Clec16a is Critical for Autolysosome Function and Purkinje Cell Survival

CLEC16A is in a locus genetically linked to autoimmune diseases including multiple sclerosis, but the function of this gene in the nervous system is unknown. Here we show that two mouse strains carrying independent Clec16a mutations developed neurodegenerative disease characterized by motor impairments and loss of Purkinje cells. Neurons from Clec16a-mutant mice exhibited increased expression of the autophagy substrate p62, accumulation of abnormal intra-axonal membranous structures bearing the autophagy protein LC3, and abnormal Golgi morphology. Multiple aspects of endocytosis, lysosome and Golgi function were normal in Clec16a-deficient murine embryonic fibroblasts and HeLa cells. However, these cells displayed abnormal bulk autophagy despite unimpaired autophagosome formation. Cultured Clec16a-deficient cells exhibited a striking accumulation of LC3 and LAMP-1 positive autolysosomes containing undigested cytoplasmic contents. Therefore Clec16a, an autophagy protein that is critical for autolysosome function and clearance, is required for Purkinje cell survival.

in the deep cerebellar nuclei. Interestingly, Purkinje cells lacking the essential autophagy genes Atg5 or Atg7 die over time 20,21 .
Given the expression of Clec16a in the cerebellum and the sensitivity of cerebellar neurons to disruptions in autophagy, we sought to identify the role of Clec16a in the CNS of mice. We found that two independent mouse lines carrying homozygous mutations in Clec16a exhibited a profound neurodegenerative disease characterized by motor impairment, loss of Purkinje cells, abnormal Golgi morphology, and disrupted autophagy. Abnormalities in Golgi structure and bulk autophagy were also observed in Clec16a mutant murine and human cells. Importantly, cultured CLEC16A-deficient cells accumulated autolysosomes despite lysosome and Golgi function being normal by multiple measures and showed normal fusion of autophagosomes and lysosomes. This demonstrated that Clec16a plays a key role in the survival of Purkinje cells in mice and the degradative function or clearance of autolysosomes.

Results
Neurologic disease and Purkinje cell loss in Clec16a mutant mice. Due to the demonstration in published data of high levels of Clec16a expression seen in Purkinje cells and neurons in the deep cerebellar nuclei (Allen Mouse Brain Atlas, http://mouse.brain-map.org/gene/show/50215) [24][25][26] , we sought to define the physiological function of Clec16a in the central nervous system of mammals by studying mice carrying a gene-trap insertion in Clec16a (Clec16a GT ; Fig. 1, Supplementary Fig. 1). Clec16a GT/GT mice on a mixed 129/ SvEv-C57BL/6 genetic background averaged 42% of the weight of control mice ( Supplementary Fig. 1b,c) and displayed motor impairment. While many of the Clec16a mutant mice, but not control mice, exhibited hind limb paralysis, we did not separately quantify this observation over time. After backcrossing to the C57BL/6-J background, B6.Clec16a GT/GT mice continued to display size dimorphism ( Supplementary Fig. 1d) and motor impairment. To compare the expression levels of Clec16a transcripts we utilized primers targeting exons 2-3 or exons 23-24 located either 3′ or 5′ of the genetrap cassette, respectively ( Supplementary Fig. 1e). While amplification of Clec16a transcripts across exons 2-3 were comparable in wild-type and B6.Clec16a GT/GT murine embryo fibroblast cells (MEFs), there was a greater than 90% reduction in Clec16a transcript using primers targeting exons 23-24 ( Supplementary Fig. 1e), indicating that the mRNA for this gene was interrupted by the GT cassette. The transcription of neighboring genes, Ciita, Dexi and Socs-1, in B6.Clec16a GT/GT MEFs was unaffected by gene trap insertion in Clec16a ( Supplementary Fig. 1e). Starting at seven to eight weeks of age, male and female B6.Clec16a GT/GT mice displayed abnormal hind limb clasping (Fig. 1a,b) 23 and were unable to maintain their balance or grip the bars of an inverted metal cage for a normal period of time (Fig. 1c) 28 . To assure that the phenotype observed in B6.Clec16a GT/GT mice was reflective of disruption of Clec16a, we studied a second independent Clec16a mutant mouse strain: Clec16a CURT mice (on the SWR/J background) carrying a spontaneous 4 base pair deletion in Clec16a ( Supplementary Fig. 1a) 29 . Transcript levels of Clec16a were significantly reduced in Clec16a CURT MEFs using primers targeting the mRNA both 5′ and 3′ of the mutation in exon 21, indicating that this mutation significantly destabilizes Clec16a mRNA(s) (Supplementary Fig. 1e). These mice also exhibited size dimorphism and motor impairment 29 . Therefore, mutation of Clec16a was consistent with the development of neurologic disease in two independent mouse strains.
