Novel somatic single nucleotide variants within the RNA binding protein hnRNP A1 in multiple sclerosis patients

Some somatic single nucleotide variants (SNVs) are thought to be pathogenic, leading to neurological disease. We hypothesized that heterogeneous nuclear ribonuclear protein A1 (hnRNP A1), an autoantigen associated with multiple sclerosis (MS) would contain SNVs. MS patients develop antibodies to hnRNP A1 293-304, an epitope within the M9 domain (AA 268-305) of hnRNP A1. M9 is hnRNP A1’s nucleocytoplasmic transport domain, which binds transportin-1 (TPNO-1) and allows for hnRNP A1’s transport into and out of the nucleus. Genomic DNA sequencing of M9 revealed nine novel SNVs that resulted in an amino acid substitution in MS patients that were not present in controls. SNVs occurred within the TPNO-1 binding domain (hnRNP A1 268-289) and the MS IgG epitope (hnRNP A1 293-304), within M9. In contrast to the nuclear localization of wild type (WT) hnRNP A1, mutant hnRNP A1 mis-localized to the cytoplasm, co-localized with stress granules and caused cellular apoptosis. Whilst WT hnRNP A1 bound TPNO-1, mutant hnRNP A1 showed reduced TPNO-1 binding. These data suggest SNVs in hnRNP A1 might contribute to pathogenesis of MS.


Introduction
Multiple sclerosis (MS) is the most common autoimmune disease of the central nervous system (CNS) in humans, whose pathogenesis remains unknown. A number of genetic and immune studies indicate dysregulated immune responses as contributors to the pathogenesis of MS [1][2][3][4][5][6][7] . Genetic analyses show an association of MS with major histocompatibility complex (MHC) Class II human leukocyte antigen (HLA)-DRB-1 and single nucleotide polymorphisms (SNPs) related to immune function 1,2,8 . Both Th1/Th17 CD4 + T-lymphocytes and immunoglobulins appear to have a causative role 1,2,9 . Immunoglobulin G (IgG) responses to myelin and non-myelin targets have differentiated some MS patients from healthy controls 9-11 . Nonmyelin antigens that are targets for immunoglobulins isolated from MS patients include neurofilaments, axonal neurofascin and RNA binding proteins, including heterogeneous nuclear ribonucleoprotein A1 (hnRNP A1) 9,12-16 .
Recently, mutations in RNA binding proteins have been shown to cause neurological disease [17][18][19][20][21] . For example, a mutation (p.D263V) in the prion-like domain (PrLD) of hnRNP A1 has been shown to cause familial amyotrophic lateral sclerosis (ALS) 22 . In addition to inherited mutations, somatic variants have also been shown to cause neurological disease 23 . hnRNP A1 performs a number of critical cellular functions related to transcription, nucleocytoplasmic transport of mRNA and translation 24,25 . In addition to the PrLD, other important functional domains in hnRNP A1 include two RNA binding domains (RBDs) and M9, its nucleocytoplasmic shuttling domain 22 . M9 binds its nuclear receptor, transportin-1 (TPNO-1, also known as karyopherin β2) and the hnRNP A1:TPNO-1 complex is transported into and out of the nucleus 3,9,16,26,27 .
