Frontotemporal dementia causative CHMP2B impairs neuronal endolysosomal traffic-rescue by TMEM106B knockdown

Abstract

Mutations in the endosome-associated protein CHMP2B cause frontotemporal dementia and lead to lysosomal storage pathology in neurons. We here report that physiological levels of mutant CHMP2B causes reduced numbers and significantly impaired trafficking of endolysosomes within neuronal dendrites, accompanied by increased dendritic branching. Mechanistically, this is due to the stable incorporation of mutant CHMP2B onto neuronal endolysosomes, which we show renders them unable to traffic within dendrites. This defect is due to the inability of mutant CHMP2B to recruit the ATPase VPS4, which is required for release of CHMP2B from endosomal membranes. Strikingly, both impaired trafficking and the increased dendritic branching were rescued by treatment with antisense oligonucleotides targeting the well validated frontotemporal dementia risk factor TMEM106B, which encodes an endolysosomal protein. This indicates that reducing TMEM106B levels can restore endosomal health in frontotemporal dementia. As TMEM106B is a risk factor for frontotemporal dementia caused by both C9orf72 and progranulin mutations, and antisense oligonucleotides are showing promise as therapeutics for neurodegenerative diseases, our data suggests a potential new strategy for treating the wide range of frontotemporal dementias associated with endolysosomal dysfunction.

Figure 1

Physiological expression of mutant CHMP2B results in a decrease of endolysosomes at the soma of cortical neurons. (A) Western blot of CHMP2B levels on brain homogenates from P0 or P1 postnatal mutant CHMP2B mice or non-transgenic controls. β-actin is shown as a loading control. Bracket indicates non-specific bands. (B) Quantification of bands shown in A relative to loading controls. (C) LAMP2 staining in mutant CHMP2B and control β-tubulin stained cortical neurons. Scale bar = 10 µm. (D) Quantification of LAMP2-positive structures in DIV 14–16 cortical neurons. n = 3, with 6–10 neurons per n. Unpaired t-test, **P < 0.01.

Figure 2

Mutant CHMP2B decreases endolysosomal trafficking. (A) Representative images of LysoTracker® labelled control and mutant CHMP2B cultures. Scale bar = 10 µm. (B) Example kymographs from 50 µm sections of dendrite labelled with LysoTracker® and imaged for 120 s. Vertical scale bar = 20 s, horizontal scale bar = 10 µm. (C) Quantification of moving and stationary LysoTracker® structures from kymographs. n = 3, with three to five neurons per n, DIV 12–20 cortical cultures. (D) Representative images of GFP-LAMP transfected control and mutant CHMP2B cortical cultures. Scale bar = 10 µm. (E) Kymographs from sections of dendrite labelled with GFP-LAMP and live cell imaged for 120 s. Vertical scale bar = 20 s, horizontal scale bar = 10 µm. (F) Quantification of moving and stationary GFP-LAMP structures from kymographs. n = 3 for mutant CHMP2B, n = 4 for non-transgenic with four to seven neurons per n, DIV 10–14 cortical cultures. Unpaired t-test, *P < 0.05.

Figure 3

Mutant CHMP2B structures are stationary and show no fluorescence recovery after photobleaching. GFP-LAP tagged mutant (A) or wild-type (B) CHMP2B in DIV 10 primary cortical cultures. (C) Insets from A and B. The structures highlighted by boxes were bleached at the indicated time point, and the recovery of fluorescence followed over time. Representative time points are shown (1, 11, 20 and 50 s). (D) Quantification of the average fluorescence recovery over time for mutant CHMP2B (circles) or wild-type CHMP2B (triangles) normalized to the starting fluorescence. (E) Representative time points of immobile GFP-LAP CHMP2B^Intron5 structures (arrows) and moving GFP-LAP CHMP2B^Wildtype structures (arrowheads). (F) Quantification of the number of traces per transfected cell generated using Trackmate in ImageJ. (G) Quantification of the average displacement of traces automatically generated in ImageJ. n = 10 DIV 10 neurons from two independent experiments. Unpaired t-test, ***P < 0.001, ****P < 0.0001. Scale bars = 5 µm.

