DISCUSSION
cDNA model system. Since discovery of the NF1 gene in 1990 research efforts have been hindered by the lack of a full-length coding cDNA. This is partially due to the size of the gene and toxicity of the human cDNA construct. We are aware of three potentially availableNF1 cDNAs: mouse Nf1 (Wallis et al., 2018) (isoform 2 with 2839 amino acids), codon optimized human NF1 (Bonneau, Lenherr, Pena, Hart, & Scheffzek, 2009) (isoform 1 with 2818 aa), and a humanNF1 with mini-intron 35-36 (Cui & Morrison, 2019) (isoforms 1 and 2). As the mNf1 cDNA is highly homologous to endogenous humanNF1 , we have developed and validated the mouse Nf1 cDNA expression system that allows us to examine the biochemical effects of any Nf1 genetic variant. We have been able to perform dose-response studies to titrate in varying amounts of mNf1 cDNA and are able to detect a clear dose-response in terms of levels of neurofibromin and repression of GTP-Ras levels and pERK/ERK ratios, giving us confidence in both the cDNA and the functional assays.
Assay Performance: We see similar dynamic ranges between the GTP-Ras and pERK/ERK experiments, with WT cDNA able to repress both GTP-Ras and pERK/ERK ratios by about half that seen with EV control. The Morrison lab reported that inactive variants (R1276P) served as better controls than empty vectors (Cui & Morrison, 2019). While our R1276Q and R1276X both displayed similar GTP-Ras levels as our EV, both showed insignificantly higher pERK/ERK ratios. Indeed, we see significant variability in the pERK/ERK ratios, leading to large error bars. Regardless, each individual cDNA performs similarly between the two assays and across multiple experiments, indicating that the results are reliable. Each variant’s ability to affect Ras signaling relative to WT cDNA is consistent between assays; however, we have noted exceptions. delM992 and C379R both have significantly different GTP-Ras levels, whereas pERK/ERK ratios fail to reach significance; and R1748X has a significantly different pERK/ERK ratio, but GTP-Ras doesn’t reach significance.
Using our previously published cDNA expression system (Wallis et al., 2018) we can evaluate functional significance of variants in individuals with NF1. These data indicate that each variant has a slightly different functional profile in terms of both protein stability and the ability to inhibit Ras signaling. We incorporated non-pathogenic variants in our assays: E1327G, Q1336R, and P2782L. These three cDNAs performed similarly to the WT cDNA in terms of neurofibromin levels and the ability to inhibit Ras signaling, adding confidence to our functional profile characterizations. Some pathogenic variants result in instability of neurofibromin but still retain GRD function. The best examples of this are with the G848R and L1957P variants, which retain the ability to repress Ras activity (Figure 5) in both of our assays, yet are unstable and produce less than 50% of the neurofibromin that is observed with WT cDNA protein (Figure 4). Some variants result in stable protein but have lost GRD function. For example, K1423E and S1997R both produce neurofibromin levels similar to WT cDNA, yet display significantly elevated Ras activity compared with WT cDNA. We also observed that certain variants demonstrated both unstable protein and loss of function; L1490P and D1623G exhibited lowered neurofibromin levels, with inability to repress Ras signaling. Nonsense variants resulting in truncated proteins have variable stability. Nonsense variants with truncations after the GRD may maintain GRD function in these overexpression assays. While both R1947X and R2550X show increased Ras signaling, it is not statistically different from WT cDNA.