Mutations in the vesicular trafficking protein annexin A11 are associated with amyotrophic lateral sclerosis

Abstract

Amyotrophic lateral sclerosis (ALS) is a fatal neurodegenerative disorder. We screened 751 familial ALS patient whole-exome sequences and identified six mutations including p.D40G in the ANXA11 gene in 13 individuals. The p.D40G mutation was absent from 70,000 control whole-exome sequences. This mutation segregated with disease in two kindreds and was present in another two unrelated cases (P = 0.0102), and all mutation carriers shared a common founder haplotype. Annexin A11-positive protein aggregates were abundant in spinal cord motor neurons and hippocampal neuronal axons in an ALS patient carrying the p.D40G mutation. Transfected human embryonic kidney cells expressing ANXA11 with the p.D40G mutation and other N-terminal mutations showed altered binding to calcyclin, and the p.R235Q mutant protein formed insoluble aggregates. We conclude that mutations in ANXA11 are associated with ALS and implicate defective intracellular protein trafficking in disease pathogenesis.

Fig. 1.. Annexin A11 mutations identified in ALS patients after stringent filtering of exome sequencing data.

(A and B) Pedigrees of U.K. Family 1 and 2, respectively, carrying the p.D40G mutation. Family members for whom DNA was available for segregation analysis are labeled with M/W for a heterozygous mutation (A > G allele) or W/W for a homozygous reference allele. The gender has been anonymized for each individual. Affected individuals with ALS are denoted by a solid black diamond, unaffected individuals by white diamonds, and unaffected carriers with a black dot. Affected family members who underwent exome sequencing as part of this study are marked with abluestar. (C) The p.G175R mutation segregates in both an index case and affected sibling. (A to C) The age at death, where data are available, is listed above each p.D40G carrier (affected and unaffected). (D) Schematic representation of the annexin A11 protein highlighting the clustering of mutations in the N terminus (first 196 residues). Mutation positions in the N terminus are fully conserved in mammals and conserved in birds, amphibians, and reptiles if the mutations are in annexin domains. Mutations in FALS are indicated by red boxes and those in SALS by blue boxes. The p.D40G mutation found in two index FALS cases and an affected sibling from each U.K. family, an Italian index case, and a U.K. SALS case (n = 6) clusters with the p.G38R mutation in the N terminus of the annexin A11 molecule. The binding site of calcyclin (CACY) is located in the N terminus (residues 50 to 62).

Fig. 2.. Annexin A11 immunohistochemical analysis in postmortem spinal cord tissue from a SALS case with the p.D40G mutation.

(A) Phospho-TDP-43-positive cytoplasmic inclusion from the anterior horn of the spinal cord. (B to E) Annexin A11-positive inclusions in motor neurons of the spinal cord include skein-like structures (B) and filamentous and tubular-shaped structures (C and D). Occasional basket-like inclusions were seen in the spinal cord (E). (F) Abundant annexin A11-positive torpedo-like neuritic structures were present in the neuropil of the motor cortex (red arrows). Representative spinal cord staining for annexin A11 inclusions in a SALS case (n =15) who is negative for ANXA11 mutations (G) and two C9ORF72 expansion-positive ALS cases (spinal cord and frontal cortex) (H and I). (J) Staining for annexin A11 inclusions was also negative in an ALS case harboring a p.D101G mutation in SOD1 (spinal cord). Similarly, frontotemporal lobar degeneration (FTLD)-TDP-43 cases (spinal cord; n = 3) (K) and Alzheimer’s disease (AD) cases (n = 3) (L) and control individuals (n = 13) (M) were also negative for staining for annexin A11 inclusions. (N and O) Double labeling of spinal cord tissue in the p.D40G SALS case for phospho-TDP-43 (green) and annexin A11 aggregates (red). (P) Costaining for ubiquitinated aggregates (red, annexin A11; green, ubiquitin) showed occasional colocalization (white arrow). Scale bars, 30 μm (A, E, G, and H), 20 μm (B and C), 15 μm (D), 50 μm (F, I, J, K, and L), 25 μm (M and N), and 50 μm (O and P).

Fig. 3.. Transfection of WT and mutant annexin A11 into mouse primary motor neurons and HEK cells.

