ERK-ERF-EGR1, a novel switch underlying acquisition of a motile phenotype

Abstract

Unlike the well-characterized checkpoints of the cell cycle, which establish commitment to cell division, signaling pathways and gene expression programs that commit cells to migration are incompletely understood. Apparently, several molecular switches are activated in response to an extracellular cue, such as the epidermal growth factor (EGF), and they simultaneously confer distinct features of an integrated motile phenotype. Here we review such early (transcription-independent) and late switches, in light of a novel ERK-ERF-EGR1 switch we recently reported in the FASEB Journal. The study employed human mammary cells and two stimuli: EGF, which induced mammary cell migration, and serum factors, which stimulated cell growth. By contrasting the underlying pathways we unveiled a cascade that allows the active form of the ERK mitogen-activated protein kinase (MAPK) cascade to export the ERF repressor from the nucleus, thereby permitting tightly balanced stimulation of an EGR1-centered gene expression program.

Epithelial sheets acquire motile phenotypes in the context of normal physiology, such as in embryogenesis and in tissue repair,¹ as well as under pathological conditions, which include organ fibrosis and metastases formation.² Motility of epithelial cells requires a substrate matrix on which cells crawl, while repeatedly engaging a cyclic process entailing formation of adhesion sites at the leading edge, and simultaneous disassembly of adhesions at trailing tails.³ To sustain this process and also invade across basement membranes and walls of lymph or blood vessels, cells apply a myriad of switches, both early transcription-independent events, and transcriptional switches involving newly synthesized mRNAs, microRNAs and proteins. The inducers of the switches are primarily soluble molecules, such as chemokines and growth factors. The latter include the hepatocyte growth factor (HGF), the transforming growth factors and bone morphogenic proteins (BMPs), along with multiple molecules of the epidermal growth factor (EGF) family. Employing mammary epithelial cells and contrasting two stimuli, EGF, which induces migration, and serum factors, which stimulate cellular proliferation, we recently reported a novel composite switch, which involves EGF-induced activation of the ERK mitogen-activated kinase (MAPK) pathway, phosphorylation and translocation of the ERF transcriptional repressor from the nucleus to the cytoplasm, as well as de novo induction of the EGR1 transcription factor.⁴ Below we review the ERK-ERF-EGR1 switch in the context of several other molecular switches, which constitute the core of motility induction (see Fig. 1).

Figure 1. Molecular switches instructing the acquisition of motile phenotypes by epithelial cells. Growth factor-induced, early and late molecular mechanisms are schematically depicted. The early switches involve activation of signal transduction pathways by a growth factor, such as EGF. For example, activation of PI3K results in the generation of specific phosphatidylinositol lipids, which recruit an adaptor called TKS and enable assembly of an actin-filled invasive structure (invadopodium). Simultaneous activation of PLC-gamma results in breakdown of another inositol lipid, phosphatidylinositol 4,5 bisphosphate, thereby activates several actin binders, such as cofilin, an actin-severing enzyme necessary for lamellipodium formation. The best-characterized late switch elevates N-cadherin and downregulates the abundance of E-cadherin, by either repressing transcription, destabilizing the transcript corresponding to E-cadherin, or targeting E-cadherin to degradation. Consequently, E-cadherin based adhesion complexes, which anchor actin filaments through catenins, undergo disassembly. Another late switch dissociates integrin-based adhesion sites by downregulating tensin 3, while enhancing expression of cten, a tensin family member that lacks the characteristic actin-binding domain. The newly described switch (depicted within the nucleus of the cell on the left side) entails activation of the ERK pathway, leading to rapid turnover of microRNA-191 and phosphorylation of the ERF transcriptional repressor. A major burst of transcription follows translocation of modified ERF molecules from the nucleus to the cytoplasm. This includes transcription of EGR1 and several downstream targets, collectively regulating a well-balanced phenotypic transition. ECM, extracellular matrix; EGF, epidermal growth factor; ERK, extracellular-regulated kinase; PLC, phospholipase C; PI3K, phosphatidylinositol 3-kinase.

