Rap GEF Family
List of Discussed GEFs
RapGEF2: PDZ-GEF1, RA-GEF, CnRasGEF, KIAA0313, nRapGEP
RapGEF3: Epac, cAMP-GEF I
RapGEF4: Epac2, cAMP-GEF II
RapGEF5: MR-GEF, Repac, GFR, KIAA0277
RapGEF6: PDZ-GEF2, RA-GEF2
RasGRP2: CalDAG-GEF I
RasGRP3: GRP3, CalDAG-GEF III, KIAA0846
GTP binding proteins are extensively used in nature to regulate biological processes. Most of these proteins act as molecular switches that transition between inactive GDP-bound and active GTP-bound conformations. The largest family of GDP/GTP switches is the Ras superfamily of small-molecular-weight GTP-binding proteins that constitute approximately 150 members in mammalian cells. This superfamily can be subdivided into Ras, Rho, Rab, Ran, and ARF families that regulate a myriad of cellular functions (Wennerberg et al. 2005). True Ras proteins play major roles in coupling cell surface receptors to intracellular signaling pathways that control proliferation, differentiation, and survival. Due to their mutational activation in over 20% of human cancers, much research has focused on H-, K-, and N-Ras (Hobbs et al. 2016). In the shadow of these oncoproteins, there are additionally over 30 Ras-related proteins that include the Rap proteins, Rap1A, 1B, Rap2A, 2B, and 2C. These proteins were initially discovered due to sequence similarity to Ras, their ability to revert the actions of Ras, and their abundance in leukocytes. More recently Rap1 has been shown to play multiple roles that include inside-out signaling to control the affinity of integrins for extracellular matrix and consequently cell adhesion and migration. Rap1 also localizes to adherens junctions where it influences cell–cell adhesion and plays a number of additional roles in cell signaling, often overlapping with those of Ras (Gloerich and Bos 2011).
So how do Ras/Rap proteins act as molecular switches and how does the extracellular environment control their activity? Ras proteins normally exist in an inactive or resting GDP-bound state. To acquire an active GTP-bound confirmation, they must exchange nucleotide. An intrinsic GTPase activity then hydrolyzes the terminal phosphate off the bound GTP enabling return to the inactive confirmation. The intrinsic rates of nucleotide exchange and GTP hydrolysis are in the order of minutes to an hour, not practical for rapid response to the extracellular environment. This means that additional factors must exist to both accelerate and tightly regulate these processes. Guanine nucleotide exchange factors (GEFs) promote the release of GDP, whereas GTPase-activating proteins (GAPs) have evolved to assist in the rapid hydrolysis of GTP (Quilliam et al. 2002). Whereas inhibition of GAPs and activation of GEFs both tip the balance in favor of Ras-GTP accumulation, nature most frequently uses GEFs to activate Ras proteins and enable biological responses. This article will focus on the nature and regulation of Rap family GEFs.
Although there are approximately equal numbers of Ras subfamily members and GEFs, there is not a simple monogamous pairing of GEFs to Ras proteins. Significantly more GEFs have evolved to regulate Rap than any other group (Fig. 1). Rap GEFs contain multiple regulatory domains suggesting that Rap proteins are under the control of diverse extracellular stimuli. Regulation includes protein–protein or protein–lipid interactions, binding of second messengers, and/or posttranslational modifications. The need for so many GEFs may be to activate Rap in different tissues or different locations within a given cell, at key points in development, or in response to different hormones/growth factors, etc., that use unique signaling mechanisms. Recent knockout studies in mice have supported the notion that Rap1A and 1B play fundamental roles in mammalian development and function, and likewise, creation of mice lacking various Rap GEFs (discussed below) support the notion that individual exchange factors play equally critical functions.
