An ANXA11 P93S variant dysregulates TDP-43 and causes corticobasal syndrome

Abstract

Introduction: Variants of uncertain significance (VUS) surged with affordable genetic testing, posing challenges for determining pathogenicity. We examine the pathogenicity of a novel VUS P93S in Annexin A11 (ANXA11) - an amyotrophic lateral sclerosis/frontotemporal dementia-associated gene - in a corticobasal syndrome kindred. Established ANXA11 mutations cause ANXA11 aggregation, altered lysosomal-RNA granule co-trafficking, and transactive response DNA binding protein of 43 kDa (TDP-43) mis-localization.

Methods: We described the clinical presentation and explored the phenotypic diversity of ANXA11 variants. P93S's effect on ANXA11 function and TDP-43 biology was characterized in induced pluripotent stem cell-derived neurons alongside multiomic neuronal and microglial profiling.

Results: ANXA11 mutations were linked to corticobasal syndrome cases. P93S led to decreased lysosome colocalization, neuritic RNA, and nuclear TDP-43 with cryptic exon expression. Multiomic microglial signatures implicated immune dysregulation and interferon signaling pathways.

Discussion: This study establishes ANXA11 P93S pathogenicity, broadens the phenotypic spectrum of ANXA11 mutations, underscores neuronal and microglial dysfunction in ANXA11 pathophysiology, and demonstrates the potential of cellular models to determine variant pathogenicity.

Highlights: ANXA11 P93S is a pathogenic variant. Corticobasal syndrome is part of the ANXA11 phenotypic spectrum. Hybridization chain reaction fluorescence in situ hybridization (HCR FISH) is a new tool for the detection of cryptic exons due to TDP-43-related loss of splicing regulation. Microglial ANXA11 and related immune pathways are important drivers of disease. Cellular models are powerful tools for adjudicating variants of uncertain significance.

Keywords: ANXA11; TDP‐43; corticobasal syndrome; variant of uncertain significance.

FIGURE 1

P93S kindred. (A) Pedigree of P93S family. Genetic testing is available from three individuals including two symptomatic and one asymptomatic; the position 93 amino acid is indicated by a letter, where P indicates proline and S indicates serine; unavailable testing is indicated by a question mark. Symptomatic individuals (blue) include patients 1, 2, 3, and 4. Diagonal lines indicate individual is deceased. Circles indicate female, squares male. (B) Representative MRI findings. T2‐weighted FLAIR MRI images from patients 1 and 2 demonstrating central‐predominant atrophy in a frontoparietal distribution and prominent white matter hyperintensities in the cerebral cortex and white matter hyperintensities on spinal cord images from patient 2, indicated by arrowheads.

FIGURE 2

ANXA11 structure. (A) Schematic of ANXA11. Low‐complexity domain (LCD) is shown in blue, annexin domains in black. The P93S variant is identified by a red line; remaining variants/mutations are indicated by black lines. (B and C) ANXA11 sequence conservation. VarSite illustration of sequence conservation with proline in position 93 indicated in blue (B) and sequence conservation of annexin A11 across vertebrate species (C) with the P93S variant indicated in red. (D) Poor performance of in silico variant prediction models for mutations in LCDs. P93S VUS with other known pathogenic mutations in common amyotrophic lateral sclerosis/frontotemporal dementia‐associated genes with LCDs is shown. The predictions of pathogenicity by color: predicted benign in green, moderate in orange, pathogenic in red, and indeterminate in blue. GERP score considered deleterious when >2. MetaLR score ranges from 0 benign to 1 deleterious. PolyPhen score considered deleterious when >0.446. CADD score considered deleterious when >30. REVEL score ranges from 0 benign to 1 deleterious. SIFT score considered deleterious when <0.05. Mutation assessor score ranges from 0 benign to 1 deleterious. EVE score considered potentially deleterious when >0.5 and deleterious when >0.7.

FIGURE 3

Functional impact of P93S variant. (A and B) Decreased colocalization of lysosomes in mutant ANXA11 neurons. (A) Representative images of iPSC‐derived neurons expressing WT and mutant ANXA11 (green) and LAMP1 lysosomal marker (red) showing colocalization of lysosomes with ANXA11 puncta indicated by arrowheads; scale bar = 25 µm, inset scale bar = 4 µm. (B) Quantification of number of lysosomes with ANXA11, well mean indicated by dot, horizontal line indicates median, p = .0009. (C and D) Decreased neuritic RNA in mutant ANXA11 neurons. (C) Representative images of fixed iPSC‐derived neurons expressing WT and mutant ANXA11 with in situ hybridization probes for β‐actin RNA using RNAscope to identify neuritic RNA indicated by arrowheads, scale bar = 25 µm, inset scale bar = 4 µm. (D) Quantification of proportion of neuritic RNA over total RNA, well mean indicated by dot, horizontal line indicates median, p = .03.

