RAPID COMMUNICATION
Sigma Nonopioid Intracellular Receptor 1 Mutations Cause Frontotemporal Lobar Degeneration–Motor Neuron Disease Agnes A. Luty, BSc,1–3* John B.J. Kwok, PhD,1–3* Carol Dobson-Stone, PhD,1–3* Clement T. Loy, MD,1–3 Kirsten G. Coupland, BSc,1 Helena Karlstro¨m, PhD,4 Tomasz Sobow, MD,5 Joanna Tchorzewska, MD,5 Aleksandra Maruszak, BSc,6 Maria Barcikowska, MD,6 Peter K. Panegyres, MD,7,8 Cezary Zekanowski, PhD,6 William S. Brooks, MD,1,2 Kelly L. Williams, BSc,9 Ian P. Blair, PhD,9,10 Karen A. Mather, PhD,11 Perminder S. Sachdev, MD,11,12 Glenda M. Halliday, PhD,1,2 and Peter R. Schofield, PhD, DSc1–3 Objective: Frontotemporal lobar degeneration (FTLD) is the most common cause of early-onset dementia. Pathological ubiquitinated inclusion bodies observed in FTLD and motor neuron disease (MND) comprise transactivating response element (TAR) DNA binding protein (TDP-43) and/or fused in sarcoma (FUS) protein. Our objective was to identify the causative gene in an FTLD-MND pedigree with no mutations in known dementia genes. Methods: A mutation screen of candidate genes, luciferase assays, and quantitative polymerase chain reaction (PCR) was performed to identify the biological role of the putative mutation. Neuropathological characterization of affected individuals and western blot studies of cell lines were performed to identify the pathological mechanism of the mutation. Results: We identified a nonpolymorphic mutation (c.672*51G>T) in the 30 -untranslated region (UTR) of the Sigma nonopioid intracellular receptor 1 (SIGMAR1) gene in affected individuals from the FTLD-MND pedigree. The c.672*51G>T mutation increased gene expression by 1.4-fold, corresponding with a significant 1.5-fold to 2-fold change in the SIGMAR1 transcript or Sigma-1 protein in lymphocyte or brain tissue. Brains of SIGMAR1 mutation carriers displayed a unique pathology with cytoplasmic inclusions immunopositive for either TDP-43 or FUS but not Sigma-1. Overexpression of SIGMAR1 shunted TDP-43 and FUS from the nucleus to the cytoplasm by 2.3-fold and 5.2fold, respectively. Treatment of cells with Sigma-1 ligands significantly altered translocation of TDP-43 by up to 2-fold. Interpretation: SIGMAR1 is a causative gene for familial FTLD-MND with a unique neuropathology that differs from other FTLD and MND cases. Our findings also suggest Sigma-1 drugs as potential treatments for the TDP-43/FUS proteinopathies. ANN NEUROL 2010;68:639–649
F
rontotemporal lobar degeneration (FTLD) is the third most common cause of dementia, after Alzheimer’s disease and dementia with Lewy bodies, and the
most common cause of dementia under the age of 65 years.1 The spectrum of FTLD phenotypes includes the co-occurrence of FTLD with motor neuron disease
View this article online at wileyonlinelibrary.com. DOI: 10.1002/ana.22274 Received May 26, 2010, and in revised form Aug 30, 2010. Accepted for publication Sep 17, 2010. Address correspondence to Dr Schofield, PhD, DSc, Executive Director and CEO, Neuroscience Research Australia, Barker St, Randwick, Sydney, NSW 2031, Australia. E-mail:
[email protected] *These authors contributed equally to this work. From 1Neuroscience Research Australia, Sydney, New South Wales, Australia; 2School of Medical Sciences, University of New South Wales, Sydney, New South Wales, Australia; 3Garvan Institute of Medical Research, Sydney, New South Wales, Australia; 4Karolinska Institute, Stockholm, Sweden; 5Department of Medical Psychology, Medical University of Lodz, Lodz, Poland; 6Department of Neurodegenerative Disorders, Polish Academy of Sciences, Warsaw, Poland; 7Neurosciences Unit, Department of Health, Perth, Western Australia, Australia; 8Neurodegenerative Disorders Research, Subiaco, Western Australia, Australia; 9Northcott Neuroscience Laboratory, ANZAC Research Institute, Concord, New South Wales, Australia; 10Faculty of Medicine, University of Sydney, Sydney, New South Wales, Australia; 11Brain and Ageing Research Program, School of Psychiatry, University of New South Wales, New South Wales, Australia; and 12Neuropsychiatric Institute, The Prince of Wales Hospital, Sydney, New South Wales, Australia. Additional Supporting Information can be found in the online version of this article.
