RAB-39 mutants that mimic dominant negative ‘GDP-locked’ RAB-39S31N and constitutively active ‘GTP-locked’ RAB-39S77L forms by site-directed mutagenesis. HEK293T cells were transiently transfected with FLAG-rasf-1 and WT HA-rab-39 or mutated HA-rab-39. WT and HA-RAB-39S77L were co-immunoprecipitated with FLAG-RASF-1, but not HA-RAB-39S31N (Fig. 3C). These results suggest that RASF-1 interacts with GTP-bound active form of RAB-39.
rasf-1 is genetically linked to rab-39 in oxidative stress response
Although mammalian Rab39 functions as a membrane-associated GTPase required for intracellular vesicular trafficking and for regulation of endo- and exocytosis in mammals, the function of C. elegans rab-39 remains unclear. To investigate the function of rab-39, we focused on oxidative stress response from the analogy of rasf-1 and treated rab-39 mutant animals with sodium arsenite. As we expected, rab-39 mutant animals were more sensitive to oxidative stress compared with wild type (WT) (Fig. 4A). To confirm the phenotypes observed in the rab-39 mutant animals, we performed feeding RNAi of rab-39 gene. Quantitative RT-PCR showed that the expression of rab-39 gene was significantly reduced by RNAi (Fig. 4B, left). Under these conditions, rab-39 knockdown animals were more sensitive to oxidative stress compared with those of control RNAi (Fig. 4B, right). Moreover, we tested whether the rab-39 has genetic linkage with rasf-1. To address this, we performed RNAi of rab-39 in rasf-1 mutant animals and measured the sensitivity to the oxidative stress. rab-39 RNAi-treated rasf-1 mutant animals did not reduce the viability in the presence of sodium arsenite compared with the control RNAi animals (Fig. 4C). These data indicate that RASF-1 and RAB-39 function in the common pathway responding to oxidative stress.
Interaction of RASF-1 and RAB-39 is conserved beyond species
As RAB-39 has its mammalian ortholog, Rab39, we next tested whether the interaction is also conserved between the mammalian counterparts. There are seven different isoforms of Rassf1 (Rassf1A to Rassf1G) that are generated by differential usage of two promoters and through alternative splicing (Dammann et al. 2005). So far, however, the biological relevance of only two isoforms, Rassf1A and Rassf1C, has been described. Therefore, Rassf1A and Rassf1C were subcloned into a mammalian expression vector to generate HA epitope-tagged mRassf1A and mRassf1C, respectively. Rab39 gene has two isoforms, Rab39A and Rab39B. Rab39A is ubiquitously expressed in human and mouse tissues, whereas Rab39B is predominantly expressed in brain (Giannandrea et al. 2010). We constructed mammalian expression vectors for FLAG epitope-tagged mRab39A and mRab39B, respectively. To determine which forms of Rassf1 interacts with Rab39, HEK293T cells were transiently transfected with each isoform of HA-mRassf1 as well as FLAG-Rab39 and subjected to co-immunoprecipitation. Rab39A was co-immunoprecipitated with Rassf1A but not Rab39B (Fig. 5A). Rab39B strongly interacted with Rassf1C but only weakly with Rassf1A (Fig. 5B). These results indicate that the interaction which we found in C. elegans is conserved in mammals.
In the present study, we first identified and examined the C. elegans homolog of the tumor suppressor Rassf, RASF-1. RASF-1 has the characteristics of mammalian Rassf1, which possesses DAG, RA and SARAH domain. The RASF-1 expression was observed mainly in the nervous system and pharynx. In spite of no visible phenotype in development, both the deletion mutant and rasf-1 knockdown animals were shown to be more sensitive to oxidative stress. Rassf1 has three major interaction domains through which it imparts its functions. Each of these domains is involved in binding to different effecter proteins. The DAG domain binds to MDM2 to stabilize p53, and the RA and SARAH domains are required to activate the mammalian Hippo pathway (Foley et al. 2008; Song et al. 2008). p53 expression is induced by stress, which in turn results in the activation of a wide range of transcriptional targets. Low-intensity stress activates p53 gene in a manner that underlies antioxidant response (Borras et al. 2011). The DAG domain is lesioned in rasf-1(tm5002) mutant, and accordingly the activity of C. elegans p53, cep-1, might be decreased in the rasf-1 mutant animals. CEP-1 is also involved in stress resistance (Derry et al. 2001). It seems possible that p53 activity might be decreased in the rasf-1(tm5002) animals, which could increase sensitivity to the oxidative stress. Based on the distribution of RASF-1, it is likely that higher sensitivity of rasf-1 mutant or knock down animals to oxidative stress is ascribed to increased neuronal cell death.
