ALS Associated Mutations in Matrin 3 Alter Protein-Protein Interactions and Impede mRNA Nuclear Export

Abstract

Mutations in Matrin 3 have recently been linked to ALS, though the mechanism that induces disease in these patients is unknown. To define the protein interactome of wild-type and ALS-linked MATR3 mutations, we performed immunoprecipitation followed by mass spectrometry using NSC-34 cells expressing human wild-type or mutant Matrin 3. Gene ontology analysis identified a novel role for Matrin 3 in mRNA transport centered on proteins in the TRanscription and EXport (TREX) complex, known to function in mRNA biogenesis and nuclear export. ALS-linked mutations in Matrin 3 led to its re-distribution within the nucleus, decreased co-localization with endogenous Matrin 3 and increased co-localization with specific TREX components. Expression of disease-causing Matrin 3 mutations led to nuclear mRNA export defects of both global mRNA and more specifically the mRNA of TDP-43 and FUS. Our findings identify a potential pathogenic mechanism attributable to MATR3 mutations and further link cellular transport defects to ALS.

Figure 1

Matrin 3 cell culture model and IP-MS workflow. (a) Domain structure of Matrin 3 including location of mutants studied in this work listed below the protein as well as other recently identified mutations shown above. (b) Representative immunofluorescence images of NSC-34 cells stably expressing Flag tagged wild-type Matrin 3 or one of S85C, F115C, P154S or T622A Matrin 3 mutants. Cells transfected with empty vector are also shown denoting endogenous levels and localization of Matrin 3. Flag expression is shown in red, Matrin 3 in green, DAPI marking nucleus in blue. (c) Western blot of whole cell lysates probed with antibodies against Flag (top) and Matrin 3 (bottom) showing expression levels of endogenous Matrin 3 and Flag tagged Matrin 3 in NSC-34 stable cells lines and quantitation. Error bars represent standard error of the mean (SEM) of three experiments. One way Analysis of Variance (ANOVA) with Dunnett’s post-test showed no significant differences between level of expression of wild-type and any of the four mutations (Flag p-values: WT vs 85: p = 0.8163, WT vs 115: p = 0.4753, WT vs 154 p = 0.0619, WT vs 622 p = 0.0715, F-value: 8.671, DF = 14, Matrin 3 p-values: WT vs 85: p = 0.9957, WT vs 115: p = 0.9722, WT vs 154 p = 0.9998, WT vs 622 p = 0.7550, F-value: 2.716, DF = 17). Full length blots are presented in Supplementary Figure 7. (d) Flow chart of IP-MS sample preparation and analysis protocols.

Figure 2

Functionally organized GO term network (ClueGO) of binding partners to wild type and mutant Matrin 3 in NSC-34 cells. Associated gene clusters and functional differences are highlighted. GO terms with a single sample frequency above 50% were color-coded: wild type (purple), Ser85Cys(green), Phe115Cys(yellow), Pro154Ser(blue), Thr622Ala(red), and unspecific (grey). Terms were considered unspecific if sample frequency was above 50% across more than one sample. Sample frequency was determined as a percentage based on the number of genes that defined that specific term. Increased size of GO term nodes inversely correlates to p-values computed by a two-sided hypergeometric test, with step-down Bonferroni correction.

Figure 3

Immunofluorescence images of NSC-34 cells transiently transfected with wild-type or mutant Matrin 3 subjected to co-localization analysis. (a,c,e,g) Representative images from immunofluorescence staining are shown. In each case Flag is shown in red marking exogenous Matrin 3 and the protein of interest (Matrin 3, Aly, Ddx39b and Sarnp, respectively) is shown in green, merged image of two signals below. Scale bar indicates 5 µm. (b,d,f,h) Average Pearson’s correlation coefficient for Flag and the protein of interest, whiskers indicate 1.5 times the interquartile range (IQR) for 40–50 cells per genotype. One way ANOVA followed by Dunnett’s post-test, (*) denotes p-value < 0.05, (**) p < 0.01 and (***) p < 0.001 compared to wild-type (Matrin 3 p-values: WT vs 85: p = 0.1524, WT vs 115: p = 0.0035, WT vs 154 p = 0.0002, WT vs 622 p = 0.5535, F-value: 12.61, DF = 205; Aly p-values: WT vs 85: p = 0.0001, WT vs 115: p = 0.8307, WT vs 154 p = 0.0001, WT vs 622 p = 0.3368, F-value: 9.284, DF = 211; Ddx39b p-values: WT vs 85: p = 0.9725, WT vs 115: p = 0.0002, WT vs 154 p = 0.0107, WT vs 622 p = 0.0364, F-value: 7.701, DF = 224; Sarnp p-values: WT vs 85: p = 0.0090, WT vs 115: p = 0.9400, WT vs 154 p = 0.9791, WT vs 622 p = 0.8224, F-value: 2.913, DF = 228).

