P-Rex2 (PIP3-dependent Rac exchanger 2, PREX2) is a Dbl-type guanine-nucleotide exchange factor (GEF) that activates the small G protein (small GTPase) Rac1, a member of the Rho family. P-Rex2 was discovered in 2004 on the basis of its homology to P-Rex1 (Donald et al. 2004; Rosenfeldt et al. 2004). Like P-Rex1, P-Rex2 is synergistically activated by the lipid second messenger phosphatidyl inositol (3,4,5)-trisphosphate (PIP3), which is generated by phosphoinositide 3-kinase (PI3K), and by the Gβγ subunits of heterotrimeric G proteins, which are released upon the activation of G protein-coupled receptors (GPCRs) (Donald et al. 2004; Li et al. 2005). Studies using genetically modified mice showed that P-Rex2 controls the dendrite morphology and synaptic plasticity of cerebellar Purkinje neurons and thus plays a role in motor coordination (Donald et al. 2008; Jackson et al. 2010). A more recent study in mice revealed that P-Rex2 is also important for glucose homeostasis and insulin sensitivity (Hodakoski et al. 2014). This is likely to be relevant in humans, as reduced levels of P-Rex2 are observed in adipose tissue of insulin-resistant patients (Hodakoski et al. 2014).
Whereas our understanding of the physiological roles of P-Rex2 is still relatively limited, P-Rex2 is widely recognized for its roles in cancer progression. P-Rex2 is the most frequently mutated Rho-GEF in cancer, particularly in melanoma and pancreatic cancer. Deregulated expression of P-Rex2 is also seen in many types of cancer (Berger et al. 2012; Waddell et al. 2015). Furthermore, P-Rex2 is a negative regulator of the tumor suppressor PTEN, independently of its Rac-GEF activity (Fine et al. 2009). Through deregulation of its expression and activity, P-Rex2 can promote the growth and invasive migration of cancer cells in vitro and accelerate tumor growth in vivo.
Several reviews have recently evaluated the P-Rex2 literature, including reference (Welch 2015). This allows us to focus here on the newest developments, such as the elucidation of novel mechanisms of P-Rex2 regulation and the discovery that cancer-associated P-Rex2 mutations render the GEF insensitive to negative regulation by PTEN.
Gene and Proteins
The P-Rex2 protein has the same domain structure as P-Rex1, consisting of an N-terminal Dbl homology (DH) domain, which harbors the catalytic Rac-GEF activity, in tandem with a pleckstrin homology (PH) domain, followed by two DEP and two PDZ protein interaction domains, and weak homology over the C-terminal half to inositol polyphosphate 4-phosphatase (IP4P). This IP4P domain is not present in the P-Rex2b splice variant and is thought to be devoid of phosphatase activity (Fig. 1a).
P-Rex2 and P-Rex2b activate the small G protein (GTPase) Rac1 (Donald et al. 2004; Joseph and Norris 2005) and possibly also other Rac isoforms, although this remains to be investigated. In vivo, P-Rex2 and P-Rex2b were shown to activate Rac1 but not the other principal Rho family GTPases Cdc42 and RhoA (Donald et al. 2004; Rosenfeldt et al. 2004). P-Rex2 promotes the release of GDP from Rac1, allowing excess free cellular GTP to bind to the GTPase and thus inducing the active Rac1 conformation that is able to engage downstream target proteins (Donald et al. 2004; Rosenfeldt et al. 2004) (Fig. 1b). Through this Rac-GEF activity, P-Rex2 controls a wide range of cellular functions, including responses that depend on the structure of the actomyosin cytoskeleton, such as cell adhesion and migration, but also others such as gene expression. In addition, P-Rex2 serves as a positive regulator of PI3K signaling, through its ability to inhibit the tumor suppressor PTEN (Fine et al. 2009). This inhibition of PTEN is an adaptor function of P-Rex2 that is independent of the catalytic Rac-GEF activity (see below). A catalytically inactive P-Rex2 mutant (E30A/N212A, GEF-dead) was recently developed (Mense et al. 2015) that will help to dissect GEF-activity dependent and independent functions in the future.
