Research ArticleCancer

Activating mutations in MEK1 enhance homodimerization and promote tumorigenesis

See allHide authors and affiliations

Science Signaling  30 Oct 2018:
Vol. 11, Issue 554, eaar6795
DOI: 10.1126/scisignal.aar6795

Know thine enemy: Mutant MEK

Many cancers, notably melanomas, are driven by activation of the RAS-ERK signaling pathway, but tumors are often resistant to pathway-targeted therapies. Yuan et al. characterized cancer-related mutations in the pathway-mediating kinase MEK1 and found that deletion mutations in a loop portion of the protein promoted MEK1 homodimerization and autophosphorylation, enabling the activation of ERK with or without upstream pathway activity. These mutants transformed cells and differentially conferred resistance to MEK inhibitors. Understanding the precise functions of these mutants and detecting them at diagnosis may help devise more effective treatment strategies for patients.


RAS-RAF-MEK-ERK signaling has a well-defined role in cancer biology. Although aberrant pathway activation occurs mostly upstream of the kinase MEK, mutations in MEK are prevalent in some cancer subsets. Here, we found that cancer-related, activating mutations in MEK can be classified into two groups: those that relieve inhibitory interactions with the helix A region and those that are in-frame deletions of the β3-αC loop, which enhance MEK1 homodimerization. The former, helix A–associated mutants, are inhibited by traditional MEK inhibitors. However, we found that the increased homodimerization associated with the loop-deletion mutants promoted intradimer cross-phosphorylation of the activation loop and conferred differential resistance to MEK inhibitors both in vitro and in vivo. MEK1 dimerization was required both for its activation by the kinase RAF and for its catalytic activity toward the kinase ERK. Our findings not only identify a previously unknown group of MEK mutants and provide insight into some key steps in RAF-MEK-ERK activation but also have implications for the design of therapies targeting RAS-ERK signaling in cancers.


The RAF-MEK-ERK kinase cascade plays a central role in cell biology, and its hyperactivation results in various human diseases, notably cancers (13). Genetic alterations that aberrantly activate this kinase cascade in cancers mainly occur in receptor tyrosine kinases (RTKs), the guanosine triphosphatase RAS, and the kinase RAF, whereas oncogenic mutations of the kinases mitogen-activated protein kinase kinase (MEK) and extracellular signal–regulated kinase (ERK) are rare. However, recent genomic sequencings showed that MEK mutations are highly prevalent in some types of cancers such as IGHV4-34+ hairy cell leukemia and non-BRAF(V600E) Langerhans cell histiocytosis (47). Characterizing oncogenic MEK mutations and determining how these mutations activate MEK have important implications for the development of therapeutic inhibitors for the treatment of MEK-driven cancers.

As a core component of the RAF-MEK-ERK kinase cascade, MEK is a dual-specific kinase that transmits a signal from active RAF to ERK via phosphorylation (8, 9). The regulation of MEK is complex and not yet completely understood. In RAF-driven cancers, active RAF mutants, such as BRAF(V600E), activate MEK by phosphorylating its activation loop (AL), which results in an abundance of phosphorylated MEK (phospho-MEK) (10). In contrast, the abundance of phospho-MEK is markedly lower in RTK/RAS-driven cancers, although they have similar amounts of MEK activity, as indicated by a comparable abundance of phospho-ERK (11, 12). This suggests that wild-type active RAF activates MEK in a different way than does BRAF(V600E). In addition, some studies have indicated that AL autophosphorylation of MEK might also contribute to its activation (1315), adding another layer of complexity to the regulation of MEK activity.

The activity of MEK is regulated not only by the AL phosphorylation/dephosphorylation but also by its interactions with other components of RAF-MEK-ERK kinase cascade (1619). The heterodimerization of RAF and MEK facilitates MEK activation (20), whereas the MEK-ERK interaction facilitates ERK phosphorylation and activation (21). Moreover, MEK has been shown to form face-to-face homodimers in which the AL of one protomer aligns with the catalytic site of the other (22), suggesting a potential intradimer transphosphorylation. In addition, MEK also forms heterodimers between its two isoforms, which regulates the duration of ERK signaling (23). However, whether and how MEK interactome alterations contribute to hyperactive ERK signaling–driven tumorigenesis is not yet known.

In this study, we conducted a survey of cancer genomic databases and identified a previously unreported group of oncogenic MEK1 mutants with in-frame deletions in the β3-αC loop, a regulatory region in various protein kinases, such as epidermal growth factor receptor (EGFR) and RAF. These MEK1 mutants exhibited an increased but differential homodimerization that promoted intradimer transphosphorylation of ALs between two protomers and was responsible for their different sensitivities to MEK inhibitors in clinical therapy or trials. Furthermore, we investigated the role of MEK dimerization in the signal transmission of RAF-MEK-ERK kinase cascade by using these MEK1 mutants with elevated but differential dimer affinity. Our data showed that MEK1 was phosphorylated by active RAF in a dimer-dependent manner and functioned as a dimer to phosphorylate ERK1/2, indicating that the dimerization of MEK is critical for its function. Together, this study not only disclosed a novel group of oncogenic MEK1 mutants but also uncovered the regulatory mechanism of MEK1 in the RAF-MEK-ERK kinase cascade. These findings had substantial implications for both targeted drug design and clinical approaches to treating RAS-RAF-MEK-ERK signaling–driven cancers.