One of the earliest mammalian Ras family GEFs to be isolated was the Crk SH 3-binding guanine nucleotide exchange factor (C3G) (Quilliam et al. 2002; Gloerich and Bos 2011). In addition to the CDC25 homology and REM domains, C3G contains central proline-rich sequences that bind to the N-terminal SH3 domain of the Crk adapter protein, a more N-terminal proline-rich motif that associated with p130Cas, c-Abl, and Hck SH3s, and a central tyrosine residue (Y504) that is phosphorylated by Src family tyrosine kinases (Fig. 1). C3G is perhaps the most ubiquitously expressed Rap GEF. It is activated by numerous cell surface molecules that include integrins, T and B cell receptors, E-cadherin, and various growth factors and G-protein-coupled receptors.
C3G was initially found to activate Rap1 but has since been reported to also promote nucleotide exchange on several other small GTPases: Rap2, R-Ras, and TC10. Interestingly, TC10 is a member of the Rho family of GTPases that are typically regulated by a distinctly different family of GEFs. TC10 plays a role in insulin-stimulated glucose uptake and in 2010 a Korean study linked mutations within C3G to type II diabetes (Hong et al. 2009). Two C3G knockout mice were created and demonstrated embryonic lethality, exemplifying its important biological role. C3G is frequently localized through its interaction with the Crk and CrkL SH2-/SH3-containing adapter proteins. For example, the Abl tyrosine kinase can recruit a CrkL–C3G complex to the immune synapse.
RapGEF3 and 4/Epac1 and 2
The second messenger cyclic adenosine monophosphate (cAMP) exerts many effects on cell biology. While these were classically known to be mediated by the cAMP-dependent protein kinase/protein kinase A (PKA) and olfactory cyclic nucleotide-gated ion channels, two cAMP-activated Rap GEFs were discovered in 1998 that helped explain the PKA-independent actions of cAMP (Quilliam et al. 2002; Gloerich and Bos 2011). They are called Epacs (exchange proteins activated by cAMP) 1 and 2 or cAMP-GEFs. While the former name is most popular (and used herein) their official gene names are RapGEFs 3 and 4.
In addition to the REM/CDC25 exchange region that acts on Rap1 and 2, Epac proteins have either one (Epac1) or two (Epac2) cyclic nucleotide-binding domains (CNBD); a Disheveled, Egl-10, Pleckstrin (DEP) domain; and an RA domain (Fig. 1). Structural studies on Epac2 indicate that the N-terminus folds over the C-terminus and hinders Epacs binding to its substrate Rap. This autoinhibition is relieved by the binding of cAMP to the CNBDs (Banerjee and Cheng 2015). Epacs’ CNBDs lack a glutamate residue found in PKA and cAMP-gated ion channels that typically interacts with the 2-OH group of the ribose of cAMP. Consequently, Epacs can bind bulky cAMP analogs such as 8-(4-chloro-phenylthio)-2′-O-methyladenosine-cAMP (8CPT) where the 2-OH group has been replaced with a O-Me to selectively activate Epac proteins versus other cAMP effectors. These compounds have proven to be very useful tools to implicate Epac in biological events and may have clinical potential (Gloerich and Bos 2011). Additional Epac antagonists have also recently become available (Banerjee and Cheng 2015).