Figure 4

Mutant CHMP2B does not recruit VPS4. (A) Representative examples of neurons co-transfected with haemagglutinin (HA) tagged CHMP2B and GFP-VPS4. Asterisk indicates an example of localization of mutant HA-CHMP2B, which is negative for VPS4. (B) Representative examples of neurons co-transfected with GFP-VPS4EQ and HA-tagged wild-type (top) or mutant CHMP2B (bottom). Arrowhead (top) indicates area of co-localization of wild-type CHMP2B with VPS4EQ. Arrows (bottom) show differential localization of mutant CHMP2B and VPS4EQ. DIV 8 cortical cultures. (C) Manders coefficient of co-localization generated using JaCoP in ImageJ. Unpaired t-test, *P < 0.05. Scale bars = 10 µm.

Figure 5

Peripheral neurite branching is increased in mutant CHMP2B neurons. (A) Representative maximum intensity projections of DIV 7 non-transgenic control and mutant CHMP2B neurons as labelled. (B) Sholl analysis of neurite branching for control and mutant CHMP2B neurons. (C) Average number of branches per Sholl intersection for 10–120, and 130–250 µm from the cell soma. Control data are shown in grey, mutant CHMP2B in black. Unpaired t-test, *P < 0.05. n = 7 for both, with four to six neurons per n. Scale bar = 10 µm.

Figure 6

Knockdown of Tmem106b with ASOs rescues neuritic branching and endolysosomal trafficking defects. (A) Quantification of Tmem106b mRNA levels in mutant CHMP2B cortical cultures following treatment with the indicated ASOs normalized to GAPDH, n = 2. (B) Blot and quantification of TMEM106B protein levels from primary cultures treated with the indicated ASOs. (C) Representative maximum intensity projections of mutant CHMP2B neurons at DIV 7 treated with 5 µM of the indicated ASOs for 7 days. Scale bar = 10 µm. (D) Sholl analysis of ASO treated neurons normalized to control ASO treated neurons. ASO 687524 data are shown in blue, ASO 687552 in red. Two-way ANOVA with Dunnett’s multiple comparisons. *P < 0.05. Blue asterisks indicate statistical significance between control ASO and ASO 687524. Red asterisks indicate significance between control ASO and ASO 687552. n = 6 for each condition, with 4–10 neurons per n.

Figure 7

Knockdown of Tmem106b with ASOs rescues endolysosomal trafficking defects. (A) Representative images of mutant CHMP2B neurons treated with 5 µM of the indicated ASOs for 7 days loaded with LysoTracker®. Scale bar = 10 µm. (B) Representative kymographs from LysoTracker® live cell recordings of neurons labelled with the indicated ASOs. Vertical scale bar = 20 s, horizontal scale bar = 10 µm. (C) Quantification of kymographs of LysoTracker® movement in DIV 14 mutant CHMP2B cortical cultures following treatment with the indicated ASOs. n = 4 with five to seven neurons per n. (D) Quantification of the number of LysoTracker® labelled structures in the sections of neurite chosen for kymograph analysis. n = 4 with five to seven neurons per n. One-way ANOVA with Newman-Keuls multiple comparison test, **P < 0.01.

Figure 8

The size, number and motility of GFP-LAP CHMP2B^Intron5 puncta is not altered in TMEM106B ASO-treated neurons. (A) Representative images of neurons transfected with GFP-LAP CHMP2B^Intron5 and treated with 5 µM of the indicated ASOs. Scale bar = 10 µm. (B) Quantification of the size of GFP puncta, and quantification of the number of GFP puncta per 50 µm section of proximal dendrite in fixed neurons. n = 3 with 3–10 DIV 10 neurons per n. (C) Quantification of the number of motile structures in live-imaged GFP-LAP CHMP2B^Intron5 transfected cells treated with the indicated ASOs. n = 3 with five DIV 10–11 neurons per n.