A unique group of MEK mutations exist in cancer genomes

To survey and characterize cancer-related MEK mutants, here, we interrogated the International Cancer Genome Consortium database, the cBioportal for cancer genomics database, and the Catalogue of Somatic Mutations in Cancer database. Together with those reported in the literature, we compiled a list of all MEK mutants in cancer genomes (table S1) and constructed the MEK mutation spectrum (Fig. 1A and fig. S1A). Through statistical analysis, we found that MEK1 was the dominant isoform of oncogenic alterations and had four hot spots, which included the negative regulatory helix A (residues 44 to 61) (I), the αC-β4 loop (II), the β7-β8 loop (III), and the β3-αC loop (IV) (Fig. 1, B and C). It is well defined that the negative regulatory helix A interacts with the kinase domain and stabilizes its inactive conformation, whose disruption will trigger the kinase activity of MEK1 (24). According to their structural distributions, we thought that MEK1 alterations or mutations in regions I, II, and III activated MEK1 likely by relieving the inhibition of helix A, because all altered residues were involved in the interaction of helix A with the kinase domain. In contrast, the in-frame deletions within the β3-αC loop, which was highly conserved across different species (fig. S1B), represented a previously unknown group of MEK1 mutations that mainly existed in Langerhans cell histiocytosis and malignant melanoma and activated MEK1 through a distinct mechanism.

Fig. 1 The β3-αC loop deletion defines a unique group of oncogenic MEK1 mutations in genomes.

(A) Localization of hot spots in the cancer-related MEK1 mutation spectrum: (I) The inhibitory helix A, (II) the αC-β4 loop, (III) the β7-β8 loop, and (IV) the β3-αC loop. (B) Schematic diagram of MEK1 (Protein Data Bank ID: 3EQI) was generated by using PyMOL software. The frequently mutated residues in helix A are colored in blue, and those in hot spots II and III are colored in red. All residues in hot spots I, II, and III are involved in the interaction of helix A with the kinase domain. The hot spot IV is colored in cyan. (C) Alignment of MEK1 mutants with in-frame deletions of β3-αC loop. In total, 601 MEK1 mutants were identified in 69,813 patient samples, which include 31 mutants with in-frame deletion of β3-αC loop. (D and E) The activity of MEK1 mutants expressed in 293T cells was detected by immunoblot for phospho-ERK1/2 (pERK1/2). The phosphorylated and total fractions in (D) were quantified by ImageJ, and their ratio was calculated and shown in (E). (F and G) The activity of MEK1 mutants purified from 293T transfectants by immunoprecipitation was measured by in vitro kinase assay and immunoblot for phospho-ERK1/2. Data were quantified and shown as described for (D) and (E). (H) Foci formation assay in immortalized fibroblasts infected by retroviruses carrying various MEK1 mutants. Cells (5 × 103) per 60-mm dish were seeded, and the medium was refreshed every other day for 12 days. Foci were calculated as detailed in the Materials and Methods. All images are representative of at least three independent experiments. Data in (E) and (G) are means ± SD from at least three independent experiments.

MEK1 mutants with in-frame deletions of the β3-αC loop are constitutively active and oncogenic

To characterize MEK1 mutants with in-frame deletions of β3-αC loop, we measured their activity in live cells and in vitro, and found that these mutants activated ERK1/2 when expressed in 293T cells or purified from 293T transfectants (Fig. 1, D to G, and fig. S1, C and D), suggesting that all of them are constitutively active. However, MEK1 mutants with analogous deletions along this loop (ΔEI versus ΔIK versus ΔPA) exhibited elevated but differential activities (fig. S1C), suggesting that both the residue context and the shortening of the β3-αC loop contribute to the activation of MEK1. To justify this notion, we mutated the residues that are deleted in active MEK1 mutants or randomly deleted residues in the β3-αC loop (fig. S1, E to G) and found that all alterations activated MEK1 by different extends, which supported our hypothesis. As a direct target, MEK1 is activated by RAF upon upstream stimulation. To further confirm the constitutive activity of MEK1 mutants with variable deletions of β3-αC loop, we expressed these mutants in wild-type or RAF-deficient immortalized fibroblasts and demonstrated that they were able to activate ERK signaling independently of RAFs (fig. S1H). Moreover, the expression of MEK1 mutants except MEK1(ΔHLEI) that had the lowest activity in fibroblasts induced foci formations or transformed morphologies (Fig. 1H and fig. S1, I and J), suggesting that they have RAF-independent oncogenic potential. Together, these data indicate that MEK1 mutants with in-frame deletions of β3-αC loop are truly constitutively active and may promote tumorigenesis.

