**3. Results**

#### *3.1. MBZ Inhibited NF1-Derived MPNST Cell Lines through Ras Inhibition*

Human MPNST cells NF90-8 and sNF96.2, both derived from NF1 patients, were treated with MBZ for 72 h at indicated concentrations, revealing favorable IC50 levels at 0.18 and 0.32 μM, respectively (Figure 1A). Because NF1-associated tumors are mainly driven by Ras hyperactivation, we studied MBZ's ability to inhibit Ras activity in the NF90-8 cell line by exposing NF90-8 cells to di fferent concentrations of MBZ (0.2 and 1 μM) for 24 h. The activated form of GTP-bound Ras, detected by GST-Raf1-RBD fusion protein binding, was reduced in MBZ-treated NF90-8 cells in a concentration-dependent manner (Figure 1B). This confirmed the Ras inhibitory e ffect of MBZ in vitro.

**Figure 1.** Mebendazole (MBZ) inhibits malignant peripheral nerve sheath tumor (MPNST) cells and Ras activity. ( **A**) IC50s of MBZ with NF90-8 and sNF96.2 cells were measured at 0.18 and 0.32 μM, respectively. Cells were incubated with MBZ or DMSO for 72 h and viable cells were determined with WST-8 and calculated as percentage of the control. Data are presented as mean ± s.d. (**B**) RASopathy

Neurofibromatosis 1 (NF1)-deficient NF90-8 cells were treated with MBZ at 0.2 and 1 μM for 24 h and cell lysates were incubated with GST-Raf1-RBD (the Ras-binding domain) coupled with glutathione resin. The pulldown products were analyzed by anti-Ras western blot, showing the activated GTP-bound Ras protein. Lysates incubated with GTPγS were used as positive controls.

#### *3.2. MBZ Delayed Tumor Formation and Improves Survival in NPcis Mice*

As reported before, *cis Nf1*+/−*;Tp53* +/− (NPcis) mice are naturally predisposed to a number of solid malignancies, which typically form ~3–5 months after birth: 77% will develop soft tissue sarcomas—of which 60–65% are MPNSTs, 20% malignant Triton tumors, 10% rhabdomyosarcomas, 10% leiomyosarcomas and fibrohistiocytomas, 14% lymphomas, 8% carcinomas, and 1% neuroblastomas [21–23]; astrocytomas have also been reported [24,26].

To determine the most e ffective and tolerable long-term MBZ dose in vivo, 60-day old male and female NPcis mice were separated into groups and provided with control feed or continuous medicated feed containing 175, 195, 215 or 250 mg/kg MBZ. This range was calculated based on our previously established maximal dose of 50 mg/kg MBZ via oral gavage and the estimated daily food intake of a mouse [17]. Mice were weighed weekly and examined for signs of toxicity over 4 weeks. In the higher MBZ dosing groups of 250 and 215 mg/kg diets, nearly all mice showed evidence of excessive toxicity, including ru ffled fur and significant weight loss between 10–15% thereby precluding the long-term use of those doses and establishing 195 mg/kg MBZ feed as the most suitable diet for long-term chemoprevention in these mice (Figure 2A,B).

**Figure 2.** Dose-dependent MBZ toxicity in *cis Nf1*+/−*;Tp53*+/− (NPcis) mice. The 60-day old NPcis mice were provided with MBZ feed at indicated concentrations. Shown is the 30-day weight of ( **A**) male and (**B**) female mice on the MBZ diet with the indicated doses. *n* = 5 mice per each MBZ dosing group.

In order to investigate the tumor-preventative e ffects of MBZ, continuous oral administration of MBZ via 195 mg/kg feed was initiated at 60 days after birth, before the formation of any malignancies. Mice were palpated weekly for the presence of any tumors. For the purpose of this study, 'Solid Malignancies' were defined as any type of sarcoma and astrocytoma, in addition to neuroblastomas and carcinomas, while 'Others' included non-solid malignancies such as lymphomas, leukemias and unknown causes of death.

MBZ treatment started at the age of 60 days significantly increased the overall median survival for male, female and combined cohorts (Figure 3A). In MBZ-treated mice, the time to tumor occurrence was significantly delayed compared to untreated control animals: 50% of all control mice had developed tumors and succumbed to disease by the age of 160 days, whereas in the MBZ-treated cohort, the tumor occurrence and median mortality was delayed by 32 days to 192 days (Figure 3B). Although observed in male and female NPcis mice alike, MBZs cancer preventative e ffect appeared to be more pronounced in males, with an increase in median survival by 34.5 days compared to 14 days in female mice (Figure 3B). Figure 3C demonstrates that MBZs chemopreventative e ffect was specific to mice with solid malignancies and did not affect the median survival of other, i.e., non-solid malignancy-related and unknown, causes of death both in male and female mice (Figure 3C). Lastly, MBZ treatment resulted in a ~25% reduction in solid cancer-related causes of death, thus demonstrating the feasibility of such a cancer prevention strategy in these NPcis mice (Figure 3D)

**Figure 3.** MBZ delays cancer onset and improves survival in NPcis mice. Shown are the Kaplan–Meier curves for MBZ-treated NPcis mice, initiated 60 days after birth, in comparison to controls for (**A**) overall survival, (**B**) solid malignancy-related mortality and (**C**) others, i.e., non-solid malignancy-related and unknown causes of mortality, analyzed as combined (males and females, left) male (middle) and female (right) cohorts. Animal numbers are provided for the specific groups in each graph and were analyzed with a two-sided log-rank test. \* *p* ≤ 0.05; \*\*\* *p* ≤ 0.001, \*\*\*\* *p* ≤ 0.0001; ns = not significant. (**D**) Percentage distribution of malignancy-related cause of death of MBZ-treated NPcis mice compared to controls, analyzed as combined (males and females, left) male (middle) and female (right) cohorts.

