Targeting ERK beyond the boundaries of the kinase active site in melanoma

Abstract

ERK1/2 constitute a point of convergence for complex signaling events that regulate essential cellular processes, including proliferation and survival. As such, dysregulation of the ERK signaling pathway is prevalent in many cancers. In the case of BRAF-V600E mutant melanoma, ERK inhibition has emerged as a viable clinical approach to abrogate signaling through the ERK pathway, even in cases where MEK and Raf inhibitor treatments fail to induce tumor regression due to resistance mechanisms. Several ERK inhibitors that target the active site of ERK have reached clinical trials, however, many critical ERK interactions occur at other potentially druggable sites on the protein. Here we discuss the role of ERK signaling in cell fate, in driving melanoma, and in resistance mechanisms to current BRAF-V600E melanoma treatments. We explore targeting ERK via a distinct site of protein-protein interaction, known as the D-recruitment site (DRS), as an alternative or supplementary mode of ERK pathway inhibition in BRAF-V600E melanoma. Targeting the DRS with inhibitors in melanoma has the potential to not only disrupt the catalytic apparatus of ERK but also its non-catalytic functions, which have significant impacts on spatiotemporal signaling dynamics and cell fate.

Overview of the ERK signaling pathway

Components of the pathway

The two major extracellular signal-regulated kinase (ERK) isoforms, ERK1 and ERK2, are the final components of the core Ras/Raf/MEK/ERK cellular signaling cascade (Figure 1A). ERK pathway signaling is activated by a variety of mitogens and other extracellular sources, such as growth factors, insulin, osmotic stress, and cytokines.^1,2 These various ligands and stimuli trigger receptor-tyrosine kinase (RTK) dimerization and activation at the cell membrane, which in turn leads to recruitment of Ras-GDP to the receptor. Ras-GDP is converted to Ras-GTP with assistance from the guanine-nucleotide exchange factors Sos-1/2 (Son of Sevenless homolog -1 and -2),³ which are recruited to the receptor via several adaptor proteins (Shc, Grb2).¹ Ras-GTP then leads to the activation of the Raf family (A-Raf, B-Raf, C-Raf), culminating in Raf dimerization.⁴ Raf serves as the primary member of the mitogen-activated protein kinase kinase kinase (MAP3K) tier in the core ERK signaling pathway. Activated Raf then phosphorylates and activates MEK1/2 (the MAP kinase kinase or MAP2K tier), which phosphorylate threonine and tyrosine on ERK1 (T202/Y204)^* and ERK2 (T185/Y187) (the MAPK tier), leading to their activation.¹

Figure 1. ERK phosphorylation cascade and feedback signaling.

The ERK signaling cascade¹ (A) begins with pathway stimulation by ligands, resulting in the activation of receptor tyrosine kinases (RTKs) that lead to Ras activation. Ras activates Raf, which begins a phosphorylation/activation cascade leading to ERK activation. ERK phosphorylates numerous substrates in the cytoplasm and nucleus that play a role in cell fate, such as RSK and the AP-1 and ETS families of transcription factors. (B) The ERK pathway has many intrinsic mechanisms of negative feedback regulation.^{10,11,152,153} ERK can exert negative feedback at each level of the pathway via phosphoregulation. ERK can also drive transcription of negative pathway regulators such as DUSP5/6 and SPRY that can and inactivate members of the ERK pathway. ^{1,10,11,153,154}

ERK1 and ERK2 are serine/threonine protein kinases that phosphorylate over 250 known substrates.⁵ These substrates include transcription factors (e.g. Ets-1, Elk-1, c-Fos, c-Jun), the RSK family of kinases, phosphatases, apoptotic proteins, and cytoskeletal proteins, among many others.¹ This wide array of substrates governs cellular processes, such as proliferation, gene transcription, cell cycle progression, migration, adhesion, survival, and metabolism.⁶

ERK isoforms

ERK1 and ERK2 share high structural and sequence similarity (85%)⁷ and are ubiquitously expressed, albeit with ERK2 typically expressed in higher levels relative to ERK1 across tissues and model systems.⁸ To date, no isoform-specific agonists or substrates have been identified, though an assay employing chemoproteomics has recently been developed to quantify independent isoform activity across different cell types.⁹ Silencing ERK1 or ERK2 using shRNA or siRNA has shown in numerous studies that loss of ERK2 has a more significant biological effect than the loss of ERK1, though this could result from a greater global reduction of ERK levels due to higher endogenous expression of ERK2.⁸ Though it is currently unclear whether or not ERK1 and ERK2 functions are redundant, data such as differing mutations between the isoforms in cancers (COSMIC database <http://cancer.sanger.ac.uk/cancergenome/projects/cosmic/>) imply that there may be context-dependent isoform-specific roles that have yet to be characterized.

Pathway feedback

Because the ERK pathway governs so many substrates and cellular processes, tight regulation of its signaling outputs is crucial. One of the ways this is achieved is through negative feedback within the ERK pathway, some examples of which are illustrated in Figure 1B. ERK can directly phosphorylate and inhibit upstream pathway components such as Raf, Sos, and some receptor tyrosine kinases, and ERK signaling can also differentially regulate MEK1/2 levels.^10–12 Additionally, ERK phosphorylates transcription factors that induce expression of negative regulators of the ERK pathway, such as DUSP5/6 phosphatases and SPRY.^13,14 Therefore, ERK activity itself can directly and indirectly exert feedback-regulation at every level of the upstream signaling pathway.

Cell cycle progression

ERK activity is implicated in the promotion of a variety of processes in different phases of the cell cycle, some of which are discussed here. First, ERK is known to have a role in regulating several foundational elements of cell cycle progression, including pyrimidine synthesis, ribosome synthesis, chromatin remodeling, and protein translation.¹⁵ Translocation of ERK to the nucleus is required for entry into the cell cycle, and elevated ERK activity must occur within the G1 phase in order for a cell to enter the S phase. In the G1/S transition, newly expressed cyclin D forms complexes with CDK4/6, which leads to the release of E2F transcription factors from pRb repression. This results in the induction of cyclin E expression which is necessary for S-phase entry (Figure 2A). In this checkpoint, ERK regulates induction of cyclin D1 expression via the Fos family of transcription factors and Myc phosphorylation (Figure 2A). ERK activity is also required for the nuclear translocation and phosphorylation of CDK2 at a positive regulatory site.¹⁵ Additionally, ERK aids cell cycle progression by suppressing anti-proliferative genes, such as JunD (Figure 2A).

Figure 2. The roles of ERK in the cell cycle and apoptosis.

(A) ERK translocation to the nucleus is required for cell cycle entry and plays many roles in cell cycle progression.^15,155 In the G1/S transition, ERK: regulates cyclin D1 induction via Fos and Myc, represses JunD transcription and is required for CDK2 nuclear translocation and phosphorylation. In the G2/M transition, ERK is one of the kinases that phosphorylate cyclin B1 for nuclear translocation. ERK phosphorylation of RSK leads to phosphorylation/inactivation of Myt1, which prevents Myt1 from activating CDK1. (B) ERK plays several roles in preventing or promoting apoptosis.^17,18,21,22 ERK can phosphorylate Bim and Bad to relieve suppression of anti-apoptotic proteins in the Bcl-2 family (Mcl-1, Bcl-2, and Bcl-XL), and ERK can phosphorylate and inhibit caspase-9. Conversely, sustained activation of ERK can lead to apoptosis via upregulation and accumulation of p53, as a result of p14ARF binding to Mdm2.

