Maturation and clearance of autophagosomes in neurons depends on a specific cysteine protease isoform, ATG-4.2

Autophagosomes in C. elegans neurons contain synaptic proteins and form near synapses in response to physiological stimuli

To examine autophagy in neurons of living animals, we imaged LGG-1 and LGG-2 in the AIY interneurons of C. elegans at larval stage 4 (L4) (Figure 1A–1E, Figure S1A-S1C, and as previously described in (Stavoe et al., 2016)). LGG-1 and LGG-2 are C. elegans homologues of Atg8 (in yeast) and GABARAP and LC3 family proteins (in mammals) that associate with immature and mature autophagic structures, and have been used as markers to track autophagosomes in living cells ((Alberti et al., 2010; Manil-Segalen et al., 2014; Melendez et al., 2003), for controls of these markers, please see Figures 1F and S1D-S1F and STAR Methods).

Autophagosomes in AIY were present in the neurite (Figures 1B–1D) and in the cell body (Figure 1E). In the neurite, the majority of autophagosome biogenesis events occur near synaptic regions (Stavoe et al., 2016). To understand how activity of neurons relates to autophagosome biogenesis in vivo, we first examined autophagy in AIY neurons under varying physiological conditions. AIY is an interneuron that predictably changes its firing frequency based on inputs from sensory neurons (Laurent et al., 2015; Luo et al., 2014; Satoh et al., 2014). As part of its role in thermotaxis behavior, the frequency of calcium transients in AIY increases for animals raised at warmer temperatures (Clark et al., 2006; Hawk et al., 2018). We observed that animals raised at warmer temperatures, in which the neuronal activity state of AIY is increased, also had a higher percentage of animals with autophagosomes in AIY, and that this could be physiologically modulated by changing the cultivation temperature of the animals ((Hawk et al., 2018) and Figure 1G–1K). For example, in animals raised at 20°C, a condition known to have lower AIY calcium activity (Hawk et al., 2018), the percentage of animals with one or more autophagosomes in either AIY neurite was 37% (n=73 animals). When animals were cultivated at 25°C for 4 hours, which is known to change their temperature preference through an increased calcium responsiveness of AIY, a concomitant increase in the percentage of animals with one or more autophagosomes in either AIY neurite was observed (81% (n=78 animals) and Figure 1G–1I). We do note that animals raised at warmer temperatures for prolonged periods of time, under conditions associated with noxious stimuli to the animal (and not known to be associated with thermotaxis behavior or increased activity states in AIY), also displayed an activity-independent increase in autophagosomes in the AIY neurite. Yet for the physiological conditions in which AIY is known to be active and required for thermotaxis behavior (exposure to higher temperatures for periods of less than 8 hours), genetically inhibiting synaptic transmission suppressed the observed increase in synaptic autophagosomes (Figure 1J). Moreover, chemo-genetic silencing of AIY through the cell-specific expression of a histamine-activated chloride channel suppressed temperature-related increases in synaptic autophagosomes (Figure 1K). Previous findings using optogenetic approaches established a link between stimulation of the neuron and autophagosome biogenesis in the neurites (Shehata et al., 2012; Soukup et al., 2016; Wang et al., 2015). Our findings are consistent with and extend these studies, demonstrating that increases in autophagosomes near synapses can be induced by activity-dependent mechanisms associated with physiological responses of the neuron, and activity-independent mechanisms associated with noxious stimuli to the system.

Does increased autophagy at active synapses act to recycle synaptic material? To address this we examined if autophagic vacuoles in the cell body, where autophagosomes are transported for degradation, co-localized with synaptic proteins. Simultaneous visualization of active zone protein SYD-1 and autophagy-associated protein LGG-1 revealed partial co-localization in the neuronal cell body (Figure 1L–1N). Similar results were observed for active zone protein SYD-2/Liprin alpha and synaptic vesicle protein synaptobrevin (SNB-1) (Figures 1O–1Q and S1G-S1I). Our findings indicate that autophagic vacuoles in the cell body contain synaptic proteins, and suggest that autophagy might degrade synaptic material such as active zone and synaptic vesicle proteins. Together, our findings are consistent with an association, in vivo, between the state of synaptic activity, presence of autophagosomes at the synapse and degradation of synaptic material in the neuronal cell body.

