MAIN 44Amyotrophic Lateral Sclerosis (ALS) is a fatal motor neuron degenerative disease 1-3 characterized 45 by a progressive loss of motor neurons in the brain, brainstem and spinal cord 1 . Most cases are 46 sporadic (sALS), although up to 10 % are familial (fALS) 2 . The first gene associated with ALS (SOD1) 47 was described in 1993 4 . Since then, over 25 genes have been associated with fALS, sALS or both, 48 with the four most common being C9ORF72, SOD1, FUS, and TARDBP 2,5 . Notably, mutations in 49FUS are linked to both fALS and sALS, and present some of the earliest and more rapidly 50 progressive forms of the disease [6][7][8] . FUS is also involved in other neurodegenerative diseases, such 51 as frontotemporal dementia (FUS-FTLD), all characterized by FUS cytoplasmic inclusions 9 . 52 FUS is a nuclear RNA-binding protein, with a C-terminal nuclear localization signal (NLS) 9,10 . The 53 protein is involved in RNA processing, DNA repair and cellular proliferation, although its functions 54 are not precisely elucidated 9 . Most of the mutations associated with neurodegenerative diseases 55 alter FUS NLS [11][12][13] . More recently, mutations in FUS 3'UTR were described in ALS patients and 56 linked to an increased level of FUS mRNA and protein [14][15][16] . Surprisingly, over-expression of wild-57 type FUS provokes an aggressive ALS phenotype in mice and fruit flies, in accordance with findings 58 in yeast and mammalian cells [17][18][19][20][21] . The mechanism of the wild-type or ALS-linked mutated FUS 59 toxicity remains unclear 3,9,22 . 60With currently non-annotated proteins being increasingly reported 23-26 , we hypothesized that the 61 toxicity resulting from wild-type FUS over-expression may come from another, unseen, actor 26 . 62These novel proteins are coded by alternative open reading frames (altORFs) that are located 63 within "non-coding" RNAs (ncRNA), within the 5' or 3' "untranslated" regions (UTR) of mRNAs, or 64 overlapping a known coding sequence (CDS) within a different frame of an mRNA 25-27 . 65 Serendipitous discoveries and ribosome profiling have recently highlighted the distribution of 66 4 altORFs throughout the human genome, and the consequences of their absence from current 67 databases 26 . For example, mass spectrometry-based proteomics has become the gold standard 68 for protein identification and has been extensively used in ALS studies 28,29 . However, if a protein 69 is not annotated, it is not included in the protein database (e.g. UniProtKB), and thus cannot be 70 detected by mass spectrometry. An estimated 50 % of mass spectra from a proteomics 71 experiment are unmatched at the end of the analysis 26,30 . 72 Genome annotations must avoid spurious ORF annotations. Thus unless functional 73 characterization has been published, they rely upon 2 arbitrary criteria: a minimum length of 100 74 codons and a single ORF per transcript. Several groups developed tools to challenge such criteria, 75 such as the sORFs repository 31 and the OpenProt database 27 , which ...