ERK is also reported to be involved in the G2/M transition, which requires the formation of cyclin B/CDK1 complexes. For example, it is one of the kinases that phosphorylate cyclin B1 in order to promote its nuclear translocation (Figure 2A).¹⁵ ERK also represses negative regulators of the G2/M transition. For instance, its phosphorylation of RSK leads to phosphorylation and inactivation of Myt1, which prevents Myt1 from inactivating CDK1 (Figure 2A).¹⁵

The duration of ERK signaling can also influence cell cycle progression. Robust early activation of ERK in the cell cycle, followed by moderate sustained activity drives G1 progression and cell proliferation. For example, c-Fos expression represses cyclin D1, but after sustained ERK activation, it is believed that c-Fos can drive cyclin D1 expression via induction of the transcription factor Fra-1 (Figure 2A).¹⁵ However, sustained ERK activity does not always positively regulate cell cycle progression. Persistent hyperactivation of ERK can paradoxically block S-phase entry.¹⁵ Hyperactivation of ERK results in a large accumulation of cyclin D1, which binds to p21CIP1 and prevents its degradation. p21CIP1 is an inhibitor of cyclin E/CDK2 complex formation, which is required for S-phase entry. This accumulation of p21CIP1 induces G1 arrest and is also associated with inhibition of CDK4 and CDK2.¹⁶ In response to hyperproliferative stimuli like oncogenic ERK/Myc signaling, p53 can also lead to G1 arrest by inducing p21CIP1 expression.¹⁷

Apoptosis

ERK signaling is also known to play a role in apoptosis. In a pro-survival context, ERK can mediate phosphorylation of Bim and Bad by multiple mechanisms, which relieves suppression of anti-apoptotic proteins in the Bcl-2 family, like Mcl-1, Bcl-2, and Bcl-XL (Figure 2B).^18–20 ERK can also phosphorylate and inhibit caspase-9, which is required for apoptosis.²¹ Alternatively, high levels of ERK signaling through Myc can result in apoptotic signaling via upregulation of p53 (versus G1 arrest as described above) that leads to p14ARF synthesis.^17,22 p14ARF binds to Mdm2, an E3 ubiquitin ligase and a transcriptional repressor of p53, and thereby prevents p53 degradation. This stabilization and the resulting accumulation of p53 due to p14ARF results in cell death (Figure 2B).

ERK pathway signaling dynamics

ERK pathway signaling is characterized by several key, often interconnected spatiotemporal factors: (1) the amplitude and (2) the duration of ERK signaling, and (3) the subcellular localization and distribution of ERK activity. Understanding the circuitry and dynamics of ERK signaling is critical for revealing how the different cellular fates described in the previous section are delegated.

ERK signaling amplitudes and thresholds

There is abundant evidence supporting the notion that ERK signaling can alter cell fates by crossing certain activity thresholds. For example, as mentioned earlier, robust sustained ERK activity drives cell cycle progression, but excessive ERK activation can lead to cell cycle arrest. In some cases, this type of behavior can be described as switch-like, where an increase or decrease in ERK phosphorylation in a cell can lead to an all-or-nothing binary response. Indeed, ERK nuclear translocation in response to graded phosphorylation has been observed to be switch-like.²³ Models of the ERK pathway, as well as clinical data, suggest that ERK activity must be suppressed to a threshold of less than 10% of its maximal level to block proliferation.^24,25 On the other hand, hyperactivation or excess expression levels of ERK have been shown to result in cellular growth arrest or apoptosis (Figure 3).^26,27 Several mutational studies have shown that gain-of-function ERK mutations can suppress proliferation.^28,29 ERK is required for Ras-induced senescence in fibroblasts, and this senescence has been attributed to ERK-dependent degradation of target proteins in a process known as senescence-activated protein degradation (SAPD).^30,31 Hong, et al. (2018) found that apoptotic signaling in response to high ERK activity could be converted to growth arrest signaling by titrating a MEK inhibitor, suggesting a threshold of ERK activity for induction of apoptosis.¹⁰ When comparing certain cell types, strong ERK signaling can produce opposite outcomes. For example, strong ERK activation in colon carcinoma cells that are detached from their extracellular matrix can promote survival signaling rather than apoptosis or growth arrest.^32,33 It is possible that the ERK1 and ERK2 isoforms may play different roles in directing these activity-induced fates, which could explain why different levels of active ERK have different effects, depending on cell type and signaling context.³⁴

Figure 3. ERK signaling amplitude helps determine cell fate.

When ERK signaling is below a certain threshold, estimated to be ~10% activity within a cell,²⁴ melanoma cell proliferation can be prevented (figure not drawn to scale). This can be achieved through inhibitors of the ERK pathway and of ERK itself. Conversely, hyperactivation of ERK can lead to cell senescence³¹ or death.²⁶ This can be achieved by events such as (1) suppression of pathway feedback such as the application of a ‘drug holiday’, or the removal of ERK pathway inhibitors,¹⁴⁵ (2) gain-of-function ERK mutations,^28,29 or (3) ERK overexpression.

ERK signaling duration

The dynamics of ERK responses to input stimuli can also affect signaling outcomes. For example, in rat PC12 cells, sustained epidermal growth factor (EGF) stimulation of ERK signaling was found to induce proliferation, while sustained nerve growth factor (NGF) stimulation resulted in differentiation.³⁵ In contrast, when pulses of low EGF concentrations were applied to the cells, they underwent differentiation instead, while pulses of higher EGF concentrations resulted in proliferation.^36,37

In MCF-10A cells, sustained EGF stimulation was observed to cause oscillations of ERK activity in single cells.³⁸ Thus, signaling results depend on cell type and context as well as the duration and magnitude of pathway input. Positive feedback and feedforward signaling dominate during sustained ERK activation, while transient ERK responses are typically characterized by negative feedback. Furthermore, positive feedback signaling can induce the switch-like behavior described above, while negative feedback, which is intended to resist or counteract signal perturbations, can result in oscillations of ERK activity. Liu, et al. (2011) have modeled oscillatory behavior of ERK activity that is independent of pathway stimuli but rather governed by ERK substrates.³⁹ In their model, ERK can accelerate degradation of a substrate that competes with phosphatases for the same binding site, and thus ERK becomes more accessible to phosphatases, resulting in the decline of its activity. This allows the accumulation of the substrate, which can then outcompete phosphatases for ERK binding, causing a rebound in active ERK. Repetition of this cycle is one explanation for oscillating levels of ERK activity when no variations in pathway stimuli occur. There is evidence that oscillation of ERK activity governs induction of a subset of target genes that differ from the gene expression profiles of cells with transient or sustained non-oscillating ERK activity.^40,41 Thus, the temporal modulation of ERK signaling is important for producing particular downstream responses.

