Inhibition of glioblastoma dispersal by the MEK inhibitor PD0325901

Therapies aimed at containing tumor cell dispersal could provide a powerful path towards extending the time of disease-free and overall survival of glioblastoma patients. Identifying drugs that can target molecular pathways involved in dispersal would provide valuable insight towards this goal. Our previous studies showed that Dexamethasone, an FDA approved drug to treat tumor-related edema in GBM, can also decrease in vitro and ex vivo dispersal of primary human GBM cells. It does so by activating ?5?1 integrin and subsequent restoration of FNMA and re-organization of cortical actin into stress fibers. In turn, these changes engender an increase in the strength of intercellular cohesion, increased attachment of tumor cells to substrate, and reduced cell motility. The net effect is an overall reduction in dispersal [1, 27]. The effects of Dex, however, are pleiotropic and the drug likely targets many pathways, which in part may explain the many side-effects associated with Dex treatment. Identifying drugs that are more specific in their targeting of dispersal-related pathways is therefore important.

In this study, we explore whether inhibition of the MAPK/ERK pathway, a critical regulator of processes underlying invasion and metastasis [28], could have similar effects on GBM dispersal. We tested the effects of the MEK inhibitor, PD0325901, on 4 primary GBM cell lines that were previously used to assess the effects of Dex on dispersal [1, 21]. Studies have shown that certain GBM lines do not respond to MEK inhibitors [20]. We therefore assessed whether our lines are responsive to PD0325901 by determining whether treatment results in a decrease in the levels of phospho-ERK. All 4 lines responded to the drug. We previously established that the cell lines were all deficient in their capacity for FNMA [1]. In contrast to Dex, treatment with PD0325901 did not result in a significant increase in FNMA. However, treatment with the MEK inhibitor resulted in a remarkable change in cell shape and in the reorganization of actin from cortical into stress fibers. This was particularly evident when actin was visualized in 3D spheroids. Given that the actin cytoskeleton is a fundamental mediator of cell and tissue stiffness [29], we posited that a shift in actin organization would correspond to a change in tissue stiffness.

Stiffening of the ECM is considered to be a hallmark of fibrotic lesions and has been demonstrated to modulate cell invasion and migration [30]. The current study focused on whether aggregate stiffness and viscosity could modulate dispersal. We quantified stiffness and viscosity using methods based on ellipsoid relaxation, specifically after the deforming external force is removed [31, 32]. The aggregate was modeled as a Kelvin-Voigt viscoelastic body [33, 34]. Unexpectedly, PD0325901 treatment only resulted in a modest increase in aggregate stiffness but not of viscosity. However, when aggregates were generated in higher concentrations of fibronectin, both stiffness and viscosity increased significantly. This is important for several reasons. First, the fibronectin gene has been shown to be up-regulated in GBM [35]. Accordingly, tumors able to respond to PD0325901 and in the presence of high concentrations of fibronectin, could, in principle, become stiffer and more viscous. Stiffer tumors have previously been shown to be less invasive and to grow more slowly [36]. Few studies have addressed the issue of tumor viscosity and those that have focus on applications of magnetic resonance elastography in liver tumors where fibrosis is a key parameter. In those studies, tumor viscosity appeared to be higher in malignant tumors [37]. In GBM, however, fibrosis is not typically observed. In GBM spheroids, the increase in viscosity in response to PD0325901 treatment was likely due to higher binding energy between the activated ?5?1 integrin and fibronectin. This would effectively increase the friction between cells and the ECM. This increase in friction could significantly reduce the capacity for dispersal of tumor cells from the primary mass.

Treatment also resulted in the localization of p-FAK at sites of cell-substrate attachment. This is consistent with the observed resistance to flow-induced substrate detachment of GBM cells, and to decreased motility. Since cells require intermediate levels of cell-ECM adhesion to be optimally motile [38], an increase in the strength of cell-ECM adhesion past this point might stabilize adhesion to substrate to a point that significantly reduces cell movement, and consequently, dispersal. Decreased motility also appears to be associated with a significant decrease in dispersal velocity of GBM aggregates. Since PD0325901 treatment did not restore FNMA, it is likely that decreased motility rather than increased cohesion is the physical mechanism that restrains the detachment of tumor cells from the mass. Indeed, cells at the leading edge of treated aggregates appear to attach tightly to substrate causing cells behind them to pile up, again pointing to reduced motility as the primary restraint for detachment. For three of the four primary GBM lines, PD0325901 also significantly reduced the ability of single GBM cells to disperse through an astrocyte-seeded scaffold. It is not possible to differentiate between the effects of PD0325901 on decreased motility and ability to disperse through the scaffold, however, it is possible that on a single cell level, the re-organization of actin into stress fibers may have effectively rendered cells less compliant and inhibited their capacity to sufficiently deform and squeeze through pores established by the physical environment established by the scaffold. It is important to note that for GBM-4, treatment did not reduce z-axis dispersal. It is possible that in this line, compliance was not effected by treatment, thus allowing cells to penetrate into the scaffold.

Lastly, MEK inhibitor treatment also appears to significantly reduce growth rate of these primary GBM lines in both conventional 2D and in 3D cultures. Other studies have demonstrated in vivo efficacy of PD0325901 in reducing tumor growth in preclinical orthotopic models of glioblastoma [18]. Our study provides compelling evidence that PD0325901 can also reduce dispersal. Growth and dispersal contribute significantly to recurrence. Accordingly, the drug has the potential to significantly delay the onset of recurrence in GBM.

Identifying agents that can contain the primary or recurrent tumor could significantly improve targeted delivery of chemotherapeutic agents and increase the likelihood of total surgical resection. We have previously identified Dexamethasone (Dex) as a potential candidate to reduce dispersal of GBM [1]. Interestingly, the doses required to elicit a dispersal inhibitory response are significantly lower than those typically used to reduce edema [1]. Clinically, MEK inhibitors are generally well tolerated. Commonly occurring toxicities include rash, diarrhea, fatigue, peripheral oedema and acneiform dermatitis. Life-threatening toxicities associated with MEKi are extremely rare. Long-term use is possible providing that adverse events are monitored and dose or treatment schedules are modified, as required [39]. The measureable outcome for MEK inhibitor studies focus on their ability to reduce tumor size. Here, we show an added benefit of one MEK inhibitor as a potential deterrent of tumor cell dispersal. Whereas Dexamethasone readily crosses the blood–brain barrier, some MEK inhibitors, including trametinib, have demonstrated limited brain distribution due to association with the P-glycoprotein efflux transporters found at the blood–brain barrier [19]. Perhaps a strategy in which MEK inhibitors are used as interstitial chemotherapy, followed by continued administration of low-dose Dex, could significantly improve prognosis of this devastating disease.