In the studies described here, we employed cultured melanoma cells freshly isolated from individual patients’ tumors as well as normal skin cells to investigate the impact of genetic variations on current therapy with PLX4032. We demonstrated that while PLX4032 inhibited ERK1/2 in BRAFV600E/K
, it activated this signaling pathway in BRAFWT
melanoma cells via stimulation of RAF1 in a RAS-independent manner. Activating mutations in NRAS and β-catenin, or loss of PTEN did not affect the responses of BRAFWT
melanoma cells to this BRAF inhibitor. PLX4032 enhanced the rate of proliferation of mitogen-dependent primary melanoma cells carrying the NRAS Q61L mutation, and decreased adhesion and increased migration, of rapidly dividing melanoma cells from advanced lesions, changes that may confer tumor advantage in vivo. Interestingly, while the proliferation of benign melanocytes isolated from a giant nevus was not affected, the drug inhibited keratinocytes. The latter results are not in contradiction with in vivo observations, i.e. an increase in the incidence of cutaneous squamous cell carcinoma in patients chronically exposed to the drug (Flaherty et al., 2009
), because the keratinocytes were isolated from basal and not squamous epithelium, which is composed of differentiated cells likely to have different growth properties. We also report for the first time inhibition of the ERK1/2 kinase MEKK3 in BRAFWT
cells treated with this PLX4032.
Activation of ERK1/2 by RAF inhibitors, such as SB-590885 and ZM 336372, has been reported before (Hall-Jackson et al., 1999
; King et al., 2006
), but the mechanism and consequences of such activation were not explored in these previous studies. In the course of peer review, two manuscripts were published that confirm our results (Hatzivassiliou et al., 2010
; Heidorn et al., 2010
). In these reports, the investigators also observed that selective BRAF inhibitors, such as PLX4720, 885-A and GDC-0879 stimulated MEK–ERK signaling in BRAF wild-type melanoma and carcinoma cells via RAF1 activation (Hatzivassiliou et al., 2010
; Heidorn et al., 2010
). Dr Marais and his collaborators went one step further, showing that mutationally inactive BRAFD594A
cooperates with oncogenic KRASK12D
in inducing melanoma in genetically engineered mice in vivo. The results of both groups support a model in which the BRAF-specific inhibitors induce RAS-GTP-dependent RAF1 activation via the formation of BRAF-RAF heterodimers or RAF1 homodimers followed by recruitment of RAF1 to the plasma membrane, triggering the MEK-ERK pathway (Hatzivassiliou et al., 2010
; Heidorn et al., 2010
). In support of this mechanism, the investigators demonstrated co-immunoprecipitation of RAF1 with BRAFWT
after treatment with 885-A (Heidorn et al., 2010
), or GDC-0879 (Hatzivassiliou et al., 2010
), RAF1 kinase-domain homodimers when co-crystallized with GDC-0879 (Hatzivassiliou et al., 2010
), and the translocation of BRAF and RAF1 to the plasma membrane accompanied by increased RAF1(S338) phosphorylation (Hatzivassiliou et al., 2010
). In addition, the BRAF-inhibitors induced about 30% increase in the proliferation of established carcinoma cells (but not melanoma cell lines) derived from different tumors (Hatzivassiliou et al., 2010
). However, as in our studies, both groups failed to detect BRAF-RAF1 heterodimers in response to PLX4720, a drug structurally related to PLX4032, and PLX4720 did not induce BRAF and RAF1 translocation to the membrane or an increase in RAF1(S338) phosphorylation (Hatzivassiliou et al., 2010
). Furthermore, crystal structure revealed distinct differences in the mode of PLX4720 and GDC-0879 interaction with BRAF. While PLX4720 binding disrupts ion pairing, shifting the αC helix which inactivates the kinase, binding of GDC-0879 maintains the ion pair, orienting the αC-helix into an active conformation (Hatzivassiliou et al., 2010
Altogether, although we have not uncovered the mechanism by which PLX4032 activated RAF1 in BRAFWT melanoma cells, our data are consistent with the published reports and suggest that PLX4032 may have a different mode of action, which is independent of RAS-GTP because mutant RAF1 R89L that does not bind RAS was also activated to the same degree as its wild-type counterpart. It is also possible that PLX4032, like PLX4720, induces weak hetero- or homo- RAF dimers, as suggested by Marais and his collaborators that is not sustained through experimental manipulations.
In our hands, the presence of mutant NRAS in BRAFWT melanoma cells was not required for PLX4032 induced ERK activation, cell detachment, loss of adherence and migration. However, so far, drug induced promotion of cell proliferation was observed in two NRAS mutant primary melanoma cell strains that did not acquire full independence from environmental growth stimuli. Therefore, dependency of growth response on NRAS mutation after treatment with PLX4032 should be further explored once growth factor-dependent BRAFWT/NRASWT melanoma cells become available. Altogether, our studies demonstrate persistence ERK activation in response to BRAF inhibition in BRAFwt melanoma cells, the consequences on several downstream targets, the upregulation of a wide scope of ERK-responsive genes and the physiological changes that can confer growth advantage to these cells.
An important factor in cancer management is the ability to choose patients for specific therapy, monitor treatment and assess response. Very frequently, the presence of the activated target is not sufficient to assume responsiveness as shown for the activated/amplified EGFR where the levels of phosphorylated downstream mediators, Akt, ERK1/2 and Stat3 served as better markers (Emery et al., 2009
). We demonstrated that melanoma cells heterozygous for V600E, but not V600K BRAF alleles were less sensitive to the inhibitory effect of PLX4032 and maybe be considered in patient selection and treatment. Our studies also suggest that monitoring the status of phospho-ERK1/2 and phospho-FAK S910 in tumor biopsies can be a good indicator for adverse effects, and that blood levels of IL8 and LIF may serve as markers, because normal cells were either inhibited or not-responsive to the drug.