Subcellular localization and cell fate

The spatial distribution of ERK in a cell can localize signaling outcomes and influence the duration and amplitude of ERK activity. As mentioned earlier, ERK must be present in the nucleus in order to phosphorylate specific substrates and transcription factors to potentiate cell cycle progression. ERK can localize to numerous organelles, but here we focus on translocation between the cytosol and nucleus as it pertains to modulation of transcription and cell fate.

ERK localization in a cell is controlled by the assembly and disassembly of protein complexes. ERK can be sequestered in the cytoplasm by numerous proteins, including the scaffolding protein PEA-15, phosphatases, the microtubule network, and MEK1/2.^42–45 Some of these proteins, like MEK and PEA-15, contain a nuclear export sequence (NES) that localizes them to the cytosol. However, MEK1/2 are also known to shuttle back and forth between the nucleus and cytosol.^44,46,47 The scaffolding protein KSR1 (kinase suppressor of Ras 1) interacts with Raf, MEK, and ERK, colocalizing them at the plasma membrane upon pathway stimulation. This facilitates not only the phosphorylation cascade but also feedback regulation, by bringing the components of the pathway into close proximity.¹¹ Arrestins are another example of proteins that directly facilitate the activation of ERK by MEK and cytoplasmic retention of ERK by colocalizing to microtubules or G-protein-coupled receptors.⁴⁸

Once ERK in the cytoplasm is activated through pathway stimulation, it redistributes between the cytosol and the nucleus to engage various downstream substrates. The release of ERK from cytoplasmic containment and its translocation to the nucleus is believed to involve several mechanisms. ERK2 has been reported to directly interact with nucleoporins, which may enable its transport into the nucleus.⁴⁹ There is also evidence that monomeric ERK can passively diffuse into the nucleus, despite an apparent ~40 kDa exclusion limit for passive transport through nuclear pores.^50,51 Diffusion of the monomer is supported by live-cell FRET imaging and fluorescence recovery after photobleaching (FRAP) experiments tracking the nuclear translocation of ERK.⁵² In the equilibrium model developed from this study, monomeric ERK shuttles freely between the nucleus and cytoplasm, while its overall distribution is governed by the formation of nuclear- and cytoplasmic- anchoring complexes.⁵³ Currently, little is known about potential nuclear-anchoring mechanisms once ERK has entered the nucleus. Dimerization of active ERK2 has been proposed to be necessary for extranuclear signaling, although the existence of endogenous ERK dimers remains a matter of some debate.^52,54–56 Molecules that target the putative dimerization interface have been shown to inhibit oncogenic ERK pathway signaling,⁵⁷ however, such molecules could potentially function by disrupting other protein-protein interactions or allosterically effecting ERK2. We have previously shown that recombinant ERK2 expressed in E. coli is a monomer in vitro and that partial dimerization may be observed due to His₆-tag association of the protein.^55,56

In summary, the localization of ERK can drive the signaling behaviors described above, such as switch-like or oscillatory ERK activity, depending on import and retention in the nucleus. The localization of ERK at different organelles suggests the possibility of subpopulations of ERK in a cell that governs particular signaling behaviors and select transcriptional profiles.

The ERK pathway in melanoma

As we have discussed above, ERK signaling dynamics are complex and highly regulated, as they play critical roles in determining cell fate. Therefore, it follows that aberrant ERK signaling is prominent in proliferative diseases such as cancer. In fact, the ERK pathway is hyperactivated or dysregulated in over 85% of cancers.⁵⁸ Somatic BRAF and NRAS mutations are the most common in cutaneous melanoma and are present in up to 52% and 28% of cases, respectively (The Cancer Genome Atlas Network).⁵⁹ Of the different reported mutations, the constitutively active BRAF-V600E is the most prevalent.⁶⁰ When BRAF-V600E is present, the ERK pathway downstream of Raf is no longer dependent on upstream signaling events, such as feedback regulation of RTKs and Ras or pathway stimulation by ligand-receptor binding. In such cells, Ras activity is low due to ERK-dependent negative feedback resulting from robust ERK activation.⁶¹ ERK also exerts feedback regulation of its own phosphorylation by driving DUSP transcription. This feedback prevents hyperactivation of ERK and thus avoids senescence or cell death, while net flux through the pathway maintains ERK activation at a level that promotes tumorigenesis (see Figure 3).¹¹

When mutant NRAS drives ERK pathway signaling and BRAF is wildtype, N-Ras is constitutively activated in its GTP-bound form. In normal melanocytes and BRAF-V600E melanoma, B-Raf is the primary isoform responsible for activating MEK. However, when oncogenic N-Ras is present, the pathway switches to utilize C-Raf to activate MEK instead of B-Raf.⁶² This could occur because active ERK can phosphorylate and inhibit B-Raf by preventing it from binding to N-Ras-GTP, so the pathway uses C-Raf to escape negative feedback.^63,64 It could also be due to the fact that N-Ras activates B-Raf more strongly than C-Raf, so C-Raf is preferred in order to avoid hyperactivation of ERK and subsequent Ras-induced senescence.⁶² Furthermore, the transcriptional profile of NRAS-mutant tumors is driven in part by other signaling pathways like PI3K/Akt.^11,65

Together, these examples illustrate that different cell states exist in melanoma, with different transcriptional profiles, as well as different steady-state signaling at each node of the ERK pathway. These different states of feedback regulation are likely related to why hot-spot BRAF and NRAS mutations are inversely correlated in endogenous melanomas and mutually exclusive on a single-cell level.^59,66 The expression and activity levels of ERK pathway members in melanoma influence responses to different treatments, as well as resistance mechanisms to treatments, as discussed in the sections below.

Current melanoma treatments

There are two standard types of treatment for melanoma: immune checkpoint inhibitors, and targeted ERK pathway inhibitors (of B-Raf and MEK). The immunotherapy combination of ipilimumab and nivolumab has shown an objective response rate of 52% in BRAF-mutant melanomas, which is superior to either monotherapy, but associated with greater toxicity.^67,68 Similarly, combinations of MEK and B-Raf inhibitors have yielded objective response rates of 67% (dabrafenib and trametinib), which improves upon the results of either agent alone.^67,69 However, these current therapies are not always successful. Immune checkpoint inhibitors can be met with toxicity issues when given concomitantly or post-treatment with ERK pathway inhibitors, and rationales for optimal sequencing and choice of agents are still in development.^70,71 Acquired resistance is also known to limit the efficacy of immunotherapies for patients with metastatic melanoma.⁷² In addition, 10–20% of BRAF V600-mutant melanoma patients fail to respond to any current MAPK pathway inhibitors, and resistance almost inevitably results in patients that do initially respond.⁷³ Single-agent B-Raf inhibitors result in acquired resistance in a median time of 6–8 months.^74,75 MEK inhibitors used alone or in combination with B-Raf inhibitors can improve responses, but efficacy is still limited by eventual resistance.^12,76,77 Thus, there is still a critical need for improvement of targeted ERK pathway therapies, and for anticipating and overcoming resistance mechanisms to these therapies. Some of the known resistance mechanisms to ERK pathway inhibitors are discussed in the following section.