P-Rex2 mRNA is detectable in many tissues (Donald et al. 2004). However, the tissue distribution of the P-Rex2 protein is more restricted, being highest in brain (particularly the cerebellum) and lung (Donald et al. 2004; Donald et al. 2008). Notably, P-Rex2 is absent from leukocytes, in contrast to P-Rex1. The splice variant P-Rex2b is found in endothelial cells and in heart (Rosenfeldt et al. 2004; Li et al. 2005). One mechanism that could explain why P-Rex2 mRNA, but not P-Rex2 protein, can be seen in some tissues is through post-transcriptional control by micro-RNAs (miRs). P-Rex2 expression was found to be regulated by the binding of miR-338-3p to the 3′ UTR of its mRNA, which limited protein production. This repression was lost in human neuroblastoma and gastric cancer cells, resulting in the upregulation of P-Rex2 (Chen et al. 2013; Guo et al. 2014). In neuroblastoma cells, this mechanism was sufficient to induce anchorage-independent cell growth and invasiveness (Chen et al. 2013). Similarly, in gastric cancer cells, blockade of the miR-338 promoter by the transcriptional suppressor MECP2 was sufficient to induce P-Rex2 expression, cell growth, and proliferation (Tong et al. 2016).
Regulators and Binding Partners
PIP3 and Gβγ
As mentioned above, P-Rex2 and P-Rex2b can be activated by PIP3 and Gβγ, both in vitro and in vivo (Donald et al. 2004; Rosenfeldt et al. 2004; Li et al. 2005). Furthermore, PIP3 and Gβγ can activate P-Rex2 either independently or synergistically (Donald et al. 2004). This synergistic mode of activation is unique to the P-Rex family and enables the GEFs to act as coincidence detectors for concomitant signaling through PI3K-coupled receptors and GPCRs.
Recently, the isolated DH domain of P-Rex2 was shown to be sufficient for the Gβγ-mediated stimulation of Rac-GEF activity in vitro, similar to previous findings for P-Rex1 (Welch 2015), and modeling was used to predict that residues Lys254 and Arg263 in the P-Rex2 PH domain are critical for the interaction with PIP3. Indeed, K254E and R263E mutations were shown to be sufficient to inhibit PIP3 binding and the PIP3 mediated stimulation of Rac-GEF activity (Barrows et al. 2015) (Fig. 2). Likewise, the equivalent residues in P-Rex1, Lys280 and Arg289, were shown by Cash et al. to be required for PIP3 binding (see P-Rex1 chapter). An intriguing early study had proposed that the PH domain of P-Rex2 may also contribute to Rac1 binding (Joseph and Norris 2005). If confirmed, this would suggest a very different mechanism of catalysis for P-Rex2 compared to P-Rex1, given the recent crystal structures of P-Rex1 which showed that the PH domain does not make contact with Rac1. However, as the DH/PH domains of P-Rex1 and P-Rex2 are quite similar, this seems unlikely.
Finally, P-Rex2 is mainly cytosolic under basal conditions, but must translocate to the cell membrane in order to activate Rac1 (Donald et al. 2004). For P-Rex1, it is known that PIP3 and Gβγ can synergistically stimulate membrane translocation as well as Rac-GEF activity. It seems likely that the same mechanism regulates the subcellular localization of P-Rex2, but this remains to be investigated.