C 2010 American Neurological Association V 639
ANNALS
of Neurology
(MND).2 FTLD is a pathologically heterogeneous disorder categorized into 2 main groups: cases with tau-positive pathology known as tauopathies, and those without, previously known as FTLD-U.3 The trans-activating response element (TAR) DNA binding protein (TDP43), a heterogeneous nuclear ribonucleoprotein (hnRNP) involved in exon splicing and transcription regulation,4 was identified as a major protein component of the ubiquitinated inclusions in many FTLD-U (FTLD-TDP) and MND patients.5,6 Mutations in the TDP-43 (TARDBP) gene have been reported in familial and sporadic FTLDU7 and MND cases.8,9 More recently, mutations in the gene encoding a related hnRNP known as fused in sarcoma (FUS) were identified in familial MND cases.10,11 FUS protein colocalizes with TDP-43 in the motor neuron inclusions.12 A neuropathological review of FTLD cases identified FUS as a major protein in the inclusions found in atypical FTLD-U patients (FTLD-FUS) without TDP-43 inclusions,12–14 and these FUS inclusions are only rarely present in the brains of FTLD-TDP-43 patients or in FTLD-MND cases.12 Neurons with cytoplasmic inclusions containing either hnRNP show a partial or total loss of normal nuclear TDP-435,6 or FUS,12,13 raising the possibility that nuclear clearing and cytoplasmic sequestration of these hnRNPs play a mechanistic role in disease pathogenesis. Genetic studies have identified FTLD-U causal mutations in the progranulin (GRN) and valosin-containing protein (VCP) genes.15 A major FTLD-MND locus is situated on chromosome 9p, with 12 pedigrees having definite or suggestive linkage to a 3.2Mb minimal disease region between markers D9S169 and D9S251.16–21 Chromosome 9p–linked pedigrees are characterized by a combination of TDP-43 immunopositive FTLD and classical MND.16–21 Progranulin and VCP mutation carriers and chromosome 9p–linked cases exhibit variation in ubiquitin, TDP-43, and FUS deposition, indicating that FTLD-U is clinically, neuropathogically, and genetically heterogenous.12–14 We previously reported on a multigenerational FTLDMND pedigree (Aus-14) with linkage to chromosome 9.21 In this study, we examined a series of candidate genes for a mutation that cosegregates with the disease phenotype in the FTLD-MND pedigree. We present functional data that the Sigma nonopioid intracellular receptor 1 (SIGMAR1) gene is a novel causative locus for FTLD-MND.
Patients and Methods Materials Commercially available Sigma-1 ligands AC915 (A 3595), opipramol (O 5889), and haloperidol (H 1512) were obtained from Sigma-Aldrich (St. Louis, MO).
640
Consent and Ethics Approval Written informed consent was obtained from the appropriate legal guardians for blood donations for genetic studies and for brain donations. Aus-14 family tissue was obtained from the South Australian Brain Bank and other FTLD and control samples from the Sydney Brain Bank at Neuroscience Research Australia. All studies were approved by the relevant institutional ethics committees.
FTLD, Familial Presenile Dementia, and Sydney Older Person Study Cohorts The FTLD-MND pedigree (family Aus-14) has been described.21 The Australian familial FTLD cohort comprises 26 pedigrees selected on the basis of a positive family history.22 All probands fulfill current diagnostic criteria for FTLD.23 All probands were negative for MAPT and GRN mutations. Two independent Polish presenile dementia cohorts comprise 158 familial cases. Cases were negative for APP, PSEN1, PSEN2, and MAPT mutations.24,25 The Polish cohort is also negative for GRN mutations. The Sydney Older Person Study (SOPS) cohort comprises 169 neurologically normal elderly individuals (mean age ¼ 88; SD ¼ 4.3 years).26 An additional 350 neurologically normal controls who did not have a family history of MND (mean age ¼ 61; SD ¼ 10.9 years)8 and 750 nondemented controls (mean age ¼ 79; SD ¼ 4.9 years) from the community-based Memory and Aging cohort were screened.27
