RasGrf1 and RasGrf2
The Ras guanine nucleotide releasing factors (RasGrfs) were initially identified as a result of a search for mammalian homolog(s) of the yeast CDC25 Ras activator protein. Whereas other mammalian RasGEFs identified (belonging to the SOS and GRP families) are more widely expressed, the RasGrf1 and RasGrf2 members of the RasGRF family are preferentially, but not exclusively, expressed in the central nervous system. RasGrf1 and RasGrf2 are large proteins composed by multiple modular domains accounting for protein-protein or protein-lipid interactions which are responsible for functional coupling to upstream and downstream signaling and for fine regulation of their intrinsic exchange activity.
Initial studies showed that both RasGrf1 and RasGrf2 are able to activate canonical Ras proteins (H-Ras, N-Ras, or K-Ras) and Rac1, but only RasGrf1 is able to activate members of the R-Ras subfamily (R-Ras, TC21, M-Ras). Furthermore, both RasGrfs are known to be activatable in response to increases of intracellular levels of calcium or cAMP, activation of heterotrimeric G-protein-coupled receptors, or NGF receptor stimulation. More recently, the characterization of genetically modified mouse models has yielded significant clues to physiological roles played by the RasGrfs. Thus, these studies have documented the implication of RasGrf1 in memory and learning (LTD responses and HFS-LTP in mice over 2 months of age), postnatal growth, pancreatic β-cell function and glucose homeostasis, neurosensory processes and photoreception or the participation of RasGrf2 in memory formation (TBS-LTP responses), T-cell signaling responses, retinal morphogenesis and predisposition to alcohol addiction (Fernandez-Medarde and Santos 2011; Hysi et al. 2010; Jin et al. 2014).
RasGrfs Protein Structure and Domain Distribution
PH1 domain – Although many PH domains mediate protein targeting to membranes, a RasGrf1 construct lacking this domain retains partial plasma membrane localization. The PH1 domain is reported to interact with G-protein β/γ subunits and is required for G-protein-induced ERK1/2 activation (Fernandez-Medarde and Santos 2011). In addition, together with the IQ and coiled coil domains, the PH1 domain is involved in interaction with scaffold proteins such as JIP2 or spinophilin or with ribosomal proteins. This domain is also necessary for complete calcium- and LPA-mediated activation of RasGrf1, and it is phosphorylated upon interaction with the TrkA receptor (Fernandez-Medarde and Santos 2011).
Coiled coil (cc) domain – All functional roles for this domain have been described in association with the flanking PH1 and IQ domains. As described above, its contribution is important for complete ERK1/2 activation upon intracellular calcium increase and for interaction with the JIP2 scaffold protein (Fernandez-Medarde and Santos 2011).
IQ domain – The main role of this domain is the interaction with Calmodulin, which is essential for activation of both RasGrfs by increases of intracellular calcium concentration. Mutations in this domain abolish ionomycin-mediated activation of RasGrf1 and stimulation of the Ras-ERK pathway (Fernandez-Medarde and Santos 2011). Recent work analyzing protein chimeras resulting from interchanging different amino terminal domains of RasGrf1 and RasGrf2 has shown that the IQ domain of RasGrf1 alone is sufficient to determine whether a RasGrf protein can induce LTD responses (Jin et al. 2014).
DH domain – The DH-PH2 tandem constitutes the canonical exchange domain for GTPases of the Rho subfamily. In the RasGrf proteins, this region is responsible for activation of Rac1. The DH domain also exerts a regulatory role on Ras activation by RasGrf1 and may also be important for Ras-independent, ionomycin-induced ERK activation. This domain also mediates protein-protein interactions of RasGrf1 with β-tubulin and SCLIP, and it is involved in homo- and hetero-oligomerization of the RasGrf1 and RasGrf2 proteins (Fernandez-Medarde and Santos 2011).
PH2 domain – A defined role for this domain remains unclear. In RasGrf1, some reports indicate that it is required for proper Ras and ERK activation, whereas studies in different cellular systems suggest that this domain is dispensable for Ras or ERK activation. Further work is needed to clarify these discrepancies analyzing RasGrf function in more physiological environments.
REM motif – This region is common to all Ras guanine nucleotide exchange factors, and it is responsible for GEF-Ras interaction. It is likely that the REM motif has a similar role in all the Sos, RasGRP, and RasGrf1 families of GEF proteins.
