The RalGEF-Ral Effector Signaling Network

Abstract

The high frequency of RAS mutations in human cancers (33%) has stimulated intense interest in the development of anti-Ras inhibitors for cancer therapy. Currently, the major focus of these efforts is centered on inhibitors of components involved in Ras downstream effector signaling. In particular, more than 40 inhibitors of the Raf-MEK-ERK mitogen-activated protein kinase cascade and phosphoinositide 3-kinase-AKT-mTOR effector signaling networks are currently under clinical evaluation. However, these efforts are complicated by the fact that Ras can utilize at least 9 additional functionally distinct effectors, with at least 3 additional effectors with validated roles in Ras-mediated oncogenesis. Of these, the guanine nucleotide exchange factors of the Ras-like (Ral) small GTPases (RalGEFs) have emerged as important effectors of mutant Ras in pancreatic, colon, and other cancers. In this review, we summarize the evidence for the importance of this effector pathway in cancer and discuss possible directions for therapeutic inhibition of aberrant Ral activation and signaling.

Introduction

The frequent mutational activation of Ras in human cancers,¹ in particular those in which there remains a dire need for new therapies (e.g., pancreatic, colon, lung, melanoma), has prompted intense research interest and pharmaceutical industry effort to develop anti-Ras inhibitors for cancer treatment.² Because Ras itself has not, to date, been a tractable target for the development of direct inhibitors, much of these efforts have involved indirect approaches for blocking either Ras membrane association, critical for Ras function, or Ras downstream effector signaling. In light of the disappointing failure of farnesyltransferase inhibitors to block the membrane association of the Ras isoforms most commonly mutated in human cancers (K-Ras and N-Ras),³ much of the current effort is now focused on inhibitors of Ras effector signaling, in particular the Raf-MEK-ERK mitogen-activated protein kinase (MAPK) cascade and phosphoinositide 3-kinase (PI3K)– AKT-mTOR effector signaling networks. In this review, we focus on a lesser studied effector pathway, the RalGEF-Ral small GTPase signaling network. Although not mutated frequently in human cancer, the validation of the importance of this pathway in cancer continues to mount. We summarize the validation of Ral in cancer and discuss approaches for targeting Ral for cancer treatment.

Effectors of Ras

A diverse spectrum of extracellular stimuli activates Ras guanine nucleotide exchange factors (RasGEFs) (e.g., Sos, RasGRF), leading to transient formation of active GTP-bound Ras.⁴ Ras GTPase-activating proteins (RasGAPs) (e.g., neurofibromin) then stimulate GTP hydrolysis, returning Ras to its inactive GDP-bound state. In human cancers, Ras proteins harbor single amino acid substitutions, most commonly at residues 12, 13, and 61, that render Ras persistently GTP bound and active.² Activated Ras-GTP binds preferentially to a spectrum of functionally diverse downstream effectors (Fig. 1). Most effectors are characterized by Ras binding (RBD) or Ras association (RA; also RalGDS/AF-6) domains that preferentially interact with Ras-GTP.^5,6 The Raf serine/threonine kinases (Raf-1, A-Raf, and B-Raf) are the best characterized effectors of Ras.^7,8 Ras-mediated Raf activation in turn activates the MEK1 and MEK2 dual-specificity kinases, which then activate the ERK1 and ERK2 MAPKs. The second best characterized effectors of Ras are the p110 (α, β, γ, and δ) catalytic subunits of class I PI3Ks.^9-11 The critical role of these 2 effector classes in Ras-mediated oncogenesis is supported by the frequent mutational activation of the genes that encode B-Raf (BRAF) (13%, COSMIC [http://www.sanger.ac.uk/genetics/CGP/cosmic/]) and p110α (PIK3CA) (19%, COSMIC) in human cancers.

Figure 1.

Effectors implicated in Ras-mediated oncogenesis. Missense mutations at Ras residues, primarily at Ras residues G12, G13, and Q61, impair GAP stimulation of intrinsic Ras GTP hydrolysis activity. Thus, mutant Ras is persistently GTP bound and active. In addition to the Raf, PI3K, and RalGEF effector families, the Tiam1 Rho family GEF and phospholipase Cϵ have also been identified as effectors important in Ras-mediated tumorigenesis. RAS mutation frequencies are compiled from COSMIC (http://www.sanger.ac.uk/genetics/CGP/cosmic/). PIP3 = phosphatidylinositol 3,4,5-bisphosphate; PIP2 = phosphatidylinositol 4,5-bisphosphate; IP3 = inositol 1,4,5-trisphosphate; DAG = diacyglycerol.

Despite the strong experimental evidence validating key roles for Raf and PI3K in Ras-mediated oncogenesis, there is substantial evidence that additional effectors must also contribute critical functions for mutant Ras in cancer growth.³ There are at least 9 additional functionally distinct effector classes identified for Ras, and of these, the validated roles of 3 additional effectors in Ras-mediated oncogenesis have been established.² Of these, substantial and rapidly accumulating evidence validates a key role for the RalGEF-Ral effector pathway in a diversity of human cancers,¹² in particular KRAS mutant pancreatic and colorectal cancer. In this review, we first provide an overview of RalGEF-Ral effector regulation and effector signaling. We then summarize the data supporting essential roles for Ral GTPases in various human cancers. Finally, we discuss possible therapeutic approaches for blocking RalGEF-Ral signaling for cancer treatment.