Dynactin and JIP3/UNC-16 promote autophagosome retrograde transport

To then investigate how autophagosomes at the synapse are transported and degraded in the cell body, we recorded and tracked individual synaptic autophagosomes in wild-type and mutant animals. In C. elegans neurons, as in vertebrate neurons, retrograde transport of autophagosomes depends on the dynein complex (Cheng et al., 2015; Ikenaka et al., 2013; Katsumata et al., 2010; Maday et al., 2012) (Figure 2A–2C). We therefore examined known regulators of transport to investigate their specific requirement in autophagosome trafficking.

We observed that while available alleles for jip-1/Mapk8Ip1, unc-116/kinesin-1 and unc-14/RUN mutant animals did not display defects in autophagosome accumulation at synapses (Figure 2G), loss of JIP3/UNC-16/Sunday Driver resulted in a significant accumulation of LGG-1/LGG-2 puncta in the neurite, but not in the cell body (Figures 2E–2G and S2A-S2D). The unc-16/jip3 mutant phenotype was similar to the phenotype seen for dnc-1/p150 dynactin complex subunit mutant animals (which affect the dynein complex) ((Ikenaka et al., 2013) and Figure 2B–2C). JIP3/UNC-16/Sunday Driver is a conserved adaptor protein that binds to kinesin and dynein (Arimoto et al., 2011; Cavalli et al., 2005) to regulate early endosome and lysosome transport (Brown et al., 2009; Byrd et al., 2001; Drerup and Nechiporuk, 2013; Edwards et al., 2015; Edwards et al., 2013; Gowrishankar et al., 2017). Mutant animals containing one of three independent alleles of unc-16 (ju146, e109, or n730) (Figure 2D) displayed an increase in the percentage of animals with LGG-1 puncta in the neurites (from 34% in wild-type animals to 100% in all unc-16 allele mutant animals examined). Moreover, while wild-type animals averaged less than one LGG-1 puncta per neurite, unc-16 mutant animals averaged ~2 LGG-1 puncta in the presynaptic regions (Figure 2G), indicating abnormal accumulation of autophagosomes. Analyses of the subcellular localization of LGG-1-containing structures in unc-16/jip3 mutant animals also revealed a higher probability for autophagosomes to be present in the distal synaptic Zone 3 region (Figures 2H and 2I). The distal accumulation of autophagosomes observed for unc-16 mutant animals was not due to a rearrangement of microtubule polarity, as we observed that the microtubule plus-end-binding protein EBP-2/EB1 (Baas and Lin, 2011; Maniar et al., 2011) was similarly oriented (plus-end-out) in wild-type and unc-16/jip3 mutant animals (Figure S2E-S2G). Our findings indicate that unc-16/jip3 is necessary to prevent the abnormal accumulation of autophagosomes in the neurite, and are consistent with a requirement for unc-16/jip3 in autophagosome retrograde transport.

Previous reports have implicated UNC-16/JIP3/Sunday Driver in the regulation of retrograde transport and retention of lysosomes, early endosomes and Golgi in the cell body ((Brown et al., 2009; Byrd et al., 2001; Edwards et al., 2013) and Figure S3). The role of UNC-16/JIP3 in transport of autophagosomes has not been examined. To directly investigate if unc-16/jip3 is required for retrograde transport of autophagosomes, we performed time-lapse imaging of individual LGG-1 puncta in wild-type neurons (n=71) and for two independent alleles of unc-16 mutant neurons (ju146, n=73; and e109, n=43). We observed that compared to wild-type animals, unc-16/jip3 mutant animals displayed decreased retrograde transport of LGG-1 puncta in neurites and increased anterograde trafficking or paused LGG-1 puncta (defined as puncta that did not traffic within a 5-minute observation window; Figure 2J–2L). In wild-type animals, LGG-1 puncta in the cell body were sometimes transported in an anterograde fashion into the neurite and returned to the cell body via retrograde transport within the 5-minute examination window. In contrast, the percentage of puncta that returned was greatly reduced in unc-16(ju146) mutant neurites (Figure S2H). Our findings are consistent with previous reports that demonstrate that another JIP family member (JIP1) is important for regulating autophagosome retrograde transport in cultured vertebrate neurons (Fu et al., 2014). We now extend those findings and demonstrate an important role for unc-16/jip3 in regulating the retrograde trafficking of autophagosomes in vivo.