P-Rex2 binds directly to the tumor suppressor PTEN that blocks PI3K signaling by converting the PI3K product PIP3 back to PI(4,5)P2 (Fine et al. 2009). This interaction between P-Rex2 and PTEN is conferred by binding of the P-Rex2 PH domain to the catalytic and C2 domains of PTEN and through additional contacts of the P-Rex2 IP4P domain with the C-terminal PDZ-binding domain of PTEN (Fine et al. 2009; Hodakoski et al. 2014) (Fig. 2). The interaction is specific to P-Rex2, as P-Rex1 does not bind PTEN. Importantly, P-Rex2 inhibits the phosphatase activity of PTEN, independently of its Rac-GEF activity. In consequence, P-Rex2 deficiency leads to increased PTEN activity and a suppression of PIP3 levels and PI3K pathway activity (see below) (Hodakoski et al. 2014). The inhibition of PTEN by P-Rex2 requires phosphorylation of the PTEN tail. However, this phosphorylation seems to occur constitutively, to maintain PTEN stability. Moreover, in mouse embryonic fibroblasts, the P-Rex2 mediated inhibition of PTEN was only seen upon insulin stimulation, whereas in the liver, it occurred constitutively (Hodakoski et al. 2014). Further research is therefore required to determine under which physiological conditions P-Rex2 can inhibit PTEN.
A recent study showed that treatment of HEK293 cells with phosphatase inhibitors reduces the sensitivity of P-Rex2 activity to PIP3 and Gβγ upon isolation of the GEF (Barrows et al. 2015), suggesting that globally high P-Rex2 phosphorylation levels are associated with low activity. Mass spectrometric analysis revealed that multiple serine and threonine residues of P-Rex2 were phosphorylated, both in basal and insulin-stimulated HEK293 cells. Among these sites, Ser1107 was phosphorylated specifically in insulin-stimulated cells and was therefore investigated further. Use of PH domain mutants and of the GEF-dead E30A/N212A mutant revealed that phosphorylation of Ser1107 depends on PIP3 binding to P-Rex2, and on its Rac-GEF activity. Hence, phosphorylation of Ser1107 occurred downstream of the P-Rex2-mediated activation of Rac1, within the PI3K signaling pathway. This led to the discovery of the Rac-GTP effector p21-activated kinase (PAK) as the kinase responsible (Barrows et al. 2015) (Fig. 2). PAK2 could phosphorylate P-Rex2 directly in vitro, and Ser1107 was identified as one of its target residues (although not the only one). Importantly, coexpression with wild-type but not kinase-dead PAK1 inhibited the Rac-GEF activity of P-Rex2 in broken-cell assays (Barrows et al. 2015). Therefore, it appears that P-Rex2 is able to catalyze its own inactivation through a Rac1/PAK-dependent negative feedback loop (Fig. 2). However, it is still unknown if phosphorylation of Ser1107 by PAK is sufficient to inhibit P-Rex2. Intriguingly, phosphorylation of the equivalent residue in P-Rex1 (Ser1169) by unknown kinases is associated with increased Rac-GEF activity (see P-Rex1 chapter). It also remains to be shown whether other pathways affect the PAK-dependent inhibition of P-Rex2 during PI3K signaling.
The serine phosphatase PP1α binds directly and constitutively to P-Rex2, through an RVxF motif (residues 1084-1087) in the C-terminal half of the GEF (Barber et al. 2012). P-Rex1 also binds PP1α and is activated by PP1α-mediated dephosphorylation of Ser1165 (Barber et al. 2012). However, it remains unknown whether PP1α can dephosphorylate the equivalent residue in P-Rex2, Ser1103. Instead, Ser1107 of P-Rex2 was recently identified as one of the target residues of PP1α (and to a lesser extent of PP2A), whereas the phosphorylation of that residue was mediated by the negative regulator PAK (Barrows et al. 2015) (Fig. 2). Altogether, it therefore seems likely that P-Rex2 activity is regulated by PP1α, but it this remains to be investigated further, particularly on the endogenous level. A P-Rex2 mutant with an RvXF motif unable to bind PP1α (V1085A/F1087A, “VAFA”) was recently generated (Barrows et al. 2015) that will enable future investigation of PP1α-dependent P-Rex2 functions. It should also be noted that, in contrast to full-length P-Rex2, the splice variant P-Rex2b does not contain the consensus RVxF motif for PP1α binding, so P-Rex2b is unlikely to be regulated PP1α.