Discovery of the RalGEF-Ral Effector Pathway

Ral (Ras-like) GTPases (46%-51% identity with human Ras) were identified initially in a search for RAS-related genes.¹³ Using oligonucleotide probes corresponding to a sequence of 7 amino acids strictly conserved in Ras, RhoA, and the yeast YPT Rab protein (Ras residues 57-63; DTAGQE/D), a screen of a cDNA library derived from B95-8 Epstein-Barr virus, immortalized simian B lymphocytes identified RALA, which encodes a 206 amino acid protein with approximately 50% sequence identity with Ras. This initial source of Ral isolation accounts for why the approved RAL gene name is “v-ral simian leukemia viral oncogene homolog” (HUGO). The simian RALA sequence was then used to isolate human RALA, and additionally, the related RALB gene, from a human pheochromocytoma library.¹⁴ As described below, although the 2 human RAL genes encode highly related proteins (82% sequence identity) (Fig. 2A and 2B), they exhibit very distinct functional roles in cancer cell biology. RAL genes are expressed ubiquitously and conserved in evolution, with related genes found in Caenorhabditis elegans and Drosophila melanogaster that encode highly related Ral GTPases (Fig. 2B).

Figure 2.

Ral small GTPases. (A) Sequence alignment of human (Hs) RalA and RalB. There is 100% identity in the switch I (SI: 41-51) and switch II (SII: 69-81) sequences that change conformation in the GDP- and GTP-bound states and are involved in effector binding. The greatest sequence divergence is in the C-terminal membrane-targeting sequences, with polybasic (green), phosphorylation site (red), and CAAX motif (blue) residues indicated. G domain = residues 15-178/9 (corresponding to Ras residues 4-166). Asterisks indicate sequence identity. (B) Conservation of Ral GTPases in evolution. Drosophila melanogaster (Dm) and Caenorhabditis elegans (Ce) possess a single Ral GTPase ortholog with strong sequence identity to human Ral. The dendrogram was generated by Clustal/W multiple sequence alignment of the indicated Ral proteins, human K-Ras4B and Rac1 (Rho family).

The first RalGEF, RalGDS (Ral guanine nucleotide dissociation stimulator), was identified in a PCR-based screen of a mouse cDNA library for genes with sequence identity to yeast RasGEFs¹⁵ (Fig. 3A). RalGDS was found to possess sequence homology with the REM (Ras exchange motif) and CDC25 RasGEF catalytic domains characteristic of Ras-GEFs (e.g., Sos, RasGRP, RasGRF). However, Ral-GDS did not display exchange activity for Ras and instead was selective for RalA and RalB.

Figure 3.

Regulators and effectors of Ral. (A) Regulators of the Ral GDP-GTP cycle. The 2 classes of RalGEFs are defined by the presence of an RA or PH domain. (B) RalGEF domain structure and evolutionary conservation. REM = Ras exchange motif; CDC25 homology domain = RalGEF catalytic domain; RA = Ras association domain.

Stimulated by the success of yeast 2-hybrid library screening that identified Raf as a Ras effector,¹⁶ a flurry of similar studies established RalGEFs as Ras effectors. Independently, human RalGDS was identified in yeast 2-hybrid screens for H-Ras-GTP,¹⁷ R-Ras-GTP,¹⁸ or R-Ras2/TC21-GTP¹⁹ binding proteins. A mouse RalGDS-like (Rgl) protein was identified in a yeast 2-hybrid screen for H-Ras-GTP–interacting proteins.²⁰ Subsequently, mouse (RalGDS-like factor [Rlf]) and human Rgl2 were identified as Rap 1A– or Rap 1B–GTP binding proteins, respectively.^21,22 Yeast 2-hybrid library screening for Rit small GTPase binding proteins identified mouse Rgl3 (Ral GEF-like 3) and additionally RalGDS, Rgl, and Rgl2.²³ Independently, Rgl3 was identified in a yeast 2-hybrid screen for R-Ras3/M-Ras²⁴ or Rap1-interacting²⁴ proteins. Both studies also established H-Ras, as well as R-Ras and Rap1, association with Rgl3. The 4 human RalGEFs that can serve as Ras effectors share a common domain structure: an N-terminal REM followed by a CDC25 homology RalGEF catalytic domain and a C-terminal RA domain (Fig. 3B).

All 4 RalGEFs can interact with Ras and the closely related (~50% identity) Rap and R-Ras small GTPases (Fig. 3A). Although ectopic overexpression studies show that RalGEFs can be activated by R-Ras and Rap family small GTPases,²⁵ whether endogenous activation of these Ras family proteins can activate Ral is not clear. For example, it was found that mutationally activated H-Ras, but not R-Ras2/TC21, transformation was dependent on Ral activation.²⁶ Whether the different RalGEFs are regulated by different Ras family GTPases and whether they differentially regulate RalA and RalB are issues that remain unresolved.²⁷

The RA domain–containing RalGEFs are conserved in evolution, with highly homologous orthologs in D. melanogaster (RGL) and C. elegans (RGL1)^28,29 (Fig. 3B). Finally, 2 other Ral-specific GEFs (Fig. 3A), RalGPS1/RalGEF2 and Ral-GPS2 (Ral GEFs with PH domain and SH3 binding motif), have been identified that lack RA domains and instead contain pleckstrin homology domains.^30-33 Because RalGPS proteins lack RA or RBD domains, they are not activated directly by Ras; limited evidence suggests activation by the Grb2 adaptor or phospholipids. Another RalGEF, Rgr, was identified originally as a transforming protein and can also activate other small GTPases.^34,35 Ral activation in RAS wild-type cancer cells may be mediated by these additional RalGEFs.