In order to characterize the specificity of the neurodegeneration seen in B6.Clec16a GT/GT mice an experienced neuropathologist analyzed multiple areas of the brain, including the pons/medulla, hippocampus, frontal lobe and basal ganglia, and found no evidence that mutation in Clec16a resulted in neurodegeneration in these regions of the brain at this time point (Supplementary Fig. 2a-d). However, while gross cerebellar architecture was normal in B6.Clec16a GT/GT mice despite decreased brain size (Fig. 2a,b), loss of Purkinje cells was readily detectable (Fig. 2c,d) at eight weeks of age before development of neurological deficits in most mice (Fig. 1b,c). Purkinje cell degeneration models such as pcd mice display motor abnormalities that were similar to those observed in B6.Clec16a GT/GT and Clec16a CURT/CURT mice 30,31 and staining with the Purkinje cell marker calbindin confirmed loss of these cells in B6.Clec16a GT/GT mice (Fig. 2d) at 8 weeks of age that was dramatically increased by 47 weeks of age ( Supplementary Fig. 2e). Bielschowsky staining revealed characteristic 'empty baskets' (basket cell processes) in areas of Purkinje cell loss, indicating that the Purkinje cells had developed normally but were then lost ( Supplementary Fig. 2f) 32 . Clec16a CURT/CURT mice also demonstrated loss of Purkinje cells by histology and calbindin staining ( Supplementary Fig. 2g-j). These data indicate that Clec16a is important for Purkinje cell survival.
Neuronal and axonal abnormalities in the cerebellum of Clec16a mutant mice. We selected eight weeks of age, before the onset of severe neurological disease, to examine cerebellar neurons in Clec16a mutant mice in greater detail. Western blots confirmed increased expression of the lipidated form of the autophagy protein LC3 in the cerebellum of B6.Clec16a GT/GT mice (Fig. 3a). While the cell bodies of Purkinje cells lie in the Purkinje cell layer, axons from Purkinje cells form terminal synapses with neurons in the deep cerebellar nuclei. Neurons in the deep cerebellar nuclei showed accumulation of the autophagy substrate p62 in both B6.Clec16a GT/GT and Clec16a CURT/CURT mice (Fig. 3b, Supplementary Fig. 3a). The accumulation of p62 also occurred in calbindin-positive axons, derived from Purkinje cells, in the same region (Fig. 3b, Supplementary Fig. 3a). As electron microscopy is the gold standard for assessment of autophagic structures, we utilized it to assess the effect of Clec16a mutation on the morphology of membrane structures and organelles in Purkinje cells and their axons in the deep cerebellar nuclei. Ultrastructural analysis of mice with Atg5 and Atg7 deficiency in Purkinje cells previously demonstrated the accumulation of abnormal membrane structures in the dystrophic axons of Purkinje cells 20,21 . We found that cell bodies of cerebellar and deep nuclei neurons from B6.Clec16a GT/GT mice contained enlarged, vacuolated membranous structures and lacked defined Golgi stacks (Fig. 3c). The morphology of synapses between neuronal processes in both the Purkinje cell layer and the deep cerebellar nuclei were similar between B6.Clec16a GT/GT mice and littermate controls (data not shown). However in myelinated axons passing through the deep cerebellar nuclei, a region dense in calbindin-positive Purkinje cell axons (Fig. 3b), we observed increased autophagosome-and autolysosome-like structures (Fig. 3e, Supplementary Fig. 3b). These abnormal dystrophic axons were not observed in control mice. The structures within Purkinje cell axons of B6.Clec16a GT/ GT mice were likely of autophagic origin as they labeled for LC3 expression by immuno-electron microscopy (Fig. 3f). Together, these abnormalities suggested a role for Clec16a in both the Golgi apparatus and in autophagy in neurons in vivo.
