The work presented herein provides evidence of frequent cross-resistance to the BRAF inhibitor vemurafenib and the MEK inhibitor AZD6244 in cell lines with primary or acquired resistance to vemurafenib, with frequent reversal of the acquired resistance by the addition of inhibitors of the AKT/mTOR pathway. In this study only cell lines with a secondary NRAS mutation, that are sensitive to MEK inhibitor, were the exceptions to the cross-resistance to BRAF and MEK inhibitors. The cross-resistance between the BRAF and MEK inhibitors is rather surprising given the exquisite dependence that BRAFV600E
mutant melanomas have demonstrated on the MAPK pathway. In paired biopsies of patients treated with vemurafenib this agent demonstrated a dose-dependent inhibition of p-ERK, suggesting that the activity of this agent as inhibitor of oncogenic BRAF relies on efficient inhibition of MAPK pathway signaling 
. The development of in vitro
acquired resistance to PLX4720, an analogue of vemurafenib, has been linked to the re-activation of p-ERK 
. In addition, acquired resistance to a different BRAF inhibitor, AZ628, was associated with alternate signaling from BRAF to CRAF again resulting in the re-activation of p-ERK 
. Combined, these data had suggested that further inhibition of the MAPK pathway with a MEK inhibitor may be a way to treat acquired resistance to the BRAF inhibitor. In fact, this concept has been taken into the clinic with ongoing clinical trials, but our data predicts that sequential single agent treatment with a MEK inhibitor after developing acquired resistance to a BRAF inhibitor will only work (partially) in a subset of cases with secondary NRAS mutations.
As the molecular mechanisms of primary and acquired resistance to vemurafenib are being studied 
it will be important to tailor the treatments to be added or sequentially tested in patients progressing on this therapy. It is becoming clear that resistance to BRAF inhibitors will not follow the pathway of resistance of chronic myelogenous leukemia (CML) to imatinib, where secondary mutations in the abl
kinase are the main mechanism of resistance 
. The study of resistance mechanisms in the sublines with in vitro
acquired resistance to vemurafenib and patient-derived resistant cell lines included in this and other studies suggest three main mechanisms of acquired resistance, the upregulation of the receptor tyrosine kinases such as PDGFR1β 
or IGF1R 
, increased expression of the cancer Osaka thyroid (COT, also known as MAP3K8) kinase 
, or secondary mutations in NRAS 
or MEK 
. Secondary mutation in NRAS or MEK, or upregulation of COT suggests acquired resistance mechanisms that maintain dependence on the MAPK pathway. In our studies, two vemurafenib-resistant cell lines with an acquired NRASQ61k
mutation secondary to their pre-existing BRAFV600E
mutation exhibited some sensitivity to a sequential treatment with a MEK inhibitor. Interestingly, these two cell lines with the secondary NRAS mutation also showed sensitivity to the combinations of drugs inhibiting both AKT and MAPK pathways. This may be due to a possible cross talk between mutated NRAS and AKT pathway. Possibility of such a cross talk holds clinical and scientific importance and would be interesting to be investigated in the future studies. Meanwhile, all other cell lines displayed resistance to the sequential treatment with the MEK inhibitor if they were resistant to vemurafenib. In this group of resistant cell lines, most have the PDGFRβ-mediated mechanism of acquired resistance 
. This information suggests that the elucidation of the specific mechanisms of resistance to vemurafenib points out to different therapies to be added or used sequentially with BRAF inhibitors.
A recent study on resistance to an analog of vemurafinib, PLX4720, suggested that only in cell lines with PTEN deletion p-AKT is induced by this BRAF inhibitor, and lack of PTEN may play a role in preventing apoptosis of melanoma cell treated with this compound 
. However, in our study we found that the PTEN null cell line M249 was very sensitive to both vemurafenib and AZD6244, which may be an outlier compared to prior reported data 
. Interestingly, by continuous exposure of this BRAFV600E
mutant/PTEN null cell line to vemurafenib an acquired resistant cell line that was mediated by a secondary mutation in NRAS causing the resistance through the reactivation of the MAPK pathway. Moreover, our results from other in vitro
acquired resistant cell lines indicated that regardless of the PTEN status, p-AKT could be induced by vemurafenib or AZD6244 treatment. These findings indicate that alterations in both MAPK and AKT pathways can be the cause of resistance to vemurafenib and induction of p-AKT in resistant melanoma cell lines is rather a more general phenomenon and not solely limited to PTEN mutant cell lines.