Recently, 2 RalGAPs have been identified, designated RalGAP1 and Ral-GAP2³⁶ (Fig. 3A). Both are large heterodimeric complexes, each consisting of a catalytic α1 or α2 subunit and a common β subunit. Rather than sharing sequence similarity to RasGAPs, the RalGAP complexes share structural and catalytic similarities with the tuberous sclerosis tumor suppressor Tsc1/Tsc2 complex, which acts as a GAP for the Rheb branch of the Ras family small GTPases. Mutational loss of function of the neurofibromin RasGAP and the Tsc1/2 RhebGAP has been identified in cancer.³⁷ Whether a similar loss of function of RalGAPs may promote oncogenesis has not been addressed.

Ral Effector Signaling

Similar to Ras,² activated Ral-GTP interacts with multiple, functionally divergent downstream effectors (Fig. 4). The first effector was identified in a yeast 2-hybrid library or cDNA expression library screens using RalA, leading to independent discovery of RalBP1 (Ral binding protein 1)/RLIP76 (76-kDa Ral-interacting protein 1)/ RIP (Ral-interacting protein).^38-40 In addition to a Ral-GTP binding domain,⁴¹ RalBP1 also contains a RhoGAP homology catalytic domain with activity for Rac1 and Cdc42 but not RhoA. GTP-bound RalA and RalB can interact with RalBP1 through a conserved Ral binding domain, regulating RalBP1 subcellular localization but not intrinsic RhoGAP activity.⁴² Rac1 activation promotes membrane ruffling at the leading edge of migrating cells, whereas Cdc42 promotes filopodia formation.

Figure 4.

Ral effectors. RalA and/or RalB have been determined to interact with a diverse spectrum of downstream effectors. Most bind preferentially to the GTP-bound protein, whereas some are nucleotide independent. Ral effector networks can regulate endocytosis, exocytosis, actin organization, phospholipid metabolism, and generation of second messengers. Ral has been found to activate various transcription factors or promote elements, to regulate gene expression, through pathways that may or may not be distinct from known effectors. PC = phosphatidylcholine; PA = phosphatidic acid.

RalBP1 can also serve as a scaffold and associate with a diversity of other proteins. Two independent yeast 2-hybrid library screening studies identified the closely related Reps1 (RalBP1-associated Eps homology (EH) domain protein 1) and Reps2/POB1 (partner of RalBP1) proteins that interact with RalBP1 C-terminal sequences distinct from the RhoGAP and Ral binding domains.^43,44 Reps1 and Reps2 contain EH domains, which are found on proteins involved in endocytosis. Reps1 associates through its EH domain with Rab11-FIP2, a Rab11 binding protein implicated in endocytosis.⁴⁵ Reps2/POB1, via its EH domain, interacts with Epsin and Eps15,^46,47 proteins that regulate receptor-mediated endocytosis. Similarly, a second RalBP1 interaction, through N-terminal sequences, is with the µ2 subunit of the plasma membrane–associated AP-2 tetrameric complex.⁴⁸ AP-2 promotes clathrin coat formation and specific recognition of membrane receptors for endocytosis. Epsin, Eps15, and Rab11-FIP2 can also interact directly with AP-2. These interactions support a role for RalBP1 in the regulation of receptor-mediated endocytosis.

RalBP1 also interacts with ARIP2 (activin receptor-interacting protein 2), which regulates endocytosis of activin type II receptors,⁴⁹ HSF1 (heat shock factor 1),⁵⁰ which regulates expression of heat shock genes in response to stress, cyclin B1 during mitosis,⁵¹ and PSD-1, a postsynaptic scaffolding protein implicated in the regulation of excitatory synaptic function.⁵²

RalBP1 was also identified independently as a transporter activity, designated DNP-SG (S-(2, 4-dinitrophenyl)glutathione) ATPase, involved in the active transport of conjugated and unconjugated electrophiles out of cells.⁵³ RalBP1 contains 2 ATP binding sites⁵⁴ that allow it to function as an ATP-dependent transporter protein and efflux pump for small molecules, including anticancer drugs and endogenous metabolites.⁵⁵ Inhibition of RalBP1 expression or function has been shown to cause regression of lung, kidney, melanoma, colon, and prostate cancer cell line xenografts, although the significance of these observations for Ral function is not clear.

Perhaps the best characterized of the Ral effectors are 2 components of the octameric exocyst complex (also called Sec6/8 complex), Sec5 and Exo84.^56,57 The octameric exocyst complex is involved in the regulation of exocytosis.^58,59 The exocyst facilitates the tethering of post-Golgi secretory vesicles to the plasma membrane prior to exocytic fusion, and exocyst function has been implicated in a variety of cellular processes including cell migration and tumor cell invasion. Ral regulates exocyst subcellular localization rather than assembly.⁶⁰ Studies with Ral effector domain mutants and/or RNA interference have implicated exocyst function in several Ral-mediated functions.¹²

Other less characterized effectors of RalA include filamin, an actin filament crosslinking protein required for RalA-induced filopodia formation.⁶¹ RalA and RalB have also been shown to directly interact with and activate phospholipase C delta 1 (PLCδ1).⁶² PLCδ1 is not a conventional effector in that interaction with RalB was nucleotide independent and required the N-terminal 11 residues of RalB. Activation of various G protein–coupled receptor signaling pathways stimulates Ral-dependent PLC activation.