Accumulation of p62, LC3 in cultured CLEC16A-mutant cells. To define the cell biology of the function of Clec16a in autophagy we chose to utilize a rapidly-growing, easily transfectable human HeLa cell line expressing GFP-LC3 to generate CLEC16A-deficient cells using CRISPR/Cas9 nuclease technology.   Fig. 4a,b). The use of non-neuronal cell lines, B6.Clec16a GT/GT MEFs and HeLa-CLEC16A∆ cells, as oppose to primary cerebellar neuronal cultures, allowed for the extensive quantification and characterization of multiple steps in autophagy by confocal microscopy in cells with abundant cytoplasm. Immunofluorescence analysis of B6.Clec16a GT/GT MEFs revealed a 5.7 fold increase in LC3+ puncta/cell at baseline compared to control cells and this did not result in a statistically significant change when blocking autolysosome degradation through treatment with chloroquine (CHQ) (Fig. 4a,b). Western blotting demonstrated an increase in p62 and the lipidated form of LC3, termed LC3-II, in B6.Clec16a GT/GT MEFs ( Supplementary Fig. 4c). Similar changes were observed in Clec16a CURT/CURT MEFs (data not shown). Immunofluorescence analysis of both independent clones of HeLa-CLEC16A∆ cells demonstrated increased p62+ and LC3+ puncta/cell compared to control cells ( Fig. 4c-f) and western blotting revealed increased p62 protein levels in these cells (Fig. 4g,h). To determine whether major changes in cellular energy metabolism might contribute to changes in autophagy, we assessed cellular ATP levels and levels of mitochondrial DNA (mtDNA). We did not detect a change in total mtDNA, or ATP levels in Clec16a-deficient cells, indicating that Clec16a-deficiency was not associated with major alterations in mitophagy in cells cultured under our conditions, and that abnormalities in autophagy were not due to inadequate intracellular energy stores ( Supplementary Fig. 4d,e). These data demonstrated impaired bulk autophagy in Clec16a-deficient human and murine cells. The high level of LC3-positive structures at baseline, combined with a failure of chloroquine to increase the number of these structures, was most consistent with Clec16a playing a role in autophagy after the formation of autophagosomes.

Functional endolysosomal system in Clec16a-mutant cells. Since autophagic flux was abnormal in
Clec16a-mutant cells we considered whether another lysosome-dependent degradation process, endolysosomal processing of cell surface proteins, was similarly affected by deletion of Clec16a 9,16,33 . Knock-down or loss of Clec16a in human and murine cells has been described to alter the morphology of late endosomes 16,33 but we did not observe alterations in the endosomal network as detected by staining of endosomes for Rab5, Rab7 or EEA1 in Clec16a-deficient cells (Fig. 5a, Supplementary Fig. 5a). Knock-down of CLEC16A in HeLa cells reduced the amount of epidermal growth factor (EGF) receptor at steady-state but did not alter the rate of endocytic internalization and endolysosomal degradation of the receptor following ligand binding (Fig. 5b,c). The rate of endolysosomal degradation of the receptor following ligand binding was also unimpaired in CLEC16A-deficient HeLa cells ( Supplementary Fig. 5b,c), indicating that a failure to observe changes in siRNA-treated cells was not due to residual low levels of protein. We also examined the cell surface expression of MHC class II which is regulated in part by endosomal recycling of cell surface protein. Presentation of a synthetic peptide antigen to T cells is a sensitive measure of cell surface expression of MHC class II. We therefore assessed the presentation of a peptide derived from Listeria monocytogenes by B6.Clec16a GT/GT macrophages. Peptide-specific CD4+ T cells co-cultured with peptide-pulsed macrophages from B6.Clec16a GT/GT mice were equivalently activated across a range of peptide doses indicating normal cell surface expression of functional MHC class II (Fig. 5d). These data are similar to those obtained upon CLEC16A siRNA knockdown in human B cells serving as antigen presenting cells 34 . These assays did not detect abnormalities in the endosomal or endolysosomal system in Clec16a-mutant cells.