MEK1 mutants with in-frame deletions of the β3-αC loop are activated by enhanced homodimerization

In-frame deletions of the β3-αC loop defined a new group of oncogenic MEK1 mutations. However, we then wanted to explore how these alterations triggered the kinase activity of MEK1. In previous studies, we and other groups have found that variable β3-αC loop deletions activate RAF kinases through promoting homodimerization (2527). Here, we assumed that these MEK1 mutants were activated through a similar mechanism, although MEK forms a face-to-face dimer (22), which is different from the side-to-side dimer of RAF kinases. To testify this hypothesis, we first measured the dimer affinity of MEK1 mutants with variable β3-αC loop deletions by using polyacrylamide gel electrophoresis (PAGE) with low SDS (0.01%) and microscale thermophoresis (MST) methods. Our data showed that MEK1 mutants with in-frame deletions of β3-αC loop had increased dimer affinity with MEK1(ΔIHLEIK) > MEK1(ΔEI) > MEK1(DD) > MEK1(ΔHLEI) > wild-type MEK1 (Fig. 2, A and B, and fig. S2, A and B). A compound mutation that includes N78G, V224G, F311A, and L314A (called GGAA below) in the dimer interface (22, 23) markedly decreased both the dimer affinity and the activity of MEK1 mutants with β3-αC loop deletions (Fig. 2C and fig. S2C), which further supported our speculation. Moreover, the deletions of β3-αC loop in MEK1 did not alter its propensity to bind ERK or kinase suppressor of Ras (KSR) (fig. S2, D and E), which excludes the possibility that this type of alterations activate MEK1 through other modes. In addition, the homodimerization of MEK1 could be improved upon stimulations with EGF, BRAF(V600E), and KRAS(G12V) (fig. S2, F and G), suggesting that it may function as a principle to govern the activity of MEK1.

Fig. 2 The MEK1 mutants with deletions of β3-αC loop are activated through homodimerization-driven transphosphorylation.

(A) FLAG-tagged MEK1 mutants were expressed in 293T cells and analyzed for oligomeric status by PAGE with low SDS (0.01%) and immunoblot for FLAG. pMEK1/2, phospho-MEK1/2. (B) FLAG-tagged MEK1 mutants were expressed in 293T cells and purified by affinity chromatography, and their dimer affinity was measured by using MST method. Kd, dissociation constant. (C) MEK1 mutants were expressed in 293T cells, and their activity was measured by immunoblot for phospho-ERK1/2. (D) Catalytic spine–fused MEK1 mutants without inhibitory helix A (Δ44–51; activators) were coexpressed with wild-type MEK1 (“receiver”) in 293T cells, and the activation of ERK1/2 was measured by immunoblotting for phospho-ERK1/2. (E and F) The allosteric activity of the MEK1(DD) or MEK1(ΔEI) mutant that has a disrupted dimer interface was measured as described in (D). The allosteric, kinase-deficient activator derived from MEK1(ΔEI) served as a positive control. (G) Kinase-deficient MEK1 allosteric activators were coexpressed with wild-type MEK1 in 293T cells, and their associations were measured by coimmunoprecipitation (IP) and immunoblot. WCL, whole-cell lysate; IgH, immunoglobulin heavy chain. (H) MEK1 mutants were immunoprecipitated from 293T transfectants, and their AL phosphorylation was detected by immunoblot. (I) The AL phosphorylation of MEK1 mutants was determined as described in (H). (J) The wild-type MEK1 was coexpressed with kinase-dead allosteric MEK1 mutants, and its AL phosphorylation was detected by immunoblot upon immunoprecipitation. (K and L) MEK1 mutants were expressed in 293T cells, and their activity was measured by immunoblot for phospho-ERK1/2 (K). The ratio of phosphorylated to total ERK1/2 is shown in (L). (M) A model of homodimerization-driven MEK activation. The homodimerization helps MEK to assemble an active conformation, which in turn induces a cross-phosphorylation of AL between two protomers. The AL phosphorylation fully activates MEK dimer and stabilizes its active conformation. Furthermore, active MEK dimer with phospho-AL phosphorylates and activates ERK. All blots are representative of at least three independent experiments. Data in (B) and (L) are means ± SD from at least three independent experiments.

The activity of MEK1 was triggered by the enhanced homodimerization, suggesting that it has both allosteric and catalytic functions as RAF kinases do. To examine the allosteric activity of MEK1 by protein kinase coactivation assay (2834), we created kinase-deficient allosteric “activators” by deleting the negative regulatory helix A (ΔE44-E51) and fusing the catalytic spine (V82F mutation) in MEK1 mutants with variable deletions of β3-αC loop (fig. S2H). MEK1(ΔEI/ΔE44-E51/V82F) was the only mutant capable of activating wild-type MEK1 when coexpressed in 293T cells (Fig. 2D). Using the same approach, we next determined whether active wild-type MEK1 had allosteric activity as well as MEK1(ΔEI). We found that the AL mutation that mimics the phospho-AL status of active MEK1 turned MEK1(ΔE44-E51/V82F) into an effective activator (Fig. 2E), suggesting that MEK1 has intrinsic allosteric activity once activated. On the other hand, the kinase-dead K97A mutant had much less ability to transactivate wild-type MEK1 in contrast to the V82F mutant in the same context (fig. S2I), indicating that the active conformation of MEK1 is critical for its allosteric activity. To further confirm that MEK1(ΔEI) and its wild-type counterpart have the dimerization-driven transactivation ability, we disrupted the dimer interface of MEK1(ΔEI/ΔE44-E51/V82F) by introducing the GGAA mutation as above and found that this alteration abolished its allosteric activity (Fig. 2F). In addition, the subtle allosteric activities of MEK1(ΔHLEI/ΔE44-E51/V82F) and MEK1(ΔIHLEIK/ΔE44-E51/V82F) toward wild-type MEK1 might arise from their weak propensities to dimerize with it (Fig. 2G). Together, these data demonstrate that MEK1 mutants with deletions of β3-αC loop and wild-type MEK1 have both catalytic and allosteric activities.