Early Switches Conferring Motile Phenotypes

To sustain their locomotion, epithelial cells extend membrane protrusions (e.g., lamellipodia and filopodia) and apply contractile forces by means of actin stress fibers.³ In parallel, spatiotemporally regulated cycles of vesicle exocytosis and endocytosis ensure turnover of adhesion sites, and supply the large amounts of plasma membrane needed for crawling of the leading edge. For example, cycling of integrins regulates cell migration by allowing rapid turnover of integrin-based adhesion sites called focal adhesions.⁵ How is cell migration induced and maintained by an extracellular cue is well exemplified by EGF, which binds with a receptor tyrosine kinase, called EGFR/ERBB-1, and simultaneously stimulates several distinct pathways necessary for cell migration (see Fig. 1).

The phospholipase C pathway

Phospholipase C (PLC) gamma, the enzyme that hydrolyzes phosphoinositol 4,5 bisphosphate [PI(4,5)P₂] into diacylglycerol and inositol triphosphate [Ins(1,4,5)P₃], is rapidly phosphorylated and activated by EGFR, and this is essential for the induction of fibroblast motility.⁶ The underlying mechanism entails displacement of several actin-modifying proteins from a PI(4,5)P₂-bound inactive state anchored at the plasma membrane. For example, gelsolin, a Ca²⁺-regulated actin filament severing, capping, and nucleating protein, undergoes de-inactivation and translocates to the cytoplasm in response to EGF stimulation.⁷ Likewise, another PI(4,5)P₂ binder and actin severing protein, called cofilin, is released from the plasma membrane and severs F-actin, which is coincident with actin polymerization and lamellipod formation.⁸

The ERK-MAPK pathway

One important function of the active ERK-MAPK cascade is the regulation of membrane protrusions.⁹ EGF-activated ERK localizes to the leading edge of protruding lamellipodia and phosphorylates cortactin¹⁰ and the WAVE2 regulatory complex (WRC), which leads to WRC binding to and activation of the ARP2/3 actin nucleator. Importantly, the ERK-MAPK cascade also controls the rapid turnover of sites confined to the trailing tail. Activation of the ERK-MAPK cascade stimulates calpain, an intracellular protease,¹¹ which cleaves spectrin and talin, along with other adhesion-related proteins.¹² Similarly, ERK activates the myosin light chain kinase (MLCK). Once phosphorylated, MLCK induces phosphorylation of the myosin light chain, polymerization of actin fibers, and protrusion of frontal membranes.¹³ Studies that employed HGF revealed yet additional functions of the ERK-MAPK cascade, namely the disassembly of cell-to-cell adherens junction¹⁴ and the regulation of cell-to-matrix adhesion sites.¹⁵ Within such adhesion sites, the focal adhesion complex protein paxillin physically associates with ERK, as well as with the upstream kinases RAF and MEK, resulting in a complex that can mediate localized ERK activation. Once phosphorylated, paxillin recruits the focal adhesion kinase (FAK) to adhesion sites, resulting in rapid turnover of these sites and lamellipod extension. Along with its kinase activity, FAK functions as an adaptor molecule that mediates the assembly of another complex consisting of calpain and ERK.¹⁶ Subsequent calpain-mediated proteolysis of FAK enables focal adhesion turnover and cell migration.

The phosphatidylinositol 3-kinase (PI3K) pathway

Although EGFR cannot directly interact with phosphatidylinositol 3-kinase (PI3K), indirect activation, as well as coupling of activated RAS molecules to PI3K under certain conditions, permit EGF to stimulate the PI3K-AKT cascade. Phosphoinisitides generated by PI3K trigger activation of AKT through direct binding to the pleckstrin homology (PH) domain and subsequent phosphorylation of AKT on two residues by the phosphoinositide dependent kinase-1 (PDK-1; Ser-308) and by other kinases (Ser-473). This pathway controls both migration, by engaging a large spectrum of effector proteins,¹⁷ and invasion, by instigating the assembly of invadopodia, actin-filled, matrix-degrading organelles that propel invasion through tissue barriers.¹⁸ Sequential action of PI3K, which generates PI(3,4,5)P₃,¹⁹ and adaptor TKS proteins, which bind the dephosphorylated product PI(3,4)P₂,²⁰ establish signposts essential for invadopodia formation at the ventral side of migrating cells.²¹ Two downstream effectors of AKT, Rho and NFκΒ, exemplify contributions to cell migration and invasion. Assembly and disassembly of actin filaments involves the RHO family of the small GTPases RAC, RHO and CDC42. It has been shown that EGF can stimulate RAC1 through a pathway involving PI3K and the SRC kinase,²² which likely enhances cell migration via the JNK MAPK pathway. Another migration-regulating effector is the NFκB transcription factor.²³ Although the exact wiring remains unclear, it is possible that PI3K/AKT phosphorylates p65/RelA, a subunit of NFκB, to directly stimulate a gene expression program controlling cell survival and migration.