(A) HA tagged annexin A11^WTand mutant annexin A11^{G38R/D40G/R235Q} constructs were transfected into mouse primary motor neurons. Transfected cells displayed (A) cytoplasmic vesicle-like structures for annexin A11^WT/G38R/D40G and (B) increased localization of smaller foci in the cytoplasm of annexin a11^R235Q transfected neurons [***P = 0.0004, one-way analysis of variance (ANOVA) and Dunnett’s post hoc test; bars represent mean and SEM]. Vesicles were defined as structures with a diameter between 0.5 and 1.9 μm (mean, 1 μm) and foci as structures with a diameter between 0.16 and 0.5 μm (mean, 0.3 μm). (C) A significant proportion of foci were associated with annexin A11^R235Q (****P < 0.0001, one-way ANOVA and Dunnett’s post hoc test); a smaller proportion of vesicles were associated with annexin A11^G38R and were absent with annexin A11^R235Q (**P = 0.0051 and ****P < 0.0001, respectively, one-way ANOVA and Dunnett’s post hoc test). (D) Transfection of HEK cells with GFP tagged annexin A11^{WT/G38R/D40G/G189E/R235Q} constructs followed by a solubility assay [showing lysate (L), soluble (S), and insoluble (I) fractions] revealed that annexin A11^R235Q formed detergent-resistant insoluble aggregates. (E) The amount of the annexin A11^R235Q insoluble fraction is statistically significant when compared with that for the WT protein (***P = 0.007, one-way ANOVA with Dunnett’s post hoc test).

Fig. 4.. Annexin A11 with the p.R235Q mutation sequesters WT annexin A11.

(A)CoexpressionofANXA11^WT tagged with HA and ANXA11^R235Q tagged with GFP in SH-SY5Y cells demonstrated colocalization by immunocytohistochemistry, suggesting that the mutant protein sequestered the WT protein. Scale bars, 10 μm. (B) To confirm this observation, we conducted solubility assays, generating lysate, soluble, and insoluble fractions, followed by Western blot. Expression of GFP-tagged ANXA11^WT and GFP-tagged ANXA11^R235Q in HEK cells replicated that seen in Fig. 3D, with aggregates seen only in the insoluble fraction (labeled as I) of ANXA11^R235Q. Coexpression of GFP-tagged ANXA11^WT or ANXA11^R235Q with an equal amount of HA-tagged ANXA11^WT in HEK cells resulted in accumulation of HA-tagged ANXA11^WT in the insoluble fraction only when coexpressed with GFP-tagged ANXA11^R235Q (indicated by a red cross) (n = 3). (C) A further IP assay also confirmed sequestering of ANXA11^WT by ANXA11^R235Q. Cotransfection of HEK cells with GFP-tagged ANXA11^R235Q and HA-tagged ANXA11^WT followed by IP with rabbit anti-GFP antibody, and probing with mouse anti-HA antibody demonstrated direct sequestering of ANXA11^WT in the IP fraction. This included higher insoluble molecular weight species compared to control empty plasmid vector and empty plasmid vector plus GFP (top). Verification of the IP was shown by staining with mouse-GFP (input total protein lysate and IP;middle)and positive staining of HA-tagged ANXA11^WT [input and nonbound bead flow-through (FT);top].

Fig. 5.. Annexin A11 mutations disrupt binding to calcyclin.

(A) Western blot demonstrating lack of binding of FLAG-tagged calcyclin to GFP-tagged mutant annexin A11 (ANXA11^D40G, ANXA11^G189E, and ANXA11^R235Q) compared to ANXA11^WT after IP (n = 3).Top: GFP intensities of input and immunoprecipitated fractions, with immunoglobulin G (IgG) heavy [50kDA (*)]and light [25 kDa (**)] chains indicated. Bottom: Input and IP of FLAG-tagged calcyclin alone. Ctrl, control. (B) Cross section of postmortem spinal cord tissue showing calcyclin expression in (i) controls and (ii) a SALS case with the p.D40G mutation. High calcyclin expression can be seen in the lateral corticospinal tracts (black arrows). (iii) SALS case with no known ALS causing mutation but displaying strong expression of calcyclin in the lateral corticospinal tracts in postmortem spinal cord (black arrows). Scale bars, 40 μm (i), 30 μm (ii), and 25 μm (iii). (C) To determine the effect of calcyclin overexpression on aggregation, we conducted an NP-40 solubility assay by coexpressing 500 ng of GFP-tagged ANXA11^WT or ANXA11^R235Q and 500 ng of FLAG-tagged calcyclin in HEK cells. Clearance of insoluble p.R235Q mutant annexin A11 aggregates by calcyclin was observed. Addition of the proteasomal inhibitor MG132 restored the aggregation event by blocking degradation of the aggregated proteins by the ubiquitin proteosomal pathway. An equivalent amount of empty GFP vector (500 ng) was added to HEK cells expressing WT or p.R235Q mutant annexin A11 as a loading control. (D) Quantification of the proportion of insoluble fractions from (C) relative to untreated ANXA11^GFP (WT) including calcyclin overexpression (+C) and calcyclin overexpression plus MG132 treatment (+C +M). Bands were quantified with ImageJ, and intensity was calculated with respect to untreated GFP-tagged ANXA11^WT. A one-way ANOVA test demonstrated significance for increased aggregation due to the p.R235Q mutation and rescue of aggregation in p.R235Q mutant cells by calcyclin overexpression and MG132 treatment (R235Q + C + M). *P < 0.05 and ***P < 0.001, respectively, one-way ANOVA and Dunnett’s post hoc test (n = 3).