Epacs, like most other GEFs and GAPs, regulate Rap proteins in a spatial and temporal manner. For instance, localization of Epac1 and 2 is regulated by their distinct RA domains. The RA domain of Epac2 specifically binds K- and N-Ras (with weaker affinity for H-Ras), enabling Ras-GTP to translocate Epac2, but not Epac1, from the cytosol to the plasma membrane. Consequently, a pool of plasma membrane-bound Rap1 can become activated upon concurrent cAMP and Ras signaling (Li et al. 2006). Meanwhile, the RA domain of Epac1 interacts with Ran, a small G-protein best known for its role in regulating nuclear transport. Ran and its binding partner RanBP2 anchor Epac1 to the nuclear pore, allowing localized Rap1 activation at the nuclear envelope upon cAMP elevation (Gloerich and Bos 2011). In a different scenario, Epac1 can be targeted to the plasma membrane via its DEP domain. This is necessary for Rap to regulate integrin-mediated adhesion at the membrane. However, in Rat1a fibroblasts, peripheral Rap1 activation by Epac1 is counteracted by high RapGAP activity, resulting in predominantly perinuclear Rap-GTP (Gloerich and Bos 2011). Furthermore, in both interphase and mitotic cells, Epac1 is targeted to microtubules by tubulin or the microtubule-associated protein 1 and may play a role in microtubule polymerization. Other reports show Epacs localizing to centrosomes, mitochondria, macrophage phagosomes, the apical epithelial membrane, and regulating the DNA damage–responsive kinase, DNA-PK, in the nucleus. Temporal expression also contributes to Epac action: As monocytes differentiate into macrophages, their Epac protein levels increase threefold and play a role in chemokine secretion.
Epacs regulate a variety of physiological processes that include secretion of insulin from pancreatic beta cells, permeabilization of vascular endothelium, transmigration of leukocytes, and regulation of cardiac calcium channels. Consequently, Epac activity has been associated with diabetes, vascular inflammation, and heart disease. Interestingly, PKA is also involved in these processes demonstrating the close partnership of Epac and PKA in mediating cAMP action.
Using Epac2 knockout mice, Seino’s lab established that cAMP potentiates glucose-induced exocytosis via Epac2 rather than PKA. A number of studies have implicated Epacs in inflammation both through regulation of leukocytes and vascular permeability (Borland et al. 2009). In cardiac myocytes, Epacs, and another Rap GEF, phospholipase Cε both play critical synergistic roles in calcium-induced calcium release downstream of β-adrenergic receptors.
RapGEF2 and 6/PDZ-GEF1 and 2
Similar in structure to Epacs are PDZ-GEF1 and 2, also called RA-GEF-1 and 2, or official names RapGEFs 2 and 6. PDZ-GEF1 has also been described as CNrasGEF or nRapGEP. Like Epacs, PDZ-GEFs have a REM-CDC25 GEF module, an RA domain, and a region sharing homology with CNBDs (Quilliam et al. 2002; Gloerich and Bos 2011) (Fig. 1). However, reports differ as to whether cAMP can bind to this latter region with high affinity. PDZ-GEFs have a PSD-95/DlgA/ZO-1 (PDZ) domain that can bind the β1 adrenergic receptor, potentially linking G-protein-coupled receptors to Ras activation. At its C-terminus, PDZ-GEFs have a proline-rich region and a PDZ-binding motif, which interact with the PDZ domains of cell junctional proteins, MAGI-1 and -2 (Sakurai et al. 2006). This links PDZ-GEF with β-catenin and contact-induced activation of Rap1 (Sakurai et al. 2006). In addition, two PY motifs at its C-terminus are responsible for binding to the WW domain of the ubiquitin protein ligase Nedd4. This interaction regulates PDZ-GEF protein turnover rate via proteasomal degradation.
PDZ-GEF1 and 2 functions have been studied through gene knockouts in both Drosophila and mice. In Drosophila, PDZ-GEF (Gef26) regulates DE-cadherin to control stem cell adhesion to its niche. Functional mutation of this GEF leads to loss of cell polarity, impaired adherens junctions, and thus reduction in stem cell number (Wang et al. 2006). Two labs found that PDZ-GEF knockout is embryonic lethal in mice; vascular development was impaired in PDZ-GEF1−/− mice at around E7.5 with embryonic lethality occurring by E9.5, while PDZ-GEF2−/− mouse embryos died of vasculature defects in the yolk sac and the allantois at ~E11.5. In addition, deletion of PDZ-GEF2 late in embryogenesis resulted in defective fetal liver erythropoiesis. However, such deletion in the adult bone marrow or specific deletion in B-cells, T-cells, hematopoietic stem cells, or endothelial cells had no impact on hematopoiesis (Satyanarayana et al. 2010). This reiterates the importance of temporal control and the role of various GEFs at different stages of development.