Resistance mechanisms to ERK pathway inhibitors

There are three observed stages of resistance in response to ERK pathway-targeted therapies in melanoma: innate or intrinsic resistance, adaptive resistance that is hallmarked by dynamic cellular reprogramming, and acquired resistance.^61,78 Intrinsic resistance is characterized by no initial response to ERK pathway inhibitors, while adaptive resistance occurs early (in as little as 24 hours of B-Raf inhibitor exposure)⁷⁹, and acquired resistance is preceded by initial tumor regression. There is no clear distinction between mechanisms underlying resistance for these three cases; the same cellular alterations can occur at different times.

Targeted kinase inhibitor resistance can arise by a variety of mechanisms, such as mutations of the target kinase itself or mutations within the kinase pathway,^12,80,81 bypass of pathway or signaling elements via alternative signaling routes,^12,82,83 epigenetic changes,^84–87 adaptation of cellular processes such as drug transport,⁸⁴ metabolic reprogramming,^88–90 and changes in the tumor niche.^61,91 Some examples of these acquired resistance mechanisms to ERK pathway inhibitors in BRAF-V600E melanoma are illustrated in Figure 4. ERK pathway reactivation is the most common resistance mechanism by far and is observed in 50–70% of melanoma tumors that have been treated with a single B-Raf inhibitor.^92–94 ERK pathway reactivation can result from numerous sources in the canonical pathway itself: RTK upregulation; mutation or upregulation of MEK, Ras, and B-Raf; alternate B-Raf splice variants; and, heterodimerization of Raf isoforms. Reactivation of the ERK pathway can also be attained by suppression of negative feedback regulation mechanisms. For example, negative feedback relief can arise from loss or inactivation of NF1, a tumor suppressor gene that inhibits Ras activity.⁶¹ Other kinases can bypass pathway elements to culminate in activation of ERK. For instance, expression of MLKs (mixed lineage kinases) can be upregulated to activate MEK when Raf is inhibited.⁹⁵ Tumor microenvironment changes can also enhance tumorigenic phenotypes in response to ERK pathway inhibitor treatments. Increased levels of soluble autocrine and paracrine factors, like HGF, EGF, NRG1, TGFβ, and VEGF, can drive RTK signaling down the ERK pathway or other alternative pathways to achieve the same signaling outcomes. Upregulation of receptor tyrosine kinases at the cell surface can also drive other pathways such as PI3K/Akt signaling to overcome ERK pathway inhibition, and this is likewise a critical result of the entire kinome remodeling in adaptive resistance responses to inhibitor treatment.^96,97 Because most of these resistance mechanisms culminate in the reactivation of ERK, ERK-targeted therapies have emerged as a viable strategy for overcoming resistance in BRAF-V600E melanoma.

Figure 4. Resistance mechanisms to BRAF and MEK inhibitors in BRAF-V600E melanoma.

Examples of mechanisms of resistance to BRAF and MEK inhibitors in BRAF-V600E melanoma are depicted.^{61,76,78,88,93,97} Inhibitors shown as blue circles. Yellow stars denote mutants present prior to drug exposure; red stars denote secondary mutations that can happen as a result of inhibitor treatment. Red arrows indicate upregulation of protein expression and/or other cellular pathways. Resistance can occur as a result of changes within the ERK pathway itself, including BRAF upregulation; upregulation and heterodimerization of other Raf isoforms or BRAF splice forms; activation of MEK by MLK; MEK mutations; upregulation or mutation of Ras isoforms; and upregulation of receptor tyrosine kinases. Suppression of negative feedback in the ERK pathway, for example, through mutation/loss of NF1, can result in resistance. Resistance can also arise by kinome remodeling at the RTK level and upregulation of alternate signaling pathways like PI3K/Akt/mTOR. Changes in the tumor microenvironment, as well as metabolic reprogramming to promote glycolysis or MITF overexpression, can also overcome ERK pathway inhibition

Targeted ERK inhibitors in melanoma

There is evidence that ERK inhibition can help avoid the resistance mechanisms that arise from MEK and B-Raf inhibition. Notably, ERK mutations have not been clinically implicated as resistance mechanisms to B-Raf and MEK inhibitors. Indeed, BRAF-V600E melanoma cell lines that become resistant to Raf inhibitors and exhibit pathway reactivation maintain sensitivity to ERK inhibitors.^14,28,98 When resistance to ERK inhibitor treatment emerges in BRAF-V600-mutant melanoma, these cells can regain sensitivity to B-Raf and MEK inhibitors, suggesting that alternating treatment regimens may be a method of overcoming resistance mechanisms.^28,99,100

When considering ERK inhibitor development, it is necessary to distinguish the different classes of kinase inhibitors (Figure 5).^101–103 Type I inhibitors reversibly bind to the ATP-binding pocket of the kinase in the active DFG-in conformation, where the conserved aspartate residue (Asp184/167 in human ERK1/2) on the kinase activation loop faces into the pocket. Type II inhibitors bind to the same ATP-binding pocket, while the kinase is in the inactive DFG-out conformation. In this conformation, the aspartate residue of the activation loop is flipped outward from the binding site. Type III-IV inhibitors are allosteric; they inhibit the kinase without blocking the ATP-site, instead binding adjacent and distal to the ATP-site, respectively.^104,105 Additional types of kinase inhibitors include bivalent inhibitors that span two sites (Type V), and covalent inhibitors (Type VI). Most biochemical and computational methods for kinase inhibitor discovery utilize ATP competition as an inhibitor detection method.¹⁰¹ Therefore, the majority of identified kinase inhibitors are ATP-competitive (Type I and II). The other types of inhibitors are often serendipitously identified during mechanistic evaluations of hit compounds from ATP-site –directed biochemical screenings.

Figure 5. Types of kinase inhibitors.

A kinase, depicted with an ATP-binding site and distal protein-binding site, can be inhibited in a variety of ways (inhibitor is shown in red).^101,103,156 Type I inhibitors target the ATP site when the kinase is active and in the DFG-in conformation, where the aspartate residue of the DFG-motif (orange) of the activation loop is facing into the pocket. Type II inhibitors target the ATP site of the inactive kinase conformation when the DFG-motif aspartate residue is facing out from the pocket. Type III and IV inhibitors are allosteric, targeting sites adjacent and distant from the ATP site, respectively. These inhibitors do not block the binding of ATP. Type V inhibitors are bivalent, targeting more than one site on a kinase usually by connecting two site-targeted inhibitors via a flexible linker. Type VI inhibitors are covalent.

Numerous ATP-competitive ERK inhibitors, including ulixertinib (BVD-523), ravoxertinib (GDC-0994), CC-90003, LY3214996, KO947, LTT462, and MK-8353, are being evaluated in clinical trials for various forms of cancer (www.clinicaltrials.gov) (Table 1). CC-90003 is a covalent ATP-competitive inhibitor that shows promise in KRAS mutant xenograft models¹⁰⁶ but has been terminated in a Phase I clinical trial for neoplasm metastasis due to limited response at the maximum tolerated dose. MK-8353 is a dual-mechanism ATP-competitive ERK inhibitor that functions by inhibiting both the activation of ERK by MEK1/2 as well as ATP binding. It was tested in patients with BRAF-V600E mutant melanoma after yielding encouraging pre-clinical data.¹⁰⁷ Partial response was observed in 3 out of 15 patients, which did not correlate well with pharmacodynamic parameters.^107,108 However, ulixertinib and ravoxertinib have shown promising results in clinical trials for various cancer types, and several of the ERK inhibitors listed in Table 1 are also being investigated in combination with other agents, such as chemotherapeutics and immunotherapies.