CBD motif – This short domain usually targets proteins for degradation by the proteasome. It is located between the REM and CDC25H domains. Its involvement with ubiquitination and proteolytic degradation has been shown experimentally only for RasGrf2.
PEST motif rich region – PEST [Proline (P), Glutamic acid (E), Serine (S), Threonine (T)] regions are known targets for calpain-type protease degradation, a role that has been experimentally demonstrated in RasGrf1. This area is also targeted in RasGrf1 for phosphorylation by PKA or upon muscarinic receptor-mediated activation by CDK5 in both RasGrfs, inhibiting the GEF activity towards Rac1 (RasGrf2) or Ras (RasGrf1) (Fernandez-Medarde and Santos 2011).
Neuronal domain – Only found in RasGrf1, it is responsible for RasGrf1 binding to the NR2B subunit of the NMDA receptors and activation of downstream pathways (Feig 2011).
CDC25H domain – This C-terminal domain contains the catalytic region responsible for GDP/GTP exchange (GEF activity) on Ras family members. It is shared by all GEFs acting on canonical Ras proteins and shows a high degree of conservation through evolution. This domain is necessary and sufficient for “in vitro” Ras activation by RasGrf1, and it is also responsible for Ras activation by RasGrf2. “In vivo” modulation of the GEF activity of the CDC25H domain in full length RasGrf proteins can be exerted through various intramolecular interactions or biochemical modifications (Fernandez-Medarde and Santos 2011).
Control of RasGrf Expression and Cellular Protein Levels
RasGrf1 is an imprinted gene expressed only after birth. In mice, the paternal allele of the RasGrf1 locus is methylated on a differentially methylated domain (DMD) located 30 kbp 5’ of the promoter. In mouse neonatal brain, RasGrf1 expression occurs exclusively from the paternal allele and accounts for ca. 90% of total Rasgrf1 expression in the adult mice. A repeat sequence located immediately downstream of the DMD controls its methylation and is therefore required to establish RasGrf1 methylation in the male germ line. CTCF (a CCCTC-binding factor) binds to the DMD in a methylation-sensitive manner, acting as an “enhancer blocker.” In the unmethylated maternal allele, CTCF is bound to DMD silencing expression, whereas CTCF cannot bind to the methylated paternal allele, thus allowing expression. The repeats and the DMD constitute a dual switch regulating RasGrf1 imprinting and timing of expression (Fernandez-Medarde and Santos 2011).
RasGrf2 is not an imprinted gene, but genomic methylation may still play a role in control of its expression. The shortage of specific studies on RasGrf2 expression determines that the mechanisms controlling expression in the postnatal brain remain largely unknown. Interestingly, in colon, pancreatic, lung (NSCLC) tumors and cell lines, hypermethylation of RasGrf2 locus and reduced protein expression are frequently associated (Fernandez-Medarde and Santos 2011).
Some external factors are also known to modulate RasGrf1 expression. Cocaine induces overexpression in dorsal and ventral striatum, whereas the Alzheimer-related amyloid precursor protein with the Swedish mutation (APPSw) or oncogenic ErbB2/Neu and luteinizing hormone repress its expression in hippocampus and mammary gland, respectively (Fernandez-Medarde and Santos 2011).
Control of Proteolytic Degradation
The intracellular concentration of the RasGrf proteins is regulated by cellular proteases. Both RasGrfs contain a type A cyclin destruction box (CDB), located between the REM and CDC25 domains (Fig. 1). These domains trigger ubiquitination and degradation of RasGrf2 by the proteasome upon Ras binding. There is no experimental evidence showing a similar role in RasGrf1, although the interaction between RasGrf1 and the deubiquitinating enzyme mUBPy results in increased RasGrf1 half-life and, in M2 melanoma cells, the actin binding protein Filamin A induces destabilization and ubiquitination of RasGrf1 with subsequent reduction of MMP9 expression (Fernandez-Medarde and Santos 2011).
