Interaction of tau with HNRNPA2B1 and N6-methyladenosine RNA mediates the progression of tauopathy

Abstract

The microtubule-associated protein tau oligomerizes, but the actions of oligomeric tau (oTau) are unknown. We have used Cry2-based optogenetics to induce tau oligomers (oTau-c). Optical induction of oTau-c elicits tau phosphorylation, aggregation, and a translational stress response that includes stress granules and reduced protein synthesis. Proteomic analysis identifies HNRNPA2B1 as a principle target of oTau-c. The association of HNRNPA2B1 with endogenous oTau was verified in neurons, animal models, and human Alzheimer brain tissues. Mechanistic studies demonstrate that HNRNPA2B1 functions as a linker, connecting oTau with N⁶-methyladenosine (m⁶A) modified RNA transcripts. Knockdown of HNRNPA2B1 prevents oTau or oTau-c from associating with m⁶A or from reducing protein synthesis and reduces oTau-induced neurodegeneration. Levels of m⁶A and the m⁶A-oTau-HNRNPA2B1 complex are increased up to 5-fold in the brains of Alzheimer subjects and P301S tau mice. These results reveal a complex containing oTau, HNRNPA2B1, and m⁶A that contributes to the integrated stress response of oTau.

Keywords: Alzheimer's disease; METTL3; RNA methylation; RNA translation; fibrils; lamin; neurodegeneration; nuclear envelope; stress granules; tau oligomerization.

Figure 1.. Optogenetic Cry2Olig drives tau oligomerization in primary cortical neurons.

(A) Schematic diagram of the optogenetic platform. (B) Images of blue light-activated assembly of mCherry::Cry2 or 4R1N Tau::Cry2 chimeras in living cells. Scale bar, 20μm. (C) Quantification of rapid clustering of mCherry::Cry2 or 4R1N Tau::Cry2 chimeras in the first and second cycles of 488λ blue light activation in (B), n=10. Linear regression XY analyses **p<0.01. (D) Quantification of long-lasting oligomers at 10 mins after the termination of light. n=10. (E) The Tau::Cry2 and mCherry::Cry2 transduced cultures were exposed to light for 0, 5, 10, 20, 40 and 60 min. S1p TBS-extracted oTau-c fraction was quantified by IB with the Tau13. (F) Quantification of the IB in (E) with GAPDH as an internal control followed by being normalized to the “no light” negative control group. N=3. (G) Images of the seeding activity of total lysate extracted from neurons expressing Tau::Cry2 exposed to 20 or 60 min light. The P3 fraction of fibrillar tau from PS19 P301S mice (9 months old) was used as a positive control. Scale bar, 50μm. (H) Quantification of seeding induced by tau aggregates from the Tau::Cry2 or mCherry::Cry2 neurons. N=6. Error bars = SEM. One-way ANOVA with Tukey’s multiple comparisons test were used for experiments with > 2 groups, *p<0.05, **p<0.01, ***p<0.005.

Figure 2.. Proteomic profiling revealed the evolution of protein-protein interactions in the process of tau oligomerization.

(A) Experimental design. Each condition was performed in triplicate. (B) Volcano plot showing the relative abundance of increased and decreased proteins that were bound to 4R1N Tau::Cry2 chimeras at 20min or 60min of light exposure in comparison to no light control group. Proteins significantly increased or decreased more than 3-fold (p≤ 0.05 and absolute log₂ FC ≥1.58) are shown in red, whereas those exhibiting significant change but less than 3-fold (p≤ 0.05 but absolute log₂ FC < 1.58) are shown in blue. Proteins showing variable changes > 3-fold are marked as green while those without significant change are marked as gray. (C) Venn diagram and gene enrichment of unique proteins binding to oTau-c at 20 and or 60 min. (D) The functional analysis of proteins that exhibit significantly enriched binding to oTau-c at early (20 mins) and late (60 mins) stages of oligomerization. (E) STRING network analysis of proteins that have significantly enriched binding to the complex of oTau-c at 20min and 60mins of light exposure. Proteins were filtered by p<0.05 and absolute log₂FC≥1.58 (more than 3-fold change). Only interactions with a STRING score ≥ 0.8 are shown. Evidence of interaction is represented by the distance between nodes, with more tightly packed nodes having a higher STRING score. Proteins that did not display interactions are not shown. Node color are linearly related to fold-change.