Transcription-Mediated Switches Conferring Motile Phenotypes

A number of distinct, transcription-controlled molecular switches are engaged in order to initiate trans-differentiation of a polarized epithelium to a collection of motile mesenchymal cells.^1,2 This process is commonly known as the epithelial-mesenchymal transition and it involves loss of epithelial markers like E-cadherin, MUC1, syndecan and laminin-1, concomitantly with gain of mesenchymal markers, such as N-cadherin, vimentin, fibronectin and several transcription factors, including ETS-1 and snail. The roles for such switches are exemplified below using E-cadherin loss and the replacement of a group of F-actin binders called tensins by Cten/tensin 4.

Loss of E-cadherin

An important feature of the transition to motile phenotypes manifests through the disassembly of polarized epithelial cell-cell junctional complexes, such as tight junctions and adherens junctions. The latter are comprised primarily of the transmembrane protein epithelial cadherin (E-cadherin), which connects to the actin cytoskeleton via catenins.²⁴ The loss of E-cadherin expression is considered the hallmark of EMT, and in some cases it is coupled to upregulation of another calcium-dependent cell adhesion molecule, N-cadherin.²⁵ One important example is the acquisition of a fibroblastoid morphology, increased N-cadherin expression, loss of junctional E-cadherin localization, and increased cellular motility in response to TGF-β.²⁶ The transcription of E-cadherin is tightly controlled, either directly by a group of repressors able to bind with the E-cadherin’s promoter (e.g., snail, ZEB1/2, E47 and KLF8), or indirectly by factors like twist, goosecoid, E2.2 and FOXC2.²⁷ Thus, the regulation of E-cadherin expression is remarkably complex. For example, on the one hand all three snail isoforms indirectly interact with chromatin remodeling factors and snail’s transcription is both auto-inhibited and trans-inhibited by EGR1,²⁸ and on the other hand, ZEB1 and ZEB2, crucial activators of the motile phenotypes, are inhibited by members of the microRNA-200 family, which they repress via a bi-directional feedback loop.²⁹

The reciprocal cten-tensin switch

The tensin family comprises four members, all localized to the cytoplasmic tails of integrins at focal adhesions. Unlike tensins 1, 2 and 3, tensin 4 (also called COOH-terminus tensin-like molecule; cten) harbors no N-terminal actin-binding domain that is present in the other tensin proteins.³⁰ Cten is upregulated and associates with a poor prognosis in breast cancer, thymomas, gastric cancers and lung cancers, and according to a recent report mutant RAS upregulates cten expression.³¹ Because tensin family proteins interact with several components of focal adhesion sites (e.g., vinculin and paxillin) they are thought to stabilize cell-to-matrix associations. Importantly, stimulation of mammary cells with EGF is followed by a transcriptional upregulation of cten, concomitant with downregulation of tensin 3.³² This reciprocal switch enables cten to displace tensins from the cytoplasmic tail of integrins, thereby disassemble focal adhesions and promote cell migration. Presumably, by altering adhesion to the matrix, the reciprocal cten-tensin switch biases a weakly adhesive type of cell migration, called ameboid migration.³³

The ERK-ERF-EGR1 Switch

To unravel yet unknown motility controlling switches, we employed human mammary epithelial cells, and contrasted a signaling cascade leading to cell migration (stimulated by EGF) and a distinct pathway culminating in cell proliferation (stimulated by serum factors).⁴ Proteomic analyses using reverse-phase protein arrays revealed that EGF strongly and persistently activated the ERK cascade, but activation of the PI3K-AKT pathway by EGF displayed a relatively weak and transient profile. Reciprocally, the proliferative signal (serum) associated with only a transient stimulation of ERK, but strong and prolonged kinetics of AKT activation. In line with the critical motility roles played by the ERK pathway, a pharmacological approach that used U0126, a MEK inhibitor, and wortmanin, a PI3K blocker, indicated that ERK rather than AKT is essential for migration. Moreover, because application of U0126 markedly stimulated the AKT pathway, we concluded that the motogenic cue polarizes intracellular signaling in favor of the ERK pathway, while suppressing AKT signaling. Presumably, a reciprocal crosstalk enables the mitogenic cue (i.e., serum) to suppress the ERK pathway through a previously reported AKT-mediated inhibition of RAF.³⁴ Consistent with polarized signals, we found that serum rapidly induced de novo transcription of two components of the AP1 transcription complex, namely FOS and FOSB, but stimulation with EGF resulted in weaker activation of these immediate early genes.