Calpain can also cleave RasGrf1, but the functional significance of this is unclear. Some reports described that this cleavage increases RasGrf1 GEF activity towards Ras proteins by releasing the C-terminus from inhibition by the N-terminal portion, whereas other studies show that phosphorylation by p35/CDK5 targets RasGrf1 for proteolysis by m-calpain, resulting in reduced Ras activation and AKT phosphorylation. Phosphorylation of RasGrf2 by p35/CDK5 also results in accumulation and increased concentration of RasGrf2 in the body of neurons (Fernandez-Medarde and Santos 2011).
Signaling Through the RasGrfs
Activation of RasGrfs in Response to Increase in Intracellular Calcium Concentration
RasGrf2 can also be activated by increased intracellular calcium concentration. In 293 T cells, calcium influx causes translocation of RasGrf2 to the cell periphery, localizing it close to membrane GTPases, through a mechanism not yet fully understood. Although RasGrf2 interacts with calmodulin through its IQ domain, a full, functionally productive interaction appears to need the additional cooperation of other N-terminal domains (Fernandez-Medarde and Santos 2011). Interestingly, a RasGrf2 construct lacking the IQ domain is still able to activate Ras, suggesting that interaction with calmodulin is not necessary for RasGrf2 GEF activity, although both the IQ domain and Calmodulin interaction with RasGrf2 are indispensable for ERK activation. These observations suggest the existence of Ras-independent mechanisms of ERK1/2 activation or the need of other interacting partners for full coupling of Ras and ERK1/2 activation (Fernandez-Medarde and Santos 2011).
Activation of RasGrfs in Response to G-Protein Coupled Receptors (GPCR)
The activity of RasGrf1 can be enhanced by stimulation with LPA or serum, but not with PDGF. This activation is inhibited by pretreatment with pertussis toxin but not with genistein, suggesting that GPCRs, but not receptor tyrosine kinases, play a role in serum activation of RasGrf1 (Fernandez-Medarde and Santos 2011). LPA treatment of NIH3T3 cells overexpressing RasGrf1 induces phosphorylation of RasGrf1 in serine residues, and calcium is also needed for full GEF activation. Overexpression of G-protein βγ subunits and LPA treatment can also induce RasGrf1 GEF activity towards Rac1, leading to JNK and c-fos promoter activation, but to fully activate Rac1, RasGrf1 needs to be phosphorylated in tyrosine by Src (Fernandez-Medarde and Santos 2011) (Fig. 2). Other GPCRs can also activate RasGrf1. Overexpression of subtype1 Muscarinic receptor induces RasGrf phosphorylation and its GEF activity upon carbachol stimulation. This activation is prevented by phosphatases and G-protein α-subunit overexpression and is constitutive when G-protein βγ subunits are overexpressed. Overexpression of the 5-HT4 serotonin receptor also induces RasGrf1 phosphorylation by PKA (at Serine 916) and IQ-dependent activation upon serotonin stimulation, suggesting that the complete activation of RasGrf1 by these receptors involves both cAMP and calcium/calmodulin-dependent signaling (Fig. 2) (Fernandez-Medarde and Santos 2011).
RasGrf2 can also be activated by GPCRs, as shown by reports indicating that α-thrombin, a potent mitogen acting through PAR1, activates N-Ras through RasGrf2 in IIC9 fibroblasts (Fernandez-Medarde and Santos 2011).
Activation of RasGrfs in Response to Receptor and Nonreceptor Tyrosine Kinases
In PC12 cells, the TrkA nerve growth factor (NGF) receptor interacts with, and induces phosphorylation of, RasGrf1 in its PH1 domain (Feig 2011) (Fig. 2). There are conflicting reports regarding how RasGrf1 potentiates NGF-induced PC12 differentiation. Whereas early work indicated that this process is dependent on H-Ras and ERK1/2 but independent of Rac1 or PI3K pathways, a recent report claims that activation of both Ras and Rac are required for neurite outgrowth (Fernandez-Medarde and Santos 2011; Talebian et al. 2012).