Table 1. ERK inhibitors in clinical trials for the treatment of cancer.

Table of ERK inhibitors currently undergoing clinical trials for different cancer types, either as single agents or in combination with other agents (www.clinicaltrials.gov). Not included: ERK inhibitors that target other kinases; absorption, metabolism, and excretion (AME) studies.

Inhibitor	Trial type	Cancer type	Trial number	ERK inhibitor structure (or description of other drugs)
BVD-523 (Ulixertinib)	Phase II (recruiting)	Metastatic uveal melanoma	NCT03417739
	Phase I/II (completed)	AML*, myelo-dysplastic syndrome	NCT02296242
	Phase I (completed)	Advanced solid tumors	NCT01781429
	Phase I (recruiting)	Advanced solid tumors, NHL†, hystiocystic disorders with MAPK pathway mutations	NCT03698994
(as one of many options)	Phase II (recruiting)		NCT03155620	Various agents
with Palbociclib	Phase I (recruiting)	Solid tumors, pancreatic cancer	NCT03454035	Selective inhibitor of CDK4 and CDK6
with Nab-Paclitaxel, Gemcitabine	Phase I (recruiting)	Metastatic pancreatic cancer	NCT02608229	Chemotherapies
MK-8353 (SCH 900353)	Phase I (terminated)	Solid tumors	NCT01358331
with Pembrolizumab	Phase I (recruiting)	Neoplasms, colorectal cancer	NCT02972034	Monoclonal antibody inhibitor of PD-1 on lymphocytes
with Selumetinib	Phase I (not yet recruiting)	Advanced/Metastatic solid tumors	NCT03745989	MEK1/2 inhibitor
CC-90003	Phase I (terminated)	Neoplasm metastasis	NCT02313012	N/A‡
KO-947	Phase I (recruiting)	Advanced malignant neoplasm	NCT03051035
GDC-0994	Phase I (completed)	Advanced or metastatic solid tumors	NCT01875705
with Cobimetinib	Phase I (completed)	Advanced or metastatic solid tumors	NCT02457793	MEK inhibitor
LTT462	Phase I (recruiting)	Advanced solid tumors	NCT02711345	N/A‡
with LXH254	Phase I (recruiting)	Non-small cell lung cancer	NCT02974725	pan-Raf inhibitor
LY3214996	Phase I (recruiting)	Advanced or metastatic solid tumors	NCT02857270
alone or with Midazolam, Abemaciclib, Nab-Paclitaxel and Gemcitabine				Various agents

^*AML = acute myeloid leukemia;

^†NHL = non-Hodgkin lymphoma;

^‡N/A = structures of ERK inhibitors are not available.

Limitations of ATP-competitive ERK inhibitors

Current preclinical ERK inhibitors are limited by toxicity and adverse effects, as ERK function is important in healthy cells, and ERK inhibition must reach >90% to achieve tumor regression.^24,109 The development of new ATP-competitive inhibitors is associated with numerous additional shortcomings that limit their potency and selectivity. Kinases have a high affinity for ATP, which is often present at millimolar-level concentrations in cells, thus ATP-competitive inhibitors must be extremely potent.^110–112 Additionally, ATP-binding sites have high structure- and sequence- homology among the various families of kinases, so ATP-competitive inhibitors must be highly selective to avoid off-target effects and promiscuity.¹¹² One alternative way to target ERK in order to complement the inhibition provided by ATP-competitive inhibitors is by targeting other locations on the surface of ERK, like protein-protein interaction (PPI) sites.

The D-recruitment site

PPI sites, or docking sites, found on cellular proteins, serve as interfaces of recognition that underlie signaling specificity within MAPK pathways. Docking motifs that bind to PPI sites on MAPKs are typically less than 20 amino acids in length and are linear, disordered regions.¹¹³ More than one motif can be present in a protein and potentially they can target more than one MAPK.¹¹⁴ MAPKs possess two main docking sites that interact with docking motifs: the D-recruitment site (DRS) and the F-recruitment site (FRS). MAPK docking sites interact with upstream and downstream effectors, positive and negative regulatory elements, as well as structural and scaffolding proteins. A single substrate of ERK can potentially access different combinations of docking sites, and thereby engage the DRS, FRS, or both, depending on different signaling directives.¹¹⁵ Here, we will specifically focus on the D-recruitment site of ERK, the significance of targeting this site, and the development of DRS inhibitors for use in melanoma.

Structure and domains of the DRS

The D-recruitment site (DRS) can be found on all MAPKs (p38, JNK, ERK1/2, ERK5) (Figure 6A) in a location distal to the ATP-site on the protein surface.^116,117 The DRS binds to proteins with complementary D-site, or D-motif, sequences that are linear regions of varying lengths. The general consensus sequence of these D-sites is ψ_1–3Χ_3–7ΦΧΦ, where ψ specifies positively charged residues, Φ specifies hydrophobic residues, and Χ can be any residue.¹¹³ The DRS on all MAPKs includes a charged region and a hydrophobic groove (Figure 6B). The charged region is a conserved acidic patch consisting of Asp or Glu residues, referred to as the common docking (CD) domain. The hydrophobic groove region of the DRS is flanked by the CD domain and another motif, the ED domain, which consists of two variable residues (ED in p38, SD in JNK, TT in ERK1/2, and EN in ERK5). Variability within the ED domain facilitates interaction specificity among the different MAPK D-recruitment sites.^118,119 The hydrophobic docking grooves are different for p38, ERK1/2 and JNK, making this region an ideal target for the design of selective inhibitors.¹¹³ The hydrophobic groove also contains a solvent-exposed cysteine residue, adjacent to the ED domain, that is conserved among all MAPKs.

Figure 6. The D-recruitment site of ERK2.

The ERK2 D-recruitment site. (A) The sequence alignment of the major domains of the DRS for all major MAPKs is shown. The DRS, as defined by ERK2, is composed of three major elements: (1) a groove formed by the hydrophobic residues of the αD and αE helices that is bordered by (2) the β7- β8 reverse turn containing the ED domain, and (3) the L16 loop containing the CD domain.¹ Key hydrophobic residues identified via ERK2 are shown in orange, while other conserved hydrophobic residues are highlighted in gold. The ED domain is shown in pink, adjacent to the conserved DRS cysteine residue (red). The DFG motif of the activation segment is shown in blue, and residues of the CD domain are shown in green. All MAPK sequences are human, obtained from the UniProt database and aligned using Clustal Omega. (B) Yellow, Red, Blue (YRB)¹⁵⁷ highlighting of ERK2 (PDB 2ERK)¹³⁶ DRS residues performed in PyMol (left panel). In the YRB color code, yellow shows carbon atoms with potential hydrophobic contacts, blue denotes nitrogen atoms of arginine and lysine side chains, and red shows oxygen atoms of glutamate and aspartate side chains. This YRB surface map indicates the major regions of the DRS (right panel): a hydrophobic groove (yellow) separating two polar regions (cyan) that contain the CD and ED domains. (C) Structural overlay of peptides bound at the DRS showing different binding modes: pepHePTP (PDB 2GPH – green),¹²⁰ pepMKK2 (PDB 4H3Q – pink),¹²¹ pepRSK1 (PDB 4H3P – yellow).¹²¹ Cys159 on ERK2 (rat numbering; PDB 2ERK)¹³⁶ is labeled in red. (D) Detail of the peptides shown in (C). The peptides interact with one or more of the charged acidic residues of the CD domain via arginine residues that are within hydrogen-bonding distance (red-dashed lines). Residues are labeled according to PDB entry 2ERK.¹³⁶

