The exact molecular pathogenesis of malignant transformation of benign tumor tissues to MPNSTs in NF1 patients is poorly understood. Since haploinsufficiency of neurofibromin activity, by NF1 mutation exists in all the cells of the NF1 patient, even in the normally functional cells, it has been suggested that the additional genetic or epigenetic changes may participate in malignant development of benign tumor tissues to MPNSTs as well as in tumorigenesis of NF1 (Dasgupta and Gutmann, 2003). Somatic loss of heterozygosity (LOH) at the NF1 locus and genomic imbalances in chromosomes 17, 19 and 22q have been mainly described to be responsible for this tumor development (Legius et al., 1993; Koga et al., 2002). Furthermore, alterations like mutations and/or gene expression changes in many genes such as CD44, CDKN2A, EGFR, PTEN, RB1, SOX9, TP53, etc. have also been
accumulation of additional loss-of-function mutations in these tumor suppressor genes in NF1-/- SCs may be required for MPNST pathogenesis. In particular, a large-scale comparison between human MPNST SCs and normal SCs revealed a relative down-regulation of the SC differentiation markers SOX10, CNP, PMP22, and nerve growth factor receptor and relative
up-regulation of the neural crest stem cell markers SOX9 and TWIST1 in MPNST SCs (Miller et al., 2006).
Furthermore, post-transcriptional modification by microRNAs has also been recently studied in NF1, and the results showed up-regulation of miR-10b (Chai et al., 2010) and down-regulation of miR34a whose expression is mediated by p53 (Subramanian et al., 2010) in MPNST cells or tissues.
Mice model studies have reported that double-hit inactivation of the Nf1 gene (Nf1-/-) in SCs leads to form benign tumors (Brossier and Carroll, 2012). Although NF1 LOH has been identified in benign neurofibromas (Sawada et al., 1996; Kluwe et al., 1999b;
Rasmussen et al., 2000; Wiest et al., 2003), a much higher frequency of NF1 LOH (>4 times) has been observed in MPNSTs than that in neurofibromas of patients with NF1 (Upadhyaya et al., 2008). The interaction between NF1-/- SCs and other types of NF1+/- cells, including fibroblasts, mast cells, and perineurial cells and elevated expression of stem cell factors in NF1-/- SCs in the tumor microenvironment have been implicated in tumor progression of PNs to MPNSTs (Le and Parada, 2007; Jouhilahti et al., 2011). In addition, NF1 inactivation in SCs in mice was reported to promote the development and malignant progression of PNs (Keng et al., 2012). These results strongly indicate that bi-allelic inactivation of the NF1 gene in SCs plays a crucial role in NF1 tumorigenesis and MPNST pathogenesis, and suggest that other genetic alterations in SCs and other types of cells in neural crest-derived tissues may be required for tumors to progress from PNs to MPNSTs.
Studies have frequently noted the hyperexpression of EGFR in NF1-associated MPNST tissues and suggested that the upregulation of EGFR in SCs may play a key role in
tumorigenesis and tumor progression of NF1 (DeClue et al., 2000; Carroll, 2012).
Furthermore, increased and extended EGFR-dependent progenitor cells have caused tumorigenesis in NF1-deficient mice (Fig. 5) (Williams et al., 2008). The EGFR is a member of the ErbB family of receptors, a subfamily of four closely related receptor tyrosine kinases, EGFR (ERBB1), HER2/c-neu (ERBB2), Her 3 (ERBB3), and Her 4 (ERBB4) (Cooper et al., 2011). Upregulated EGFR stimulate the downstream activation of the Ras-mitogen-activated protein kinase (MAPK) cascade (Karnoub and Weinberg, 2008). Hence, upregulation of the EGFR expression in MPNSTs is thought to be a major factor for the malignant progression from benign PNs to MPNSTs (Dasgupta and Gutmann, 2003). Despite robust studies that have been carried out to find the clues explaining the upregulation of EGFR in MPNSTs, the molecular mechanisms have yet to be elucidated.
Crucially, I found that the expression level of EGFR protein was inversely related to the expression level of neurofibromin protein in the SCs and primary cells tested (Figs. 7 and 8). Particularly, the EGFR expression level was exactly correlated with neurofibromin expression levels. The finding of the neurofibromin-mediated transcriptional expression of EGFR (Fig. 8) led to focusing my further study on Sp1, the known transcription factor of EGFR gene. The site-specific chromatin immunoprecipitation targeting the pSp1-binding site
of EGFR gene in the NF1-depleted cells showed the direct evidence that neurofibromin itself participates in the Sp1-mediated EGFR transcriptional regulation (Fig. 13). Finally, I found that the Ras/Erk/Sp1-signaling pathway mediates EGFR upregulation in MPNST cells (Figs.