Another effector of Ral is phospholipase D (PLD1),⁶³ which leads to the generation of lipid second messengers, including phosphatidic acid, lysophosphatidic acid, and diacylglycerol. However, PLD1 does not function classically as a Ral effector in that PLD1 association is not regulated by GDP/GTP cycling and interaction is through Ral N-terminal sequences distinct from the switch I and II sequences involved in GTP-dependent effector binding. Instead, RalA activation of PLD1 includes the additional association with the Arf6 small GTPase.⁶⁴

Although RalA and RalB share 100% sequence identity in residues involved in effector interaction (Fig. 2A) and, where studied in vitro, can interact with the same effectors, RalA and RalB can exhibit strikingly different roles in normal and neoplastic cell function. These functional differences are largely because of their distinct subcellular membrane locations. Whereas RalA is found at the plasma membrane and with endosomes, RalB is primarily endosome associated.⁶⁵ However, Ral subcellular localization is dynamic and can be regulated by its activation state and by phosphorylation.^66,67 Ral subcellular localization in turn influences specific effector interactions. Some examples of RalA and RalB differences in oncogenesis are summarized below.

Ral Regulation of Gene Expression

In addition to the specific effector functions described above, RalGEF-Ral signaling can regulate gene expression. RalGEF and/or Ral activation is involved in Ras-mediated activation of various transcription factors (Fig. 4) that include phosphorylation and activation of c-Jun through JNK MAPK activation,⁶⁸ ATF2,⁶⁹ STAT3 through Src tyrosine kinase activation,⁷⁰ NFAT,⁷¹ and AFX (FOXO4).⁷² Ral activation also stimulated transcriptional activation of the c-Fos promoter through the ternary complex factor,^73,74 NF-κB binding sites for the human cyclin D1 promoter,^75,76 the urokinase plasminogen activator receptor promoter AP-1 binding motifs,⁶⁹ and the binding site for the Ras-responsive element binding protein 1 (RREB-1).⁷⁷

The effector signaling that regulates these transcription factors is not well characterized. However, activation of NF-κB has been investigated and found to be regulated downstream of the Ral effector Sec5 and not through PLD1 or RalBP1.^75,76 Active RalB signaling causes the association of Sec5 with TBK1 (TANK-binding kinase), an atypical IκB kinase (IKK)–related protein kinase. The recruitment of TBK1 into a complex with Sec5 resulted in increased TBK1 catalytic activity.⁷⁵ TBK1 can directly phosphorylate and promote the nuclear localization and activation of the c-Rel NF-κB family member.⁷⁸

Global gene expression profiles have also identified Ral-regulated transcription factors and gene targets.^77,79 Microarray analyses of RalA siRNA– and/or RalB siRNA–depleted UM-UC-3 bladder cancer cells, which harbor an activating KRAS mutation, found that a majority of genes identified were dependent on both RalA and RalB (547 genes), but additionally, subsets of RalA (77 genes) isoform– or RalB (85 genes) isoform–specific upregulated or downregulated genes were also identified. Computation analysis identified enrichment of NF-κB as well as RREB-1 binding sites in the Ral gene expression signature. One gene identified with RalA- and RalB-dependent upregulation was CD24, which encodes the metastasis-associated protein CD24,⁸⁰ a glycosyl phosphatidyl inositol–linked surface protein. Suppression of CD24 expression in UMUC-3 and other tumor cells reduced anchorage-independent proliferation and survival, suggesting that CD24 upregulation may be an important component of Ral-mediated oncogenesis. Taken together, these observations suggest that Ral-regulated genes may be fertile ground for the identification of candidate anti-Ral therapeutic targets.

RalGEF-Ral Signaling and Oncogenesis

The RalGEF-Ral pathway was characterized initially to play a relatively minor role in Ras transformation of rodent fibroblasts.^81,82 However, subsequent studies by Counter et al. established a very significant role for this effector pathway in Ras transformation of human cells.⁸³ Key support for RalGEF-transforming function came from the use of Ras effector domain binding mutants, in particular the E37G effector domain mutation, which is impaired in Raf and PI3K activation yet exhibits transforming activity in part through retained binding to RalGDS.^84-87 Additionally, White et al. identified critical but distinct roles for the related RalA and RalB isoforms in normal and human cancer cell line growth.⁸⁸ siRNA suppression of RalA impaired tumor cell anchorage-independent but not -dependent growth, whereas suppression of RalB caused tumor but not normal cell apoptosis. Finally, mouse model studies showed that homozygous deletion of RalGDS (a RalGEF) caused resistance to H-Ras–induced skin squamous cell carcinoma formation.⁸⁹ Ral GTPases have now been implicated in a variety of human cancers. Below, we summarize the data validating Ral GTPases in human cancers associated with frequent RAS mutation (pancreatic, colorectal, and melanoma) as well as human cancers in which RAS mutational activation is infrequent (bladder, prostate).

Ral and Pancreatic Cancer

The most comprehensive and compelling evidence for the importance of RalGEF-Ral signaling in cancer has been seen for pancreatic ductal adenocarcinoma (PDAC), where KRAS is the most frequently mutated gene (>90%) seen.^90,91 Although the majority of Ras effector–targeted therapies currently under clinical evaluation are focused on the Raf and PI3K effector pathways,³⁷ it is of interest that the RalGEF-Ral pathway, rather than Raf or PI3K, was found to be more consistently activated in pancreatic patient tumors.⁹² A panel of 18 matched and unmatched patient pancreatic tumor samples exhibited high levels of both RalA and RalB activation but surprisingly not phosphorylation of ERK1/2 and Akt.⁹² In addition, the expression of the RalGEF Rgl2 was elevated in matched pancreatic patient tumors, and suppression of Rgl2 expression impaired PDAC growth.⁹³ These studies validate the critical importance of this pathway in patients harboring K-Ras–driven and -dependent pancreatic tumors.