Because the ETS family of transcription factors is the main effector of growth factor-activated ERK-MAPK pathway,³⁵ we systematically inactivated individual members and identified the repressors ERF and TEL/ETV6 as a stimulator and an inhibitor, respectively, of cell migration. In line with a previous report,³⁶ on stimulation with EGF, ERF underwent phosphorylation on multiple sites and subsequently exited the nucleus. Possibly, ERF represses expression of migration-promoting genes, or directly promotes migration once in the cytoplasm, a scenario supported by yeast 2-hybrid analyses, which identified adhesion and cytoskeletal proteins as potential ERF partners. Importantly, growth factor-induced, post-translational modifications and protein translocations, as exemplified by ERF, inevitably lead to waves of newly synthesized mRNAs and micro-RNAs (reviewed in ref. 37). The earliest wave comprises the immediate early genes (IEGs), a group of proto-oncogenic transcription regulators. Like in earlier steps, EGF and serum factors differed: while transcription of the IEGs FOS and FOSB followed stimulation by serum and this was essential for cell proliferation, EGR1 represented the first EGF-induced IEG, the transcription of which permitted cell migration.

A combination of DNA arrays, chromatin immunoprecipitation, reporter gene assays and RNA interference identified 14 target genes of EGR1, the majority of which (11/14) were physically bound by this transcription factor. Out of the 14 genes, 10 demonstrated significant and reproducible effects on cell migration, primarily a decrease when disabled, but only one gene, interleukin 8 (IL-8), displayed an effect on cell proliferation. For example, the list of EGR1 targets includes two components of the plasminogen activator (PLAUR and SerpinB2), along with soluble mediators of cell migration like the parathyroid hormone-related protein (PTHrP) and transforming growth factor α (TGF-α), but the list also included inhibitors of cell migration, such as the MAPK phosphatase DUSP4, EGR3 and the repressor BHLHB2. Interestingly, the latter three genes function as negative feedback regulators of EGR1: DUSP4 inhibits the upstream ERK-ERF pathway, whereas EGR3 and BHLHB2 directly inhibit EGR1 transcription. In addition, we found that EGR1 binds to its own promoter and the respective transcript is downregulated by microRNA-191, which undergoes downregulation via the ERK pathway immediately after EGF stimulation.³⁸ Notably, EGR1 has been implicated in early responses to a wide spectrum of mitogens and stressors. In addition, along with reports associating EGR1 with cell proliferation and oncogenesis, the majority of relevant studies linked this transcription factor to pathways leading to apoptosis and tumor suppression.³⁹ Hence, the uncovered rich control of EGR1, the ability to both suppress and induce gene expression, as well as the migration promoting or suppressing actions of EGR1’s targets, are all congruent with critical, but complex regulatory roles.

In summary, the ERK-ERF-EGR1 switch embodies both nuclear and cytoplasmic events culminating in a robust transition to a motile cellular state. However, while the ERK pathway is relatively well understood, the ERF-to-EGR1 part of the switch, along with the downstream targets of EGR1, raise many questions, such as the exact nuclear and cytoplasmic functions of ERF and the identity of the second tier of EGR1’s target genes. Although EGR1 has long been considered a tumor suppressor, a wealth of new evidence shows that this transcription factor promotes progression of prostate and other types of tumors. Remarkably, EGR1 controls on the one hand several hubs of apoptosis (e.g., p53 and PTEN), and on the other hand it induces transcripts promoting angiogenesis (e.g., the vascular endothelial growth factor) and cell migration (e.g., TGF-β1 and fibronectin). Hence, future works will address the paradoxes of EGR1 and its large agenda, which is likely achieved by the choice of specific target genes.⁴⁰ Conceivably, in addition to EGR1, yet unknown but similarly complex switches transform extracellular signals into complex gene expression programs essential for tumor progression.

Tarcic G, Avraham R, Pines G, Amit I, Shay T, Lu Y, et al. EGR1 and the ERK-ERF axis drive mammary cell migration in response to EGF. FASEB J. 2012;26:1582–92. doi: 10.1096/fj.11-194654.