Brain Tissue Analyses For western blot analyses, frozen brain tissue from 3 SIGMAR1 c.672*51G>T carriers (Fig 1; III:2, III:3, III:12) and 3 agematched controls was used. Each sample was assayed twice. Crude total brain protein was extracted from the frontal cortex using methods described.28 Specific proteins were visualized using commercially available antibodies for TDP-43 (BC001487 [ProteinTech Group, Chicago, IL; diluted 1:2000] and ab57105 [Abcam, Cambridge, UK; diluted 1:1000]), Sigma-1 (ab53852 [Abcam; diluted 1:500] or sc-22948 [Santa Cruz Biotechnology, Santa Cruz, CA; diluted 1:200]), and bactin (ab8226 [Abcam; diluted 1:2000]) and enhanced chemiluminescence according to the manufacturer’s instructions (Supersignal West Pico Chemiluminescent Substrate; Thermo Scientific, Rockford, IL). Immunohistochemistry was performed on formalin-fixed paraffin-embedded sections of the superior frontal cortex and hippocampus from SIGMAR1 c.672*51G>T carriers, an FTLD case with a GRN Arg493X mutation,29 a FTLD-FUS case, and age-matched controls. Sigma-1 protein was visualized using a citrate buffer antigen retrieval and immunoperoxidase procedure (sc-22948 [Santa Cruz Biotechnology; diluted 1:25]).21 Specificity of the reaction was confirmed by omitting the primary antibodies. Double immunofluorescence was performed using mouse anti-phospho TDP-43 (pS409/410 [Cosmo Bio Co, Tokyo, Japan; diluted 1:5000]) antibody, rabbit anti-FUS antibody (ab70381 [Abcam; diluted 1:500]), and rabbit anti-Sigma-1 antibodies (sc-22948 [Santa Cruz
Volume 68, No. 5
Luty et al: SIGMAR1 Mutations in FTLD-MND
FIGURE 1: Three pedigrees with FTLD and SIGMAR1 30 -UTR mutations. Black figures indicate neuropathological and/or clinical diagnosis of presenile dementia. Gray figures indicate presence of MND. Half-gray figures indicate the presence of dementia and MND. Note that individual III:8 from Aus-14 has since died of MND but is believed to have been a phenocopy.17 DNA samples available for analyses are indicated by asterisks. Probands are indicated by arrows. The presence of familial SIGMAR1 mutations are indicated (1).
Biotechnology; diluted 1:25]). Antibodies were visualized with the relevant combination of Alexa Fluor 594 donkey anti-mouse, Alexa Fluor 594 donkey anti-rabbit, Alexa Fluor 488 donkey anti-mouse and Alexa Fluor 488 donkey anti-goat secondary antibodies (Invitrogen, Carlsbad, CA) on a confocal microscope (C190; Nikon Corporation, Tokyo, Japan). To ensure specificity of the immunohistochemical reactions, a section without primary antibodies was included for each staining procedure as a negative control. Additionally, a mixture of the secondary antibodies was applied to sections with only 1 primary antibody incubated on each section.
Mutation Screen of Candidate Genes Intronic polymerase chain reaction (PCR) primers were designed to amplify each noncoding and coding exon and flanking intronic sequence of the candidate genes miR876, miR873, NCRNA00032, UBE2R2, DNAJA1, PAX5, CNTNAP3, GDA, DNAI1, CNTFR, DCNT3, ILIIRA, GALT, CCL19, CCL21, CCL27, ARID3C, TLN1, MOBKL2B, HINT2, AQP3, UBAP1, ALDH1B1, PLAA, IFNK, P23, UNIQ470, UBAP2, TOPORS, NDUFB6, APTX, BAG1, and SIGMAR1 using the ExonPrimer program accessed using the University of California Santa Cruz (UCSC) Genome Bioinformatics Site (http://genome.ucsc.edu; primer sequences available
November, 2010
on request). PCR products amplified from genomic templates were sequenced using BigDye chemistry and a 3730 DNA Analyzer (Applied Biosystems, Foster City, CA).
Luciferase Reporter Assay A 1223bp promoter fragment was amplified from the SIGMAR1 gene using the oligonucleotides SIGMAR1-PromF (50 -CTGGGGAGTAGGACCATTGTTTC-30 ) and SIGMAR1PromR (50 -TATCTCTTCGCGCTGGAAGACG-30 ) and subcloned into a pGL3 vector (Promega, Sydney, Australia) containing the luciferase reporter gene. A 1104bp genomic fragment was amplified corresponding to the entire 30 -untranslated region (30 -UTR) of the SIGMAR1 gene using the oligonucleotides SIGMAR1-3UTRF (50 -ACTGTCTTCAGCACCCAGGACT-30 ) and SIGMAR1-3UTRR (50 -ACCATGAATCACACAGCAAGAG-30 ). Genomic DNA from subjects with the c.672*51G>T or c672*26C>T alleles or from normal subjects was used as a template. Wild-type and mutant alleles were subcloned into a modified pGL3 vector containing the wild-type SIGMAR1 promoter. The c.672*47G>A mutation was introduced into the luciferase reporter construct with the wild-type SIGMAR1 promoter and wild-type 30 -UTR by site-directed mutagenesis (Stratagene, Cedar Creek, TX). Each recombinant vector was transfected into human neuroblastoma SK-N-MC
641
ANNALS
of Neurology
(ATCC HTB 10) or human embryonic kidney cells HEK293 (CRL-1573) using Lipofectamine 2000 reagent according to the manufacturer’s instructions (Invitrogen). Cells were lysed after 48 hours, and luciferase activities were assayed using the Readi-Glo reagent according to manufacturer’s instructions (Promega).