The Src tyrosine kinase can mediate transduction of G-protein-dependent signals from RasGrf1 to Rac1 proteins, but not to Ras canonical proteins (Fernandez-Medarde and Santos 2011). Other tyrosine kinases, such as ACK1 and Lck, are also able to phosphorylate RasGrf1, resulting in enhanced Ras GEF activity. In particular, ACK1 is known to be activated by Cdc42, which in its inactive GDP-Cdc42 conformation is also able to inhibit RasGrf1 activation of Ras (Fernandez-Medarde and Santos 2011). A similar mechanism may be applicable to RasGrf2, as the expression of dominant negative Cdc42 (Cdc42N17) abolishes RasGrf2 recruitment to the plasma membrane, Ras activation, and ERK phosphorylation. RasGrf2 and RasGrf1 are activated upon T-cell receptor stimulation, a process requiring the contribution of tyrosine kinase(s) of the Src family. Activation of RasGrf2 induces Ras-dependent and PLC-γ1-mediated signaling pathways, producing the activation of NF-AT, a transcriptional factor crucial for T-cell activation and differentiation (Fernandez-Medarde and Santos 2011).
GEF Activity-Independent Functional Roles of RasGrf1 and RasGrf2 in Cell Signaling
A number of functional roles recently described for the RasGrfs cannot be explained only by their biochemical, guanine nucleotide exchange activities. For example, both RasGrf1 and RasGrf2 have been shown to bind to, and prevent, the activation of CDC42, a key regulator controlling specific steps in processes of cytoskeletal actin dynamics, cell polarity establishment, or cellular and organelle movement that are relevant in physiological neurogenesis and in tumorigenesis (Calvo et al. 2011). The RasGrfs have also been reported to interact with the Golgi matrix protein GM130 forming a complex that regulates the amount of Golgi-associated CDC42 to control cell polarity (Baschieri et al. 2014). In addition, both RasGrfs can act as adaptors that bind PLCγ1 to mediate IL-1 induction of calcium-mediated ERK activation and MMP-3 expression in the regulation of extracellular matrix remodeling induced by IL-1 (Wang et al. 2013).
Functional Studies Using Genetically Modified Mouse Models of the RasGrfs
Phenotypic characterization of a number of independently generated RasGrf1 KO strains has contributed to a better understanding of the functional “in vivo” roles of RasGrf1. In particular, the RasGrf1 null mice have longer lifespan than WT animals (Borras et al. 2011) and display reduced or retarded postnatal growth and defective abilities regarding memory consolidation and learning, glucose homeostasis, eye lens formation, and light photoreception (Fernandez-Medarde and Santos 2011).
An initial report on RasGrf1-KO mice described LTP defects and impairment of amygdala-dependent learning, whereas later studies on a separate KO strain reflected LTD defects and impairment of hippocampus-dependent learning (Feig 2011). The discrepancy may be due to different gene-targeting strategies or different mouse genetic backgrounds. More recently, RasGrf1 has also been implicated in olfactory learning and memory (Drake et al. 2011), the formation of dendritic spines (DiBattista et al. 2015), and in adult neurogenesis in the dentate gyrus (Darcy et al. 2014a).
Regarding drug addictions, the studies analyzing cannabinoid tolerance in RasGrf1 KO mice showed reduced tolerance to Δ9-tetrahydrocannabinoid, probably through alterations in cannabinoid receptor- and cAMP-mediated signaling (Feig 2011). In addition, elimination of RasGrf1 also causes attenuation of the locomotor sensitization and conditioned place preference observed upon cocaine treatment, possibly through lower ERK activation in the striatum (Feig 2011).
RasGrf1 may also be important to maintain normal photoreception as the RasGrf1 KO mice develop light perception defects which worsen with age (Fernandez-Medarde and Santos 2011) and also show smaller and heavier lenses suggesting a role for this particular GEF in lens morphogenesis (Hysi et al. 2010).
Regarding the control of postnatal growth, adult RasGrf1 null mice are 15–25% smaller than wild-type controls, a phenotype probably associated with lower levels of growth hormone and circulating plasma insulin in RasGrf1-deficient mice. This hypoinsulinemia together with reduced pancreatic beta-cell mass in the KO mice supports a significant role of RasGrf1 in control of beta cell proliferation and neogenesis (Fernandez-Medarde and Santos 2011).
Although RasGrf2 is dispensable for mouse development, postnatal growth, fertility, and aging, and the RasGrf2 KO mice are morphologically indistinguishable from wild-type controls (Fernandez-Medarde and Santos 2011), recent studies support its relevance in several physiological processes. Thus, RasGrf2 KO mice show lower levels of ERK activation upon NMDA-induction and defective LTP in the CA1 region of the hippocampus (Feig 2011). Additionally, RasGrf2 has also been shown to be important for development of LTP and survival of newborn neurons in the dentate girus (Darcy et al. 2014b), for morphogenesis of the retina (Jimeno et al. 2016) and for the control of dopaminergic and noradrenergic responses to alcohol (Stacey et al. 2012).
