INTRODUCTION
Neurofibromatosis type 1 (NF1) is one of the most common autosomal dominant neurological disorders and results in a multifaceted phenotype which includes: bone dysplasia, learning disabilities, benign nerve sheath tumors, and malignant tumors.  NF1 is caused by pathogenic variants in the neurofibromin (NF1) gene, and the clinical phenotypes of individuals with NF1 can vary widely, even among family members with the same genotype. NF1 functions as a GTPase-Activating Protein (GAP), simulating conversion of active Ras-GTP to the inactive form, Ras-GDP (Figure 1).  Almost 2,900 pathogenic variants have been reported in the Human Gene Mutation Database (http://www.hgmd.org); most lead to lack of expression of the NF1 gene product. NF1 does not exhibit a mutagenic hotspot, with variants occurring throughout the gene. A subset (17%) of these variants consists of missense (MS) variants which may result in an unstable or dysfunctional protein (Koczkowska et al., 2019).  MS variants in NF1 are not confined to any specific region of the gene. Some occur in the GAP-related domain (GRD) and might be expected to interfere with GAP function. MS variants occurring outside of this domain may result in other NF1 dysfunction, including protein instability, cellular mis-localization, or the disruption of neurofibromin interaction with other proteins in the cell. Another subset of variants (up to 29% (Kang et al., 2020)) includes nonsense (NS) variants, which result in premature termination of translation, and, in most cases, nonsense-mediated decay of the transcript.
The most critical role of neurofibromin appears to be the regulation of Ras signaling, with loss of neurofibromin function resulting in increased signaling. NF1 tumor cells (e.g., Schwann cells in neurofibromas) exhibit increased Ras signaling as a consequence of loss of function of both NF1 alleles (one in the germ line and one somatically-acquired (Figure 1; right) (Cichowski & Jacks, 2001). Therapeutic interventions to date have focused on inhibition of upregulated Ras signaling (e.g., MEK inhibition with selumetinib (Figure 1, right)). While MEK inhibitors have demonstrated effectiveness, not all patients benefit, plexiform neurofibromas do not completely disappear, and there can be significant side effects (Baldo, Magnolato, Barbi, & Bruno, 2021). Therefore, additional treatments that can be used in conjunction with MEK inhibitors are needed.
Only a small portion of the NF1 protein, the GRD, directly interacts with Ras. Some of the key NF1 residues involved in the NF1-Ras interaction include R1276 and K1423 (Yan et al., 2020). R1276, a highly conserved arginine residue, also termed the arginine “finger”, aids in the stabilization of the GTP/GDP-NF1 complex (Scheffzek et al., 1997). Mutation of this residue has little effect on Ras binding, but results in complete loss of GAP activity (Ahmadian, Stege, Scheffzek, & Wittinghofer, 1997; Sermon, Lowe, Strom, & Eccleston, 1998). NF1 K1423 forms a salt bridge with Ras D38 to stabilize the protein-protein interaction. In addition to dysregulated Ras signaling due to variants located in the GRD, interference with neurofibromin’s localization to the cytoplasmic membrane can result in abnormal interaction with Ras (Stowe et al., 2012). MS variants that occur within the SPRED1-binding domain of NF1 can reduce the affinity for SPRED1 such that neurofibromin fails to traffic to the membrane (Dunzendorfer-Matt, Mercado, Maly, McCormick, & Scheffzek, 2016; Hirata et al., 2016; Yan et al., 2020). However, some pathogenic NF1 variants fail to co-immunoprecipiate with SPRED1, do not directly interfere with Ras binding, and do retain GAP activity (Dunzendorfer-Matt et al., 2016).
An additional complexity to NF1 interactions is its dimerization activity (Figure 1, left) (Carnes, Kesterson, Korf, Mobley, & Wallis, 2019; Sherekar et al., 2019). While the mechanistic significance has yet to be determined, dimerization provides a potential explanation for the phenotypes observed with many NF1 disease variants. Often heterozygous NS or frameshift (FS) variants are observed in NF1, where there is still a single copy of wild type (WT) NF1 allele present. The total amount of full-length neurofibromin in some affected individuals with these variants may be considerably less than the predicted 50% of WT levels (Anastasaki, Woo, Messiaen, & Gutmann, 2015). Neurofibromin levels could be drastically lowered if mutant protein dimerizes with WT and then is targeted to the proteasome for degradation (Figure 1). The ubiquitin-proteasome pathway (UPP) controls NF1 levels and both the amplitude and duration of Ras-mediated signaling (Cichowski, Santiago, Jardim, Johnson, & Jacks, 2003). Excessive proteasomal degradation and genetic loss results in NF1 inactivation in sporadic gliomas (McGillicuddy et al., 2009). Proteasomal degradation of NF1 is partially regulated by the binding of both the SAG-SKP1-CUL1-FBXW7 and RBX1/2-CUL3-KBTBD7 complexes with NF1 (Figure 1; left) (Hollstein & Cichowski, 2013; Tan et al., 2011). The CUL3/KBTBD7 complex has been implicated in the pathogenic destabilization of neurofibromin in glioblastomas (Hollstein & Cichowski, 2013).
Here, we further validate our heterologous mNf1 cDNA expression system and multiple assays for neurofibromin function and use them to evaluate patient-specific variants (Wallis et al., 2018). We have included many MS variants with known genotype-phenotype correlations to assess their functional effects on NF1 levels as well as Ras signaling. We demonstrate that effects on stability and Ras signaling can be mutually exclusive functions. We suggest that such stratification of variant effects will have implications for mutation-targeted therapeutics for NF1, neurofibromin-driven breast cancers, and potentially other phenotypes. These assays may also have utility in classifying newly identified variants of unknown significance. As new Ras-independent functions are discovered for neurofibromin, it will be important to assay variants for their effects in new functional assays.