11–14). These results indicate that the neurofibromin deficiency-mediated EGFR increase is a crucial factor in MPNST pathogenesis.
Although an intact NF1 gene dose is critical for EGFR expression (Fig. 7A, B), the neurofibromin level also influenced the EGFR overexpression (Fig. 7C, D). These results suggest that alteration in EGFR expression level may be caused by somatic expression changes in the intact NF1 locus and not by somatic NF1 mutation. The reduced expression of neurofibromin in the MPNSTs having one normal NF1 locus has already been reported (Basu et al., 1992). However, no hypermethylation of the NF1 gene have been detected in MPNSTs (Harder et al., 2004). We also investigated the methylation levels in the promoter region of the NF1 gene in the primary MPNST and normal cells by DNA methylation chip analysis using the GoldenGate Methylation Cancer Panel I (Illumina) but did not detect hypermethylation in both the cell types (data not shown).
These results cannot clearly explain the reason why the complete loss of NF1 gene (NF1-/-) in SCs is not sufficient to development of MPNSTs from PNs (Brossier and Carroll, 2012), suggesting the presence of other genetic factors that may be associated with EGFR regulation or other molecules involving the MPNST pathogenesis. In this study, I suggest the anti-apoptotic Bcl-xL protein as another key candidate molecule contributing the MPNST development of PNs in addition to EGFR. The role of Ras-signaling pathway has been closely implicated in malignant transformation and drug resistance in many types of cancers (McCubrey et al., 2006). Notably, NF1 deficiency promotes carcinogenesis by inducing heat shock factor 1 (HSF1), which is mediated by aberrant Ras/MAPK signaling (Dai et al., 2012).
HSF1 overexpression and activation are observed in NF1-deficient MPNST cells and tumor resections from patients with NF1(Dai et al., 2012). Intriguingly, NF1 deficiency contributes to the epithelial-mesenchymal transition (EMT) in NF1 (Arima et al., 2010). In that study,
the expression levels of the EMT-related transcription factors Snail, Twist, and ZEB1 were significantly up-regulated in the sNF96.2 MPNST SCs compared with those in normal HSCs.
The EMT is involved in cancer metastasis via the Ras/MAPK signaling pathway (Edme et al., 2002).
Therefore, I investigated if NF1 deficiency was directly involved in Bcl-xL upregulation in MPNST cells and found that Bcl-xL was significantly upregulated in the tested MPNST cells. I also found that the elevated Bcl-xL expression was closely linked to the increased resistance to anticancer drugs in MPNST cells (Figs. 17–20). I further demonstrated that the decreased transcriptional expression of the NF1 gene caused an overexpression of Bcl-xL in MPNST cells (Fig.17), as in the neurofibromin-mediated EGFR expression. Finally, I demonstrated that the Ras/Erk/Sp1-signaling pathway also mediated the Bcl-xL upregulation in MPNST cells (Fig. 21).
Since STAT3, Ets and NFκB, as well as Sp1, the downstream proteins in the Ras-signaling pathway, are well known as the main transcriptional factors for the BCL2L1 gene (Sevilla et al., 2001; Lee et al., 2009), further study will be needed to investigate whether deficiency in neurofibromin can cause the promotion of STAT3, Ets, and NFκB expression.
Although the molecular mechanisms for somatic loss of NF1 in the MPNST cells have not been elucidated, loss of neurofibromin may directly contribute to Bcl-xL overexpression through the activation of the pathway and may further contribute to malignant development of benign tumor tissues or normal tissues to MPNSTs.
All of the results suggest that neurofibromin is critical for the regulation of EGFR and Bcl-xL expression via a transcription-regulating function. It is also critical for the regulation
of Ras activation via a GTPase-activating function. I found the hyperexpression of both EGFR and Bcl-xL, not only in MPNST cells but also in MPNST tumor tissues from patients with NF1 (Fig. 22). Taken together, my results demonstrate that neurofibromin-mediated EGFR and Bcl-xL expression are controlled at the transcriptional level via the Ras/MAPK signaling pathway. Nf1 deficiency-mediated Ras activation has been identified in a subpopulation of SCs (Nf1-/-) but not fibroblasts (Nf1-/-) in mice with neurofibromas (Sherman et al., 2000). My results suggest that a neurofibromin deficiency in SCs caused by bi-allelic inactivation at the NF1 locus enhanced Ras signaling, which consequently lead to the expression of EGFR and BCL2L1 transcription factors.