Figure 4

Immunoprecipitation followed by western blot from NSC-34 cell lines and human lumbar spinal cord tissue. (a) Immunoprecipitation using Flag antibody followed by western blot which was probed with Flag and Matrin 3 to confirm efficient pull down of Matrin 3, or TREX components Aly, Ddx39b and Sarnp; representative blots are shown, and all experiments were performed a minimum of three times with similar results. (b,c) Matrin 3 IP performed on endogenous Matrin 3 in untransfected NSC-34 cells. Immunoblots are probed with Aly (b) or Ddx39b (c). (d) Reverse immunoprecipitation experiments using antibodies against Aly and Ddx39b followed by western blot probed with either Aly or Ddx39b confirming pull-down of the target and Flag to measure the amount of mutant Matrin 3 bound, representative blots shown. (e,f) Quantification of Aly and Ddx39b IP-WB experiments; values are expressed as Flag signal over signal of the bait protein (Aly or Ddx39b respectively) to control for IP efficiency, Aly IP values from five replicates, Ddx39b IP values from four replicates. Values are expressed as fold change over wild-type and error bars represent SEM. One way ANOVA followed by Dunnett’s post-test, (*) denotes p-value < 0.05 (Aly p-values: WT vs 85: p = 0.6561, WT vs 115: p = 0.9670, WT vs 154 p = 0.2250, WT vs 622 p = 0.9611, F-value: 1.011, DF = 24; Ddx39b p-values: WT vs 85: p = 0.0133, WT vs 115: p = 0.9944, WT vs 154 p = 0.0760, WT vs 622 p = 0.9993, F-value: 6.025, DF = 19).(g) Matrin 3 IP performed in human lumbar spinal cord nuclear lysates of controls n = 3 and ALS patients n = 3. Immunoblot is probed with Ddx39b and Matrin 3. Arrow indicates IgG heavy chain band. Patient demographics can be found in Supplemental Table 2. Full length blots presented in Supplementary Figure 7.

Figure 5

RNA-FISH and cellular fractionation followed by RT-PCR show defects in RNA export. Experiments performed in NSC-34 cells transiently transfected with either wild-type or mutant Matrin 3. (a–e) mRNA signal of RNA-FISH experiment shown in red, immunofluorescence staining of cells using actin to mark the cell body (white), Flag to mark transfected cells (green) and DAPI to mark the nucleus (blue) (representative images). (f) Nuclear to cytoplasmic mRNA ratio of transfected (T) vs. untransfected (UT) cells for wild-type Matrin 3 and each mutant expressed as fold change over untransfected cells on the same slide, 31–34 cells were measured per genotype collected from three independent experiments. Whiskers indicate 1.5(IQR). One way ANOVA followed by Bonferroni post-test, WT p-value: 0.0697, 85 p-value: 0.0005, 115 p-value: 0.4100, 154 p-value: 0.0073, 622 p-value: 0.0596). (g,h) Cell fractionation followed by RT-PCR on nuclear and cytoplasmic fractions of HEK-293 cells. Values are expressed as average nuclear to cytoplasmic ratio of either TDP-43 or FUS mRNA, normalized to tRNA-Lys for the nuclear fraction and cytochrome b for the cytoplasmic fraction. Error bars represent mean and SEM of three replicates. Experiments were each performed three times, graphs show representative experiment. One way ANOVA followed by Dunnett’s post-test (TDP-43 p-values: WT vs Vector: p = 0.8529, WT vs 85: p = 0.0001, WT vs 115: p = 0.1003, WT vs 154 p = 0.6840, WT vs 622 p = 0.0001, F-value: 74.31, DF = 17; FUS p-values: WT vs Vector p = 0.9999, WT vs 85: p = 0.0054, WT vs 115: p = 0.1336, WT vs 154 p = 0.0045, WT vs 622 p = 0.0001, F-value: 35.21, DF = 17, (*) denotes p-value < 0.05, (**) p < 0.01 and (***) p < 0.001 for both RNA FISH and RT-PCR data.