Essential but distinct functions for RalA and RalB have been seen in PDAC growth. Sustained shRNA suppression of RalA but not RalB in 10 of 10 KRAS mutant PDAC cell lines diminished anchorage-independent growth in vitro and primary tumor xenograft growth in immunocompromised mice.⁹² Conversely, cell lines expressing RalB-specific shRNA exhibited impaired Matrigel (BD Biosciences, Franklin Lakes, NJ) invasion in vitro and lung colony formation in an experimental metastasis model. These studies suggest that RalA is required for the early stages of Ras-driven pancreatic tumorigenesis, whereas RalB is required for later stages of malignant growth.⁹²

Although the basis for these distinct functions, and the effectors involved, remains to be determined, clues are provided by the consequences of Aurora A phosphorylation of RalA. The phosphorylation of RalA at Ser194 by Aurora A kinase disrupted plasma membrane association and resulted in cytoplasmic translocation. In addition, phosphorylation led to preferential association with and activation of RalBP1 and decreased Cdc42 and Rac activity.⁶⁶ The exocyst, through lipid raft microdomain exocytosis, may also be an important effector of RalA-supported anchorage-independent growth.⁹⁴

Ral and Colorectal Cancer

Genetic sequencing has verified that KRAS is the most frequently mutated oncogene in colorectal cancer (CRC) (40%-50%).^95,96 A critical but distinct role for Ral GTPases has also been described for CRC. As in pancreatic cancer, Ral-GTP levels were found elevated in CRC tissue and cell lines.⁹⁷ Stable shRNA suppression found that RalA was necessary for the anchorage-independent growth of 8 of 8 CRC cell lines, independent of KRAS mutation status. Surprisingly, stable suppression of RalB was found to greatly enhance the anchorage-independent growth of all 8 CRC tumor cells. This activity was specific because ectopic restoration of RalB expression suppressed soft agar growth. Finally, using Ral effector binding mutants and RNAi, it was determined that both Ral isoforms required the ability to bind RalBP1 but utilized distinct components of the exocyst to mediate their effects on anchorage-independent growth. Specifically, Exo84 binding was required for RalA, while RalB required Sec5 engagement.

That these 2 highly related proteins could serve opposing functions in transformed growth has been described previously in studies with Ras-transformed, immortalized human embryonic kidney epithelial cells.⁹⁸ Interestingly, these results contrast with observations made with transient RalB suppression in CRC cells that caused apoptosis.^88,99 These divergent observations suggest different consequences of short-term versus prolonged loss of Ral function and may reflect compensatory events due to sustained loss. Consistent with this possibility, in CRC tumor cells, loss of one Ral isoform led to an increase in the GTP loading of the other isoform, indicating potential crosstalk between RalA and RalB.

Ral and Melanoma

Oncogenic NRAS mutations occur (15%-30% frequency) in cutaneous malignant melanoma.^100,101 It has long been appreciated that oncogenic mutations (V600E) in BRAF are very common (up to 70%) in melanoma and are often mutually exclusive of NRAS mutations, suggesting that the ERK MAPK pathway downstream of Ras is the critical pathway regulating melanoma tumorigenesis.^100,102-104 Despite the lack of evidence for mutations in RalGEFs in melanoma,¹⁰⁵ recent evidence suggests that the RalGEF-Ral pathway does play an important role in melanoma tumor progression. Genetically engineered, immortalized p19 Arf-deficient primary mouse melanocytes were used to assess the contribution of the 3 major downstream Ras-activated pathways in melanoma tumorigenesis.¹⁰⁶ Interestingly, Arf –/– melanocytes expressing constitutively activated Rgl2, but not activated Raf or PI3K, demonstrated the same robust anchorage-independent growth capacity as mutant N-Ras–expressing melanocytes. Furthermore, Ral dominant-negative inhibition of the RalGEF pathway in melanocytes expressing mutant N-Ras impaired growth in soft agar. RalGEF activation alone also phenocopied mutant N-Ras morphological transformation and Matrigel (BD Biosciences) invasion in vitro. Thus, while activated Raf or PI3K also promoted aspects of N-Ras–mediated melanocyte transformation, RalGEF activation was surprisingly the most significant effector pathway.

It was also recently demonstrated that a majority of human melanoma cells with wild-type NRAS and those cells harboring mutations in NRAS and BRAF have high levels of RalA but not RalB activation.¹⁰⁷ Expression of RalA-specific shRNA in NRAS and BRAF mutant cell lines, and to a lesser extent NRAS and BRAF wild-type cells, inhibited tumorigenesis in vivo. These observations suggest a role for the RalGEF-Ral pathway in melanoma tumorigenesis, regardless of the mutational status of NRAS and BRAF.