Finally, P-Rex2 migrates on gels as two distinct bands, and the mobility of these bands can be affected by phosphorylation. Cell fractionation was used to show that the lower-migrating band, which reflects an overall dephosphorylated state, is found preferentially in the membrane fraction, whereas the upper band was seen only in the cytosolic fraction (Barrows et al. 2015). This suggested that phosphorylation of P-Rex2 not only regulates Rac-GEF activity, but might also contribute to the control of subcellular localization. This possibility also requires further study. In general, the mechanisms that regulate P-Rex2 membrane translocation remain to be elucidated.
Like P-Rex1, P-Rex2 and P-Rex2b also directly interact with the serine kinase mTOR, a key regulator of cell growth and of a plethora of other cell responses (Hernández-Negrete et al. 2007). However, it is unknown if P-Rex2 and P-Rex2b can modulate mTOR signaling or vice versa (Fig. 2).
Neurons and Behavior
P-Rex2 is highly expressed in cerebellar Purkinje neurons, which control motor coordination (Donald et al. 2008). Dendrite morphology was shown to be altered in Purkinje neurons of Prex2−/− mice, and the mice developed a motor coordination defect that worsens with age, particularly in females (Donald et al. 2008). Prex1−/−Prex2−/− double-deficient mice showed an exacerbated phenotype, with a more severe impairment of motor activity, as well as posture and gait defects. These impairments were consistent with cerebellar dysfunction and were evident in both males and females from a young age (Donald et al. 2008). Furthermore, patch clamp electrophysiology showed that Prex1−/−Prex2−/− Purkinje neurons compensate well for the dendrite morphology defect, as their passive membrane properties and basal synaptic transmission were normal. However, these neurons had impaired synaptic plasticity, as they were unable to maintain long-term potentiation, which is required for learned motor coordination skills (Jackson et al. 2010).
Glucose Homeostasis and Insulin Sensitivity
P-Rex2 controls glucose homeostasis and insulin sensitivity. This was first demonstrated in Prex2 −/− mice, which showed an impaired ability to control blood glucose levels upon glucose or insulin challenge (Hodakoski et al. 2014). PIP3 levels and the activities of several insulin pathway components were suppressed in the adipose tissue and liver of Prex2−/− mice, whereas PTEN activity was elevated (Hodakoski et al. 2014). Therefore, it was proposed that P-Rex2 controls insulin signaling, at least in part, through its inhibition of PTEN. However, the relative contribution of GEF-dependent and -independent functions of P-Rex2 to insulin signaling and glucose homeostasis remains to be elucidated. Importantly, the role of P-Rex2 in glucose homeostasis seems to be relevant in humans, as P-Rex2 protein levels were shown to be reduced in adipose tissue of insulin-resistant patients (Hodakoski et al. 2014). As insulin resistance can develop into type 2 diabetes, it seems possible that reduced P-Rex2 expression may be involved in the development of metabolic syndrome, although this remains to be investigated.
Knockdown of P-Rex2b in human umbilical vein endothelial cells (HUVECs) showed that this Rac-GEF is required for the sphingosine 1-phosphate stimulated migration of endothelial cells (Li et al. 2005). Moreover, a recent study, which employed an shRNA knockdown screen in HUVECs, identified P-Rex2 as a potential regulator of mechanical-force-induced orientation of endothelial cells, a response which is relevant in vivo for the integrity of blood vessel walls (Abiko et al. 2015). Therefore, P-Rex2 and P-Rex2b appear to control endothelial cell function, but a full characterization of the physiological roles of these Rac-GEFs in vascular biology remains to be done.
The roles of P-Rex2 in cancer have recently been widely reviewed elsewhere; please see reference (Welch 2015) for citations of detailed reviews. Here we provide largely an update.