Golgi apparatus structure and function in Clec16a-mutant cells. Given our novel findings of abnormal Golgi appearance in the neurons of mice lacking Clec16a, and the role of the Golgi in production of lysosomes, we next examined the Golgi apparatus in cultured B6.Clec16a GT and HeLa-CLEC16A cells (Fig. 6).  (Fig. 6a,b, Supplementary Fig. 6a,b). We detected no alteration in endoplasmic reticulum morphology (Fig. 6c), as expected from prior studies of Clec16a-deficient murine islets 16 . Staining with other markers such as the trans-Golgi network protein, TGN-46, and the Golgi tethering protein Giantin confirmed the dispersion of the Golgi in CLEC16A-mutant cells (Supplementary Fig. 6c). These abnormalities were not due to defective autophagy in Clec16a-deficient cells since lack of the essential autophagy gene Atg5 did not lead to altered Golgi or ERGIC morphology ( Supplementary Fig. 6d).
Since Clec16a-deficient islet cells are reported to exhibit defective insulin secretion in response to glucose challenge 16 , we quantified protein secretion using HeLa cells expressing a GFP-tagged FKBP mutant (F36M) linked to human growth hormone (hGH) 35 . This protein aggregates in the ER until ligand is added causing solubilization of aggregates and secretion into the extracellular space. In control siRNA-treated cells, intracellular GFP levels declined over time as expected due to hGH secretion (Fig. 6e). Loss of RAB1A/B impaired secretion (Fig. 6d,e), serving as a positive control 36,37 . However knock-down of CLEC16A did not affect secretion (Fig. 6d,e).
We next assessed the addition of N-linked glycans to glycoproteins as a measure of Golgi function. Western blotting of EndoH-and PNGase-digested lysates of B6.Clec16a GT/GT MEFs and HeLa-CLEC16A∆ cells demonstrated that the highly glycosylated lysosomal membrane protein LAMP-1 was present at equivalent levels in wild-type and CLEC16A-deficient cells. Further, LAMP-1 acquired appropriate N-linked glycans 38,39 (Fig. 6f,  Supplementary Fig. 6e). Together these data indicated that the Golgi functions normally in CLEC16A-deficient cells as measured by these assays.
Clec16a is necessary for autolysosomal function. We next investigated the stage in the autophagic process at which Clec16a functions. While in ema-deficient Drosophila cells there is no observed defect in fusion between lysosomes and autophagosomes, reduced fusion has been suggested in studies of mitophagy induction in islet cells lacking Clec16a 9,16 . Fusion of LAMP-1+ lysosomes and LC3+ autophagosomes results in the generation of the late autophagic structure, the autolysosome. These structures typically mature into LC3 negative but LAMP-1 positive structures and contain cytoplasmic debris which is ultimately degraded by lysosomal enzymes. B6.Clec16a GT/GT MEFs and CLEC16A-mutant HeLa cells contained large numbers of single membrane bound structures containing cellular material not seen in control cells (Fig. 7a, Supplementary Fig. 7a,b). These structures were most consistent with being autolysosomes rather than autophagosomes as they had a single membrane. This was confirmed by labeling these structures with both LC3 and LAMP-1 by immuno-electron microscopy (Fig. 7b, Supplementary Fig. 7c).
