Seeding competent TDP-43 persists in human patient and mouse muscle

Abstract

TAR DNA-binding protein 43 (TDP-43) is an RNA binding protein that accumulates as aggregates in the central nervous system of some neurodegenerative diseases. However, TDP-43 aggregation is also a sensitive and specific pathologic feature found in a family of degenerative muscle diseases termed inclusion body myopathy (IBM). TDP-43 aggregates from ALS and FTD brain lysates may serve as self-templating aggregate seeds in vitro and in vivo, supporting a prion-like spread from cell to cell. Whether a similar process occurs in IBM patient muscle is not clear. We developed a mouse model of inducible, muscle-specific cytoplasmic localized TDP-43. These mice develop muscle weakness with robust accumulation of insoluble and phosphorylated sarcoplasmic TDP-43, leading to eosinophilic inclusions, altered proteostasis and changes in TDP-43-related RNA processing that resolve with the removal of doxycycline. Skeletal muscle lysates from these mice also have seeding competent TDP-43, as determined by a FRET-based biosensor, that persists for weeks upon resolution of TDP-43 aggregate pathology. Human muscle biopsies with TDP-43 pathology also contain TDP-43 aggregate seeds. Using lysates from muscle biopsies of patients with IBM, IMNM and ALS we found that TDP-43 seeding capacity was specific to IBM. Surprisingly, TDP-43 seeding capacity anti-correlated with TDP-43 aggregate and vacuole abundance. These data support that TDP-43 aggregate seeds are present in IBM skeletal muscle and represent a unique TDP-43 pathogenic species not previously appreciated in human muscle disease.

Fig. 1.. Confirmation of TDP-43 seeding in patient brain and muscle samples using a FRET sensor line.

(A) TDP-43 FRET biosensors express two independently tagged c-terminal fragments (aa262-414) of human TDP-43 fused to mruby and mclover. When TDP-43 aggregation occurs, the two fluorophores are close enough for an energy transfer to occur and the signal is detected by flow cytometry. (B) Representative fluorescent images of the FRET sensor line and the flow cytometry gating strategy using lipofectamine alone, 100nM TDP-43 monomeric protein, or 100 nM TDP-43 preformed fibrils (PFF). Note only TDP-43 PFF induces a FRET signal. (C) A dose response curve of the TDP-43 biosensor line treated with buffer, lipofectamine, TDP-43 monomer, or increasing concentrations of recombinant TDP-43 PFF. (D) The RIPA-insoluble fraction of human ALS patient autopsy brain tissue was added to the FRET sensor line with lipofectamine and analyzed by flow cytometry after 72 hours. Note the increase in FRET signal from ALS brain tissue. (E) Representative immunohistochemical images of TDP-43 staining from human muscle biopsies with TDP-43 inclusions (arrows). Black dotted lines outline individual muscle fibers. (F) The RIPA-insoluble fraction from control muscle or muscle biopsies confirmed to have positive TDP-43 aggregate staining were added to the FRET sensor line to confirm seeding. (G) Samples from muscle and brain lysates with established TDP-43 seeding ability were added to an α-synuclein FRET sensor line to test for TDP-43 seeding specificity. In contrast to TDP-43 biosensors, no α-synuclein seeding is present. FRET signal is measured as integrated FRET density, % positive cells x median fluorescence intensity and normalized to a lipofectamine only control. *P<0.05, **P<0.01, ***P<0.001 by one-way ANOVA followed by Dunnett’s multiple comparisons test to lipofectamine control.

Fig. 2.. Mice expressing cytoplasmic hTDP-43 develop a myopathy with premature death.

(A) HSA-rtTA mice were crossed with tetO-hTDP-43^∆NLS mice to create doxycycline-inducible muscle-specific expression of hTDP-43^∆NLS. (B) A cohort of mice were studied over the course of a 4 week dox treatment (shaded in green on the survival curve), tracking total body weight, strength, and survival time. N = 13 for survival curve, N = 11 for weight and grip strength, N = 7 for hanging wire as 4 mice were too frail to perform the test. (C) After 4 weeks of dox treatment, H&E staining showed mild myopathic changes with varied fiber sizes. Non-nuclear eosinophilic staining in H&E and pale regions on NADH staining (arrows) highlight sarcoplasmic inclusion. (D) Mice treated with dox chow for 4 weeks compared to age-matched no dox controls showed decreased hind limb muscle weights as well as decreased total body weight compared to untreated controls. (E) Example x-ray images of spinal kyphosis in HSA-hTDP-43^∆NLS mice after 3 weeks on dox chow and the calculated kyphosis index. *P>0.05, **P<0.01, ***P<0.001; unpaired Student’s t-test.

Fig. 3.. HSA-hTDP-43^∆NLS mice develop abundant sarcoplasmic insoluble phosphorylated TDP-43 aggregates which disrupt proteostasis.