Figure 3.. Tau oligomerization elicits striking change in HNRNPA2B1 localization.

(A) Validation of the mCherry IP samples by IB, including HNRNPA2B1, HNRNPH and Eif3l. (B, C and D) Quantification of the HNRNPH, HNRNPA2B1 and Eif3l band intensities. The integrated intensity of each RBP band was normalized to the band intensity in the corresponding mCherry labeling and then compared to the control group with no light exposure. N=3. (E) Translocation and co-localization of HNRNPA2B1 with tau oligomers at 0, 5, 10, 20, 40, and 60 mins of light exposure. Scale bar, 10μm. (F) Quantification of tau-HNRNPA2B1 co-localization by the ratio of yellow to total green. N=5. (G) Quantification of cytoplasmic granular HNRNPA2B1 intensity. N=5. (H) Images showing the translocation and co-localization of HNRNPA2B1 (green) with tau oligomers (TOMA2, red) in the lateral entorhinal cortex (LEnt) of PS19 P301S tau transgenic mice at age of 3-month, 6-month and 9-month. Scale bar, 5μm. (I) Quantification for the accumulation of tau oligomers in PS19 P301S mice compared to age-matched WT control by the total fluorescence of TOMA2 labeling. N=5. (J) Quantification for the co-localization of HNRNPA2B1 with tau oligomers by the co-efficiency Pearson’s R value of green (HNRNPA2B1) to red (TOMA2). N=5. (K) Proximity ligation assay (PLA) show the co-localization of HNRNPA2B1 and tau aggregates in 9-month aged PS19 P301S mouse brain in comparison to WT control. HNRNPA2B1 was probed using a rabbit antibody and tau aggregates were probed with the mouse Tau 13 antibody. The λ_ex 554 nm; λ_em 576 nm (Cyanine 3; Zeiss Filter set 20) fluorescence activity reflects the co-localization of protein molecules within 40nm. Scale bar 10μm. (L) Quantification of the PLA assay as shown in (K). The data were collected based on the total orange fluorescence intensity and normalized to the fold-change of WT control group. Error bars = SEM. Two-way ANOVA with Tukey’s multiple comparisons test or two-tailed Welch’s t-test, *or #p<0.05, **p<0.01. ****p<0.001.

Figure 4.. m⁶A co-localizes with HNRNPA2B1 and oligomeric tau in P301S tau transgenic mice brain.