The AL phosphorylation is an indicator of MEK fully activation (8). To explore the molecular basis of MEK1 activation by enhanced homodimerization, we assessed the AL phosphorylation status of MEK1 mutants with variable deletions of β3-αC loop and found that the AL was phosphorylated in MEK1(ΔEI) and MEK1(ΔHLEI) but not in MEK1(ΔIHLEIK) (Fig. 2A, bottom). Moreover, their AL phosphorylation was blocked by either catalytic spine fusion or dimer interface disruption (Fig. 2H). This data suggested that the homodimerization facilitates MEK1 to assemble an active conformation, which further drives its AL autophosphorylation. As to MEK1(ΔIHLEIK), it might mimic an intermediate locked in a transitional status by its extremely strong homodimerization. A partial disruption of its dimer interface by the AA mutation (F311A and L314A) led to the AL phosphorylation of MEK1(ΔIHLEIK), although the level of AL phosphorylation was still lower than that of MEK1(ΔEI) (Fig. 2I). To further determine whether the AL autophosphorylation occurs in cis or trans, we purified the receiver (wild-type MEK1) from MEK1 coactivation transfectants and found that its AL was barely phosphorylated (Fig. 2J), suggesting that the homodimerization drives an intradimer cross-phosphorylation of two protomers. This finding was also supported by that the nonphosphorylatable AL mutation (S217A/S221A) reduced the activity of MEK1(ΔEI) and MEK1(ΔHLEI) by 60 to 70% but not in that of MEK1(ΔIHLEIK) (Fig. 2, K and L). Overall, our data demonstrate that in-frame deletions of β3-αC loop activate MEK1 through homodimerization-driven, intradimer cross-phosphorylation (Fig. 2M).

MEK1 mutants with in-frame deletions of the β3-αC loop exhibit differential resistance to MEK inhibitors in vitro and in vivo

Given that MEK1 mutants with in-frame deletions of the β3-αC loop were identified from various types of cancers and exhibited strong oncogenicity (Fig. 1H and fig. S1, I and J), we next determined whether MEK inhibitors that are used in the clinic or currently in trials (35) could be used to target these mutants. By using 293T transfectants and stable fibroblast lines expressing these MEK1 mutants, we found that, among all inhibitors that were tested in this study, trametinib was the most effective at blocking the activity of MEK1(ΔEI), whereas MEK1(ΔIHLEIK) was resistant to all inhibitors (Fig. 3A and fig. S3, A to C). To further evaluate the ability of trametinib to target MEK1 mutants with β3-αC loop deletions in vivo, we here constructed xenografted melanomas by using MeWo melanoma cell lines that express these MEK1 mutants. Together with xenografted fibroblastomas derived from fibroblasts that stably express different MEK1 mutants (see Fig. 1H and fig. S1H), we found that all tumors harboring MEK1(ΔIHLEIK) had a robust resistance to trametinib treatment in contrast to those harboring MEK1(ΔEI) or MEK1(DD) (Fig. 3, B to D, and fig. S3, D to G).

Fig. 3 MEK1 mutants with in-frame deletions of β3-αC loop exhibit a different inhibitor resistance in vitro and in vivo, which arises from their differential dimer affinity.

(A) Fibroblasts stably expressing different MEK1 mutants were treated with MEK inhibitors as indicated, and the activation of ERK1/2 in cells was measured by immunoblot for phospho-ERK1/2. The ratios of phosphorylated to total ERK1/2 (calculated in and graph generated by GraphPad Prism 6) are shown below. (B and C) Fibroblasts expressing different MEK1 mutants were subcutaneously injected into nonobese diabetic–severe combined immunodeficient (NOD-SCID) mice, which were treated with or without trametinib as described in Materials and Methods, and xenografted tumors were harvested (B) and weighted (C) at the experimental end point (n = 5 to 7 per group; ****P < 0.0001 by two-tailed Student’s t test). n.s., not significant. (D) Immunohistochemistry staining analysis of xenografted tumors from (B). The activation of ERK1/2 and cell proliferation were assessed by staining for phospho-ERK1/2 and Ki67, respectively. Scale bar, 0.2 mm. (E) The MEK1 mutants were expressed in 293T cells, and their activity was measured by immunoblot for phospho-ERK1/2. (F) The 293T transfectants of MEK1 mutants from (E) were treated with trametinib as indicated, and their drug sensitivity was examined as in (A). All images are representative of at least three independent experiments. Data in (A) and (F) are means ± SD from at least three independent experiments.

Although both MEK1(ΔEI) and MEK1(ΔIHLEIK) were activated through a same mechanism, they exhibited distinct drug sensitivities. To explore the molecular basis underlying this phenomenon, we introduced different mutations into MEK1(ΔIHLEIK) to decrease its dimer affinity or to mimic its AL phosphorylation because it had higher dimer affinity and nonphosphorylated AL in contrast to MEK1(ΔEI). A partial disruption of the dimer interface by GG (N78G and V224G) or AA (F311A and L314A) mutations reduced the activity of MEK1(ΔIHLEIK) as predicted, whereas the DD (S218D and S221D) mutation slightly enhanced its activity (Fig. 3E). Furthermore, both mutants with decreased dimer affinity (GG and AA) but not that with the mimicry of phospho-AL (DD) lost their resistance to the MEK inhibitor trametinib (Fig. 3F), suggesting that the drug sensitivities of MEK1 mutants with variable deletions of β3-αC loop are determined by their dimer affinity.

