ABSTRACT: To discover RBPs with increased insolubility in a human ALS model, we applied a well established dual-SMAD inhibition-based protocol (Fang et al., 2019; Markmiller et al., 2021; Markmiller et al., 2018; Martinez et al., 2016) to generate iPSC-MN from six control iPSC lines, from four iPSC lines originating from two sALS patients, and from two iPSC lines originating from fALS patients with pathogenic variants in the TARDBP gene (Table S1; Figure S1A). No difference in differentiation capacity was observed (Figure S1B-G), resulting in average 40% ISL1+ MN (Figures S1G), comparable to numbers observed in large scale MN differentation studies (Baxi et al., 2022). The susceptibility of ALS MN to sodium arsenite-induced stress was not changed (Figure S1H and I). Next, we asked which proteins exhibit an increased insolubility in our ALS iPSC-MN. We fractioned iPSC- MN by lysis in radio-immunoprecipitation assay (RIPA) buffer, followed by ultracentrifugation and solubilization of RIPA insoluble proteins in urea buffer. The ultracentrifugation-cleared RIPA insoluble protein fraction is widely used to study protein insolubility in the context of neurodegeneration (Nuber et al., 2013; Walker et al., 2015). Label-free mass spectrometry of the insoluble protein fraction was utilized to identify proteins that are insoluble in sALS and fALS, relative to control iPSC-MNs (Figure 1A). Gene ontology (GO) analysis of the 100 proteins (top 2.9% of all detected proteins) with the highest label free quantification (LFQ) intensities in controls (Figure S1J) revealed that ‘unfolded protein binding’ (corrected P = 7.95 x 10-16) and ‘structural constituent of cytoskeleton’ (corrected P = 1.47 x 10-10) were among the 10 most significantly enriched GO terms, indicating enrichment of insoluble proteins (Figure S1K). Principle component analysis of the insoluble fractions did not distinguish ALS from control samples, suggesting that the overall insoluble proteome is not changed (Figure S1L). At threshold P ≤ 0.05 (Welch’s t-test) and fold change ≥ 1.5, we identified 88 proteins enriched in the insoluble fraction in ALS samples relative to control (Figure 1B). When the sample labels were randomly shuffled, we observed an average of 7.5 proteins (~12-fold lower) as differentially enriched at the same statistical thresholds, indicative of an ALS-specific protein insolubility pattern (Figure 1C). The 88 candidate proteins included cytoskeletal components and motor proteins, functional categories associated with prominent ALS in vitro phenotypes (Akiyama et al., 2019; Egawa et al., 2012; Fazal et al., 2021; Guo et al., 2017; Kreiter et al., 2018) (Figure 1D). Notably, 5 RBPs, NOVA1, ELAVL4, FXR2, RBFOX2, and RBFOX3 were also enriched (Figure 1D). The NOVA1 paralog NOVA2 was significantly enriched (P = 0.03) but did not meet our enrichment threshold (fold change = 1.36). Interestingly, insoluble TDP-43 protein was not significantly different in ALS and control (P = 0.98; fold change = 0.97). Western blot analysis confirmed the increase in insolubility of NOVA1, NOVA2, ELAVL4, RBFOX2 and RBFOX3 (Figure 1E and 1F). The soluble protein levels of NOVA1 and NOVA2 were also increased (Figure 1F). In conclusion, we identified 5 RBPs with elevated insoluble protein levels of ALS-iPSC-MNs.
Karsten Boldt 