Next, we isolated RAB-39 as a binding partner of RASF-1 by yeast two-hybrid screening. Rassf proteins interact with Ras-like proteins through its RA domain. For example, Rassf1A has been reported to interact with Rap1A Ras-like GTPase in vitro (Verma et al. 2011). Therefore, it is reasonable to assume that RASF-1 interacts with GTP-bound form of RAB-39 as we demonstrated also through its RA domain. Functionally, as in the case of rasf-1, rab-39 deletion mutant or rab-39 RNAi-treated animals were shown to be more sensitive to oxidative stress than WT animals. Moreover, genetic linkage was observed between rasf-1 and rab-39 in oxidative stress response, indicating that RASF-1 and RAB-39 function in a common pathway. Each Rab protein is specifically localized to intracellular membrane and regulates vesicular trafficking pathways, behaving as membrane-associated molecular switches (Stenmark 2009). It has been reported that Rab39 localizes at Golgi-associated organelles and is involved in cellular endocytosis in human dendritic cells (Chen et al. 2003). The Golgi apparatus is a pivotal organelle in cell metabolism and participates in modifying, sorting and packaging macromolecules for cell secretion or use within the cell. It is inevitably involved in the process of oxidative stress, which can cause modification and damage of lipids, proteins, DNA and other structural constituents (Jiang et al. 2011). Rassf1A functions as a scaffold protein in multiple signal transductions (Hwang et al. 2007). Thus, RASF-1 might be recruited to Golgi-associated organelles by RAB-39 and be involved in oxidative stress response there.
We demonstrated that the interaction that we found in C. elegans is conserved in mammals. Although the biological relevance of these interactions in mammalian cells remains unclear, our results may provide some insights into the role of Rab39 in functional differentiation of Rassf1 isoforms. We observed that Rassf1A and Rassf1C differently interact with each isoform of Rab39. Rassf1A and Rassf1C are two ubiquitously expressed isoforms (Zhou et al. 2012), but they have different functions. Rassf1C is not a tumor suppressor as Rassf1A in breast cancer cells, or Rassf1A and Rassf1C have opposite effect on beta-catenin degradation (Estrabaud et al. 2007; Reeves et al. 2010). The promoter of Rassf1A is inactivated in many cancers, whereas the expression of another major isoform, Rassf1C, is not affected (Estrabaud et al. 2007). Our observations suggest that the differential interaction of each Rassf1 isoform with Rab39 proteins might be related to different cellular localizations and functions. It would be intriguing to explore these aspects of Rassf1 and Rab39 along with their involvement in oxidative stress response in mammalian cells.
C. elegans strains and culture
All strains were maintained at 20 °C or 16 °C (temperature sensitive strains) on nematode growth medium (NGM) as described (Brenner 1974). Strains used in this study were N2 Bristol (wild-type), rasf-1(tm5002) (T24F1.3) and rab-39 (tm2466).
Assays for stress sensitivity
To assay the sensitivity to oxidative stress with arsenite, well-fed 1-day adult hermaphrodites were respectively picked and transferred to NGM agar plates containing 5 mm sodium arsenite. Worms were incubated at 20 °C, and the number of surviving worms was counted at the indicated times.
Feeding RNAi experiments were performed basically as described (Kamath et al. 2001). Fragments designated for RNAi were obtained by polymerase chain reaction (PCR) from each cDNA and were cloned into the L4440 feeding vector (pPD129.36). HT115 strain of E. coli was transformed with the RNAi vectors and cultured with IPTG before feeding.
The rasf-1 promoter region, a fragment containing 3 kb upstream of the initiation ATG, was amplified by PCR and subcloned into pPD95.75 GFP vector by the HindIII - XbaI sites (pPD95.75::Prasf-1). For constructing Prasf-1::rasf-1::gfp, a PCR fragment of the rasf-1 gene with the XbaI - BamHI sites was subcloned into pPD95.75::Prasf-1. rasf-1 and rab-39 cDNA were obtained from C. elegans 1st strand cDNA by PCR. Then, they were subcloned into pFLAG-CMV and pCMV-HA vector, respectively. Mouse Rassf1s and Rab39s cDNA were obtained from mouse cDNA by PCR. Then, they were subcloned into pCMV-HA and pME-FLAG vector, respectively.
The plasmid containing Prasf-1::rasf-1::gfp was injected, in combination with the rol-6 morphologic marker plasmid pRF4, into the distal gonad of young adult WT hermaphrodites at a concentration of 50 ng/μL for Prasf-1::rasf-1::gfp and 50 ng/μL for rol-6. Transgenic animals were identified by their characteristic movement (a roller phenotype).
Total RNA was isolated from young adult worms using Sepasol-RNA I Super (Nacalai Tesque, Inc., Kyoto, Japan). Reverse transcription was performed with ReverTra Ace reverse transcriptase (TOYOBO, Osaka, Japan), followed by quantitative real-time PCR using ThunderBird SYBR qPCR mix.
For the co-immunoprecipitation assay, HEK293T cells were transfected with plasmids by X-treme GENE9 (Roche, Basel, Switzerland). Forty-eight hours after transfection, cells were lysed with lysis buffer (50 mm NaH2PO4 at pH 7.5, 150 mm NaCl, 0.5% NP-40, protease inhibitor cocktail (Nacalai Tesque) and phosphatase inhibitor cocktail (Nacalai Tesque)) and immunoprecipitated with Protein G-sepharose (Invitrogen, CA) and anti-HA (12CA5) (Roche) or anti-FLAG (M2) (Sigma, MO) antibodies. The beads were washed three times with PBS, and the immunoprecipitates were eluted in SDS sample buffer.
C. elegans strains were provided by the Caenorhabditis Genetics Center, which is funded by the NIH National Center for Research Resources (NCRR) and National Bio Resource Project (NBRP, Japan). We thank members of the Hori laboratory for technical comments and helpful discussion.