In order to quantify the increase in autolysosomes seen by electron microscopy we performed colocalization studies with GFP-LC3 and LAMP-1 in CLEC16A-deficient cells. CLEC16A-deficient cells have increased colocalization of LC3 and LAMP-1 in puncta compared to control cells, consistent with the increased number of autolysosomes observed by confocal microscopy (Fig. 7c,d). Importantly, while treatment of control cells with bafilomycin A1 increases the colocalization of LC3 and LAMP-1+ in puncta consistent with its known role in inhibiting lysosome and autophagolysosome function, bafilomycin A1 (BAF) treatment did not increase the colocalization of these markers in CLEC16A-mutant cells (Fig. 7c,d). These data indicate that CLEC16A deficiency results in a failure of autolysosome function or clearance at a step in autophagy downstream of autophagosome and lysosome fusion.

Lysosomal biogenesis and function in Clec16a-mutant cells.
Given the failure of autolysosomal clearance, we considered whether Clec16a played a role in lysosomal biogenesis. As described above, we observed normal expression of the lysosomal protein LAMP-1 by western blot analysis in Clec16a-mutant cells (Fig. 6f,  Supplementary Fig. 6e). Further, the distribution and number of lysosomes as marked by the lysosomal membrane protein LAMP-1 was not altered by loss of CLEC16A expression in human cells (Fig. 8a,b). We next determined if lysosomal pH is altered in B6.Clec16a GT MEFs using a pH-sensitive fluorescent sensor. B6.Clec16a GT/GT MEFs and control cells had a comparable lysosomal pH of 4.8 and as expected this was elevated by treatment with chloroquine (Fig. 8c). Lysosomal acid hydrolases traffic from the Golgi to the lysosome after addition and subsequent enzymatic modification of mannose 6-phosphate (Man-6-P) 40 . Mutations to the first enzyme in this pathway, UDP-GlcNAc: lysosomal enzyme N-acetylglucosamine-1-phosphotransferase, lead to extracellular secretion of hydrolases due to inappropriate glycosylation 41 . However, intracellular and extracellular enzymatic activity of both α -Mannosidase and β -Hexosaminidase was normal in CLEC16A-deficient Hela cells, and these enzymes were not secreted (Fig. 8d,e), indicating normal trafficking of these enzymes to lysosomes. These data indicated that defects in autolysosome clearance are not due to major disruptions in lysosomal biogenesis, pH, or hydrolase enzyme trafficking or activity.

Discussion
We report here that CLEC16A, which has been linked by GWAS studies to multiple sclerosis and other autoimmune diseases, has an important role in the survival of cerebellar Purkinje neurons in two mutant mouse models. These results show that one dominant effect of Clec16a deficiency in the whole mouse is related to a role in Purkinje cells. Importantly, the mutations in the gene analyzed here are likely more disruptive of Clec16a function than alleles associated with human disease risk 1,2,16,33,42 . Thus these data provide, in a small animal model, insight into the effects of full disruption of the gene in the CNS. Investigation of these cells revealed abnormalities consistent with a role for Clec16a in bulk autophagy, a finding confirmed in both human and murine cells. The block in autophagy occurred after fusion of lysosomes to autophagosomes to create autolysosomes, resulting in accumulation of these late autophagic structures despite normal lysosomal biogenesis and pH. We considered the possibility that abnormalities in autolysosome clearance or function were secondary to abnormal Golgi function or lysosomal biogenesis or function which might correlate with the novel finding of abnormal Golgi morphology detected in our studies, both in neurons and in in vitro cell assays. However, protein secretion machinery from the Golgi was unaltered, glycosylation and levels of the lysosomal protein LAMP-1 were normal, the trafficking and activity of lysosomal hydrolases was preserved, and the number and pH of lysosomes was normal in Clec16a-deficient cells. Abnormal Golgi morphology in cells lacking CLEC16A may hint at a possible role of CLEC16A in maintaining or assisting other Golgi apparatus proteins in organelle structure. Together these data argue for abnormal autolysosome accumulation, rather than disrupted Golgi morphology, as the proximate cause of Clec16a-dependent autophagic abnormalities reported here. The specificity of this function of Clec16a for autophagy was further supported by the lack of detectable changes in another lysosomal degradation pathway, the recycling of membrane proteins through the endolysosomal system. We do not know the mechanism by which Clec16a deficiency impairs clearance of autolysosomes. We speculate that the protein is involved in the recycling of autophagic membranes as has been observed to occur late after induction of autophagy 43 . Our findings regarding the loss of Purkinje cells and impaired autophagy point to a potential in vivo side-effect of pharmacological inhibition of CLEC16A protein function or autophagy, and support the conclusion that Clec16a has a unique role in the function and clearance of autolysosomes.