Quantification of SIGMAR1 and TDP-43 Transcript Levels Total RNA was extracted from immortalized lymphocytes of affected individuals (Fig 1; III:2, III:6, III:7, and III:12) and nonmutation carriers (Fig 1; III:1, III:9, III:10, and III:11) from Aus-14 or from transfected cells using the SV Total RNA extraction system (Promega). RNA quality was determined by the integrity of the 28S and 18S ribosomal RNA bands upon gel electrophoresis. RNAs were reverse-transcribed using a polydT primer. SIGMAR1 transcript levels were determined by SYBR-green-chemistry quantitative PCR using SIGMAR1-RTF (50 -ACCATCATCTCTGGCACCTT-30 ) and SIGMAR1-RTR (50 -CTCCACCATCCATGTGTTTG-30 ) primers. TDP-43 transcripts levels were determined using 2 different primer sets: TDP43RT1F (50 -AAGAGCAGTCCAGA AAACATCC-30 ) and TDP43RT1R (50 -CCTGCACCACATAA GAACTTCTCC-30 ); and TDP43RT2F (50 -GATAGATGGACGATGGTGTGAC-30 ) and TDP43RT2R (50 -TCATCCTCAGTCATGTCCTC TG30 ). Transcript levels between samples were normalized using primers that amplify the housekeeping genes, succinate dehydrogenase complex subunit A (SDHA) or b-actin (ACTB). Mean values were derived from duplicate measurements.
(ATCC HTB 11) cells using Lipofectamine 2000 (Invitrogen). Cells were left for 48 hours prior to western blot analyses of TDP-43 and FUS protein levels. For total protein extraction, cells were lysed in 1" lysis buffer (50mM Tris HCl, pH 7.4; 150mM NaCl; 1 mM phenylmethylsulfonyl fluoride [PMSF]; 1" complete cocktail protease inhibitor; Roche Diagnostics GmbH, Mannheim, Germany; and 0.05% Triton X-100). Subcellular fractions were isolated using a NE-PER Nuclear and Cytoplasmic extraction kit (Thermo Scientific) according to the manufacturer’s instructions. Lysates underwent electrophoresis on a 10% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) gel and were transferred to a nitrocellulose membrane (Trans-blot transfer medium; Bio-Rad, Hercules, CA). Densities of chemiluminescence bands were quantified using the Bio-Rad Chemidoc system. Rabbit polyclonal antibodies were used to detect the TDP-43 protein (BC001487; ProteinTech Group; diluted 1:2000) and FUS protein (ab23439; Abcam; diluted 1:2000). Difference in protein levels between different samples were normalized using b-actin levels (ab8226; Abcam; diluted 1:2000).
Statistical Analyses Mean differences in quantitative measures were compared using 2-tailed Student t tests. Analysis of covariance (ANCOVA) and regression analyses were performed using the SPSS13 statistical package (SPSS, Inc., Chicago, IL), taking into account individual data for age and disease status. Mean and standard error of the mean (SEM) are reported for all variables.
SIGMAR1 Expression Constructs
Results
A full-length wild-type SIGMAR1 complementary DNA (cDNA) was constructed by reverse-transcription (RT)-PCR of lymphocyte RNA using the primers SIGMAR1-F (50 -AAAA GCTTATGCAGTGGGCCGTGGGC-30 ) and SIGMAR1-R (50 -AGGATCCTGGTGGGGAGGAGGTGGGAA-30 ) and subcloned into the expression vector pcDNA3.1 (Invitrogen) to generate the pcDNA-SIGMAR1(wt) plasmid. The presence of a FLAG motif at the amino-terminal end of the Sigma-1 protein was introduced using the primers SIGMAR1-FLAGF (50 AAAAGCTTATGGATTACAAGGATGACGACGATAAGCAG TGGGCCGTGGGC-30 ) and SIGMAR1-FLAGR (50 -AGGA TCCTGGTGGGGAGGAGGTGGGAA-30 ) to generate the pcDNA-FLAG-SIGMAR1(wt) plasmid. A Stealth RNA interference (RNAi) oligonucleotide (HSS145543; Invitrogen) was used to knock down endogenous SIGMAR1 gene expression in cell lines. Either a high GC negative RNAi control (12935400; Invitrogen) or SIGMAR1 RNAi (20pmol/ll stock) was transfected into cell lines according to the manufacturer’s instructions.
