Axonal autophagosome maturation defect through failure of ATG9A sorting underpins pathology in AP-4 deficiency syndrome

Abstract

Adaptor protein (AP) complexes mediate key sorting decisions in the cell through selective incorporation of transmembrane proteins into vesicles. Little is known of the roles of AP-4, despite its loss of function leading to a severe early onset neurological disorder, AP-4 deficiency syndrome. Here we demonstrate an AP-4 epsilon subunit knockout mouse model that recapitulates characteristic neuroanatomical phenotypes of AP-4 deficiency patients. We show that ATG9A, critical for autophagosome biogenesis, is an AP-4 cargo, which is retained within the trans-Golgi network (TGN) in vivo and in culture when AP-4 function is lost. TGN retention results in depletion of axonal ATG9A, leading to defective autophagosome generation and aberrant expansions of the distal axon. The reduction in the capacity to generate axonal autophagosomes leads to defective axonal extension and de novo generation of distal axonal swellings containing accumulated ER, underlying the impaired axonal integrity in AP-4 deficiency syndrome.Abbreviations: AP: adaptor protein; AP4B1: adaptor-related protein complex AP-4, beta 1; AP4E1: adaptor-related protein complex AP-4, epsilon 1; ATG: autophagy-related; EBSS: Earle's balanced salt solution; ER: endoplasmic reticulum; GFAP: glial fibrillary acidic protein; GOLGA1/Golgin-97/GOLG97: golgi autoantigen, golgin subfamily a, 1; GOLGA2/GM130: golgi autoantigen, golgin subfamily a, 2; HSP: hereditary spastic paraplegia; LC3/MAP1LC3B: microtubule-associated protein 1 light chain 3 beta; MAP2: microtubule-associated protein 2; MAPK8IP1/JIP1: mitogen-acitvated protein kinase 8 interacting protein 1; NEFH/NF200: neurofilament, heavy polypeptide; RBFOX3/NeuN (RNA binding protein, fox-1 homolog [C. elegans] 3); SQSTM1/p62: sequestosome 1; TGN: trans-Golgi network; WIPI2: WD repeat domain, phosphoinositide interacting protein 2.

Keywords: AP4B1; AP4E1; AP4M1; AP4S1; ER-phagy; SPG47; SPG51; mAtg9; reticulophagy; swelling; varicosities.

Figure 1.

ap4e1 null mice recapitulate neuroanatomical features of AP-4 deficiency. (a) Schematic of promoter driven ap4e1 KO tm1b allele, showing removal of critical exon 3. (b) Western blot prepared from AP-4 line showing total loss of AP4E1 protein and concurrent loss of AP4B1 at protein level in ap4e1 KO animals. (c, d) Sections prepared from animals at 1 and 4 month timepoints stained against RBFOX3/NeuN and GFAP showing lateral ventricular enlargement in KO. Scale bar: 200 μm. (D) Quantification of relative area of lateral ventricle (n = 3/5 animals 1 and 4 months). (e, f) Axons stained using FluoroMyelin at 1 month and 4 months showing thinning of corpus callosum tracts in KO. Scale bar: 200 μm. (f) Quantification of thickness of corpus callosum (n = 3 animals). (g, h) Commissural crossing axons at 1 month, stained against Neurofilament-200 (NEFH/NF200). Scale bar: 100 μm. (h) Quantification of thickness of corpus callosum (n = 5 animals). Quantified data expressed as mean ± SEM. Statistical analysis: Two-tailed unpaired Student’s t-test, *p < 0.05, **p < 0.01 and ***p < 0.001. CC – corpus callosum.

Figure 2.

ATG9A handling is affected in vivo in ap4e1 KO mice. (a) Endogenous co-immunoprecipitation of AP4E1 with ATG9A from mouse brain, showing interaction between AP-4 and ATG9A (n = 3 independent experiments). (b) Western blot of ATG9A in hippocampus at 1 month showing increase in KO animals. (c) Quantification of relative ATG9A protein (n = 3 animals). (d) Sections prepared from AP-4 line at 1 month stained against ATG9A and NEFH/NF200, showing increased immunoreactivity and accumulation of ATG9A within neuronal cell layers. Scale bars: 200 and 50 μm (e) High magnification showing accumulation of ATG9A within cortex and (f) CA1 of hippocampus. Scale bar: 20 μm (n = 5 animals). Quantified data is expressed as mean ± SEM. Statistical analysis: Two-tailed unpaired Student’s t-test, **p < 0.01.

Figure 3.

ATG9A accumulates within the TGN in ap4e1 KO neurons. (a) DIV-8 cultured hippocampal neurons stained against ATG9A and MAP2. Inset panels show zoomed regions indicated by dashed boxes of cell body and accumulation of ATG9A. Scale bars: 20 μm; crop: 5 μm. (b) Quantification of relative ATG9A in neuronal soma (n = 40/20 neurons WT/KO). (c) SIM of DIV-8 cultured hippocampal neurons stained against ATG9A, cis-Golgi marker GOLGA2/GM130 and trans-Golgi marker GOLGA1/GOLG97. Dashed boxes indicate region in magnified panels, showing ATG9A overlapping GOLGA1/GOLG97 in KO. (d) Intensity linescans demonstrate ATG9A retention within the TGN in KO neurons. Scale bars: 5 μm; crop: 0.5 μm (n = 3 experimental repeats). (e) Reconstitution of AP-4 complex with expression of MYC-tagged FL AP4E1 and AP-4 deficiency associated AP4E1^V454X Ap4e1 constructs (refer to Fig S3B). Dashed boxes indicate region in magnified panels. Scale bars: 5 μm; crop: 2 μm. (f) Quantification of relative ATG9A in soma of KO neurons in rescue conditions (n = 26/26/23 neurons FL AP4E1/AP4E1^V454X/UT). (g) Western blot of lysates prepared from cortical neurons after 4, 8 and 12 days in culture probed against ATG9A. (h) Quantification of relative levels of ATG9A (n = 3/4 embryos WT/KO). Quantified data is expressed as mean ± SEM. Statistical analysis: (b, h) Two-tailed unpaired Student’s t-test, (f) Kruskall-Wallis test, *p < 0.05 and ***p < 0.001.