The different domains of the ERK2 DRS and their engagement with examples of different D-site peptides are illustrated in Figure 6C. Substrates can occupy the DRS of MAPKs in different binding modes,^113,114 as illustrated in Figure 6 C and D for peptides derived from the phosphatase HePTP, RSK1, and MEK2.^120,121 ERK binding partners not only occupy different regions of the DRS but can also bind in both N- to C-terminal orientations. For example, PEA-15, RSK1, and MNK1 have all been found to bind to the ERK2 DRS in a reverse orientation to canonical D-sites.^{53,113,122,123} These different binding modes and D-site motifs highlight the range of potential interactions accommodated by MAPK D-recruitment sites. Furthermore, binding at the DRS has the potential to induce conformational changes within the MAPK that provides an additional layer of complexity.¹¹⁶ For example, it has been shown that docking at the DRS can cause the activation loop of ERK2 to adopt a conformation where Thr183/Tyr185 (rat numbering; corresponds to Thr185/Tyr187 in human ERK2) become solvent exposed and primed for phosphorylation.^120,124,125 This conformation matches neither the fully active nor inactive conformations of ERK2, but likely represents an intermediate conformation that prepares ERK for phosphorylation by MEK or dephosphorylation by phosphatases. In summary, the DRS of ERK encodes numerous unique structural features that can be selectively targeted by inhibitors.

Functions of the DRS

The DRS of ERK engages in direct interactions with substrates (RSK1,^121,126,127 MAPKAPKs like MNK1 and MSK1,¹¹⁸ and caspase-9¹²⁸), phosphatases (MKP-3 and HePTP),^120,129,130 activators (MEK1/2),^121,131 and scaffolding proteins such as PEA-15.¹³² Many of these interactions influence subcellular localization of ERK, such as binding to PEA-15 and MEK1/2 which both sequester ERK in the cytoplasm. Some DRS-binding interactions are observed to take place regardless of ERK activation, as in the case of PEA-15 and Ets-1, further suggesting a noncatalytic regulatory role of the docking site.^53,55 The DRS and the FRS also play a critical role in catalysis though they are located outside of the active site. Substrates utilize these binding sites to increase the effective local concentration of their phosphorylation sequences (Ser/Thr-Pro motif) in the vicinity of the active site in order to facilitate catalysis. Thus, MAPK docking sites are regulators of what is termed ‘proximity-induced catalysis.’¹³³ They provide the basis for substrate recognition, as the very short phosphorylation consensus sequence itself does not elicit strong selectivity of substrate interactions for MAPKs.

Current DRS inhibitors

Numerous small molecule inhibitors of the ERK DRS have been identified, mainly through computer-aided drug design (CADD) and virtual screening. The reported DRS inhibitors interact with ERK by using different binding modes, which could potentially be utilized to target the different types of DRS-mediated interactions described above.

Binding near the CD domain

In 2005, Hancock, et al. used CADD to detect small molecules that bind to the cleft between the CD and ED domains of the DRS of unphosphorylated ERK2.¹³⁴ From molecular docking models, these compounds were predicted to interact with the acidic residues of the CD domain. Additionally, the hit compounds were found to inhibit proliferation and colony formation of multiple cancer cell lines. In 2006, the same group published findings for a CADD screening for inhibitors that target active, phosphorylated ERK2.¹³⁵ The activation of the enzyme induces conformational changes in various regions of ERK2, though many of the same hit compounds were identified for active and inactive ERK2, suggesting that changes in the DRS are minimal. Comparison of crystal structures for inactive (PDB 1ERK)¹³⁶ and active (PDB 2ERK)¹³⁷ ERK2 also confirm that the DRS does not significantly change conformation upon activation.¹³⁵ These DRS inhibitors were further investigated by virtual screening of analogs (Boston, et al.)¹³⁸ and use of synthetic chemistry to evaluate structure-activity relationships (Li, et al.).¹³⁹ The resulting inhibitors identified by Boston, et al. consisted of a thiazolidinedione core with an aminoethyl side-group. They were found to activate the intrinsic apoptosis pathway in cells, inhibiting proliferation and RSK1 and Bad phosphorylation.¹³⁸ The compounds also showed the potential to preferentially target transformed cells over non-transformed cells.

Another research group, Kinoshita, et al., identified small-molecule DRS inhibitors by in silico methods, which they validated by displacement of a peptide (PEP) that was designed based on the STAT3-ERK2 interaction.¹⁴⁰ Through computational docking studies, these compounds were predicted to not interact with the DRS hydrophobic pocket, but rather the acidic CD domain.

Our group has developed a high-throughput biochemical assay to identify small molecules that reversibly target the DRS (Sammons, et al., under revision, ACS Chemical Biology). By measuring fluorescence anisotropy, we are able to detect the displacement of a fluorescently labeled peptide derived from yeast Ste7 from ERK2 by competing molecules. Like Kinoshita, et al., we identified a structural class of compounds that interact mainly at the CD domain of the DRS. These compounds contain one or more arginine-like guanidinopropyl functional groups that form ionic interactions with the acidic CD residues, as evidenced by the X-ray crystal structure of the compound 2507–8 (Figure 7A). The crystal structure suggests that 2507–8 also interacts with key residues of the hydrophobic groove between the CD and ED domains. A hit compound from this study, 2507–1, was also shown to prevent ERK2 activation by MEK1 and to partially inhibit Ets-1 phosphorylation by ERK2 in vitro with IC₅₀ values of approximately 9.9 and 5.6 µM, respectively.

Figure 7. Inhibitor binding modes at the DRS.

Inhibitors occupying different binding domains within the ERK2 DRS. Key residues of the ED domain (purple), CD domain (blue) and the solvent-exposed Cys159 (red) are highlighted. (A) Crystal structure of the cyclic guanidino -based compound 2507–8 in complex with ERK2 (PDB 6NBS, Sammons, et al., under revision, ACS Chemical Biology). Portions of 2507–8 engage the key CD domain residues as well as hydrophobic residues that are adjacent to the ED domain and Cys159. (B) Docking pose from molecular dynamics simulation of BI-78D3 bound at the DRS predicts engagement at the ED domain and Cys159 (PDB 4ERK). Unlike 2507–8, BI-78D3 is not predicted to interact with key residues of the CD domain.