Identification of a Putative Mutation in SIGMAR1Gene We previously reported an FTLD-MND pedigree (Aus14) with neuronal TDP-43-immunopositive cytoplasmic inclusions (Fig 1) and maximal linkage at marker D9S1817 on chromosome 9p (logarithm of odds [LOD] 3.41).21 A mutation screen of the affected individuals did not reveal any pathogenic nucleotide changes in the known dementia and MND causative genes (APP, PSEN1, PSEN2, MAPT, VCP, GRN, CHMP2B, TARDBP, FUS, and SOD1). We therefore commenced examination of approximately 200 genes within the candidate disease region (D9S169 to D9S1845) defined in Aus-14, initially focusing on genes with possible involvement in neurodegeneration. Of the initial 30 genes examined, SIGMAR130 was the only gene that had a nonpolymorphic nucleotide change (c.672*51G>T) that cosegregated with the disease haplotype in the Aus-14 pedigree (Fig 2A,B). The c.672*51G>T nucleotide substitution is located in the 30 UTR and was not detected in a cohort of 169 elderly normal controls from the SOPS26 or in an additional cohort of 1100 neurologically normal controls.8,27 We further screened an Australian cohort of FTLD probands from 26
Determination of Total TDP-43 and FUS Levels and Subcellular Localization by Western Blotting Recombinant vectors were transfected into SK-N-MC (ATCC HTB 10), HEK293 (ATCC CRL 1573), and SK-N-SH
642
Volume 68, No. 5
Luty et al: SIGMAR1 Mutations in FTLD-MND
SOPS cohort26 and the 1100 neurologically normal controls8,27 as well. An additional 50 normal controls of Polish descent were screened to confirm the absence of the c.672*47G>A substitution.
pedigrees that were negative for MAPT and GRN mutations,22 and 2 independent Polish presenile dementia cohorts comprising 158 unrelated familial cases that were negative for APP, PSEN1, PSEN2, GRN and MAPT mutations.24,25 A nucleotide substitution in the 30 -UTR (c.672*26C>T) was identified in another Australian pedigree with FTLD (Aus-47) (see Figs 1 and 2A,B). Another 30 -UTR substitution was identified in another family (Pol1) from the Polish cohorts (c.672*47G>A) (Figs 1 and 2A,B). No additional affected individuals were available from these 2 pedigrees to demonstrate segregation of disease phenotype with the inheritance of the nucleotide substitutions. However, the c.672*26C>T and c.672*47G>A substitutions were absent in the normal controls of the
Effect of SIGMAR1 Mutations on Gene Expression The SIGMAR1 30 -UTR alterations may alter the stability of the transcript.31 We constructed chimeric reporter vectors in which the entire 30 -UTR was placed downstream of the luciferase gene. Luciferase activity from lysates of transfected SKN-MC and HEK293 cells provided a measure of the stability of the chimeric transcripts. The c.672*51G>T nucleotide substitution significantly increased luciferase activity by 1.2fold (p ¼ 0.014) and 1.4-fold (p ¼ 0.001) in SK-N-MC and HEK293 cells, respectively (see Fig 1C). Moreover, quantitative RT-PCR showed that the endogenous SIGMAR1 gene was overexpressed by 2-fold (p ¼ 0.001) in lymphocytes from c.672*51G>T carriers when compared to wild-type, neurologically normal controls from the same family, after adjustment for age as a covariate (see Fig 2D; Supporting Fig 1A). Conversely, the c.672*26C>T nucleotide substitution significantly decreased luciferase activity by 0.8-fold in HEK293 and SK-N-MC cells (p ¼ 0.003 and 0.006, respectively; see Fig 2C). A similar decrease was observed for the c.672*47G>A substitution with a 0.8-fold (p < 0.001) and 0.9-fold (p ¼ 0.001) decrease in luciferase activity in HEK293 and SK-N-MC cells, respectively (see Fig 2C). Quantitative PCR confirmed that the endogenous SIGMAR1 gene was underexpressed by #0.9-fold in lymphocytes from
3
November, 2010
FIGURE 2: Mutations in the SIGMAR1 gene and their role in altering gene expression. (A) SIGMAR1 gene comprises 4 coding exons, with translation start and stop sites indicated by open arrows. Nucleotide changes detected in the Australian FTLD-MND pedigrees Aus-14 and Aus-47 from the FTLD cohort (up-pointing arrows) and the Polish presenile dementia cohort (down-pointing arrow). (B) Electropherograms of the 30 -UTR nucleotide substitutions (arrows) in probands compared with a normal individual. (C) The presence of the c.672*26C>T, c.672*47G>A, and c.672*51G>T mutations within the 30 -UTR of SIGMAR1 affect transcript stability. Chimeric luciferase reporter constructs comprising the SIGMAR1 promoter sequence, the luciferase cDNA, and the entire SIGMAR1 30 -UTR sequence were transfected into human neuroblastoma SKN-MC (gray columns) and HEK293 (black columns) cells. Luciferase activity indicated that both 30 -UTR mutations significantly increased levels of the chimeric transcripts compared with the wild-type (wt) sequence. Values are mean 6 SEM from 5 transfections. Significance of p < 0.05 (*) and p < 0.001 (**) are indicated. (D) Relative levels of SIGMAR1 transcript in lymphocyte in c.672*51G>T carriers (n 5 4) compared with related wild-type individuals (n 5 3) (left) and Sigma-1 protein in frontal cortex of c.672*51G>T carriers (n 5 3) compared with normal unrelated controls (n 5 4) (right), after adjusting for age as a covariate.