Conditional knockout of PDZ-GEF showed other crucial functions of this GEF in neural migration and in splenocyte responses. Meanwhile, PDZ-GEF2 and Rap1 mediate TNFα-induced M-Ras activation in order to activate the integrin, lymphocyte function-associated antigen 1 (LFA-1), and subsequently, cell aggregation in response to inflammation (Yoshikawa et al. 2007).
Others also reported PDZ-GEF2 interacts through its PDZ domain with junctional adhesion molecule-A (JAM-A), which also interacts with Afadin/AF6 in human colonic epithelial cells. JAM-A and AF6 both act upstream of a signaling pathway that specifically activates Rap1A but not Rap1B in order to regulate β1 integrin and mediate cell migration (Severson et al. 2009). This is a rare report of differential signaling to Rap1A versus Rap1B. Double knockout of PDZ-GEFs 2 and 6. Recent mouse studies support the involvement of both PDZ-GEFs in maintaining the apical surface of adherens junction structures in neural progenitor cells (Maeta et al. 2016).
MR-GEF was characterized by numerous groups, as a Rap-specific GEF (acting on Rap1 and 2) (Quilliam et al. 2002; Gloerich and Bos 2011). Due to sharing highest homology to Epacs, it was referred to as Repac or alternatively as MR-GEF (M-Ras regulated) due to the presence of an RA domain that selectively bound to M-Ras-GTP. M-Ras overexpression inhibited Rap1 activation, but based on the experiences of us and others, it is likely that M-Ras is specifically targeting the GEF to activate a plasma membrane pool of Rap1 at the expense of the activity of bulk GTPase (Li et al. 2006). An apparent splice variant that swaps the first 70 amino acid residues for an alternative 208 residue sequence (NP_036426) places a DEP domain at the extreme N-terminus that may play a role in membrane localization. A DEP domain is also found in Epacs a similar distance from the REM domain (Fig. 1). However, unlike Epacs, MR-GEF does not contain an intervening cAMP-binding motif.
MR-GEF expression was induced by exposure to anthrax and expression is also turned on in developing rodent GABAergic neurons. Interestingly, MR-GEF expression is also altered in individuals with bipolar disorder. Correlation of the percentage of MR-GEF expressing neurons and 2D neuronal density between cortical layers II and IV in bipolar disorder support a growing body of evidence for its contribution to defects in cortical organization and communication in this disease (Bithell et al. 2010).
In addition to RapGEFs 1–6, several other Rap1 exchange factors exist, if not implicitly acknowledged by their official gene names. The four RasGRP (Ras guanyl releasing proteins) or CalDAG-GEF gene products contain N-terminal REM and CDC25 homology regions that are followed by two tandem Ca2+-binding EF hands similar to those found in calmodulin (Quilliam et al. 2002; Gloerich and Bos 2011). Farther C-terminal is a C1 domain similar to the diacylglycerol (DAG)/phorbol ester-binding domain found in classical and atypical protein kinases C – hence the CalDAG moniker. Both RasGRP2/CalDAG-GEF1 and RasGRP3 act as GEFs for Rap proteins. GRP2 is specific for Rap1/2 and activates it in a Ca 2+-dependent manner (although an N-terminally myristoylated splice variant was reported to also act on N- and K-Ras but to be inhibited by Ca 2+ elevation). In contrast, RasGRP3 has the broadest substrate specificity of all Ras GEFs, acting on true Ras, R-Ras, and Rap subfamilies (Quilliam et al. 2002). RasGRPs 1 and 4 are Ras-specific.