Unexpectedly, I also found that Bcl-xL, the downstream molecule of neurofibromin, reversely regulated the neurofibromin expression level. The manipulation of Bcl-xL expression in IMR90 cells revealed that Bcl-xL regulates the neurofibromin protein level (Fig. 23). In addition, I found that Bcl-xL regulates the ubiquitin-mediated proteolysis of neurofibromin by modulating the ETEA binding to neurofibromin (Fig. 24). Cichowski and colleagues reported that neurofibromin stability is regulated by ubiquitin-mediated proteolysis (Cichowski et al., 2003). Serum, EGF, PDGF, and LPA trigger neurofibromin degradation in IMR90 cells. Furthermore, ubiquitin-mediated neurofibromin degradation is regulated by ETEA, a negative regulator of neurofibromin that directly binds through the UBX domain of ETEA (Phan et al., 2010). Downregulation of Bcl-xL mediates the decreased binding of ETEA to neurofibromin, indicating that Bcl-xL regulates the proteolysis of neurofibromin via the modulation of ETEA binding to neurofibromin (Fig. 24).
These results demonstrate that Bcl-xL negatively regulates neurofibromin and positively
regulates EGFR by modulating the ubiquitin-mediated proteolysis of neurofibromin (Figs. 25 and 26). Similarly, the high expression of ERBB2 was reported in Chinese hamster ovary cells that overexpressed Bcl-xL (O'Connor et al., 2009).
Although surgical resection is the primary treatment for MPNSTs, its limitation due to tumor location and tumor multiplicity has led to the development of a drug treatment approach. The anti-apoptotic Bcl-2 family member proteins Bcl-2, Bcl-xL, Bcl-w, Mcl-1, Bfl1/A-1, and Bcl-B bind to and inactivate BH3-domain pro-apoptotic proteins (Kang and Reynolds, 2009). High expression levels of these proteins are found in various cancers and related to the development of chemoresistance in malignant tumor cells (Kostanova-Poliakova and Sabova, 2005; Kang and Reynolds, 2009; Karnak and Xu, 2010).
Chemoresistance in NF1-associated MPNSTs has been poorly discussed so far. Apoptotic Bcl-2 family proteins, Bcl-2, Bcl-xL, Mcl-l and Bcl-w, which contribute to tumorigenesis, tumor progression and tumor chemoresistance are known to be hyperexpressed in many cancers (Kirkin et al., 2004). Particularly, Bcl-xL overexpression is involved in general resistance to a large numbers of cytotoxic agents in human cancer cell lines (Amundson et al., 2000). Bcl-xL overexpression contributes to block the action of many chemotherapeutic drugs (Kostanova-Poliakova and Sabova, 2005; Karnak and Xu, 2010). Therefore, Bcl-xL has been studied as one of the promising targets for overcoming drug resistance by enhancing apoptosis in malignant tumor cells.
In this study, I found that increased Bcl-xL expression, but not that of Bcl-2 or Mcl-1, caused an increase in resistance to Doxorubicin in MPNST cells (Figs. 17-19). By manipulating Bcl-xL expression levels, it was demonstrated that reduced apoptosis
sensitivity of MPNST cells was caused by Bcl-xL overexpression (Fig. 20). Reducing Bcl-xL expression restored apoptosis sensitivity to Doxorubicin in MPNST cells (Fig. 20), leading to the suggestion of a reasonable therapeutic strategy for patients with NF1 and MPNSTs through increased chemosensitization of MPNST cells by modulating Bcl-xL expression level. Inhibition of Bcl-xL by antisense olignucleotides or siRNAs significantly enhanced chemosensitivity to Cisplatin (Littlejohn et al., 2008; Brotin et al., 2010). In addition, various non-peptidic small molecule inhibitors against Bcl-2 family proteins have been developed, and preclinical or clinical trials for various cancer therapies have also been performed (Azmi and Mohammad, 2009).
Among them, ABT-737, a mimetic of the BH3-only protein BAD, and its modified form ABT-263 are drawing attention as good candidates to selectively target cancer cells and are in the phase I/II of clinical trials for various cancer therapies (Richardson and Kaye, 2008). ABT-737 selectively inhibits Bcl-2, Bcl-xL and Bcl-w and synergizes with conventional chemotherapeutic drugs to promote apoptosis in multiple cancer types (Ni Chonghaile and Letai, 2008; Richardson and Kaye, 2008; Reynoso et al., 2011). ABT-737 has been demonstrated to enhance synergistic chemosensitivity when used in combination with Doxorubicin in other MPNST cells (Lee et al., 2012), chondrosarcoma cells (van Oosterwijk et al., 2012) and hepatoblastoma cells (Lieber et al., 2010). I thus tested apoptosis sensitivity of sNF96.2 MPNST SCs by combined treatment of ABT-737 and Doxorubicin. As a result, ABT-737 synergistically enhanced apoptosis sensitivity to Doxorubicin in MPNST cells (Fig. 27). A very low dose of ABT-737 enhanced the cytotoxic effect of Doxorubicin.