Our lab has performed extensive studies on the role of autoimmunity to hnRNP A1 in MS and human T-lymphotropic virus type 1 (HTLV-1) associated myelopathy/tropical spastic paraparesis (HAM/TSP), a viral-induced model and clinical mimic of MS 3,28-30 . Initially, we discovered that HAM/TSP patients develop antibodies to hnRNP A1 that cross-react with HTLV-1-tax, indicative of molecular mimicry 29,31 . Next, the epitope of the HAM/TSP IgG response (AA 293-304 ) was localized to M9 (AA 268-305 ) 32 . M9 is a bipartite phenylalaninetyrosine nuclear localization sequence (PY-NLS) that requires binding to TPNO-1 for hnRNP A1 to shuttle between the nucleus and cytoplasm 16,31 Because of the similarities between MS and HAM/ TSP, we hypothesized that MS patients would also develop antibodies to hnRNP A1. In fact, antibodies isolated from MS patients, in contrast to healthy controls and Alzheimer's patients, were also found to immunoreact with the identical hnRNP-A1-M9 epitope (AA 293-304 ) 16 . Subsequent studies indicated that the IgG was biologically active and potentially pathogenic. For example, mono-specific antibodies to hnRNP A1 isolated from patients caused decreased neuronal firing using neuronal patch clamp in rat brain sections 31,33 . Further, neurons exposed to anti-hnRNP A1-M9 293-304 specific antibodies resulted in neurodegeneration and neuronal death 16,34 . The anti-hnRNP A1-M9 293-304 specific antibodies also caused changes in neuronal RNA expression that correlate with the clinical phenotype of MS and HAM/TSP patients (ie. spastic paraparesis), which was subsequently confirmed in neurons isolated from the brains of MS patients 16 . Additional studies showed that anti-hnRNP A1-M9 293-304 specific antibodies entered neurons via clathrin-mediated endocytosis and caused apoptosis in a neuronal cell line 34 . Anti-hnRNP A1-M9 293-304 specific antibodies also caused a redistribution of hnRNP A1 in neurons from nuclear to an equal distribution of nuclear and cytoplasmic localization, suggesting the antibodies interfered with M9, which is required for hnRNP A1s nuclear import 34 . Considering: (1) the role of hnRNP A1 in cellular function; (2) variant forms of hnRNP A1 cause neurodegenerative disease, and (3) hnRNP A1 is an autoimmune target in MS patients, we hypothesized that MS patients would contain novel genomic DNA single nucleotide variants (SNVs) in hnRNP A1-M9, which when expressed, would alter cellular function and contribute to cell death. Preparation of human peripheral blood monocytes (PBMCs) and isolation of genomic DNA Human PBMCs were isolated from fresh blood by Ficoll-Paque gradient centrifugation and washed with PBS. Genomic DNA was isolated from PBMCs using the QIAmp blood kit (Quiagen Inc., Chatsworth, CA, U.S.A.) according to manufacturer's protocol. All DNA samples were quantified using Nanodrop (Quawell) and restriction enzyme digestion methods.
PCR amplification and subcloning One microgram of genomic DNA was amplified in a reaction mixture containing the primers and KOD Hot Start DNA polymerase (Novagen). Use of this DNA polymerase has a mutation frequency of 0.10% 36 . Before adding enzyme, the reaction mixture was heated at 95°C for 2 minutes. Amplification was carried out for 35 cycles of denaturation at 95°C for 20 s, annealing at 57°C for 10 s, and extension at 72°C for 15 s, followed by terminal elongation at 70°C for 20 s. The resulting PCR product was cloned into the pCR2.1-TOPO

Amendments from Version 1
We appreciate the reviewer's comments and have addressed them in the revised manuscript. None of the reviewers required further experiments. The reviewers requested we include data indicating that SNVs may be the result of oxidative stress or clonal expansion of T-cells. They also suggested that SNVs in CNS targets other than neurons might be important in the pathogenesis of MS and that future studies should include more genes and control groups. All of these suggestions were specifically addressed in the discussion.   