The human PREX2 locus lies in a genomic region that is often amplified in melanoma, breast, prostate, and colorectal cancers (Fine et al. 2009; Berger et al. 2012). In melanoma, the locus also undergoes frequent chromosomal translocations and rearrangements (Berger et al. 2012). Hence, multiple mechanisms seem to contribute to the deregulation of P-Rex2 expression in cancer. Furthermore, although it is very unusual for Rho-GEFs to be mutated, P-Rex2 mutations are frequently observed in cancer, particularly in melanoma, where the Rac-GEF is mutated in 14% of cases (Berger et al. 2012), and in pancreatic ductal adenocarcinomas (PDACs; 10%) (Waddell et al. 2015). Both the deregulated expression and mutations of P-Rex2 have been shown to promote tumor growth.
In breast cancer, P-Rex2 expression is correlated with wild-type PTEN status and with activating PI3K mutations (Fine et al. 2009). Knockdown of P-Rex2 in breast cancer cells, which have normal PTEN levels (e.g. MCF7), showed that the Rac-GEF is required for Akt activation, as well as for cell growth (Fine et al. 2009). Furthermore, comparison of various breast cancer cell lines that either do or do not express PTEN suggested that P-Rex2 may control cell growth at least in part through its inhibition of PTEN (Fine et al. 2009). Finally, aberrant tyrosine phosphorylation of P-Rex2 was recently proposed as another mechanism through which the Rac-GEF might promote breast cancer, from a phospho-proteome screen of murine mammary tumors driven by loss of the tumor suppressor p53 (Ali et al. 2014).
Many different P-Rex2 mutations are seen in human melanomas. These mutations include truncations and missense mutations that are distributed throughout the length of the coding region (Berger et al. 2012). Several of these mutations have been evaluated in murine xenograft and transgenic models of melanoma. Expression of the P-Rex2 mutants K278*, E824*, G844D, or Q1430* in melanocytes that also bore an activating N-Ras mutation accelerated the tumor growth of xenografts and reduced tumor-free survival (Berger et al. 2012). Similarly, reduced tumor-free survival was seen in a transgenic mouse strain with melanocytes that inducibly expressed the P-Rex2 mutant E824* as well as an activating N-Ras mutation, and tumors from this mouse exhibited increased cell proliferation (Lissanu Deribe et al. 2016). Tumors from the xenograft and transgenic models also showed decreased levels of the cell cycle inhibitors CdcN1C (Kip2) and CdcN1B (Kip1), respectively (Lissanu Deribe et al. 2016). Furthermore, increased levels of Rac1-GTP and Akt phosphorylation were observed in melanocytes that expressed the P-Rex2 mutants K278*, E824* or Q1430* in combination with an activating N-Ras mutation (Lissanu Deribe et al. 2016). Such increased Rac1-GTP levels could either be caused by lost intramolecular inhibition of P-Rex2, or through loss of inhibition by PTEN. As described above, the P-Rex2 G844D mutant evaded PTEN inhibition through reduced binding, and the Rac-GEF activity of the P-Rex2 V432M mutant was insensitive to PTEN (Mense et al. 2015) (Fig. 3). Inversely, the increased levels of Akt phosphorylation were thought be a consequence of continued PTEN inhibition by P-Rex2 mutants. However, the K278* mutant (which is essentially an isolated DH domain and thus should not be able to bind and inhibit PTEN) still increased Akt phosphorylation (Lissanu Deribe et al. 2016), suggesting that P-Rex2 can increase PI3K signaling in melanoma also independently of PTEN. Altogether, it appears that P-Rex2 mutations can promote melanoma growth through a combination of mechanisms which need to be delineated further in the future.