(A) Representative immunofluorescent images of hindlimb cross sections of HSA-hTDP-43^∆NLS mice after 4 weeks of transgene activation by dox chow, showing abundant TDP-43 (red upper panel) and phosphorylated TDP-43 aggregates (red lower panel). Dashed white lines indicate muscle fiber outlines. (B) Gastrocnemius muscle lysates from HSA-hTDP-43^∆NLS mice were processed for insoluble fractionation followed by western blot to detect total, soluble, and insoluble levels of TDP-43 and pTDP-43 which showed an increase over time. The total GAPDH can be found in part D. (C) Mice treated with dox chow for 4 weeks were evaluated for TDP-43 transgene expression in non-skeletal muscle tissue. Notably, TDP-43 was increased in cardiac tissue. (D) Gastrocnemius muscle lysates from HSA-hTDP-43^∆NLS mice processed at the indicated time points were evaluated using antibodies against ubiquitin, p62, VCP, optineurin, LC3, Lamp2, HSP70, DNAJB6, and GAPDH. (E) Representative dual fluorescent imaging of pTDP-43 (red) and p62 (green upper panel) or ubiquitin (green lower panel) at 2 weeks of dox treatment. Dashed white lines indicate muscle fiber outlines.

Fig. 4.. TDP-43 aggregates are cleared and proteostasis resolved following removal of doxycycline.

(A) Representative immunofluorescent images of quadriceps femoris muscles stained for pTDP-43 (red) in HSA-hTDP43^∆NLS mice at the indicated timepoints when treated with dox chow for 2 weeks followed by a return to a regular chow diet to turn off transgene expression. By three weeks no pTDP-43 is seen. (B) Insoluble fractionation western blot of HSA-hTDP43^∆NLS mouse gastrocnemius muscles shows accumulation and then resolution of insoluble TDP-43 and pTDP-43 over the time course of two weeks doxycycline followed by 3 weeks of recovery. (C) Gastrocnemius muscle lysates from HSA-hTDP43^∆NLS mice processed at the indicated time points were evaluated using antibodies against ubiquitin, p62, VCP, optineurin, LC3, Lamp2, HSP70, DNAJB6 and GAPDH. (D) Representative H&E and NADH images of quadriceps femoris muscle after 3 weeks of recovery as compared to no dox controls. Note the presence of smaller fibers with centralized nuclei (arrows).

Fig. 5.. Ultrastructural analysis of HSA-hTDP-43^∆NLS mouse muscle.

(A) Control TA muscle fiber with myonuclei. (B-D) HSA-hTDP-43^∆NLS mice treated for two weeks with doxycycline. Note large granular amorphous inclusions that are subsarcolemmal and myonuclei-adjacent (arrows). (D) Higher magnification of the granular structure of an aggregate from a mouse on dox treatment for 2 weeks. (E-F) HSA-hTDP-43^∆NLS mice treated for two weeks with doxycycline and then changed to normal chow for three weeks. Note autophagic debris and vacuolation in and around the inclusion (arrowheads).

Fig. 6.. Cryptic exon inclusion is an early event in TDP-43 proteinopathy development.

(A) Disrupted TDP-43 RNA processing leads to the inclusion of cryptic exons which are normally repressed by TDP-43. (B) Agarose gel electrophoresis of products from the RT-PCR amplification of exons 1–2 of Sh3bgr from TA muscles of HSA-hTDP-43^∆NLS mice treated for one week with doxycycline and one week after a return to regular chow. A more slowly migrating band is consistent with exon inclusion. (C) A similar experiment as in B using mice treated or recovered for the indicated times. (D) A similar experiment as in B using mice treated out to four weeks on dox chow.

Fig. 7.. TDP-43 seeding in HSA-hTDP-43^∆NLS mouse muscle is detected by FRET assay.

(A) The insoluble fractions from muscle of HSA-hTDP43^∆NLS mice either treated with no dox or 3 weeks of dox chow were applied to the TDP-43 biosensor and an α-synuclein biosensor. Recombinant TDP-43 and α-synuclein PFF were used for positive controls for each cell line. The insoluble fractions from mice treated with dox chow for three weeks showed robust seeding in the TDP-43 biosensor and none in the α-synuclein biosensor, indicating aggregate specificity. (B) Immunoblot from HEK293 cells expressing a tetracycline-inducible mcherry-tagged TDP-43^ΔNLS. Expression of TDP-43^ΔNLS was induced in the cell line for 16 hours and the insoluble fraction from 4 week on dox HSA-hTDP43^∆NLS mouse muscle was added to the cells and incubated for 72 hours. The cell lysates were then prepared for insoluble fractionation western blot. The smear of insoluble protein in the lanes of 4 week dox treated muscle indicates successful aggregate seeding. (C) HSA-hTDP43^∆NLS mice were treated with dox for two weeks and allowed to recover with regular chow for three weeks, as in Figure 4. Seeding peaked 1 week following the return to dox chow and persisted by three weeks. (D) Follow-up study demonstrating recovery out to 8 weeks where seeding was still present.

Fig. 8.. TDP-43 seeding is present in sIBM patient muscle but does not correlate with histological pathology.

(A) The insoluble fraction from muscle biopsies of 24 sIBM patients, along with 5 healthy controls, 10 with IMNM, and 10 with ALS were added to the TDP-43 FRET sensor line. Of the sIBM samples, 14 were statistically significant. *P<0.05, **P<0.01, ***P<0.001 One-way ANOVA with Dunnett’s multiple comparisons test to lipofectamine only control. (B) Representation of each patient’s average integrated FRET density (IFD). P*<0.05 with one-way ANOVA and Dunnett’s multiple comparisons test to healthy controls. The amount of seeding activity had a negative correlation with histological hallmarks of IBM such as TDP-43 aggregates (C) and rimmed vacuoles (D). (E) Representative images of a disease control, low seeding sample, and high seeding sample with their corresponding FRET values (samples indicated by arrows in (A). Scalebars represent 50 µm.