GST pull-down assay SK-N-SH cells were cultured in Dulbecco's Modified Eagle's medium (BD Biosciences) supplemented with 10% fetal bovine serum, 100 U/mL penicillin G and 100 µg/mL streptomycin, at 37°C under 5% CO 2 . Cells were harvested and lysed with CytoBuster™ Protein Extraction Reagent (Millipore), containing inhibitor cocktail, homogenized for a few seconds with a handheld homogenizer and spun at 16,000 × g for 5 minutes. Supernatants were used for GSTpull down assays. Glutathione-Sepharose 4B beads coupled with GST-hnRNP A1 (WT or variant), which includes the Transportin 1-binding domain, were incubated for 1 h at 4°C with 600 µL of the cell lysates in CytoBuster™ Protein Extraction Reagent and protease inhibitors. After washing the beads three times with 600 µL of 10 mM PBS (10 mM Na 2 HPO 4 , 140 mM NaCl, 2.7 mM KCl, 1.8 mM KH 2 PO 4 , pH 7.4) and protease inhibitors, proteins bound to the beads were analyzed by 8-16% SDS-PAGE followed by immunoblotting with rabbit polyclonal GST antibody (1:1000, Millipore, catalog #06-332), mouse monoclonal Transportin 1 antibody (1:1000, Millipore, catalog #05-1515) and mouse monoclonal TDP-43 antibody(1:1000, Millipore, catalog #MABN45). The immunoreactive bands were visualized using enhanced chemiluminescence. Of the six HCs, zero SNVs resulted in an amino acid substitution within the TPNO-1 binding domain, MS IgG epitope or M9 from the 481 clones that were sequenced (Table 1, Supplement 2, Supplement 3). One individual had a likely benign variant which did not result in a change in the associated amino acid (c.900A>G, p.R300R) (Supplement 2), and three others had SNVs in the C-terminal region
In the in-vitro experiments, SG formation in SK-N-SH cells formed within several hours of transfection. When we waited overnight (approximately 24 hours) the cells containing mutant hnRNP A1 developed apoptotic blebs, which contained hnRNP A1 ( Figure 2C, arrows). Apoptosis was confirmed by active caspase-3 staining ( Figure 2D). As shown in Figure 2D, SK-N-SH cells transfected with mutant hnRNP A1 showed a cytoplasmic hnRNP A1 distribution, stained positive for active caspase-3 and contained fragment nuclei, confirming apoptosis.
In summary, in contrast to WT hnRNP A1, mutant hnRNP A1 showed markedly reduced binding to its co-receptor TPNO-1, co-localized with TDP-43 within cytoplasmic SGs of cells and caused apoptosis, indicative of the potential pathogenic nature of these diseaseassociated SNVs in MS patients.

Discussion
Recent studies indicate that in addition to cancer, somatic variants can cause neurological disease 23 . In this study, we discovered novel somatic genomic DNA SNVs in MS patients. Nine were contained within the 'core' TPNO-1 binding domain of hnRNP A1-M9 (AA 268-289 ). Three additional SNVs (c.793A>G, p.N265D (in two patients); c.787T>C, p.F263L) included amino acids within the PrLD -M9 overlap region (AA 263-267 ), which also bind TPNO-1 45 . These variants were in a region of hnRNP A1 that are adjacent to mutations shown to cause ALS (p.D262V, p.N267A). Interestingly, 8 of these 12 SNV's that involved hnRNP A1-M9 binding to TPNO-1 occurred in PPMS patients. In addition, two hnRNP A1 SNVs were contained exclusively within the PrLD (c.755G>A, p.S252N; c.775A>G, p.S259G). There were also six novel SNVs that resulted in an amino acid substitution within the MS IgG epitope of M9 (AA 293-304 ), five of which segregated to patients with SPMS. Finally, there were nine SNV's in the C-terminal of hnRNP A1 (AA 306-320 ), occurring with similar frequency in HCs and MS patients. The overall somatic SNV rate (based on the number of clones sequenced) for the M9 target sequence was: PPMS -2.21%, SPMS -1.69%, RRMS -0.56%, HC -0%. If one includes the PrLD (a domain shown to be critical to hnRNP A1 function), the rates increase in PPMS, SPMS and RRMS to 2.84%, 2.25% and 0.84% respectively. None were identical to somatic mutations in the COSMIC database (n = 981,720 samples, n = 1,292,597 unique variants). We utilized a PCR -cloning technique that has been finetuned for more than a decade and shows a mutation rate of approximately 0.1% in more than 46,000 clones that were examined 36 . The rates in progressive MS patients exceed this error rate by more than a log. In addition, under identical conditions, there were no mutations in the M9 target sequence or the PrLD domain in the HCs we examined. Thus, these results are unlikely to be due to PCR errors. Importantly, there was little or no overlap with either SNVs or SNPs reported in four different databases.