Binding the ED domain

We have recently identified the compound BI-78D3 as a selective, covalent inhibitor of the ERK DRS (Kaoud, et al., under revision, Nature Communications). This compound was originally identified as a weak reversible inhibitor of DRS interactions between JNK and JNK-interacting protein-1 (JIP-1).¹⁴¹ In contrast, this compound engages the ED domain of ERK and forms an irreversible tetrahedral adduct at the DRS cysteine, Cys159 (rat numbering; corresponds to Cys161 of human ERK2). Though this cysteine is conserved among the different MAPKs, the selectivity of BI-78D3 for ERK likely arises from its interactions at the non-conserved ED domain residues (Figure 7B). We found that this compound potently inhibits ERK activation and downstream phosphorylation of RSK1 at concentrations as low as 1 µM in A375 melanoma cells, which harbor the BRAF-V600E mutation. BI-78D3 also inhibited A375 cell proliferation and anchorage-independent and -dependent colony growth, and induced apoptosis. BI-78D3 additionally inhibited tumor growth and ERK signaling in a mouse xenograft model derived from the A375 cell line. Importantly, BI-78D3 also showed growth inhibition in melanoma cell lines that are resistant to B-Raf inhibitors (451 Lu-R and MEL 1617-R).

Possible impacts of DRS inhibitors

The reported reversible small-molecule inhibitors of the CD and hydrophobic groove discussed above have shown potential for ERK pathway inhibition in cells. There is evidence that they inhibit proliferation and activate the intrinsic apoptotic pathway, and preferentially target oncogenically transformed cells. However, there is still further room for the development of improved inhibitors with drug-like biological activities. The current reversible inhibitors show relatively high dissociation constants and IC₅₀ values in vitro, with the most tightly binding inhibitors at 5–15 µM,^134,135,140 and show significant activity against downstream ERK targets in cells at high concentrations of 50–100 µM.¹³⁸

However, the covalent inhibitor of the ED domain, BI-78D3, illustrates that potent and selective inhibition of ERK signaling via the DRS can be achieved. The different effects on ERK signaling that DRS inhibitors can cause, such as those potentially underlying the anti-tumorigenic properties of BI-78D3, are discussed below. Considering that substrates utilize the DRS and FRS to increase productive encounters of their phosphorylation motifs with the ERK active site, a DRS inhibitor can have several effects on downstream effectors of ERK (Figure 8A). If a substrate docks solely at the DRS, a DRS inhibitor can fully block phosphorylation of that substrate by ERK. On the other hand, phosphorylation of substrates like Elk-1 and Ets-1 that engage both the DRS and FRS may only be partially inhibited, while substrates that interact at the FRS alone can potentially remain unaffected by a DRS inhibitor. It is also likely that inhibitor binding at the DRS can allosterically affect the activation segment as well as the FRS. As mentioned earlier, binding of peptides at the DRS has been shown to induce a conformational change in the activation segment, such that it adopts an intermediate structure between the fully active and inactive conformations.^120,124 The FRS is exposed upon activation of ERK2 when the activation segment is in the DFG-in conformation. Thus, an inhibitor binding to the DRS may result in allosteric obstruction of the FRS pocket and changes to the active site of ERK that may prevent interactions that occur at these other sites. Certain outcomes of ERK signaling will likely be affected by each particular type of DRS inhibitor (CD, hydrophobic groove, or ED) more than others.

Figure 8. The effects of DRS inhibitors on ERK signaling.

Impacts of DRS inhibitors (purple) on ERK signaling. D-sites are shown in red and F-sites in yellow, while phosphorylation consensus sequences are shown as blue squares. (A) DRS inhibitors may fully block phosphorylation of substrates with D-sites that only dock at the DRS (1). DRS inhibitors may also partially block binding of substrates that contain D- and F-sites or they can allosterically alter the activation segment and partially obscure the FRS, preventing F-site binding (2). It is also possible that in some cases, DRS inhibitors may not affect F-site binding. (B) DRS inhibitors can block association of ERK with D-site-containing scaffolding proteins and proteins that anchor ERK at specific subcellular locations. This can result in a redistribution of ERK within a cell. (C) The DRS is utilized by MEK to dually phosphorylate (labeled in orange) and activate ERK and is also used by certain phosphatases (labeled as P-ase) to dephosphorylate ERK. These reactions are shown with black arrows. ERK DRS inhibitors can disrupt MEK binding to prevent ERK activation (blue arrows), and can also disrupt phosphatase binding (red arrows) to alter ERK activity levels within a cell.

DRS inhibitors can also affect interactions between ERK and scaffolding proteins, as well as proteins that sequester ERK at different subcellular locations (Figure 8B). For example, a DRS inhibitor could prevent cytoplasmic sequestration of ERK by PEA-15, thus promoting the nuclear translocation of ERK; although DRS inhibitors have the potential to affect nuclear translocation and nuclear-anchoring interactions as well. This illustrates that some of the inhibitory effects on ERK signaling by DRS-targeting molecules are not necessarily dependent on the catalytic activity of ERK.

Similarly, phosphatases and MEK1/2 that utilize the DRS to control ERK phosphorylation and activation can be blocked by DRS inhibitors (Figure 8C). This not only affects the amplitude of ERK activity in a cell but also disrupts pathway feedback signaling. For the case of BI-78D3 in BRAF-V600E-mutant A375 cells, ERK activation and downstream signaling are irreversibly suppressed. This loss of ERK activation could be attributed to the inability of MEK to phosphorylate ERK, relative to phosphatase activity against ERK. Though inhibitors at the DRS can competitively inhibit phosphatase and MEK1/2 binding, they are likely to have allosteric effects on both the activation segment and the adjacent FRS that could further prevent binding events and ERK phosphorylation or dephosphorylation.

A benefit of irreversible inhibitors like BI-78D3 is that they fully inhibit interactions that require their target binding sites, subject only to the turnover of protein synthesis and degradation in a cell. ERK has a long cellular half-life of over 50 hours,¹⁴² so irreversible inhibitors of ERK have potential to remain effective for long durations. Irreversible inhibitors are also unaffected by shifts in equilibria of competitive-binding proteins. This renders irreversible DRS inhibition as a potential mechanism of overcoming acquired resistance to targeted MAPK pathway inhibitors in melanoma.

Potential resistance mechanisms to DRS inhibitors

Though targeting ERK can overcome resistance to B-Raf and MEK inhibitors, resistance to ERK inhibitors can occur as well. Resistance to ATP-competitive ERK inhibitors has been shown to occur through multiple mechanisms. Resistance to SCH772984 (a precursor analog of MK-8353) treatment was found to arise by a single mutation in the DFG motif of ERK1 (G186D) in KRAS-mutant colorectal cancer cells (HCT-116)⁹⁹ A similar mutation in ERK2 (G169D) was observed in cells that were resistant to MK-8353, and it is suspected this mutation functions in the same way as ERK1 (G186D).^29,99 The ERK1 (G186D) DFG mutation and several other ERK1/2 mutations were observed in melanoma cells exhibiting ERK inhibitor (VX-11e) resistance.²⁸ ERK1/2 mutations, amplification, and overexpression of ERK2, and RTK overexpression were observed as resistance mechanisms to five tested ATP-competitive ERK inhibitors in several BRAF and RAS-mutant cancer cell lines.¹⁰⁰ Accordingly, potential resistance mechanisms to ERK docking site inhibitors must be anticipated. One possible avenue for DRS inhibitor resistance is mutations in the CD domain of ERK, which are described below.