To understand how UNC-16-dependent retrograde transport contributes to the degradation of autophagic vacuoles, we then tested if defects in unc-16 mutant animals altered the total number of LGG-1 puncta in neurons. We observed that the total number of LGG-1 puncta in neurons of unc-16/jip3 mutant animals was higher than that seen for wild-type animals (average ±95% confidence interval from 4.1±0.6 in wild-type neurons (n=65) to 8.2±0.8 in unc-16(ju146) mutant neurons (n=64); significant by Student’s t test p<0.0001). Our findings indicate that the increased number of autophagic vacuoles observed in the neurites of unc-16/jip3 mutant animals did not result from a simple redistribution of a fixed number of autophagosomes. Instead, retrograde transport, promoted by unc-16/jip3, appears to be necessary for the effective degradation of autophagic vacuoles, and defects in transport in unc-16/jip3 mutant animals therefore result in an abnormal accumulation of autophagic vacuoles. Our findings underscore the importance of regulated trafficking in the degradation of synaptic autophagic vacuoles.

Mutant allele ola316 enhances the number of autophagic vacuoles in the neurites of unc-16/jip3 mutant animals

Degradation of autophagosomes occurs upon fusion with lysosomes, an important and final step that is tightly regulated but poorly understood (Itakura et al., 2012; Nakamura and Yoshimori, 2017). To better understand the molecular machinery underlying coordinated trafficking and degradation of synaptic autophagic vacuoles, we performed a visual forward genetic screen in the unc-16/jip3 mutant background. We reasoned that molecules important for regulated lysosomal fusion and degradation would result in an enhanced accumulation of autophagosomes, and screened for mutant animals with increased numbers of LGG-1 puncta in the AIY synaptic regions. We uncovered a mutant allele, ola316, in which animals displayed a dramatic accumulation of LGG-1 puncta in the neurites. While unc-16(ju146) mutant animals displayed an average of 2.6 LGG-1 puncta abnormally accumulated in the synaptic regions (n=64 neurites), ola316; unc-16(ju146) double mutant animals enhanced this phenotype and displayed an average of 9.4 LGG-1 puncta in the synaptic regions (n=67 neurites) (Figure 3A–3D). We note that in these double mutant animals with dramatic accumulation of autophagosomes, some LGG-1-positive structures were observed outside the AIY neurite (Figure S4A-S4C). It is possible that these LGG-1 structures might be expelled from the neurite in response to an overburdening of the endo-lysosomal system, a phenomenon that has been observed in the context of neurodegenerative diseases (Borland and Vilhardt, 2017; Miranda et al., 2018; Ojha et al., 2017) (Movie S1).

We then examined if animals containing the ola316 allele displayed a phenotype independent of the unc-16/jip3 mutant lesion. To achieve this, we outcrossed the ola316; unc-16(ju146) animals to remove the unc-16/jip3 lesion and examined LGG-1-containing structures in the ola316 single mutant animals. We observed that the ola316 single mutant animals did not display a significant accumulation of LGG-1 puncta in the synaptic regions when compared with unc-16(ju146) mutant or unc-16(ju146); ola316 double mutant animals (Figure 3D). However, we did observe an increased number of LGG-1 puncta in the cell bodies of the ola316 single mutant animals (Figures 3E, 3H and 3I; we observed similar results for LGG-2, Figures S4D, S4E, S4G and S4H). We also observed an increase in the percentage of neurites with one or more LGG-1 puncta in the synaptic regions (Figure S4I). The gene affected in the ola316 allele is important for most neurons in the nematode’s nervous system, as abnormal panneuronal accumulation of LGG-1 puncta was observed for both the single mutant animals and the ola316; unc-16(ju146) double mutant animals (Figure S4J-S4Q). Together our data indicate that the ola316 allele affects a gene that is important for preventing the abnormal accumulation of autophagosomes both in the neuronal soma and, through cooperation with UNC-16/JIP3, in the neurites.