Ral and Bladder Cancer

KRAS and HRAS mutations are found in approximately 13% of bladder cancers.^108-110 Microarray gene expression analyses of 65 bladder tumors and 15 normal bladder tissue determined that RalA mRNA expression increased with tumor grade, and expression of RalA and RalB protein levels increased in metastatic patient samples. In contrast, RalB mRNA expression did not significantly associate with higher tumor grade.⁷⁹ Variable levels of RalA- and RalB-GTP were seen in a panel of bladder carcinoma cell lines.⁷⁹ Studies with the KRAS mutant UM-UC-3 bladder carcinoma cell line showed that transient siRNA suppression of RalB but not RalA impaired transwell motility.¹¹¹ However, concurrent suppression of RalA and RalB did not impair motility. Additionally, ectopic expression of constitutively activated RalA impaired motility, while activated RalB stimulated motility. Thus, RalA and RalB play opposing roles in the motility of UM-UC-3 bladder cancer cells. Finally, siRNA suppression of RalB alone or together with RalA, but not RalA alone, disrupted actin stress fibers, and concurrent suppression of both RalA and RalB was required to reduce UM-UC-3 anchorage-dependent growth.

The RalGEF-Ral pathway also appears to be important in anchorage-independent growth and cell survival in HRAS mutant T24 bladder cancer cells. Studies demonstrate that PLD1 activity is dependent on Ras and RalA in bladder cancer cell lines. In addition, expression of RalA-specific shRNA induced apoptosis in response to serum withdrawal, a similar effect to that seen in cells expressing PLD-1–specific shRNA.¹¹²

Ral and Prostate Cancer

Overall, RAS mutations are found in 16% of prostate cancers (8% KRAS, 6% HRAS, 2% NRAS; COSMIC). The importance of the RalGEF-Ral pathway in prostate cancer is demonstrated in part by correlation between Ral expression and metastasis. Metastatic patient tumors exhibit increased RalA mRNA expression as compared to primary tumors.⁷⁹ Higher protein expression of RalA was also seen in patient tumors and was used as part of a gene signature to predict tumor aggressiveness in patients.¹¹³

Progression of prostate cancers to androgen independence is an important step in prostate cancer tumorigenesis. RalA appears to have a key role in androgen-independent signaling and gene expression. RalA is activated upon androgen deprivation in a ROS (reactive oxygen species) dependent manner. RalA activation upregulates the transcription of the angiogenic factor VEGF-C.¹¹⁴ Furthermore, loss of NKX3.1 and expression of constitutively activated RalA Q72L synergistically enhanced VEGF-C transcription.¹¹⁵ Androgen-independent tumors rely on growth factor signaling (e.g., epidermal growth factor [EGF]) as opposed to androgen signaling. The RalA effector RalBP1 interacts with REPS2/POB1, which is involved in the endocytosis of the EGF receptor (EGFR). REPS2/POB1 was seen to be downregulated in androgen-independent prostate cancer cell lines and xenografts. Overexpression of REPS2/POB1 in prostate cancer cell lines induced apoptosis and inhibited growth factor signaling.¹¹⁶

Ral GTPases have been implicated in prostate cancer cell migration. In one study, antagonistic roles for RalA and RalB were described for human prostate cancer cell motility. RAS wild-type DU-145 prostate cancer cells expressing RalB but not RalA shRNA had a significant reduction in transwell migration.¹¹¹ However, concurrent expression of both RalB- and RalA-specific shRNA restored normal motility, indicating RalA dominance over RalB. In a second study of Dunning rat prostate tumor cells, suppression of RalA or RalB altered cell morphology and reduced migration.⁶⁰ Loss of Ral disrupted subcellular localization of the exocyst complex to paxillin-positive focal complexes, leading to altered migration.

Prostate cancers preferentially metastasize to the bone, but the mechanisms that mediate this specificity are not fully understood. It is apparent that the Ral-GEF-Ral pathway regulates the ability of prostate cancer cells to grow in the bone microenvironment. It was recently demonstrated that the RalGEF-Ral pathway is both necessary and sufficient for prostate cancer bone metastasis.¹¹⁷ Expression of the Ras E37G effector domain mutant that primarily activates the RalGEF-Ral pathway or constitutively activated RalGEF in nonmetastatic DU-145 cells promoted metastasis to the bone. Furthermore, expression of RalA-specific shRNA in the RAS wild-type PC3 metastatic cell lines inhibited bone metastasis but not primary tumor growth. From these studies, it appears that RalA is more important in the growth of the tumor at the secondary site as opposed to the homing and initial colonization.

Ral and Other Cancers

Aberrant Ras activation in malignant peripheral nerve sheath tumors (MPNSTs), which arise from peripheral nerve Schwann cells, is caused by loss-of-function mutations in the neurofibromin RasGAP.¹¹⁸ Farassati et al. recently showed that Ral was overactive in 5 of 5 neurofibromin-deficient mouse MPNST cell lines when compared to nontransformed mouse Schwann cells. Ras-dependent Ral activation was also seen in 2 human MPNST cell lines and 3 tumors. shRNA depletion of RalA was found to inhibit the anchorage-dependent proliferation, Matrigel (BD Biosciences) invasion, and subcutaneous tumor xenograft growth of 35-1-2 mouse MPNST cells.¹¹⁹

Roles for Ral GTPases have been described for ligand-stimulated activities in a number of other cancers. In multiple myeloma (MM) cells, it was found that RalB but not RalA promoted the CXCL12/SDF-1–induced Ral activation and migration of MM cells.¹²⁰ In MCF-7 breast cancer cells, EGF stimulated Ral activation, and RalA was determined to be important for EGFR promotion of estrogen-independent proliferation.¹²¹ More recently, lysophosphatidic acid–stimulated invasion of MDA-MB-231 breast tumor cells was associated with Ral activation, and invasion was found to depend on RalA and RalB protein expression.¹²²