643
ANNALS
of Neurology
affected c.672*47G>A carriers when compared to wild-type, neurologically normal controls from the same family (Supporting Fig 1B). Sigma-1 occurs as a high-molecular-weight complex (#100kDa), with the monomeric form (31kDa) accounting for less than 5% of total protein (Supporting Fig 2). Western blot analysis of brain lysates revealed a significant 1.6-fold (p < 0.001) increase in total Sigma-1 levels in the prefrontal cortex of the c.672*51G>T carriers compared with normal controls, after adjustment for age as a covariate (see Fig 2D). Neuropathological Characterization of Affected Members of the FTLD-MND Pedigree The neuropathological features of these cases were previously described in detail21 (refer to supporting data of Luty et al.21 for additional information on neuropathology of cases). Additional immunohistochemical examination of Sigma-1 and FUS in association with TDP-43 was performed in 2 previously detailed cases from family Aus14,21 2 unrelated FTLD cases (1 with FUS inclusions and an Arg493X progranulin mutation carrier with TDP-43 inclusions29), and 3 age-matched controls without neurological or psychiatric features or other significant histopathology. In control and unrelated FTLD brain tissue, Sigma-1 was localized on membranes within the cytoplasm of most neurons (Fig 3A), astrocytes, and myelinated oligodendroglia (not shown). In 2 cases with c.672*51G>T SIGMAR1 nucleotide substitution (III:2 and III:3), Sigma1 was also concentrated within the nucleus of some neurons in regions with significant neuronal degeneration (CA1 and dentate gyrus; see Fig 3B,C). In the 2 unrelated FTLD cases, either TDP-43 (see Fig 3D,F inset) or FUS (see Fig 3I inset) immunoreactivity was observed in all inclusions, consistent with recent reports.13,14 However, in the cases with c.672*51G>T SIGMAR1 nucleotide substitution, both TDP-43 (see Fig 2D–F) and FUS immunopositive inclusions (see Fig 3G–I) were observed in affected regions (see supporting data of Luty et al.21 for detailed description), although in different neurons (see Fig 3E, F, H, and I). There was a similar density of TDP-43 and FUS inclusions in the different cases. No TDP-43 immunoreactive inclusions occurred in neurons with enhanced nuclear Sigma-1 immunoreactivity (see Fig 3D), although rare FUS immunoreactive inclusions occurred in neurons with enhanced nuclear Sigma-1 immunoreactivity (see Fig 3G). Brain tissue was not available from carriers of the c.672*26C>T or c.672*47G>A nucleotide substitutions. Alteration of SIGMAR1 Gene Expression in Transfected Cells Constitutive expression of wild-type SIGMAR1 cDNA in transfected cell lines resulted in increased levels of Sigma644
1 protein, while a commercially validated RNAi effectively knocked down endogenous Sigma-1 (Supporting Fig 3). The level of total TDP-43 in transfected cells that overexpressed SIGMAR1 increased by 1.7-fold in SK-N-MC (p ¼ 0.092) and HEK293 cells (p ¼ 0.042) (Fig 4A) compared with cells that overexpressed the control LacZ gene. No significant changes in total FUS levels were observed in these cells (see Fig 4A). Conversely, knockdown of endogenous SIGMAR1 transcripts resulted in a significant decrease in both TDP-43 and FUS levels in SK-N-MC (TDP-43: 0.7-fold, p < 0.001; FUS: 0.8fold, p ¼ 0.049) and HEK293 cells (TDP-43: 0.7-fold, p ¼ 0.026; FUS: 0.6-fold, p ¼ 0.013). Related changes in levels of both the 35kDa TDP-43 cleavage product and full-length 45kDa TDP-43 were observed (Supporting Fig 3A), indicating that there was no preferential increase in proteolytic cleavage of TDP-43 with altered SIGMAR1 expression. Quantification of TDP-43 transcript levels did not reveal any significant difference in transfected cells overexpressing or underexpressing SIGMAR1 (data not shown), indicating that the effect of SIGMAR1 on TDP-43 protein levels is not at the level of gene transcription. We next explored the effect of SIGMAR1 overexpression on the subcellular localization of TDP-43 and FUS in transfected cells by measuring the levels of these ribonuclear proteins in the cytoplasmic and nuclear subcellular fractions by western blot analysis. Overexpression of SIGMAR1 significantly increased the ratio of cytoplasmic to nuclear TDP-43 and FUS in SK-N-MC (TDP43: 1.3-fold, p ¼ 0.005; FUS: 1.6-fold, p ¼ 0.044) and HEK293 cells (TDP-43: 2.3-fold, p ¼ 0.037; FUS: 5.2fold, p ¼ 0.013) relative to control transfections with the LacZ cDNA (see Fig 4B). We verified this result by immunohistochemistry of transfected neuronal SK-N-SH cells using a FLAG-tagged SIGMAR1 cDNA expression construct (Supporting Fig 4). We observed that higher SIGMAR1 transcript levels were positively correlated (r2 ¼ 0.852, p ¼ 0.013) with increased levels of the ratio of cytoplasmic to nuclear TDP-43 (Supporting Fig 5). Finally, quantification of nuclear TDP-43 and FUS levels in SK-N-MC cells revealed that a consistent effect of both overexpression and knockdown of SIGMAR1 was a decrease in nuclear levels of these hnRNPs relative to control cells (see Fig 4C). A significant decrease in nuclear FUS levels was observed when SK-N-MC cells were transfected with SIGMAR1 expression constructs (0.49fold, p ¼ 0.016) or with RNAi against SIGMAR1 (0.76fold, p ¼ 0.017), and a similar 0.75-fold trend (p > 0.05) was observed for TDP-43 for both overexpression and knockdown of SIGMAR1. Volume 68, No. 5