ola316 is a dominant allele of the autophagy gene atg-4.2

Genetic characterization of animals containing the ola316 allele revealed that the phenotype results from a single genetic dominant lesion (Figure 3E, green and red striped bar). To identify the causative lesion in animals containing ola316, we performed single-nucleotide polymorphism (SNP) mapping, whole genome sequencing, independent allele analyses and rescue experiments. We SNP-mapped ola316 to a 2.2 Megabase (Mb) region between 10.1 and 12.3 Mb on chromosome IV. We then performed whole genome sequencing (Minevich et al., 2012; Sarin et al., 2008) and uncovered a missense lesion in the autophagy gene atg-4.2. Animals containing either of two independent alleles of atg-4.2 (gk430078 or gk628327), which result in early stop codons (Figure 3F), phenocopied the ola316 single mutant phenotype (Figure 3E (maroon and peach bars) and 3J). Moreover unc-16(ju146); atg-4.2(gk628327) double mutant animals also phenocopied the unc-16(ju146); ola316 double mutant animals isolated from the genetic screen (Figure 3D). Consistent with ola316 being an allele of atg-4.2, AIY-specific expression of the wild-type atg-4.2 cDNA in ola316 mutant animals rescued the ola316 mutant phenotype (Figures 3E (brown bar), 3I and 3K). While this rescue experiment might suggest that the dominant allele was acting through happloinsufficiency, we also observed that overexpression of mutant atg-4.2(ola316) cDNA in wild-type animals was sufficient to induce the phenotype (Figures 3E (grey bars), 3L and 3M), indicating that the atg-4.2(ola316) allele acts as a dosage-dependent dominant negative, and that it is required cell autonomously in AIY. Our data indicate that ola316 is an allele of the autophagy gene atg-4.2, and that atg-4.2 acts in neurons to regulate accumulation of autophagic structures.

Atg4s are a conserved family of cysteine proteases that bind to and cleave Atg8/GABARAP/LC3/LGG-1/LGG-2 family proteins. Comparison with the human ATG4B-LC3 crystal structure (Kumanomidou et al., 2006) revealed that the atg-4.2(ola316) lesion identified in our screen neighbors a conserved regulatory loop important for LC3 (the vertebrate homologue of LGG-2) recognition by ATG4 (Wu et al., 2012), and is sterically close to the cysteine active site (Figure 3G). Based on its position, and our in vivo genetic data, we hypothesize that the dominant lesion in the atg-4.2(ola316) allele might interfere with ATG-4.2 activity and its interaction with LGG-1/LGG-2.

Atg4s cleave Atg8/GABARAP/LC3/LGG-1/LGG-2 family proteins to achieve two functions: 1) prime LGG-1 (which contains a C terminal tail) for conjugation onto the autophagosomal membrane; and 2) remove LGG-1 and LGG-2 from the autophagosomal membrane (Alberti et al., 2010; Kirisako et al., 2000; Maruyama and Noda, 2017). In yeast one ATG4 gene performs both the priming and delipidation events on Atg8/LGG-1/LGG-2 (Kirisako et al., 2000; Maruyama and Noda, 2017), while in higher metazoans multiple genes encode distinct Atg4 cysteine proteases, some with unknown function, specificity, or redundancy (Kauffman et al., 2018; Li et al., 2011; Marino et al., 2003; Wu et al., 2012; Zhang et al., 2016). Consistent with these, a previous study in C. elegans embryos determined that the two C. elegans atg-4 genes (atg-4.1 and atg-4.2) (Figure 4A) were largely redundant (Wu et al., 2012), with ATG-4.1 displaying enhanced proteolytic activity on a soluble pro-form of LGG-1/Atg8 as compared to ATG-4.2. The specific roles of ATG-4.2 in autophagy, if any, remain unknown.

ATG-4.2, but not ATG-4.1, is required to prevent autophagosome accumulation

The two C. elegans Atg4 homologs display 44% amino acid similarity, including conserved catalytic and regulatory sites (Wu et al., 2012). A previous study investigating the degradation of protein aggregates in embryos found that atg-4.1 is more efficient at the first LGG-1 cleavage event (priming), and more important for removal of protein aggregates than atg-4.2 (Wu et al., 2012). To examine if atg-4.1 also acts as atg-4.2 to prevent autophagosomal accumulation in neurons we examined animals containing a confirmed null allele of atg-4.1(bp501) (Wu et al., 2012) or containing the allele atg-4.1(gk127286) (early stop codon) (Figure 4A). Surprisingly we did not detect LGG-1 or LGG-2 puncta accumulation in the atg-4.1 mutant animals (Figures 4C, 4E, S4F-S4H, and S4R).