FIGURE 4

Decreased nuclear TDP‐43 and formation of cryptic exons. (A and B) Decreased nuclear TDP‐43 in mutant ANXA11 neurons. (A) Representative images of fixed iPSC‐derived neurons expressing wild‐type (WT) and mutant ANXA11 (green) stained with TDP‐43 (magenta) and Hoechst nuclear counterstaining (blue) demonstrating nuclear clearing of TDP‐43 in mutant ANXA11 cells; scale bar = 25 µm, inset scale bar = 1.75 µm. (B) Quantification of mean TDP‐43 intensity in WT compared to mutant ANXA11, well mean indicated by dot, horizontal line indicates median, p < .0001. (C and D) Detection of STMN2 cryptic exon formation in TDP‐43 KD neurons. (C) Representative images of fixed iPSC‐derived CRISPRi neurons with control non‐targeting and TDP‐43 knockdown guides demonstrating detection of native STMN2 RNA (green) and cryptic RNA (magenta) using HCR FISH probes with Hoechst nuclear counterstaining (blue), scale bar = 25 µm, inset scale bar = 4 µm. (D) Quantification of cryptic exon counts per cell in non‐targeting and TDP‐43 KD cells for STMN2, well mean indicated by dot, horizontal line indicates median, Mann–Whitney p < .0001. (E and F) Increased STMN2 cryptic exon formation in mutant ANXA11 neurons. (E) Representative images of fixed iPSC‐derived neurons expressing WT and mutant ANXA11 (green) with HCR FISH probes for native STMN2 RNA (red) and cryptic RNA (magenta) with Hoechst nuclear counterstaining (blue); scale bar = 25 µm, inset scale bar = 4 µm. (F) Quantification of cryptic exon counts per cell in WT and mutant ANXA11 cells for STMN2, well mean indicated by dot, horizontal line indicates median, p <0.0001.

FIGURE 5

Transcriptomic signature of P93S. (A–C) Expression map of ANXA11 in human brain cells. (A) Single‐nucleus sequencing data from control brains demonstrating the two cell types that highly express ANXA11, microglia, and neurons. Unsupervised clustering of 34 cell types in human brain. UMAP projections illustrating different cell types identified are shown on left (A), and ANXA11 expression levels are shown on right (B). Scale represents log₂ normalized average gene expression levels. (C) Dot plot illustrating mean normalized expression of ANXA11 by cell type. Scale represents percentage of total cell population expressing ANXA11, with color representing mean normalized expression per cell type. (D and E) Differential gene expression between WT and mutant ANXA11. Bland–Altman mean difference plots where each dot indicates a gene for which there are counted reads from scRNAseq in iPSC‐derived neurons (D) and iPSC‐derived microglia (E). The x‐axis is average normalized counts, the y‐axis is log₂ fold change. Neuronal differential expression yields few significant genes related to transcriptional regulation and endocytic vesicles (D) while many more differentially expressed in microglia related to transcriptional regulation and endocytic vesicles (E). (F and G) Differential microglial gene expression interferon signaling pathway. (F) STRING diagram of differentially expressed microglial genes in interferon signaling pathway and innate immune response. Colors indicate p _adj. (G) Gene Ontology terms for biological processes of microglial gene expression hits indicating interferon response and innate immunity. Dot size indicates number of genes in each GO grouping, and color indicates −log₁₀(FDR).

FIGURE 6

Proteomic signature of P93S. (A) Volcano plot of proteomic changes in neurons identifying significant changes in proteins involved in vesicle transport and SNAREs, corroborating functional assays indicating that P93S disrupts the proper functioning of ANXA11. (B) Volcano plot of proteomic changes of mutant compared to WT microglia implicating inflammatory pathways, as well as transcriptional regulation and vesicular transport. Red dots indicate upregulated proteins (linear FC > 1.5) and blue dots indicate downregulated proteins (linear FC < ‐1.5). The dotted line demarcates a −log₁₀ of p _adj value of 1.3. (C–H) Unique microglia proteomic signature. (C) Venn diagram of significant differentially expressed proteins in neuron (n = 359) compared to microglia (n = 448). Significance is defined as p _adj > .05. (D–F) Differentially selected microglial peptide gene ontology molecular function (D), cellular component (E), and biological process (F) terms sorted by p value with brighter colors indicating more significant values. See Supplementary Material for detailed results. (G and H) Differentially selected microglial peptide reactome (G) and KEGG pathway (H) terms sorted by p value, with brighter colors indicating more significant values. See Supplementary File for expanded results.