RasGRPs are highly abundant in the brain: RasGRP2 is most highly expressed in the basal ganglia whereas RasGRP3 is found primarily in glial cells of the cerebral and cerebellar white matter. Both GRPs are also highly expressed in cells of hematopoietic origin. GRP2/Cal-DAG1 is particularly abundant in platelets (that are also replete with Rap1b) where they play a major role in coupling chemoattractant receptors to αIIbβ3 integrin activation during “inside-out” signaling as well as in thromboxane A2 release. Leukocyte adhesion deficiency (LAD) syndrome III, characterized by an inability of leukocytes to adhere and migrate during inflammatory and host defense reactions, was initially attributed to mutations in RasGrp2, but recent studies suggest that kindlin-3 rather than GEF mutation is the true culprit behind this rare disease.
Like other phospholipases C (PLCs), PLCε contains X and Y regions that make up the phospholipase catalytic domain. This enzyme cleaves phosphatidyl inositol 4,5 bisphosphate (PIP2) into the second messengers inositol 3,4,5 trisphosphate (IP 3) and DAG. PLCε similarly possesses a PH domain, C2 domain, and EF hands that bind to phospholipids and Ca 2+ (see Fig. 1). However, unlike other PLC isozymes, PLCε is also a Rap GEF. PLCε has both an N-terminal REM/CDC25 Rap GEF module and tandem RA domains located at its C-terminus (Fig. 1). The GEF function of PLCε helps maintain persistent Rap1-GTP levels following G-protein-coupled receptor stimulation (Suh et al. 2008). The C-terminal RA domain (RA2) interacts with activated Ras and Rap1 enabling PLCε recruitment to either the plasma membrane or the perinuclear area, respectively, following growth factor stimulation. Meanwhile, the other RA domain confers protein stability and possibly also autoinhibition.
Significant crosstalk between PLCε and other RapGEFs has been reported and further scenarios can readily be imagined. For example, Ca 2+ and DAG generated by PLCε activity might promote the activation of GRP2 or 3. Additionally, upon adrenaline or prostaglandin E2 stimulation, G-protein-coupled receptors elevate cAMP levels and activate Epac. This results in Rap2B activation that in turn associates with RA2 of PLCε and induces PIP 2 hydrolysis. PLCε can also be activated by other ligands such as lysophosphatidic acid or sphingosine 1-phosphate that couple to Gα12 and Gα13. In addition, Gα12 and Gα13 can activate various RhoGEFs. GTP-loaded RhoA can bind directly to the phospholipase Y domain and stimulate PLCε activity.
PLCε knockdown or mutation is embryonic lethal in Caenorhabditis elegans due to its role in epidermal morphogenesis, while PLCε−/− mice exhibit multiple cardiac defects. These include ventricular dilation, aortic and pulmonary valve defects, and stenosis due to the thickening of valve leaflets. Additionally, cardiac myocytes from PLCε null mice have a decrease in contractile response to acute β-adrenergic stimulation. Human kidney development also requires PLCε, and truncating mutations in PLCε were found in nearly 30% of children having the nephrotic syndrome, diffuse mesangial sclerosis (Suh et al. 2008). However, loss of PLCε can be advantageous to mice, resulting in reduced susceptibility to carcinogen-induced skin tumor formation (Suh et al. 2008). This is likely the result of PLCε mediating both direct agonist-dependent proliferation and an indirect inflammatory response. PLCε both mediates PDGF-, EGF-, and Rho-dependent cell growth and inhibits EGF receptor downregulation via PIP2-derived second messengers. Furthermore, PLCε can transduce mitogenic signals through its Rap1 GEF activity. On the other hand, PLCε-null mice have a reduction in phorbol ester-induced edema, granulocyte infiltration, and expression of the proinflammatory cytokine, interleukin-1α.
Three closely related GEFs, RASGEFs 1A, 1B, and 1C, whose sequence was originally described in Quilliam et al. (2002) have since been found to be specific for Rap2 (Yaman et al. 2009). This contrasts with most other Rap GEFs that do not discriminate between Rap1 or Rap2. A RasGEF2 has been described but bares less homology, and its activity/specificity has not been determined. RasGEF1A-C share most homology to the Epac family but are quite small and lack any identifiable regulatory domains.