We then examined autophagosomes in the neuronal cell bodies of these mutant animals by performing electron microscopy. Consistent with our GFP∷LGG-1 and GFP∷LGG-2 in vivo assays, we observed a dramatic accumulation of abnormal multilamellar autophagosome-like vacuoles in the neuronal somas of atg-4.2(ola316) mutant animals (Figures 4F–4K and S5A-S5P). Our findings indicate that atg-4.1 and atg-4.2 have distinct functions in vivo that result in different phenotypes regarding the accumulation of autophagosomes in neurons. Importantly, our findings indicate that atg-4.2 is specialized for the clearance of autophagic vacuoles.

While we observed distinct phenotypes for atg-4.1 and atg-4.2 in the accumulation of autophagic structures, we also note that both atg-4.1 and atg-4.2 single mutant animals contained LGG-1/LGG-2 puncta in neurons (Figures 4C–4E and S4E-S4H). Comparisons with the corresponding lipidation controls (LGG-1(G116A); explanatory diagrams in Figure S1) suggest that nearly all of the observed LGG-1 puncta in atg-4.1 and atg-4.2 mutant animals were lipidated (Figure S5Q, S5T-S5V). These observations indicate that either protease is able to partially compensate for the absence of the other protease and perform the first priming cleavage necessary to conjugate LGG-1 onto the nascent autophagosome. They also localized similarly throughout the neuronal cytoplasm, consistent with the ability to act either in the synaptic region or the cell body (Figure S6A-S6B). Also consistent with these two proteases performing partially redundant roles, we have previously shown that in AIY, where autophagy is also required for synapse development, the synaptic assembly phenotype was not seen for atg-4.1 or atg-4.2 single mutant animals, but was evident in the atg-4.1; atg-4.2 double mutant animals (Figure S6C-S6G and (Stavoe et al., 2016)), and that the GFP∷LGG-1 phenotype is enhanced in atg-4.1; atg-4.2 double mutant animals (Figures 4E and S6H-S6I). Our results indicate that while atg-4.1 and atg-4.2 can act in a partially redundant fashion, they also display distinct phenotypes in LGG-1/LGG-2 puncta accumulation. Importantly, our findings indicate that atg-4.2 is specifically necessary to prevent abnormal accumulation of autophagosomes in the neuronal soma.

ATG-4.2 promotes autophagosome maturation

Our observations that lesions in atg-4.2 result in animals with an abnormal accumulation of autophagosomes in the neuronal cell body are consistent with a model where atg-4.2 acts late in the autophagy pathway to promote autophagosome maturation and autophagic degradation. Autophagosome maturation into an acidic autolysosome is mediated by fusion between autophagosomes and late endosomes or lysosomes in a process that requires the small GTPase Rab7 (Gutierrez et al., 2004; Hyttinen et al., 2013; Jager et al., 2004). We reasoned that if atg-4.2 mutant animals have a defect at this step, then loss of rab-7 function should phenocopy the atg-4.2 mutant phenotype.

Since null lesions in rab-7 are lethal, we examined animals mutant for the rab-7 activator sand-1/Mon1 (Hegedus et al., 2016; Poteryaev et al., 2007), and the rab-7 effector, epg-5 (Wang et al., 2016). We observed that indeed sand-1 and epg-5 mutant animals displayed increased autophagosome accumulation in the cell body compared to wild type-animals, similar to atg-4.2(ola316) mutant animals (Figure 5A–5C, 5G). The observed phenotypes are specific to late blocks in autophagy, as mutant animals for genes involved in early events prior to autophagosome closure or trafficking, such as atg-9(wy56), atg-2(bp576), or unc-16(ju146), did not display autophagosome accumulation in the cell body (Figure 5G). Also consistent with atg-4.2 acting in a late step of autophagy to promote autophagosome maturation, we found that atg-4.2(gk628327); epg-5(tm3425) double mutant animals phenocopied either single mutant for the cell body accumulation phenotype, suggesting that they act in the same pathway in this process (Figure S6J). We do note however that the atg-4.2(gk628327); epg-5(tm3425) double mutant animals also displayed an enhanced accumulation in the presynaptic region (Figure S6K), suggesting that they might perform distinct roles at presynaptic sites, which we did not further examine in this study. Together our findings indicate that atg-4.2, like sand-1 and epg-5, acts in a late stage of autophagy prior to degradation.