(A) Co-localization of HNRNPA2B1 (green) and m⁶A (red) in the lateral entorhinal cortex (LEnt) of PS19 P301S tau transgenic mice at age of 6-month and 9-month, respectively. Scale bar, 20<m. (B, C) Quantification of the increase of m⁶A and HNRNPA2B1 in PS19 P301S mice compared to age-matched WT control based upon the total fluorescence of m⁶A labeling. N=5. (D) Proximity ligation assay (PLA) show the co-localization of HNRNPA2B1 protein with m⁶A transcripts in 9-month aged PS19 P301S mouse brain in comparison to WT control. HNRNPA2B1 was probed with rabbit antibody and m⁶A transcripts were probed with mouse anti-m⁶A antibody. Orange fluorescence intensity reflects the co-localization of HNRNPA2B1 and m⁶A transcripts within 40nm. Scale bar 10μm. (E) Quantification of the PLA assay as shown in (D). The data were collected with the total orange fluorescence intensity and normalized to the fold-change of WT control group. N=5. (F) Translocation and co-localization of m⁶A (red) with oTau (labeled by TOMA2 antibody, green) in the lateral entorhinal cortex (LEnt) of PS19 P301S tau transgenic mice at age of 6-month and 9-month, respectively. Scale bar, 20<m. (G) Quantification of the co-localization of TOMA2 with m⁶A based upon the raw co-localized pixels. N=5. (H) Quantification for the percentage of m⁶A co-localized to tau oligomers by the co-efficiency Pearson’s R value of red (m⁶A) to green (TOMA2). N=5. (I) PLA assay showed the co-localization of m⁶A transcripts with tau aggregates in 9-month aged PS19 P301S mouse brain in comparison to WT control. Tau aggregates were probed with a mouse antibody and m⁶A transcripts were probed with a rabbit anti-m⁶A antibody. Orange fluorescence intensity reflects the co-localization of m⁶A transcripts and tau aggregates within 40nm. Scale bar 10μm. (J) Quantification of the PLA assay as shown in (D). The data were collected based on the total orange fluorescence intensity and normalized to the fold-change of WT control group. N=5 (K) Immunoprecipitation of m⁶A from 6-month PS19 P301S brain cortex lysate. IgG antibody was used to exclude the non-specific binding. The amount of HNRNPA2B1 and tau bound to m⁶A was detected by immunoblot. RNase A or FTO groups were used to deplete the m⁶A in the lysate to confirm the specific binding of HNRNPA2B1 and tau to m⁶A. (L-M) Quantification of HNRNPA2B1 and tau band intensity as shown in (K). N=3. (N-O) The quantity of m⁶A from Tau13 or HNRNPA2B1 immuno-precipitation samples were measure by ELISA. Data were collected from 5 mice brain in each group. Error bars = SEM. Two-way ANOVA with Tukey’s multiple comparisons test was used, Error bars = SEM. Two-way ANOVA with Tukey’s multiple comparisons test or two-tailed Welch’s t-test, **p<0.01, ***p< 0.005, ****p<0.001.

Figure 5.. m⁶A co-localizes with oligomeric tau in post-mortem human AD brain tissue.

(A) Accumulation and co-localization of m⁶A (green) and cytosolic HNRNPA2B1 (magenta) with tau oligomers (red, labeled with the TOMA2 antibody) in post-mortem human AD brain cortical tissue in comparison to different Braak stages. The Plot Profile provides a spatially resolved graph indicating the fluorescence intensity for each channel. Scale bar, 20<m. (B-D) Quantification of oTau accumulation in post-mortem human AD brain tissue of each different Braak stages based on the total fluorescence intensity of TOMA2 labeling (B). Quantification for the increase of m⁶A in post-mortem human AD brain temporal tissue by the total fluorescence intensity of labeling (C). Quantification of HNRNPA2B1 accumulated in the cytosol in which nucleus fraction was subtracted by using DAPI as mask (D). Data were collected from 3 cases of each stage and 5 images of each case. (E-G) The co-localization of TOMA2, HNRNPA2B1 and m6A were quantified and analyzed by the total over-lapping intensity as shown in (A). Data were collected from 3 cases of each stage and 5 images of each case. (H) Immunoprecipitation of HNRNPA2B1 from post-mortem human AD brain tissues or age-matched controls. IgG antibody was used to exclude the non-specific binding. The amount of oligomeric tau bound to HNRNPA2B1 was detected by IB using TOC1 antibody. The amount of pull-down HNRNPA2B1 was also probed by HNRNPA2B1 antibody. Quantification of oligomeric tau bound to HNRNPA2B1. Band intensity of TOC1 was normalized to the HNRNPA2B1 band. N=7 human age-matched control and 7 AD cases. (I) The quantity of m⁶A from Tau13 IP samples was measured by ELISA. Data were collected using samples of human cortical grey matter from individuals at each Braak stage (4 individuals per stage). Data are shown as mean ± SEM. One-way ANOVA with Tukey’s multiple comparisons test or two-tailed Welch’s t-test were used, **p< 0.01, ***p< 0.005, ****p<0.001.