Role of autophagy in Clec16a-dependent neurodegeneration.
In several in vivo models, neurons, and in particular Purkinje cells, demonstrate a striking requirement for functional autophagy in order to survive. For example, the conditional deletion of the autophagy genes Atg5 or Atg7 in Purkinje cells leads to degeneration of these cells 20,21 . This phenotype is remarkably similar to what we observed here for mutation of Clec16a. Consistent with impairment of autophagy at different steps, axons in the deep cerebellar nuclei of mice lacking Atg5 and Atg7 in Purkinje cells displayed increased membranous and vacuole-like structures while Purkinje cell axons in the deep cerebellar nuclei of B6.Clec16a GT/GT mice accumulated autolysosome-like structures 20,21 . Consistent with a protective role of autophagy in neurons, over-expression of Beclin 1 significantly improves motor symptoms in a mouse model of Machado-Joseph disease or spinocerebellar ataxia type 3 44 , and expression of Atg5 protects neurons against virus infection-induced cell death 45 . Roles for several additional autophagy genes in the preservation of other neuronal populations in the CNS include the role of FIP200 in the neural stem cell population of the lateral ventricles and hippocampal dentate gyrus 22 and of Epg5 in cortical layer 5 pyramidal and spinal cord motor neurons 23 . While it is possible that the death of Purkinje cells in mice studied here is due to a non-autophagic role for Clec16a, the similarities between these multiple models of autophagy gene deficiency, together with the role for Clec16a in autolysosomal function and clearance demonstrated here, supports the concept that it is abnormal autophagy, likely as a result of abnormal autophagic flux, that leads to neurodegeneration when Clec16a is mutated. Intriguingly, the Drosophila homologue of Clec16a, ema, has been shown to colocalize with Spinster (spns1), which has recently been implicated in abnormal autolysosomal function 46 . Zebrafish with a loss-of-function mutation in spns1 demonstrate sensitivity to starvation-induced death and accumulate autolysosomes with suboptimal acidification 46 . It is interesting to speculate that Clec16a and other autophagy genes might, in addition to playing a role in survival of specific neurons under normal conditions, participate in survival of neurons in different forms of stress, as for example during auto-immune inflammation in the brain.