#### *3.3. MBZ Reduced pERK Activity in Tumors In Vivo*

In NPcis mice, the loss of *Nf1* leads to the hyperactivation of Ras, with the subsequent activation of the downstream effector ERK that is reflected by elevated levels of phosphorylated ERK (pERK) in MPNSTs and other related tumors. Immunohistochemistry showed that continuous MBZ treatment with a 195 mg/kg diet reduced pERK levels in sarcomas of NPcis mice compared to untreated mice (Figure 4). An analysis of the DAB staining intensity in three independent MBZ-treated tumor samples confirmed these results, with a reduced mean optical density (OD) of 0.02 in MBZ-treated samples compared to 0.05 in controls, while ERK staining was similar between both groups, with mean intensities of 0.07 and 0.08 for MBZ-treated and untreated tumors, respectively (Figure 4).

**Figure 4.** MBZ reduces ERK (pERK) in treated NPcis mice. Representative images of tumors from untreated controls (left) and MBZ-treated NPcis mice (left) were stained for pERK1/<sup>2</sup> (upper row) and ERK1/2 (lower row). pERK staining was visualized in brown in untreated controls but reduced in tumors of MBZ-treated mice. Each scale bar represents 100 μm.

#### *3.4. Cancer-Preventative E*ff*ects of CXB and MBZ Are Similar in NPcis Mice*

The antitumor effect of selective COX-2 inhibitors, such as sulindac (SUL) and celecoxib (CXB), has been shown in several malignancies and cancer predisposition syndromes. In the NPcis mouse model, we found that MBZ-treated mice had a longer overall median survival of 199 days compared to CXB, with 193 days; however, this difference was not statistically significant (Figure 5A). When compared to untreated controls, CXBs effect on median survival was statistically increased in male NPcis mice with solid malignancies, while female mice showed a notable, but statistically insignificant, increase in survival compared to controls. Furthermore, CXB was substantially more effective in delaying the onset of malignancies than SUL, which showed a median survival of 171.5 days and failed to demonstrate any effect in male or female mice compared to controls (Figure 5A,B). Like MBZ, neither SUL nor CXB had an effect on the survival of non-cancer related causes (Figure 5C). Consistent with our findings, we also noticed a ~25% decline in cancer-related cause of death in CXB-treated mice (Figure 5D).

**Figure 5.** MBZ and cyclooxygenase-2 (COX-2) inhibitor celecoxib (CXB) are similarly effective in preventing cancer in NPcis mice. Shown are the Kaplan–Meier curves for CXB and sulindac (SUL)-treated NPcis mice, initiated 60 days after birth, in comparison to MBZ-treated mice and untreated controls for (**A**) overall survival, (**B**) solid malignancy-related mortality and (**C**) other, non-solid malignancy-related and unknown causes of mortality, analyzed as combined (males and females, left) male (middle) and female (right) cohorts. Animal numbers are provided for the specific groups in each graph and analyzed with two-sided log-rank test. \* *p* ≤ 0.05; \*\* *p* ≤ 0.01; ns = not significant. (**D**) Percentage distribution of solid malignancy-related cause of death by CXB and SUL treated NPcis mice compared to controls, analyzed as combined (males and females, left) male (middle) and female (right) cohorts.

#### *3.5. MBZ Is More E*ff*ective than Combined MBZ with CXB*

Combined treatment with MBZ and CXB significantly increased median survival in NPcis mice compared to controls. However, the observed overall survival benefit appeared inferior to the effect achieved by MBZ or CXB alone, however, the difference is not statistically significant (Figure 6A). When investigating gender-specific effects, we found that dual use of MBZ and CXB in female NPcis mice was successful in delaying solid cancer occurrence and substantially enhancing the median survival beyond what was achieved by each agen<sup>t</sup> alone and untreated controls (Figure 6B). This stands in contrast to male mice with solid malignancies, who did not experience any additional survival benefits from the combination treatment in comparison to single agen<sup>t</sup> MBZ or CXB (Figure 6B). Figure 6C demonstrated that the combination therapy of MBZ and CXB resulted in an earlier mortality from non-solid cancer-related causes, particularly for male mice, indicating possibly the presence of toxicity, which we had assessed beforehand for each agen<sup>t</sup> separately but not in combination (Figure 6C). However, the number of mice who died in the MBZ/CXB cohort due to other, non-solid malignancy-related and unknown causes, were small and thus, limiting our ability to conclusively interpret these results. When analyzing cause of death in MBZ/CXB-treated mice, we observed a reduction in solid cancer-related causes in comparison to the controls, as expected, which was largely comparable with what was seen with single agen<sup>t</sup> use (Figure 6D).

**Figure 6.** Combination of MBZ and CXB enhances survival in female NPcis mice. Shown are the Kaplan–Meier curves for NPcis mice treated with combined MBZ and CXB, initiated 60 days after birth, in comparison to CXB and MBZ alone for (**A**) overall survival, (**B**) solid malignancy-related mortality and (**C**) other, non-solid malignancy-related and unknown causes of mortality, analyzed as combined (males and females, left) male (middle) and female (right) cohorts. Animal numbers are provided for the specific groups in each graph and analyzed with two-sided log-rank test. \* *p* ≤ 0.05; \*\* *p* ≤ 0.01; ns = not significant. (**D**) Percentage distribution of solid malignancy-related cause of death by combined MBZ/CXB treated NPcis mice compared to MBZ, CXB and controls, analyzed as combined (males and females, left) male (middle) and female (right) cohorts.