A recent study that used whole-genome sequencing and copy number variation analysis of 100 PDACs identified P-Rex2 as a new candidate driver of pancreatic cancer (Waddell et al. 2015). P-Rex2 was found to be mutated in 10% of PDAC patients, and these mutations included frame shift, splice site, and missense mutations. As described above, the Parsons lab recently investigated one of these pancreatic cancer-associated mutants, P-Rex2 P948S, which showed reduced sensitivity to inhibition by PTEN in vitro and caused increased invasive cell migration (Mense et al. 2015) (Fig. 3b). Furthermore, another recent study showed that overexpession of P-Rex2 also occurs in pancreatic cancer and that P-Rex2 is required for the growth and invasive migration of pancreatic cancer cells (Yang et al. 2016). P-Rex2 thus seems to be emerging as an important factor in pancreatic cancer progression. However, the effects of deregulated expression and mutation of P-Rex2 on pancreatic cancer growth and metastasis require extensive future study.
P-Rex2 is also overexpressed in several other types of cancers. As described above, expression of P-Rex2 in gastric cancer and neuroblastoma is caused through loss of repression by miR-338-3p, driving the growth of gastric cancer cells (Guo et al. 2014; Tong et al. 2016) and the growth and invasive migration of neuroblastoma cells (Chen et al. 2013). In liver cancer, overexpression of P-Rex2 is correlated with hepatitis-B virus status and stimulated by the chemokine CXCL9. Relevantly, P-Rex2 is required for invasive migration of hepatocellular carcinoma cells (Lan et al. 2014; He et al. 2016). Furthermore, a recent transcriptomic analysis of acute myeloid leukemias revealed upregulation of P-Rex2 mRNA in a subset of these cancers that features rearrangements in the gene for the transcription factor EVI1. It will be interesting to learn if P-Rex2 protein is produced in these leukemias, as P-Rex2 is not normally detected in leukocytes (Lavallee et al. 2015). Finally, a SNP (rs4512367) in PREX2 that causes a homozygous intronic variant was recently found to be strongly associated with the risk of tobacco habitues developing oral squamous cell carcinoma. This risk was increased further when the P-Rex2 SNP was co-inherited with SNPs in the Ras-GEF RASGRP3 and the glutamate receptor GRIK2 (Multani et al. 2016). In conclusion, P-Rex2 appears to drive the growth and invasive migration of cells from many different types of cancers, although for most of these cancers, a causal role for! P-Rex2 still remains to be evaluated in vivo.
Important recent discoveries in P-Rex2 research were the description of negative P-Rex2 regulation by PAK and PTEN, the molecular characterization of cancer-associated P-Rex2 mutations, and the discovery that cancer-associated P-Rex2 mutants can escape from PTEN inhibition. Many fundamental aspects of P-Rex2 biology are still unknown, however. Future research should include an elucidation of the precise substrate specificity and the regulation of subcellular localization, as well as structural analysis. Searches should be conducted to identify further P-Rex2 interacting proteins for evaluation as potential regulators or effectors. It will also be interesting to discover new physiological roles of P-Rex2 and to define specific functions for P-Rex2b, for example the respective roles of the two P-Rex2 isoforms in vascular biology. Furthermore, it will be important to elucidate the contribution of Rac-GEF activity dependent and independent roles of P-Rex2, both in cancer and elsewhere, and to explore further the importance of the P-Rex2 interaction with PTEN for the functions of both proteins. More research with insulin resistant and type-2 diabetic patients is also required to elucidate the possibility that deregulation of P-Rex2 contributes to the development of metabolic syndrome.
The importance of P-Rex2 in cancer makes the Rac-GEF a valid therapeutic target. However, the catalytic activity of Rac-GEFs is not easily targetable. For P-Rex2, alternative opportunities might be found in the development of small-molecule compounds that mimic the negative regulation of P-Rex2 by PAK or PTEN. In the development of such drugs, one would have to be mindful, however, to control also the P-Rex2-dependent inhibition of PTEN, in order to preserve the ability of the tumor suppressor to downregulate pro-survival PI3K signaling.