Several mutations in PPI sites of ERK have been studied for their transformative potential; for example, DRS mutations are observed in cervical, head, and neck cancers and can be found in the COSMIC database (Glu322 and Asp321).²⁸ The sevenmaker mutant (D321N) of ERK2 is a known gain-of-function mutation that reduces ERK recognition by phosphatases in cells.¹⁴³ Interestingly, though MEK binds at the DRS as well, ERK recognition by MEK in cells is largely unaffected by the sevenmaker mutation.¹⁴⁴ Other gain-of-function mutations in non-catalytic regions of ERK2 are also observed to cluster in the DRS, as detected by mutant library screens. These include mutations in the residues Glu81 and Glu322, which are grouped in the acidic common docking domain near Asp321 and are therefore termed ‘sevenmaker-like’ mutations.²⁹ The ‘sevenmaker-like’ mutations are similarly characterized by abrogation of DUSP6 binding and dephosphorylation of ERK2, leading to elevated activity in cells. In contrast, mutations in the FRS tend to result in loss-of-function, likely due to interference in ERK activation.²⁹

Goetz, et al. (2014) found that ERK1/2 overexpression reduced viability and induced growth suppression in some melanoma cell lines in a manner that depended on ERK kinase activity.²⁸ The most extensive growth suppression in A375 cells with resistance to B-Raf/MEK inhibitors was observed for ERK1/2 with activating mutations in the αC helix /CD domain and the activation segment. Brenan, et al. (2016) also observed the same growth suppression phenotype for gain-of-function ERK2 mutations in A375 cells.²⁹ Asp321 and Glu322 gain-of-function mutations in ERK2 were observed to cause acquired resistance to MEK and B-Raf inhibitors in A375 cells,^28,29 while the analogous ERK1 mutants did not. While this alludes to possible isoform-specific functions of the DRS, varied levels of active ERK could explain the phenotype differences. The results of this study also suggest that mutations in the ERK2 CD domain could contraindicate MEK or Raf inhibitor treatments. Because these mutations could potentially develop in response to CD domain-targeted inhibitors, it may be important to design and investigate inhibitors that form key interactions with other regions of the DRS, such as the ED domain.

Future applications of ERK DRS inhibitors in melanoma treatments

ERK inhibitors represent the next frontier in the MAPK pathway-targeted treatment of BRAF-V600E melanoma. Studies of several pre-clinical ERK inhibitors show that mutations arising from ATP-competitive ERK inhibition do not confer cross-resistance to MEK and B-Raf inhibitors, and vice versa.^14,28,98 This implies that alternating ERK pathway therapies may overcome the numerous resistance mechanisms that culminate in the ERK pathway reactivation. Similarly, ERK inhibitors can also be therapeutically useful in the context of a ‘drug holiday’, where once an ERK pathway inhibitor is removed from melanoma cells, flux through the signaling pathway increases to yield hyperactivation of ERK that can result in cell death.¹⁴⁵ This concept, where both ERK inhibition and ERK hyperactivation can result in tumor suppression, but intermediate levels of ERK activation yield tumor growth, is an example of the ‘Goldilocks principle.’²⁶ This illustrates the complexity of ERK signaling, yet introduces broader possibilities for manipulation of the ERK pathway to yield therapeutic benefit in melanoma.

There is growing evidence that targeting multiple MAPK signaling components in cancers like BRAF-V600E melanoma may be advantageous. This includes drug combinations that act within a single kinase pathway, multiple sites of a single kinase, parallel pathways, or upon kinases and alternative targets that converge on a common cancer phenotype.^146–149 Docking site inhibitors, like the ERK DRS inhibitors described here, allow for the possibility of targeting a single kinase in multiple ways, providing tunable effects on cell signaling outcomes as well as a countermeasure to resistance mechanisms.

It is also possible to build upon the functions of DRS inhibitors to increase their utility and therapeutic potential. For example, DRS inhibitors that reversibly bind near the ED domain and Cys159 of ERK2 (rat numbering; corresponds to Cys161 of human ERK2) can potentially be functionalized to covalently react with Cys159 in the same manner as BI-78D3. DRS inhibitors can also be modified with linkers connecting to ATP-competitive ERK inhibitors, creating bivalent inhibitors. For example, Lechtenberg, et al. (2017)¹⁵⁰ devised a bivalent inhibitor consisting of the ATP-competitive ERK inhibitor FR180204 connected to a peptide derived from RSK1, which showed higher potency against ERK than either individual fragment. DRS inhibitors and other types of ERK inhibitors can also be functionalized with linkers and appended molecules that recruit other cellular proteins to ERK for potential therapeutic benefit. For example, PROTACs (proteolysis-targeting chimeric molecules) consist of a target ligand, such as an ERK inhibitor, linked to a ligand for E3 ubiquitin ligase.¹⁵¹ This proximity facilitates the ubiquitination and degradation of drug-targets like ERK. PROTACs consisting of DRS inhibitors could be beneficial when the goal of targeting ERK is to fully inhibit its signaling versus to solely inhibit signaling events that require the DRS. Targeting E3 ubiquitin ligase to ERK could also amplify senescence-activated protein degradation (SAPD) of ERK targets when ERK is hyperactivated and further prevent proliferation.^31,152 Inhibitors of the ERK DRS, therefore, must be further developed and studied, as they may prove to be useful tools in the investigation of ERK signaling events as well as in the treatment of melanoma and other cancers.

Abbreviations:

MAP

kinase or MAPK mitogen-activated protein kinase

Shc

Sh2-containing collagen-related proteins

Grb2

growth factor receptor-bound protein 2

Raf

rapidly accelerated fibrosarcoma protein

MEK

mitogen-activated protein kinase/ ERK kinase

Elk-1

Ets-like transcription factor -1

RSK

ribosomal s6 kinase

DUSP

dual specificity phosphatase

SPRY

sprouty homolog

CDK

cyclin-dependent kinase

pRb

retinoblastoma protein

Myt1

myelin transcription factor 1

Fra-1

Fos-related antigen -1

p21CIP1

cyclin-dependent kinase inhibitor 1A

Bcl

B-cell lymphoma protein family

Mcl-1

myeloid cell leukemia -1 protein

Mdm2

murine double minute -2

PEA-15

phosphoprotein enriched in astrocytes -15

FRET

Förster (fluorescence) resonance energy transfer

NF1

neurofibromin 1

HGF

hepatocyte growth factor

NRG1

neuroregulin 1

TGFβ

transforming growth factor beta

VEGF

vascular endothelial growth factor

PI3K

phosphoinositide 3 kinase

Akt

protein kinase B

JNK

c-Jun N-terminal kinase

HePTP

hematopoietic protein tyrosine phosphatase

MNK1

MAP kinase-interacting serine/threonine-protein kinase 1

MSK1

mitogen and stress-activated protein kinase 1

MKP-3

mitogen-activated protein kinase phosphatase 3

STAT3

signal transducer and activator of transcription 3

MKK

mitogen-activated protein kinase kinase

MITF

melanogenesis-associated transcription factor