MEK1 transduced a signal from active RAF to activate ERK in a dimer-dependent manner

The interactions among different components play a crucial role in the signal transmission of RAF-MEK-ERK kinase cascade (1618). Previously, we have demonstrated that active RAF kinases function as dimers to phosphorylate MEK and that the MEK-binding ability of both protomers in RAF dimers is required for this process (27). To better understand the molecular basis of MEK activation by RAF, here, we first wanted to determine whether the dimerization of MEK is required for its AL phosphorylation by active RAF. Thus, we coexpressed BRAF(V600E) or KRAS(G12V) with either wild-type MEK1 or MEK1 mutants with altered (impaired or enhanced) dimerization. We found that both BRAF(V600E) and KRAS(G12V) induced the AL phosphorylation of wild-type MEK1 and dimeric MEK1(ΔIHLEIK) but not that of dimer-impaired MEK1 mutant (GGAA), although the amount of AL phosphorylation differed (Fig. 4A). These data suggest that MEK dimerization is essential for its AL phosphorylation by active RAF. Given that the activation of MEK also requires its heterodimerization with RAF and that the heterodimer interface on MEK largely overlaps with its homodimer interface (20), it is possible that the impaired AL phosphorylation of MEK1(GGAA) induced by BRAF(V600E) or KRAS(G12V) results from a loss of heterodimerization with RAF. To assess this possibility, we examined the interaction of MEK1 mutants with BRAF(V600E) by coimmunoprecipitation and found that both MEK1(GGAA) and MEK1(ΔIHLEIK) showed a reduced interaction with BRAF(V600E) compared to the wild-type counterpart (Fig. 4B), suggesting that the impaired AL phosphorylation of MEK1(GGAA) may not only relate to the loss of heterodimerization with RAF and that the enhanced MEK homodimerization may bypass the requirement of RAF-MEK heterodimerization in the activation of MEK by BRAF(V600E). However, the weaker AL phosphorylation of MEK1(ΔIHLEIK) induced by KRAS(G12V), but not by BRAF(V600E) compared to that of wild-type MEK1, indicates that the heterodimerization of MEK1 with RAF is still critical for its activation by wild-type RAF.

Fig. 4 The dimerization of MEK is critical for its phosphorylation by active RAF and for its catalytic activity toward ERK.

(A) MEK1 mutants were coexpressed with either BRAF(V600E) or KRAS(G12V) in 293T cells, and their AL phosphorylation was detected by immunoblot upon immunoprecipitation. (B) MEK1 mutants were coexpressed with BRAF(V600E), and their interaction was detected by coimmunoprecipitation and immunoblot. MEK1(M308/310A) that has an impaired ability to dimerize with RAF kinase served as a control. (C) BRAF mutants were expressed in 293T cells, and phospho-MEK1/2 and phospho-ERK1/2 in 293T transfectants were detected by immunoblots. Asterisk (*) represents the R462E/I617R/F667A mutation. R462E/I617R/F667A mutation. (D) A375 cell line was treated with 5 μM trametinib or 20 μM AZD6244 for 2 hours, and the phospho-MEK1/2 was detected by immunoblot. (E) BRAF(V600E) or its mutant was coexpressed with MEK1 in 293T cells and treated with or without 10 μM trametinib for 6 hours. The association of BRAF(V600E) or its mutant with MEK1 was detected by coimmunoprecipitation and immunoblot. (F) FLAG-tagged and hemagglutinin (HA)–tagged MEK1 mutants were coexpressed in 293T cells and treated with or without 10 μM trametinib or 20 μM AZD6244 for 6 hours. The homodimerization of MEK1 mutants was detected by coimmunoprecipitation and immunoblot. DMSO, dimethyl sulfoxide. (G) MEK1 mutants were expressed in 293T cells, and their activity was measured by immunoblot for phospho-ERK1/2. (H) ERK2 mutants were coexpressed with either MEK1(DD) or BRAF(V600E) in 293T cells, and their AL phosphorylation was detected by immunoblot upon immunoprecipitation. All images are representative of at least three independent experiments.

To strengthen the speculation above, we thoroughly investigated the activation of MEK by BRAF(V600E) using mutagenesis methods and pharmaceutical inhibitors. We first found that the compound mutation (R462E/I617R/F667A) that disrupts the heterodimerization of BRAF with MEK (27) abolished the activity of BRAF(V600E) toward MEK (Fig. 4C), which further confirmed the essential role of RAF-MEK heterodimerization in the activation of wild-type MEK1. The faster migration of BRAF(V600E) mutant that cannot bind MEK might arise from its lower phosphorylation status, although a further study was needed for clarification. Furthermore, the MEK inhibitor trametinib that has been shown to break the CRAF/MEK heterodimerization in contrast to other MEK inhibitors, such as AZD6244 (12), effectively blocked the phosphorylation of MEK by BRAF(V600E) in cultured A375 melanoma cells (Fig. 4D), but only slightly impaired BRAF(V600E)/MEK1 heterodimerization (Fig. 4E), suggesting that the inhibition of BRAF(V600E)-mediated MEK phosphorylation by trametinib may mainly arise from the disruption of MEK homodimerization by virtue of the largely overlapped interfaces. In contrast to AZD6244, trametinib was able to impair the homodimers of MEK1(ΔEI), although they both blocked the activity of this mutant (Fig. 4F and fig. S4). Together, these data indicate that the dimerization of MEK1 is indispensable for its phosphorylation by active RAF. In addition, AZD6244 blocked the AL phosphorylation of MEK1(ΔEI) (fig. S4), but not that of wild-type MEK1 by BRAF(V600E) in A375 cells (Fig. 4D), further supporting our conclusion that the AL is autophosphorylated in MEK1(ΔEI).