Luty et al: SIGMAR1 Mutations in FTLD-MND
FIGURE 3: Sigma-1, TDP-43, and FUS in the hippocampus. Sigma-1 immunoperoxidase staining of hippocampal neurons (A) in a control and (B, C) in 2 cases with SIGMAR1 c.672*51G>T nucleotide substitutions. Asterisks indicate nuclei. Sigma-1 normally localizes to cytoplasmic membranes (A), while in the c.672*51G>T carriers some remaining (B) dentate granule and (C) CA1 pyramidal cells show intense Sigma-1 immunoreactivity in the nucleus (white asterisks). Double immunofluorescence labeling of phosphorylated TDP-43 (red in D, green in E, F, H, I) and FUS (red in E–I) in dentate granule cells of 2 cases with SIGMAR1 c.672*51G>T nucleotide substitutions (D–I), a FTLD-TDP-43 case with a progranulin mutation (D, F insets), and a FTLD-FUS case (I inset). All nuclei indicated by white asterisks. Only phosphorylated TDP-43 inclusions (arrowheads in D, F insets) were observed in the progranulin case, and only FUS inclusions (arrowhead in I inset) were observed in the FTLD-FUS case. In cases with SIGMAR1 c.672*51G>T nucleotide substitutions, both types of inclusions were observed, (E, F, H, I) with TDP-43 inclusion-bearing neurons maintaining nuclear FUS immunoreactivity but (H, I) reduced nuclear FUS staining observed in FUS inclusion-bearing neurons. Double immunofluorescence labeling of Sigma-1 (green) and either TDP-43 (red) or FUS (red) indicated that most neurons with nuclear localization of Sigma-1 did not contain either (D) TDP-43 or FUS immunoreactive inclusions, (G) although FUS immunoreactive inclusions were infrequently observed in such neurons.
Sigma-1 Ligands and TDP-43 Translocation We used 3 Sigma-1 ligands, AC915 (N-(2-(3,4-dichlorophenyl)acetoxy)-ethylpyrrolidine (specific Sigma-1 antagonist),32 haloperidol (Sigma-1 antagonist),33 and opipramol (specific Sigma-1 agonist)34 to determine whether these small molecules can mimic the effect of altered expression of SIGMAR1 on TDP-43 subcellular localization (see Fig 4D). We observed that both AC915 and November, 2010
opipramol had a significant effect on TDP-43 localization, whereby 50nM of AC915 (antagonist) or 15nM of opipramol (agonist) significantly decreased (1.5-fold, p ¼ 0.038) or increased (1.5-fold, p ¼ 0.020), respectively, the relative level of TDP-43 in the cytoplasm compared with untreated cells. Haloperidol, a therapeutically significant compound, was also effective in modulating the localization of TDP-43, where both 24nM and 2.4nM 645
ANNALS
of Neurology
FIGURE 4: Altered expression of SIGMAR1 in human cell lines affect TDP-43 and FUS expression and subcellular location. (A) Modulation of SIGMAR1 expression by transfection of expression constructs with the full-length wild-type SIGMAR1 cDNA under the control of the constitutive CMV promoter or RNAi against endogenous SIGMAR1 transcripts led to altered total TDP-43 and FUS levels. Western blot analysis of endogenous full-length TDP-43 (black columns) or FUS (gray columns) in transfected cells (left) showed differences in protein levels in SK-N-MC and HEK293 cells after normalization to b-actin levels (right). Values are mean 6 SEM from 5 transfections. Significance of p < 0.05 (*) is indicated. (B) SIGMAR1 cDNA expression constructs transfected into either HEK293 or SK-N-MC cells followed by extraction of cytoplasmic (cyto-) or nuclear (nuc-) fractions. Western blot analysis shows TDP-43 and FUS protein levels in the 2 subcellular fractions of transfected cells overexpressing SIGMAR1 compared with LacZ. Chemiluminescent band intensities were quantified and the levels of proteins were expressed as a ratio of cytoplasmic versus nuclear levels. Values are mean 6 SEM from 5 transfections. Significance of p < 0.05 (*) is indicated. (C) Quantification of nuclear TDP-43 or FUS in SK-N-MC cells transfected with SIGMAR1 cDNA constructs or RNAi against endogenous SIGMAR1 transcripts. Western blot analysis of endogenous full-length TDP-43 (black columns) or FUS (gray columns) in transfected cells showed differences in protein levels compared with control treatments after normalization to b-actin levels. Values are mean 6 SEM from 5 transfections. Significance of p < 0.05 (*) is indicated. (D) Effect of Sigma-1 ligands on TDP-43 subcellular localization. Western blot analysis of TDP-43 protein levels in subcellular fractions of cells exposed to 3 concentrations of opipramol (agonist), AC915 (antagonist), and haloperidol (antagonist). Chemiluminescent band intensities were quantified and the levels of TDP-43 were expressed as a ratio of cytoplasmic (cyto-) vs nuclear (nuc-) TDP-43. Values are mean 6 SEM from 5 transfections. Significance of p < 0.05 (*) is indicated.