SmgGDS (small-molecular-weight G-protein guanine nucleotide dissociation stimulator) is relegated to near last place in this article but was the first mammalian Rap (or Ras family) GEF to be discovered in 1990, reviewed in Quilliam et al. (2002). It is comprised of ARM/armadillo repeats (similar to, e.g., catenins or importins) rather than having a CDC25 homology domain (Fig. 1), and selectivity is based on the presence of positively charged residues in C-terminal tail of Ras proteins. While early reports suggested it promoted nucleotide exchange on a variety of substrates that included Rap1, K-Ras, Ral, Rac1, and RhoA that each have multiple lysine residues in their C-terminal hypervariable regions, one study suggested that it only acts on Rho (Hamel et al. 2011). SmgGDS (gene name RAP1GDS1) expression is critical during development and appears to promote the malignant phenotype of certain cancer cells. However, as noted above, whether it acts as a true GEF in vivo has always been in question. A recent study suggests that two splice variants of smgGDS increase the activity of polybasic-region-containing GTPases such as Rap1 by facilitating their posttranslational modification by lipids and subsequent transit to the plasma membrane (Williams 2013).
Additional Potential Rap GEFs
Dock4, a member of the Dock180/CZH family of Rho GEFs and in partnership with ELMO, is known to regulate the activity of Rac1. While Dock4 was reported to effectively activate Rap1 in a mouse osteosarcoma cell line, there was no direct demonstration of RapGEF activity in vitro. Since Rap1 is intimately associated with Rac and adherens junction regulation, the effect of Dock4 reported here may have been indirect.
NSP1–3 (novel SH2-containing proteins) have a C-terminal CDC25 homology-like domain (Quilliam et al. 2002) (Fig. 1). They have various alternate names such as BCAR3, AND-34, SHEP1, and Chat and have been reported to associate with Rap1. While exchange activity was attributed to NSP2/BCAR3/AND34 in one study, this was likely indirect due to its association with Crk/Src to influence Rap1 and Rac GEF activity. NSPs were identified in breast cancer (BCAR3 = breast cancer antiestrogen resistance gene 3) or associated with receptor tyrosine kinases (EGFR for NSP1, EphB2 for NSP3/SHEP1/Chat). All 3 NSPs associate with the scaffold protein p130 Cas, implicating them in adhesion. Surprisingly the CDC25 homology domain is responsible for this interaction. Thus although NSPs likely play a pivotal role in coupling adhesion receptors and tyrosine kinases to actin cytoskeletal organization, their CDC25-like domains act as adapter modules rather than as Rap GEFs.
Multiple Rap GEFs have been characterized at both the molecular level and frequently also in mouse models of development/disease. However there is much still to be learned from animal models. The number of GEFs and variety of regulatory domains suggest that Rap plays key roles in cell signaling that need to be activated in unique spatiotemporal scenarios. While Rap1 plays many roles in cell biology and likely represents a poor drug target, targeting Rap GEFs might result in specific regulation of Rap1 in certain diseases. The 8CPT-cAMP molecule represents a specific activator of Epacs that may impact signaling events in vascular disease and diabetes. The discovery also of inhibitors for Epac and the Ras GEF Sos (Cox et al. 2014; Banerjee and Cheng 2015) suggests that RapGEF modulation to treat disease is a future possibility.
- Maeta K, Edamatsu H, Nishihara K, Ikutomo J, Bilasy SE, Kataoka T. Crucial role of Rapgef2 and Rapgef6, a family of guanine nucleotide exchange factors for Rap1 small GTPase, in formation of apical surface adherens junctions and neural progenitor development in the mouse cerebral cortex. eNeuro. 2016;3(3):1–17.CrossRefGoogle Scholar