We next investigated whether active MEK functioned as a dimer and required dimeric ERK as substrate for phosphorylation as active RAF did. We thus introduced the GGAA mutation in the dimer interface to dissociate the dimers of active MEK1(DD) as above and found that it impaired the activity of MEK1(DD) toward ERK (Fig. 4G, lane 5). Glutathione S-transferase (GST) is a strong dimeric protein whose fusion has been shown to restore the catalytic activity of RAF mutants with low dimer affinity by enhancing dimerization in vitro (27). Here, we further showed that such a GST fusion restored the activity of monomeric MEK1(DD/GGAA) (Fig. 4G, lane 6), suggesting that active MEK really acts as a dimer. However, wild-type and monomeric ERK2(H176E/L4A) (36) were phosphorylated equally when coexpressed with BRAF(V600E) or MEK1(DD) in 293T cells (Fig. 4H), indicating that the dimerization of ERK is not necessary for its activation by MEK.


In this study, we systematically characterized cancer-related MEK mutations and found that MEK mutants could be classified as two groups: (i) mutants in which the inhibitory interaction between helix A and kinase domain is relieved and (ii) mutants with in-frame deletions of β3-αC loop that enhance homodimerization. The former group of MEK1 mutants have been shown sensitive to MEK inhibitors (5, 6, 37, 38), whereas the latter had different sensitivities arising from their elevated but differential dimer affinity. The MEK1 mutants, such as MEK1(ΔIHLEIK), had a highest dimer affinity and exhibited a robust resistance to all inhibitors in clinical therapy or trials, which appeals a development of novel inhibitors that are able to target this type of unique mutants. Because the important cases of MEK1 mutations were identified in both primary cancers and drug-resistant cancers by genomic sequencings in past years, this study provided a guideline for the precision therapy of cancers harboring MEK1 mutations and had important implications in clinical practice.

The dimer affinity is a key factor that regulates the activation and inhibitor resistance of RAF kinases (1618). By characterizing MEK1 mutants with variable deletions of β3-αC loop, here, we have demonstrated that it also plays a critical role in the function and drug resistance of MEK. MEK inhibitors have a good efficacy in BRAF(V600E)-driven cancers, but only a marginal activity in Ras-mutated cancers (1012). Mechanistic studies have shown that MEK is highly phosphorylated in BRAF(V600E)-harboring cancers and activates downstream ERK independent of other molecules. In contrast, MEK is much weaker phosphorylated in Ras-mutated cancers, and it forms a complex with RAF kinases (39). Our finding that the inhibitor resistance of MEK1 mutants correlated with their dimer affinity but not the phosphorylation status raises a possibility that the assembly of RAF-MEK complex in Ras-mutated cancers strengthens the MEK-MEK interaction and thus weakens the efficacy of MEK inhibitors, which will be investigated in our future study.

The regulation of the RAF-MEK-ERK kinase cascade is very complex, which involves in interactions among its components (1618, 20). The RAF-MEK and RAF-RAF dimerizations have been shown indispensable for the activation of this kinase cascade. However, whether the MEK/MEK dimerization is also required for this process remains unclear. In this study, we used the MEK mutant with different dimer affinity to address this question and demonstrated that the dimerization of MEK was critical for both its activation by RAF and its catalytic activity toward ERK, which resembles the regulation of RAF. Together with previous findings, we proposed that the signal transmission in the RAF-MEK-ERK kinase cascade followed this procedure (Fig. 5A): (I) In quiescent cells, RAFs form face-to-face heterodimers with MEKs. (II) Upon stimulation, active RAS (also dimers) recruits RAF-MEK heterodimers to the plasma membrane through the RAS-binding domain (RBD) of RAFs, where RAF-MEK heterodimers form transient tetramers by the back-to-back dimerization of RAFs. (III) The back-to-back dimerization of RAFs activates RAFs by inducing cis-autophosphorylation of the AL and on the other hand, loosens RAF-MEK heterodimers to facilitate the assembly of MEK homodimers on their surface. (IV) Active RAF dimers phosphorylate one or both protomers of MEK dimers docking on their surface, and the phospho-MEK protomers can also cross-phosphorylate the other protomers in the context of MEK homodimers. (V) Once both protomers are phosphorylated, MEK dimers can be released from the RAF dimers and phosphorylate ERKs. Semi/nonphosphorylated active MEK dimers docking on the surface of active RAF dimers can also directly phosphorylate ERKs. Constitutively, active RAS mutants in cancer cells activate this kinase cascade same as active RAS does in normal cells. Previously, BRAF(V600E), the dominant RAF mutant in cancer genomes, had been thought to function as monomers to activate downstream pathway, because it is resistant to the central R509H alteration on dimer interface and sensitive to the first-generation RAF inhibitors compared with its splicing variants with high dimer affinity (40). However, this speculation has been challenged by recent findings (27, 29, 4143): First, BRAF(V600E) has been shown to exist as dimers or oligomers in cells, and the R509H alteration is not able to dissociate the dimers of BRAF(V600E) completely, although it blocks the dimerization-driven transactivation of RAF molecules under most conditions. Second, a compound mutation (P622A in APE motif and R509H) that completely blocks dimerization abrogates the activity of BRAF(V600E). Last, although the inhibitor resistance arises from the high dimer affinity of RAF mutants, it does not correlate directly with the oligomeric status of RAF mutants. Together with these findings, our study further demonstrates that like other constitutively active RAF mutants, BRAF(V600E) also activates MEK in a dimer-to-dimer manner (Fig. 5B). According to our model, RAF dimers can serve as platforms to facilitate MEK dimerization, and on the other way, MEK dimers can also function as stages to promote RAF dimerization, which can explain well a recent important finding that MEK drives RAF activation through binding with RAF or KSR (44). Our model indicates that allosteric inhibitors that disrupt RAF-MEK and MEK-MEK dimerization will have a high efficacy to block hyperactive ERK signaling in cancers harboring genetic alterations in MEK or upstream of MEK and be ideal next-generation drugs for cancer therapy.