Figure 4.

Defective autophagosome maturation in ap4e1 KO neurons. (a) Endogenous ATG9A vesicles in axons and dendrites, stained using NEFH/NF200 and MAP2 markers respectively. (b) Quantification of ATG9A vesicles per 10 μm² in axons and dendrites. Scale bar: 5 μm. (Axon; n = 19/12 WT/KO, Dendrite; 28/24 WT/KO). (c) Live imaging of nascent autophagosome biogenesis in distal most 75 μm of axon, showing aberrant autophagosome maturation in KO axons. Movies generated over 6 min from cultured hippocampal neurons at DIV-6/7 transfected with RFP-LC3. First frames and kymographs shown with pseudo-coloring of RFP-LC3 signal. X-axis scale bar: 10 μm, Y-axis represents time (1 px/1.5 s). Quantification of; (d) Total and motile RFP-LC3 tracks (e) absolute retrograde displacement, (f) anterograde and retrograde run length per motile autophagosome, (g) total distance travelled per motile autophagosome, (h) proportion of time spent stationary, moving anterogradely or retrogradely per motile autophagosome and (i) velocity of motile autophagosomes. (n = 227/117 motile autophagosomes from 46/36 neurons WT/KO). (Refer to Movies S1 and S2). (j) Autophagic flux assay in DIV-8 cultured cortical cultures. Neurons were treated in the presence of 100 nM bafilomycin for durations (in hours) indicated with either EBSS or 250 nM rapamycin to induce autophagy (n = 6 embryos). (k) Quantification of relative LC3-II and (l) SQSTM1/p62 levels relative to own 0 h control. (m) Western blot of lysates prepared from KO hippocampus at 1 month showing no alteration in endogenous LC3 levels nor processing. (n) Quantification of LC3-II:I ratio, LC3-II and LC3-I (n = 3 animals). (o) Blot of endogenous SQSTM1/p62 in KO hippocampus at 1 month showing reduction in AP-4 KO. (p) Quantification of relative protein level of SQSTM1/p62 (n = 3 animals). Quantified data is expressed as mean ± SEM. Statistical analysis: (b, n and p) Two-tailed unpaired Student’s t-test, (d – i, k and l) Mann-Whitney test, *p < 0.05, **p < 0.01 and ***p < 0.001.

Figure 5.

Axon specific defects in ap4e1 KO neurons. (a) Cultured GFP-filled DIV-14 hippocampal neurons stained against GFP showing neuronal morphology. Scale bar: 100 μm. (b) Analysis of dendritic complexity using 10 μm concentric sholl intersections. Quantification of (c) total dendritic length and (d) total branches per neuron (n = 12/17 neurons WT/KO). (e) Cultured GFP-filled DIV-4 hippocampal neurons stained against GFP revealing neuronal morphology and entire axonal extension from individual neurons. Scale bar: 50 μm. (f) Quantification of total axonal length and (g) axonal branches. (n = 22/18 neurons WT/KO). (h) Quantification of nascent dendritic processes in total length and, (i) total branches (n = 22/27 neurons WT/KO). (j) Inset magnified panel from (e) of distal axonal regions, red arrows indicating axonal swellings. Scale bar: 20 μm. (k) Quantification of number of swellings per neuron (n = 20/17 neurons WT/KO). (l) Distal axonal regions of KO neurons transfected with GFP alone, or in combination with HA-ATG9A, showing a reduction in the number of axonal swellings upon expression of ATG9A, scale bar: 20 μm, (m) Quantification of number of swellings per neuron (n = 29/26 neurons WT/KO). Quantified data is expressed as mean ± SEM. Statistical analysis: (b) Two-way ANOVA with Bonferroni post-hoc test, (c, d, f, h and i) Two-tailed unpaired Student’s t-test, (g, k and m) Two-tailed Mann-Whitney U test, *p < 0.05, **p < 0.01 and ***p < 0.001.

Figure 6.

Distal generation of axonal swelling comprised of ER accumulations in AP-4 deficiency. (a) Long-term imaging of DIV-6 hippocampal neurons transfected with GFP enabling tracking of growing axons. Axonal swellings are de novo generated distally in close proximity to the growth cone in KO axons, indicated by red arrows. Swellings disassemble over time, as indicated by blue arrows (Refer to Movies S3 and S4). Scale bar: 5 μm. (b) Swellings formed in WT axons disassemble rapidly, whereas KO swellings persist at their site of deposition for dramatically longer durations (formation indicated by red arrow, disassembly by blue arrow) (Refer to movies S5 to S8). Scale bar: 5 μm. (c) Cultured DIV-4 hippocampal neurons were identified by DIC (b) for CLEM, showing correlation of axonal region with swellings, scale bars: 20 μm, 10 μm. Please refer to Fig S6B for WT example. Individual swellings (d) and (e) shown at higher magnification, scale bar: 2 μm. Cropped regions of (d) indicated by blue and green dashed boxes, (di) and (dii) respectively, showing accumulation of ER within axonal swelling. (e) Swelling showing incorporation of ER into double membraned autophagosome within axonal swelling, cropped regions indicated by red and yellow dashed boxes, (ei) and (eii). Scale bar: 0.5 μm. (n = 3 experimental repeats).