Figure 2. Transfection of WT and mutant forms of hnRNP A1 into SK-N-SH cells. A. Localization of WT and
hnRNP A1 was one of the first RNA binding proteins shown to shuttle into and out of the nucleus 37,38 . Nucleocytoplasmic transport is dependent upon binding between the M9 domain (AA 268-305 ) of hnRNP A1 to TPNO-1, in order for this complex to pass through the nuclear pore. M9 acts as both an NES and NLS. M9 is a bipartite PY-NLS whose three-dimensional structure and binding contacts with TPNO-1 are well characterized 26,27,43 . Specifically, M9 contains three binding epitopes ( Importantly, none of the SNVs contained within hnRNP A1 -M9 or the PrLD has been reported previously. For decades, the only certain genetic risk factor for MS was with MHC Class II HLA-DRB1 8 . Genome Wide Association Studies (GWAS) have uncovered novel genetic associations with MS 1,2,45 including with the interleukin-2 receptor-α and interleukin-7 receptor genes 45 . Subsequent studies using several thousand MS cases and controls, which analyzed hundreds of thousands of autosomal SNPs, confirmed the association of MS with major MHC Class II HLA-DRB1 (DRB1 *15:01, *15:03, *13:03) and the protective effect MHC Class I HLA-A*02 1,8 . Additional studies showed a total of 48 new and 49 known non-MHC SNPs associated with MS 2 . Interestingly, the functions of the vast majority of the SNP's were related to CD4 + T-lymphocyte and immune regulation 1,8 . This is important, considering the role that T-lymphocytes and the immune response play in the pathogenesis of MS. A few were potentially associated with neurodegeneration 46 . Further, >95% of the SNPs were intronic or intergenic, with only a few SNPs involving exons, in contrast to the somatic SNVs discovered here 1,2 . In addition to GWAS, whole exome sequencing (WES) is being used to examine differential gene expression in MS patients. In contrast to GWAS, which detects known SNPs and utilizes statistical analyses designed to reveal common variants, WES is designed to discover novel, rare pathologic variants 8 . One of the genes identified by GWAS was CYP27B1, which encodes an enzyme of the same name that converts 25-hydroxyvitamin D to 1,25 hydroxyvitamin D, the biologically active form of vitamin D 1,8  This manuscript by Lee and Levin continues work by Levin on the role of an immune response to hnRNP-A1 in human demyelinating disorders, including multiple sclerosis. Previously, Levin had shown that an antibody response to HTLV-I Tax cross-reacted with hnRNP-A1. Now, they go on to show that there are mutations in hnRNP-A1 that affect the localization of this protein and how it could affect neuronal survival in multiple sclerosis.In general, the experiments appear to be well performed. One might be interested to know whether testing in another cell line such as a oligodendroglial cell line would have similar effects on cell survival as in the neuroblastoma cell line. This could also be highly relevant in MS, as in addition to neuronal loss, there is demyelination and loss of oligodendrocytes.This work attempts to address whether mutations in hnRNP-A1 could contribute to MS pathogenesis. It would be interesting to note whether antibody responses to the protein correlated with mutations (i.e. does the mutation in the protein affect tolerance). Another issue that needs to be addressed is whether the mutations in the hnRNP-A1 gene are due to increased mutational frequency that can be observed in replicating cells, similar to the observation of increased hprt mutations made by Allegretta many years ago in MBP-specific T cells.Overall, this is an interesting study which certainly increases the interest in non-myelin targets in diseases such as MS.

I have read this submission. I believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.
No competing interests were disclosed. We appreciate Professor Racke's critique, including that the studies were "well performed." The following concerns expressed by Professor Racke were addressed: One might be interested to know whether testing in another cell line such as a oligodendroglial cell line would have similar effects on cell survival as in the neuroblastoma cell line. This could also be highly relevant in MS, as in addition to neuronal loss, there is demyelination and loss of oligodendrocytes.
We used SK-N-SH neurons as a model system to examine the effect that mutant forms of hnRNP A1 might have on target cell function. Considering that cells other than neurons are clearly involved in the pathogenesis of MS, we plan to add oligodendrocyte cell lines to future studies of mutant hnRNP A1. We addressed this in the sixth paragraph of the discussion.
This work attempts to address whether mutations in hnRNP-A1 could contribute to MS pathogenesis. It would be interesting to note whether antibody responses to the protein correlated with mutations (i.e. does the mutation in the protein affect