Fig. 5 A model for the activation of RAF-MEK-ERK kinase cascade in both normal and cancer cells.

(A) The activation of RAF-MEK-ERK kinase cascade by upstream stimuli. In quiescent cells (I), RAF and MEK form a face-to-face heterodimer in cytosol. Upon stimulation (II), RAF-MEK heterodimers are recruited to the plasma membrane by active RAS through the RBD domain of RAF, where they form a transient tetramer through the side-to-side RAF dimerization. The side-to-side RAF dimerization (III) not only activates RAF but also loosens its face-to-face heterodimerization with MEK and therefore facilitates the homodimerization of MEK on the surface of RAF dimer. Active RAF dimer (IV) activates the MEK dimer upon docking to its surface by phosphorylating its AL, and MEK dimer can also activate itself through homodimerization-driven AL autophosphorylation. Active MEK dimer docking to or release from the surface of RAF dimer phosphorylates ERK (V). (B) The activation of RAF-MEK-ERK kinase cascade by BRAF(V600E) mutation. The dimeric BRAF(V600E) exists in cytosol, which serves as a platform to facilitate MEK homodimerization. Once homodimerized on the surface of BRAF(V600E) dimer, MEK is phosphorylated by BRAF(V600E), which fully activates MEK. Then, phosphorylated MEK dimer is released from the BRAF(V600E) dimer and activates ERK, whereas the BRAF(V600E) dimer recruits and activates “new” MEK proteins.


Antibodies, biochemicals, cell lines, and plasmids

Antibodies used in this study include: anti–phospho-ERK1/2 (#4370), anti–phospho-MEK1/2 (#9154), anti-MEK1/2 (#9124), anti-HA (#3724), and anti-FLAG (#14793) (Cell Signaling Technology); anti-FLAG (F3165) and anti–β-actin (A2228) (Sigma-Aldrich); anti-HA (MAB6875, Novus Biologicals); anti-ERK1/2 (A0229, AB clonal); anti-Ki67 (ab16667, Abcam); and horseradish peroxidase–labeled secondary antibodies (the Jackson Laboratory). All antibodies were diluted according to the manufacturers’ recommended protocols. Trametinib, GDC0623, cobimetinib, AZD6244, and binimentinib were purchased from MedChemExpress. All other chemicals were obtained from Sigma-Aldrich.

Wild-type BRAF−/− and CRAF−/− fibroblasts were generated in previous study (45, 46). MeWo and A375 melanoma cell lines were obtained from American Type Culture Collection.

Plasmids encoding RAS, RAF, MEK, ERK, and their mutants were constructed by Gibson assembly. pCDNA3.1(+) vector (Invitrogen) was used for transient expression, retro- or lentiviral vectors (Clontech) for stable expression, and pET-28a (Novagen) for bacterial expression.

Protein expression and purification

6xHis-tagged ERK2(K52A) was expressed in BL21(DE3) strains and purified by using a nickel column (Qiagen) as described before (47). FLAG-tagged MEK1, MEK1(ΔEI), and MEK1(ΔIHLEIK) were expressed in 293T cells and purified by using anti-FLAG affinity gel and 3xFLAG peptide (Sigma-Aldrich) and following the manufacturer’s protocol.

MST analysis

The dimer affinity of MEK1 and its mutants was measured using the Monolith NT.115 from NanoTemper Technologies. The MST analysis was carried out as described before (48, 49). Briefly, proteins were labeled with a fluorescent dye NT-647 (cysteine reactive) according to the manufacturer’s protocol. Then, a series of protein solutions were prepared by consecutive twofold dilutions in buffer containing 50 mM tris (pH 7.5), 150 mM NaCl, 0.1 mM tris(2-carboxyethyl)phosphine, and 1% NP-40. The labeled and unlabeled protein solutions were mixed with a volume ratio of 1:1 and loaded into silica capillaries after a short incubation at room temperature. The measurements were performed at 25°C by using 20% light-emitting diode power and 40% MST power. The laser-on and laser-off intervals were 30 and 5 s, respectively. The NanoTemper Analysis software (v. 2.2.4) was used to fit the data and to determine the apparent Kd values.