Finally, a tumor suppressor rather than oncogene function for RalA is suggested from studies of squamous cell carcinoma.¹²³ H-Ras–transformed human HaCaT keratinocytes show increased proliferation but not invasion. Instead, further suppression of E-cadherin function is required for invasion. Loss of E-cadherin expression was associated with decreased RalA and RalB expression. Ectopic expression of RalA but not RalB to levels found in parental noninvasive cells reversed the effects of E-cadherin loss, reducing invasion. Using a bioengineered tissue model reflective of the early steps in Ras-induced human squamous cell carcinoma of the skin, RNAi depletion of RalA in H-Ras–transformed HaCaT cells decreased E-cadherin expression and led to enhanced invasion. The potential relevance of these observations for human cancers was suggested by the Oncomine database analyses in which RalA was significantly downregulated (~30%) in head and neck squamous cell carcinoma (HNSCC) when compared to normal oral mucosa. However, a tumor suppressor function for RalA in squamous cell carcinoma is seemingly at odds with the requirement for RalGDS in H-Ras–induced skin carcinomas in mice.⁸⁹ Further studies of Ral function in human squamous cell carcinoma tumors and cell lines will be needed to clarify its involvement in this cancer type.

Evolutionary Conservation of RalGEF-Ral Signaling in D. melanogaster and C. elegans

Both Drosophila melanogaster and Caenorhabditis elegans express single Ras orthologs (Ras1 and LET-60, respectively) that share extensive identity with human Ras proteins, including 100% identity in the core effector binding regions.^124,125 Therefore, Ras1 and LET-60 probably signal through similar suites of effectors as human Ras proteins. Likewise, both flies and worms contain single genes encoding the Ras-dependent RalGEF (RGL and RGL-1) and Ral (Ral and RAL-1), and the predicted proteins are similarly highly conserved. Flies also express an ortholog of Ral-GPS, while worms do not.

In classic eye development experiments, Raf was shown to be the canonical Ras effector in flies.¹²⁶ However, the relationship of Ras to RalGEF signaling is murky in D. melanogaster. Coexpression experiments with dominant-negative Ral and mutationally activated Ras or Rap (the Ras relative) suggest that Rap, not Ras itself, activates RalGEF-Ral signaling in bristle formation and eye development.²⁸ However, the data are also consistent with multiple Ras and Rap effector pathways signaling in parallel with diverging or competing outcomes. In separate D. melanogaster studies, Ral function was implicated in immune response,⁷⁵ sensory cell apoptosis,¹²⁷ and polar cell fate and survival,¹²⁸ but the role of Ras1 and RGL relative to Ral in these processes is unclear.

During C. elegans embryonic morphogenesis, RGL-1-RAL-1 signaling through Sec5 and Exo84 components of the exocyst complex functions redundantly with another Ras family small GTPase, RAP-1, to regulate trafficking of the cadherin adhesion complex to cell junctions,¹²⁹ but the role of LET-60/Ras in this process is unknown. The canonical worm LET-60/Ras effector pathway is LIN-45/Raf¹³⁰ that induces the 1° vulval cell fate. While Ras-Raf signal drives 1° vulval fates, Ras-RalGEF-Ral drives the antagonistic 2° vulval fate in support of Notch signaling. The switching of Ras effector utilization from Raf to RalGEF during vulval development is mediated by restriction of RAL-1 expression to presumptive 2° cells²⁹ and concomitant Notch-dependent 2°-specific expression of LIP-1/MAPK phosphatase to quench Ras-MEK-ERK pro-1° signaling.¹³¹

Such interplay between Notch and Ras signaling is a common theme in developmental biology,¹³² and Notch and Ras interplay is also observed in mouse pancreatic cell differentiation and cancer development. Whether this pancreatic Ras-Notch interplay depends on K-Ras activation of the RalGEF-Ral pathway is not known, but it is intriguing that Ral-GEF but not Raf is preferentially activated in pancreatic cancer cells, and Ral activation is necessary for pancreatic cancer growth.

Therapeutic Approaches for Blocking RalGEF-Ral Signaling

Like Ras, Ral GTPases are GTP binding proteins and are therefore not considered tractable targets for inhibitor development. While there has been some success in developing inhibitors of GEFs,³⁷ to date, the most promising directions for inhibitors of RalGEF signaling involve inhibitors of Ral posttranslational processing and effector signaling. Additionally, because protein kinases are tractable targets for drug discovery, the identification of protein kinases that regulate or mediate Ral function also suggests more classic directions for blocking Ral function.

Similar to Ras, Ral GTPases terminate with C-terminal CAAX (C = cysteine, A = aliphatic amino acid, and X = terminal amino acid, which dictates prenyltransferase specificity) tetrapeptide motifs that signal for posttranslational modifications essential for Ral membrane association and subcellular localization¹³³ (Fig. 5). Ral GTPases are substrates for the geranylgeranyltransferase-I (GGTase-1)–catalyzed addition of a C20 geranylgeranyl isoprenoid group to the cysteine residue,¹³⁴ followed by Rce1 endoprotease cleavage of the AAX residues and carboxylmethylation of the now terminal lipid-modified cysteine residue. Similar to K-Ras4B, sequences upstream of the CAAX motif are rich in basic residues that likely function as a second signal essential for full membrane association. RalA and RalB show greatest divergence in their C-terminal sequences (Fig. 2A), which contribute to their distinct subcellular distributions to the plasma membrane and endomembranes, respectively.⁶⁵ Inhibitors of GGTase-I (GGTI) were shown to cause growth inhibition of MIA-PaCa2 cells in part by induction of RalB-dependent apotosis and RalA cell cycle perturbation.¹³⁵ There is currently one GGTI in a phase I clinical trial (GGTI-2418) that is well tolerated with minimal side effects (http://www.tigrispharma.com/), and others are under preclinical evaluation.¹³⁶ It was originally thought that GGTIs would possess severe off-target normal cell toxicity because of the important role of other GGTase-I substrates in normal cell physiology. However, the recent demonstration of genetic ablation of GGTase-I in mutant KRAS-driven mouse models of cancer argues that GGTI therapy may be feasible and hence a useful approach for anti-Ral therapy.^137,138

Figure 5.