The evolutionarily-conserved role of Clec16a in membrane trafficking. The Drosophila CLEC16A
homolog ema is required for efficient recruitment of the Golgi complex protein Lava lamp (Lva) to autophagic structures in fat body cells and for proper endosomal maturation in Garland cells 9,47 . Flies mutant for ema exhibit decreased turnover of p62 and decreased mitophagy 9 . Importantly, the phenotype of flies mutant in ema can be complemented by expression of human or mouse CLEC16A, indicating that the function of this protein is highly conserved 47 . These studies supported the concept that Ema played a role in autophagosomal growth but that fusion between the autophagosome and the lysosome was unaltered in the absence of the gene 9 . The reasons for the apparent differences between the findings in the fly and our studies are not clear, and will require identification of the molecular mechanisms operating in each system. In contrast, a recent study suggested that impaired mitophagy was due at least in part to reduced fusion between the autophagosomes and lysosomes 16 . It is interesting that this function of Clec16a was attributed to its physical interaction with the E3 ubiquitin ligase Nrdp1 which regulates Parkin. The lack of reported neurodegeneration in parkin-deficient and Nrdp1 mutant mice [17][18][19] makes it unlikely that this interaction explains our finding of neurodegeneration in Clec16a mutant mice.
Role of CLEC16A in human disease. Studies reported here suggest that a failure to clear autolysosomes and to properly digest their contents plays a role in neuronal survival which may contribute to central nervous system disease. However, single nucleotide polymorphisms (SNPs) in CLEC16A that are associated with autoimmunity may correlate with either increased or decreased CLEC16A protein expression in humans, complicating proposals to link the role of Clec16a in autophagy directly to human disease 1,2,16,33,42 . For example, the presence of the diabetes-associated CLEC16A SNP rs12708716 is reported to result in about a two-fold decrease in CLEC16A mRNA expression in human islets 16 . In contrast, a two-to four-fold increase in CLEC16A mRNA was detected in multiple sclerosis patients compared to controls in peripheral blood mononuclear cells and white matter, though this was not related to the multiple sclerosis risk-associated SNP rs7200786 33 . The importance of these changes in CLEC16A mRNA expression for protein expression and function await analysis. Importantly, the majority of studies on Clec16a function, like our own, have focused on loss of Clec16a function. However, Clec16a plays a fundamentally important role in autophagy by acting at a stage in the process for which specific required genes have not previously been identified. Thus the linkage of this cellular role of Clec16a in disease will be important to delineate.

Materials and Methods
Ethics Statement. All animal research was reviewed and approved by the Washington University in St.
Louis Institutional Biological and Chemical Safety Committee. All methods were carried out in accordance with the approved guidelines. In order to perform tissue collection, tissues were collected after sacrifice of animals using methods approved by the Animal studies committee approval #20140244. Behavioural analysis. Tests for hind-limb clasping were performed starting in 4.5 week old B6.Clec16a GT mice in which a score of 1 was awarded for temporary clasping of 1 or 2 hind limbs and a score of 2 for immediate and sustained clasping of 2 hind limbs over 30 seconds 23 . Data for the hind limb clasping test was then plotted to represent the initial age that individual mice score a 1 or a 2 (whichever occurred first). A drop test was also performed in which mice either dropped immediately when inverted or remained hanging inverted on the cage lid in the first 20 seconds 28 . Drop test data was plotted to represent the age that mice first drop.

Mice and cells. Clec16a
Histology and immunofluorescence studies. B6.Clec16a GT and Clec16a CURT mice were perfused with 4% paraformaldehyde and brains were embedded in paraffin prior to H&E staining. For calbindin and p62 brain staining, aged Clec16a GT , B6.Clec16a GT and Clec16a CURT mice were perfused with 10% formalin and brains were embedded in paraffin prior to sectioning. Slides were deparaffinized, rehydrated, incubated in ethanol solutions (decreasing from 100-80%) and stained with antibodies. For immunofluorescence studies, B6.Clec16a GT , Clec16a CURT MEFs and HeLa-CLEC16A cells were fixed in 4% paraformaldehyde, permeabilized in methanol, saponin or 0.05% TritonX-100, blocked in 1% BSA and 1% normal goat serum in PBS followed by antibody incubations. Antibodies utilized are listed in Supplementary Information. For quantification of LC3+ , LAMP-1+ or p62+ puncta/cell and colocalization studies, collective images of at least 100 cells/sample were analyzed using ImageJ or Volocity software. For Golgi organelle morphology scoring, collective images of at least 60 cells/ sample were analyzed visually to obtain the percent of cells compared total cells/field with an abnormal Golgi morphology.