Cell culture, transfection, and transduction

All cell lines were maintained in Dulbecco’s modified Eagle’s medium with 10% fetal bovine serum (Hyclone). Cell transfection was carried out by using the Biotool transfection reagent. To generate stable cell lines that express MEK1 mutants, viruses were prepared and applied to infect target cells according to our previous studies (47, 50). Infected cells were selected by using antibiotics.

Immunoprecipitation, in vitro kinase assay, and Western blotting

Immunoprecipitations were performed as described previously (2830). Briefly, whole-cell lysates were mixed with either anti-HA (E6779) or anti-FLAG beads (A2220) (Sigma), rotated in cold room for 60 min, and washed three times with radioimmunoprecipitation assay buffer. For in vitro kinase assays, the immunoprecipitants were washed once with kinase reaction buffer [25 mM Hepes, 10 mM MgCl2, 0.5 mM Na3VO4, and 0.5 mM dithiothreitol (pH 7.4)] and then incubated with 20 μl of kinase reaction mixture (2 μg of substrate and 100 mM adenosine triphosphate in 20 μl of kinase reaction buffer) per sample at room temperature for 10 min. Kinase reaction was stopped by adding 5 μl per sample of 5xLaemmli sample buffer. The immunoblotting was carried out as described before (47). The PAGE analysis with low SDS (0.01%) was carried out as the regular SDS-PAGE except using the native PAGE sample buffer and the running buffer and gel with 0.01% SDS.

Foci formation assay

The foci formation assay was performed as described before (28). Immortalized mouse embryonic fibroblasts (MEFs) infected with retroviruses encoding target proteins were plated at 5 × 103 cells per 60-mm dish and fed every other day. Twelve days later, cells were fixed with 2% formaldehyde and stained with Giemsa solution (Sigma).

Animal studies

For xenograft experiments, female NOD-SCID mice (6 to 8 weeks) were subcutaneously injected with 3 × 106 cells per mice in 1:1 Matrigel (Corning). Tumor volumes were monitored by digital calipers twice a week and calculated using the formula: volume = (width)2 × length/2. Trametinib was administered orally (2 mg/kg) every other day when tumors reached an average volume of ~50 to 60 mm3. At the experiment end point, mice were euthanized, and tumors were harvested for ex vivo analysis and subsequent histology. All operations were approved by the Animal Ethics Committee of National Cancer Centre Singapore (NCCS).

Immunohistochemistry staining

Tumors were fixed in 10% buffered formalin overnight and embedded according to standard procedures. Tumor sections were cut to 4 μm thickness, mounted on glass slides, and air dried at room temperature. After antigen retrieval, tumor sections were stained with antibodies and then with hematoxylin. Images of tumor sections were taken with a bright light microscope at ×10.

Statistical analysis

All statistical analysis in this study was performed using GraphPad InStat (GraphPad Software, CA, USA). Statistical significance was determined by two-tailed Student’s t test in animal studies, and error bars represent SD to show variance between samples in each group or by one-sample t test in other experiments and to show variance between independent experiments.


Fig. S1. Cancer-related MEK mutants and their oncogenic potential.

Fig. S2. The MEK1 mutants with in-frame deletions of β3-αC loop are activated through homodimerization-driven transphosphorylation.

Fig. S3. MEK1 mutants with in-frame deletions of β3-αC loop exhibit differential inhibitor resistance in vitro and in vivo.

Fig. S4. The MEK inhibitors trametinib and AZD6244 block the AL phosphorylation of MEK1(ΔEI) and its activity.

Table S1. MEK1 mutations in cancer genomes.


Acknowledgments: We thank K. M. Hui, K. Sabapathy, P. Lam, D. Virshup, M. Wang, and their laboratories for help in experimental approaches and comments on this manuscript. We also thank A. Shaw and S. Taylor for assistances. Funding: This study is supported by NCCRF startup grant (NCCRF-SUG-JH), NCCRF bridging grant (NCCRF-YR2016-JUL-BG1), NMRC seeding grants (NCCSPG-YR2015-JUL-14 and NCCSPG-YR2016-JAN-17), Duke-NUS Khoo Bridge Funding Award (Duke-NUS-KBrFA/2017/0003), Asia Fund Cancer Research (AFCR2017/2019-JH), and SHF Research grant (SHF/FG692S/2016). Author contributions: J. Yuan and J.H. designed the study. J. Yuan and J.H. searched databases/literatures for MEK mutations in cancer genomes. M.B. prepared RAF knockout cell lines. J. Yuan, W.H.N., Z.T., J. Yap, and J.H. carried out molecular biology, biochemistry, and cell biology experiments. J. Yuan constructed mouse xenograft models and performed immunohistology analysis. M.B., Z.C., and J.H. supervised all experiments and interpreted experimental data. J.H. wrote the manuscript. M.B. revised manuscript, and all authors commented and approved the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data required for supporting the conclusion in the paper are present in the main text or the Supplementary Materials.

Stay Connected to Science Signaling

Navigate This Article