significantly decreased (0.5-fold, p < 0.001 and p ¼ 0.004, respectively) the relative level of TDP-43 in the cytoplasm compared with untreated cells (see Fig 4D). 646
Discussion Significant progress in the understanding of FTLD and MND has occurred with the recent discoveries of the TDP-43 and FUS proteins as the major components of Volume 68, No. 5
Luty et al: SIGMAR1 Mutations in FTLD-MND
TABLE 1: Summary of In Vitro and In Vivo Observations of SIGMAR1 Mutations
Family
SIGMAR1 Mutation
Gene Expression In Vitro
Gene Expression In Vivo
Predicted Pathological Mechanism
Neuropathology
Aus-14
c.672*51G>T
:
: Lymphocyte; : brain
: Cytoplasmic hnRNP;; nuclear hnRNP
Frequent TDP-43 and FUS immunopositive cytoplasmic inclusions
Aus-47
c.672*26C>T
;
NA
; Nuclear hnRNP
NA
Pol-1
c.672*47G>A
;
; Lymphocyte
; Nuclear hnRNP
NA
FUS, fused in sarcoma gene; hnRNP, heterogeneous nuclear ribonucleoprotein; NA, not available for testing.
ubiquitinated inclusion bodies in both of these diseases and that mutations in the genes encoding these proteins cause these diseases. We determined that the neuropathology in our largest pedigree (family Aus-14) is unique in that it includes similar densities of TDP-43 and FUS inclusions in different neurons in vulnerable cortical and hippocampal (see Fig 3) regions. While a recent study showed colocalization of these proteins in spinal motor neuron inclusions in MND,12 the majority of TDP-43 and FUS inclusions in family Aus-14 were not colocalized but found in separate neurons. This novel cortical neuropathology differs from all other FTLD and MND cases,12 including other FTLD-MND cases previously linked to a major FTLD-MND region on chromosome 9.13 As summarized in the Table, our data are consistent with the marked effect of SIGMAR1 overexpression on the subcellular localization of both TDP-43 and FUS (see Fig 4) and provide evidence that the aberrant pathological activity of TDP-43 and FUS could be ameliorated by the modulation of Sigma-1 activity. The c.672*51G>T nucleotide substitution identified in the Aus-14 pedigree increased gene expression by 1.4fold as determined by luciferase reporter assays (see Fig 2C), corresponding with a significant 1.5-fold to 2-fold change in SIGMAR1 transcript or Sigma-1 protein in lymphocyte or brain tissue of c.672*51G>T carriers (see Fig 2D). We postulate that the 30 -UTR nucleotide substitution is a mutation that alters transcript stability and hence gene expression. There is increasing evidence that 30 -UTRs contain regulatory elements that have an important role in posttranslational control of gene expression31 and can bind not only to proteins but to small noncoding regulatory RNAs (microRNAs) as well.35 Indeed, 30 -UTRs can be hotspots for mutations, such as for the transcription factor gene GATA4 in which 60% of all putative mutations fall within the 30 -UTR.36 We observe that nucleotide substitutions that decrease SIGMAR1 expression (c.672*26C>T and c.672*47G>A) (see Fig 2; Table ) may be deleterious as well, consistent with previous studies that demonstrated November, 2010
the neurotoxic effect of Sigma-1 antagonists and SIGMAR1 RNAi.37 However, as the effect of the c.672*26C>T and c.672*47G>A on gene expression was small (