Ral posttranslational processing. Ral is synthesized initially as an inactive cytosolic protein. The cytoplasmic (GGTase-I)– or endoplasmic reticulum (Rce1 and ICMT)–associated enzymes recognize the Ral C-terminal CAAX motif, leading to covalent addition of a geranylgeranyl isoprenoid to the cysteine residue of the CAAX motif, followed by proteolytic cleavage of the AAX residues and carboxylmethylation of the now terminal prenylated cysteine. Sequences upstream of the CAAX motif are rich in K or R basic amino acids. These polybasic stretches comprise a second membrane-targeting signal. Reversible phosphorylation of S194 in RalA modulates the targeting activity of the C-terminus, altering subcellular location. CAAL = cysteine–aliphatic amino acid–aliphatic amino acid–leucine; the terminal amino acid of the CAAX tetrapeptide motif dictates prenyltransferase specificity, with leucine recognized by geranylgeranyltransferase-I (GGTase-I); K/R = basic amino acid–rich sequences; Aur-A = Aurora A; GGTI = GGTase-I inhibitor; PKI = protein kinase inhibitor; Rce1 = Ras-converting enzyme 1; Icmt = isoprenylcysteine carboxyl methyltransferase.

An emerging theme in the regulation of small GTPase function involves protein kinase phosphorylation and regulation of subcellular localization and function. This additional level of GTPase regulation beyond the GDP-GTP cycle is demonstrated most dramatically with the finding that protein kinase C alpha (PKCα)–mediated phosphorylation converts K-Ras4B from a growth-promoting protein to an apoptosis-inducing protein.¹³⁹ PKCα phosphorylation of S181 within the C-terminal polybasic region immediately adjacent to the CAAX motif promotes rapid dissociation of K-Ras4B from the plasma membrane and association with intracellular membranes, including the outer membrane of mitochondria where phosphorylated K-Ras4B interacts with Bcl-XL, causing apoptosis. A similar mode of regulation of RalA has been described in which Aurora A phosphorylation of Ral A¹⁴⁰ at a conserved C-terminal S194 residue absent in RalB causes a relocation from the plasma membrane to endomembranes, where RalA associates preferentially with RalBP1/RLIP76.⁶⁶ The fact that a phospho-deficient S194A mutant of RalA cannot support RalA-dependent PDAC cell anchorage-independent growth and tumorigenicity argues that inhibitors of Aurora A may cause RalA-selective inhibition. The importance of RalA S194 phosphorylation is also supported by the finding that this residue, together with S183, is dephosphorylated by the serine-threonine protein phosphatase 2A tumor suppressor.¹⁴¹

Recently, RalB was shown to be phosphorylated by PKC. PKC-mediated phosphorylation of RalB at the C-terminal S198 residue translocated RalB from the plasma membrane to the perinuclear region of the cell. T24 (mutant HRAS) and UM-UC-3 (mutant KRAS) tumor cells expressing a phosphorylation-deficient (S198A) RalB mutant exhibit impaired anchorage-independent growth, migration, and lung colonization in an experimental metastasis mouse model. Furthermore, enhancement of tumor growth by constitutively activated RalB G23V required S198 phosphorylation.¹⁴² As described above, an important component downstream of RalB is Sec5-mediated TBK1 activation.⁷⁵ TBK1 was subsequently identified in shRNA library screening for synthetic lethal partners of mutant KRAS.⁹⁹ RalB also showed a synthetic lethal association with mutant KRAS. Thus, inhibitors of TBK1 may be one approach for anti-RalB–selective therapies.

Another protein kinase linked to Ral function is cyclin-dependent kinase 5 (CDK5).¹⁴³ CDK5 was found to be widely active in pancreatic cancer cells. Inhibition of cyclin-dependent kinase 5 (CDK5) significantly inhibited PDAC cell line tumorigenic growth. CDK5 inhibition correlated with decreased RalA and RalB activation, and expression of constitutively activated Rgl2 (Rlf-CAAX; membrane-targeted Ral-GEF) rescued the reduction in anchorage-independent growth and migration observed with CDK5 inhibition. This suggests that CDK5 mediates PDAC tumorigenesis through Ral-dependent mechanisms. Thus, inhibitors of CDK5 may serve as inhibitors of both RalA and RalB in pancreatic cancer.

Conclusion

The RalGEF-Ral effector signaling network has emerged as an important effector pathway in oncogenic Ras-driven cancer growth. Further studies of RalGEF and Ral function in oncogenesis using mouse models of mutant KRAS-driven pancreatic, lung, colon, or NRAS-driven melanoma will be needed to extend these findings. One striking observation has been the distinct function of RalA and RalB in oncogenesis. The precise basis for their different roles and the effector functions important for Ral-dependent tumor growth remains to be determined. This information will be important for the development of Ral-targeted therapies. It is likely that the combined inhibition of multiple effector pathways will be the most effective avenue for blocking mutant Ras for cancer treatment.