Noticeably, the concentrations of both ABT-737 and Doxorubicin required for effective
apoptotic cell death were much lower in sNF96.2 cells than those in sNF02.2 cells (Fig 28A).
sNF96.2 SCs are a NF1 LOH strain (NF1-/-), whereas sNF02.2 SCs harbor one intact NF1 allele (NF1+/-), suggesting that combining ABT-737 and Doxorubicin may increase the additive effects of the combined treatment in NF1-/- SCs. Considering that NF1-/- SCs play a major role in MPNST pathogenesis in NF1 (Le and Parada, 2007; Carroll, 2012) and a high frequency of NF1-/- SCs is detected in NF1-associated MPNST tissues (Upadhyaya et al., 2008), ABT-737 and Doxorubicin seem to be a good combination to effectively treat NF1-associated MPNSTs with minimal side-effects.
Previously, many cancer cell types have been proven to be refractory to ABT-737 because of high expression of Mcl-1 (van Delft et al., 2006; Touzeau et al., 2011). However, the tested MPNST cells did not exhibit a higher Mcl-1 expression level compared to the benign and normal cells (Figs. 17 and 19), thereby suggesting a beneficial effect of ABT-737 in NF1-associated MPNSTs. Doxorubicin is a clinically used anticancer drug that functions as topoisomerase II inhibitor and forms covalent DNA adducts. In fact, Doxorubicin has been recently studied as a chemotherapeutic agent in combination with ifosfamide in patients with NF1-associated MPNSTs (Kroep et al., 2011; Moretti et al., 2011) and in combination with ABT-737 in promyelocytic leukemia and chondrosarcoma cells (Ugarenko et al., 2010;
van Oosterwijk et al., 2011). My results suggest that the combination of ABT-737 and Doxorubicin is very effective in enhancing chemotherapeutic sensitivity in the NF1-associated MPNST cells.
The proteins involved in the EGFR/Ras signaling and mTOR pathways have been the main chemotherapeutic targets for MPNSTs (Gottfried et al., 2006; Gottfried et al., 2010). In
my results, EGFR and Bcl-xL were upregulated in MPNST cells. Thus, EGFR may be good candidate targets for drug treatment for MPNSTs. Erlotinib, a low-molecular-weight inhibitor that binds to the kinase domain of EGFR and HER2, has already been implicated in NF1-associated MPNSTs (Dilworth et al., 2008; Holtkamp et al., 2008). I assessed the effect of Erlotinib on the cytotoxicity of MPNST cells when combined with Doxorubicin and found that Erlotinib could not enhance the apoptotic cell death induced by Doxorubicin (Fig. 28B).
Next, I tried the triple combination of Erlotinib, ABT-737, and Doxorubicin. Doxorubicin has a lot of advantages as an anticancer drug, but its clinical application is limited due to serious side effects (Chen et al., 2011). Furthermore, side effects in platelet function have recently been reported when using ABT-737 at a 10 μM dosage (Schoenwaelder and Jackson, 2012). Hence, I fixed the non-cytotoxic concentration of Doxorubicin and ABT-737 at 0.1 μg/ml and 0.1 μM, respectively, for the experiments (Fig. 28C). Erlotinib showed a notable synergistic cytotoxic effect in sNF96.2 MPNST cells in a dose-dependent manner (Fig. 28C).
These results suggest that the pharmacological inhibition of both EGFR and Bcl-xL in combination with anticancer drug inducing apoptosis may be a potential therapeutic strategy for the treatment of NF1-associated MPNSTs.
In conclusion, I have demonstrated that an NF1 deficiency-mediated elevation in the Ras/ERK/Sp1-signaling pathway caused the hyperexpression of EGFR and Bcl-xL in MPNST cells. Upregulation of Bcl-xL negatively regulates neurofibromin by modulating the ubiquitin-mediated proteolysis of neurofibromin and is closely associated with drug resistance in MPNST cells. EGFR inactivation by Erlotinib and Bcl-xL inactivation by ABT-737 in combination with Doxorubicin is a potential therapeutic strategy for the treatment of
the NF1-associated MPNSTs. Further studies are necessary to evaluate this combined therapy in preclinical models of MPNSTs. Consequently, this study will have a significant impact in the field of familial cancer and will be helpful for investigators studying chemotherapy and molecular mechanisms involved in the pathogenesis of NF1-associated MPNSTs.