Figure 6.. HNRNPA2B1 knockdown reduces oTau-c induced translational stress, DNA damage and association with N⁶-methyladenosine RNA.

(A) The effects of HNRNPA2B1 knockdown on the association between oTau-c and m⁶A. Neurons were transduced with Tau::Cry2 (red) plus siRNA directed against HNRNPA2B1 or scrambled control. The neurons were then exposed to 488 nm light for 0, 20 or 60 min, fixed and labeled for HNRNPA2B1 (purple), m⁶A (green) or DAPI (blue). Scale bar 10 μm. (B, C) Quantification of siRNA mediated knockdown of HNRNPA2B1. B: IB of lysates run on a 12% reducing gel and probed with anti-HNRNPA2B1 antibody and actin antibody as the internal control. C: N=5. (D) Quantification of the colocalization between oTau-c and m⁶A. Image analysis shows that HNRNPA2B1 knockdown elicited a strong reduction in the fraction of oTau-c colocalized with m⁶A, which is apparent in the reduction of yellow pixels evident visually or by scatterplot (bottom row). (E) Quantitative imaging demonstrates that HNRNPA2B1 knockdown produced a statistically significant reduction in nuclear m⁶A puncta after 60 min of light exposure. (F) IB of puromycin showing the newly synthesized proteins in conditions with siRNA towards HNRNPA2B1 or scrambled control in conditions with no light or 30min light. (G) Quantification of puromycin band intensities, which were internalized by GAPDH band intensity before being normalized to the control group without light or HNRNPA2B1 knockdown. N=4. (H) TUNEL labeling show that 60mins of continuous tau oligomerization induced DNA damage that can be ameliorated by HNRNPA2B1 knockdown. Red shows the mCherry::Cry2Olig or 4R1N Tau::mCherry::Cry2Olig positively transfected neurons. Green represents the TUNEL signaling in the nucleus and DAPI was used for the labeling of nucleus. Scale bar 50μm. (I) Quantification of TUNEL labeling intensity. (J) Cleaved caspase3 labeling show that HNRNPA2B1 knockdown abrogates toxicity induced by extended light exposure in Tau::Cry2 neurons. Red shows the mCherry::Cry2Olig or 4R1N Tau::mCherry::Cry2Olig positively transfected neurons. Green represents cleaved caspase 3 reactivity, and DAPI labels nuclei. Scale bar 20<m. (K) Quantification of cleaved caspase 3 labeling intensity. Data were normalized to the fold-change to their compared control group. Data are shown as mean ± SEM. Two-way ANOVA with Tukey’s multiple comparisons test or unpaired T-test with Welch’s correction, two-tailed, *p<0.05, **p<0.01.

Figure 7.. HNRNPA2B1 is required for tau-mediated neurodegeneration in vivo.

(A) The design for the in vivo experiments. (B) The accumulation and co-localization of oligomeric tau (by TOMA2 antibody, red), m6A (magenta) and HNRNPA2B1 (grey) in the experimental mice brain of CA3 brain region. The efficiency of lentivirus transduction are shown by the GFP-tag (green) co-expressed in the viral construct. Scale bar 50μm, (C-E) Quantification of the HNRNPA2B1, TOMA2 and m⁶A fluorescence intensity in the CA3, respectively. N=5 Data are shown as mean ± SEM. Two-way ANOVA with Tukey’s multiple comparisons test was used, *p<0.05, ***p<0.005, ****p<0.001. (F) Cell apoptosis by cleaved caspase 3 labelling (magenta) and it’s co-localization to oligomeric tau (by TOMA2 antibody, red) and HNRNPA2B1 (grey) in the experimental mice brain of CA3 brain region. The efficiency of lentivirus transduction is shown by the GFP-tag (green). Scale bar 50μm. (G) Quantification of fluorescence cleaved caspase 3 intensity in the CA3, respectively. N=5 Data are shown as mean ± SEM. Two-way ANOVA with Tukey’s multiple comparisons test was used, *p<0.05, ***p<0.005.