Additional insights into the role of RasGrf2 “in vivo” have been obtained through the analysis of mice harboring combinations of a disrupted RasGrf2 locus with null mutations for other GEFs. Thus, elimination of both RasGrf2 and RasGrf1 results in higher sensitivity to the neurotoxic effects of ischemia in the mouse brain. Furthermore, analysis of RasGrf2/Vav3 and RasGrf2/Vav1 null mice suggest a role for RasGrf2 in T-receptor signaling responses in lymphocytes (Fernandez-Medarde and Santos 2011).
Implication of the RasGrfs in Pathologies
In addition to the functional roles uncovered for the RasGrfs by studies of KO animal models (RasGrf1 implication in predisposition to myopia and refractive errors of vision; link of RasGrf2 depletion to defective retinal cone photoreception; reduced tolerance of RasGrf1 KO mice to cannabinoid/cocaine treatments; or predisposition of RasGrf2 KO mice to addictive alcohol abuse), an increasing number of studies suggest their implication in other pathological conditions. Thus, an altered methylation has been long detected in the RasGrf2 promoter in several malignancies (Fernandez-Medarde and Santos 2011). Other reports document the functional implication of RasGrf1 in metastatic human alveolar rhabdomyosarcoma (Tarnowski et al. 2012), epilepsy (Zhu et al. 2012), or increased sensitivity to candidiasis (Gavino et al. 2015).
The RasGrfs are the main GEF activators of mammalian Ras GTPases in the adult central nervous system. In vivo evidence indicates that both RasGrfs are able to activate the canonical Ras proteins (H-, N-, and K-Ras) and Rac1. RasGrf1 and RasGrf2 are large, highly homologous proteins sharing a modular structure composed of multiple functional domains which regulate their GEF activity and modulate their participation in signal transduction, connecting different upstream signals to their downstream targets and elicited cellular responses. The GEF activity of the RasGrfs becomes activated in response to several cellular signals including LPA, increased cytosolic concentration of calcium or cAMP, and activation of cell surface receptors for NMDA, AMPA, serotonin, muscarinic agonists (G-protein coupled receptors), or NGF (trkA) (Fig. 2). Known downstream effects of the participation of the RasGrfs in cellular signaling pathways include control of cellular shape and nuclear organization, neurite extension, neuronal synaptic plasticity, and induction of LTP or LTD. Analysis of genetically modified animals has uncovered the specific roles of RasGrf1 in memory and learning, postnatal growth, pancreatic beta cell proliferation, lens development, retinal photoreception, and neuroprotection against ischemia. Likewise, RasGrf2 has also been implicated in memory formation, neuroprotection, retinal morphogenesis, and immunological responses in lymphocytes. The participation of the RasGrfs in human diseases is also suspected. Increased RasGrf2 gene methylation is frequently observed in various human tumors and cancer cell lines. Other observations suggest the implication of RasGrf1 in cancer, vision defects, epilepsy, drug addiction, and Alzheimer-like neurodegenerative diseases.
In spite of the vast amount of published information on the RasGrf proteins, a number of key questions still remain unanswered. Because of their prevalent expression in the CNS, most functional studies on the RasGrfs have been restricted to neural tissues and cell lineages. However, as both RasGrfs are also expressed outside the CNS, their functional roles at those locations remain less defined and require further studies. Another poorly understood area is the functional significance of the variety of small transcripts and peptides detected for both RasGrf1 and RasGrf2 in many tissues and/or states of development. An interesting hypothesis would be that those isoforms may contribute to the fine-tuned regulation of the activation of their cellular Ras/Rho targets at the spatial and temporal level. A better understanding of the mechanisms linking functional observations made for the RasGrfs at the cellular level with those made at the organism level would also be desirable. This pertains questions such as: Is the participation of RasGrf1 in cytoskeleton remodeling necessary for neuritogenesis? Is there a connection between the role of RasGrf1 in neuritogenesis and its contribution to memory formation processes? Finally, future work efforts should be aimed at getting definitive answers to the current hints linking the RasGrf proteins to the development of different human illnesses and pathological processes.