There is clear evidence of multiple levels of cross-talk between MAPK and PI3K/AKT pathways, and it has been shown that ERK can be phosphorylated by the AKT pathway (Figure S4
. Therefore, it is likely that treatments to inhibit alternative survival signaling in melanoma cells resistant to MAPK inhibitors will require co-inhibition of the PI3K/AKT pathway. The concept of simultaneous inhibition of the MAPK and the PI3K/AKT/mTOR pathways has been widely considered to treat altered oncogenic signals 
, and at least one clinical trial combining a MEK inhibitor with an AKT inhibitor is currently underway (NCT01021748). Given the frequent cross-talk and feedback regulation between both pathways we explored the effects of vemurafenib or AZD6244 on p-AKT and its downstream factors as the key signaling molecules in cells with primary or acquired resistance to vemurafenib. Our approach was also based on the evidence that cells with resistance to PLX4720, an analogue of vemurafenib, have a MEK-independent survival drive that can be blocked by inhibitors of the PI3K/AKT/mTOR pathway 
. In addition, in cell lines with acquired resistance to BRAF inhibitors through upregulation of IGF1R, resistance can be inhibited by the co-administration of a combination of a MEK and a PI3K inhibitor 
. Indeed our experiments demonstrated a differential effect on the AKT/mTOR/S6K pathway in vemurafenib-sensitive and -resistant cells both when exposed to vemurafenib or AZD6244. The most profound effect was the persistence of p-p70 S6K1 in cross-resistant cell lines treated with either drug, but it was particularly more evident with the MEK inhibitor AZD6244. Genetic inhibition of both p70 S6K1 and S6K2 with siRNAs showed additive effects with either of the drugs to further decrease the phosphorylation of the downstream protein S6. It should be noted that S6 can be phosphorylated at Ser235 by p-ERK as well. Therefore, changes in phophorylation of S6 can be the result of alterations in activity of p-ERK or p-P70 S6K1 or both, and that can be the reason for the lack of direct correlation between phosphorylation of S6K and S6 in our pharmacological inhibition studies.
In this study, siRNA knockdown of RICTOR decreased p-AKT Ser473 and also exhibited additive effects with vemurafenib or AZD6244 in further decreasing p-S6 and p-4EB-P1. These data suggest that S6 and 4EB-P1 are also potential cross-talk points between the AKT and MAPK pathways. Therefore, at least in this case, inhibition of both pathways is necessary to overcome the resistance to vemurafenib and AZD6244. Moreover, the inhibitory effect of RICTOR knockdown on growth of resistant cells suggests that activation of AKT by mTORC2 feedback may play a role in maintenance or even induction of cell growth and therefore can be one of the causes of resistance to MAPK pathway inhibitors. AKT activation by the feedback mechanism could be the cause of growth inducing effect of vemurafenib in resistant cell lines. This feed back mechanism occurs through the induction of mTORC2, which contains RICTOR, and causes higher levels of S473 p-AKT. It should be mentioned that the ability of a combination of MAPK pathway and AKT/mTORC inhibitors to reverse resistance to single agent MAPK inhibitors was not absolute and inefficient for some of the cell lines with primary and acquired cross-resistance to vemurafenib and AZD6244. Growth inhibition assays indicated that combinations of chemical inhibitors of MAPK and AKT pathways can decrease growth rates of some of the vemurafenib resistant cell lines. However these decrease in growth rates of resistant cell lines were not accompanied by the induction of apoptosis in these cell lines, particularly when vemurafenib alone or in combination was used. In the resistant cell lines, inhibition of MEK by AZD6244 could cause some induction cleaved caspase 3 in comparison to the vehicle treated samples. However, levels of cleaved caspase 3 were not increased further by the combination of AZD6244 and inhibitors of AKT pathway. This discrepancy between the growth and apoptosis assays perhaps indicate that these drug combinations may inhibit the growth by mechanisms other than apoptosis or through the ways which do not cause the induction of cleaved caspase 3.
Other investigators have provided convincing evidence that the effects of targeted inhibitors on melanoma cell lines is different in 2-dimension and 3-dimension models 
, with higher resistance to BRAF inhibitors in 3 dimension models mediated by the PI3K/AKT pathway 
. This was not tested in our studies, and may further underscore the importance of co-targeting both the MAPK and the PI3K/AKT/mTOR pathways for more profound antitumor effects in cells with acquired resistance to single agent BRAF inhibitors. Another possibility to expand on our studies would be the testing of siRNA or an isoform-specific inhibitor of AKT3, which has been previously described to be important in melanoma 
. The fact that the particular inhibitor used by us (which preferentially blocks AKT1 and AKT2, but at higher concentrations also blocks AKT3) had synergistic effects with vemurafenib or AZD6244 in several cells with cross-resistance to either single agent underscores the promise of co-targeting both pathways as means to treat acquired resistance to BRAF inhibitors.
In conclusion, our data suggest that single agent MEK inhibitor has low activity in vemurafenib-resistant melanoma and perhaps restricted to a subset of cases with secondary oncogenic mutations in NRAS. However, upon progression the addition of an AKT or an mTOR inhibitor to the continued therapy with vemurafenib, or switching to a combination of a MEK inhibitor plus an AKT or an mTOR inhibitor, may provide additional inhibitory activities. Our data strengthens the results from other groups that have previously demonstrated the superior antitumor activity of combining MAPK and PI3K/AKT/mTOR pathway inhibitors in BRAFV600E
mutant cell lines 
, by testing this concept in isogenic pairs of sensitive and acquired resistant cell lines, and in cell lines established directly from patients progressing after a response on vemurafenib. Therefore, the elucidation of the molecular mechanisms that result in primary or acquired resistance to vemurafenib and sensitivity to combined MAPK and AKT/mTOR pathway inhibition, would provide useful biomarkers to rationally choose the most appropriate therapy in BRAFV600E
mutant melanomas resistant to vemurafenib.