Electron microscopy and immunogold labeling. For transmission electron microscopy, B6.Clec16a GT and Clec16a CURT mice were transcardially infused with saline followed by 4% paraformaldehyde and brain tissue sections were fixed and prepared for grids. A minimum of 100 Purkinje cells or neurons of the deep cerebellar nuclei/grid were evaluated. A primary evaluation of ultrastructural morphology was performed on Purkinje cell bodies, axons and cell bodies of neurons of the deep cerebellar nuclei to identify any abnormal pathology. Additionally, the morphology of synapses between neuronal processes were evaluated in both the Purkinje cell layer and deep nuclei to assess any reductions in the number of synapses in the tissues or ultrastructual differences in the synapse. For unbiased evaluation of intracellular membranes in cerebellar neurons, 2 ultrastructural experts were provided with 30 unlabeled electron microscopy images of Purkinje cell bodies and cell bodies of neurons in the deep cerebellar nuclei at a resolution which the morphology of mitochondria, ER, Golgi and vesicles could be evaluated. For transmission electron microscopy of tissue culture cells, cells were fixed, postfixed, dehydrated, embedded and mounted as above for murine tissue. For immunogold labeling of LC3 in murine tissues, B6.Clec16a GT and Clec16a CURT mice were pericardially infused with saline followed by 4% paraformaldehyde/0.05% glutaraldhyde (Polysciences Inc., Warrington, PA) in 100 mM PIPES buffer and brain tissue sections were embedded in gelatin, infiltrated with PVP-sucrose overnight and frozen in liquid nitrogen. Sections labeled for LC3 were visualized with transmission electron microscopy. For immunogold labeling of LC3 and LAMP-1, HeLa-CLEC16A cells were prepared as above for murine tissues. Sections were labeled for GFP and LAMP-1.
Secretion, glycosylation assays. HeLa C1 cells 35  Lysosomal and mitochondrial function assays. For EGFR degradation assays, protein levels of EGFR were determined by western blot in HeLa cells untransfected, transfected with siRNAs, serum starved and/or stimulated with 100 ng/ml EGF (Sigma-Aldrich, St. Louis, MO) for the indicated times. For measurement of lysosomal pH, B6.Clec16a GT MEFs were incubated with 4 uM Lysosensor Yellow/Blue DND-160 (Life Technologies, Cat No. L-7545) and, following measurement by plate reader, lysosomal pH was determined with the following ratio of the fluorescence at two different wavelengths: (F 340nm /F 380nm ). Enzymes activity assays, for β -hexosaminidase, α -mannosidase and β -glucuronidase, were performed as described 48 . For antigen presentation co-culture assays wild-type (BL6) and Clec16a-deficient bone-marrow derived macrophages were peptide pulsed (Listeriolysin O (190-205), 0-10 μM). Splenic CD4 T cells from LLO56tg mice, whose TCR is specific for LLO 190-205 49 , were isolated (CD4+ T Cell Isolation Kit II -mouse, Miltenyi Biotech) and added to the macrophages at a density of 2.5 × 10 5 T cells/well. After 24 hours, CD4 cells (eFluor450, Clone RM4-5, eBioscience) were assayed for activation by staining for expression of CD69 (PECy7, Clone H1.2F3, BioLegend). For measurement of cellular ATP levels, B6.Clec16a GT MEFs and HeLa-CLEC16A cells were seeded in 96 well plates. Following addition of CellTiter-Glo Luminescent Cell Viability Reagent (Promega Corporation, Madison, WI) to the plate luminescence was measured using a Molecular Devices SpectraMax M2 plate reader. For mtDNA quantification, qPCR was performed using total DNA from B6.Clec16a GT MEFs as described 50 . Results were comparable with 100 ng, 10 ng and 1 ng of input DNA.