Atg-4.2 genetically interacts with unc-16 in double mutant animals to enhance the number of autophagosomes observed at the synapse. To test if sand-1 also genetically interacts with unc-16, we created sand-1(or552); unc-16(ju146) double mutant animals. We observed that like atg-4.2; unc-16 double mutant animals, sand-1; unc-16 double mutant animals also displayed dramatically enhanced accumulation of autophagosomes in the synaptic regions (Figure 5D–5F and 5H). Importantly, early autophagy mutants like atg-2 (required for autophagosome closure) (Velikkakath et al., 2012) and atg-4.1 (required for LGG-1 priming) (Wu et al., 2012) failed to enhance the neurite accumulation phenotype observed in unc-16 mutant animals (Figure 5H). Our findings indicate that the observed enhancement of unc-16 with either sand-1 or atg-4.2 is a specific phenotype resulting from a genetic interaction between defects in retrograde transport and defects in autophagosome degradation.

We then tested the acidification state of autophagic vacuoles by employing a tandem label strategy in which LGG-1 was fused to both mCherry and GFP (Figure 5I; lipidation controls in Figure S7A-S7D and (Chang et al., 2017; Kimura et al., 2007)). Briefly, in this strategy immature autophagosomes are labeled with both GFP and mCherry, but because GFP is preferentially quenched in acidic environments, autophagic structures lose their GFP signal and display solely mCherry signal when mature (schematic, Figure 5J). Comparisons with the tandem marker lipidation control (Figure S7A-S7D), suggest that while GFP/mCh positive structures are lipidated, some mCh positive/GFP negative puncta accumulate in acidified lysosomes (see Figure S7, Figure Legend and STAR Methods). Using this approach in vivo and in single neurons of wild-type animals, we observed that 88% of synaptic LGG-1-labeled structures were positive for mCh and GFP signal (Figures 5K–5M and S7H). In contrast, examination of LGG-1-labeled structures in the cell body region of the neurons demonstrated that over 70% of the structures were positive for mCherry, but not for GFP (Figures 5N–5P). Our findings in wild-type animals are consistent with studies that demonstrate that active lysosomes are preferentially enriched in the neuronal cell soma (Hollenbeck, 1993; Kaasinen et al., 2008), and suggest that most autophagosomes at synapses mature as they are transported to the neuronal cell body.

To better understand the ability of the tandem marker to assess the maturation state of autophagosomes in mutant animals, we first examined it in sand-1/Mon1 (Hegedus et al., 2016; Poteryaev et al., 2007), and epg-5 (Wang et al., 2016) mutant animals, which are known to block autophagosome acidification. We observed that as compared to wild-type animals, the percentage of mature LGG-1 structures (mCh positive, GFP negative) decreases (Figure 5W), as expected given the known mechanism of action of these genes in autophagosome maturation.

We then used the tandem marker to examine the maturation state of autophagosomes in atg-4.2 mutant animals. For two independent alleles of atg-4.2, we observed that a higher percentage of autophagic vacuoles were immature (positive for both GFP and mCh signal) compared to wild-type or atg-4.1 mutant animals (Figure 5Q–5W). The immature state of the autophagic puncta in atg-4.2 mutant animals could be a result of failure to remove LGG-1 from the outside of the autophagosome (where it would not be acidified), failure to undergo internal acidification, or both, possibilities that would need to be further distinguished through biochemical assays (see below). Nevertheless, these findings with the tandem marker indicate that atg-4.2, but not atg-4.1, is required for late stages in the progression of autophagy.

ATG-4.2 delipidates LGG-1 and is necessary for the formation of autolysosomes that degrade synaptic proteins

Why is atg-4.2, but not atg-4.1, required for the late stages of autophagy? Atg4s are cysteine proteases that cleave Atg8/GABARAP/LC3/LGG-1/LGG-2 to: 1) activate LGG-1 for conjugation onto the autophagosomal membrane (early event) or 2) remove LGG-1/LGG-2 from the autophagosomal membrane (late event) ((Kirisako et al., 2000; Maruyama and Noda, 2017) and Figure 6A). We wondered if ATG-4.2 might have distinct biochemical activity towards LGG-1 in these steps.

We tested ATG-4.2 activity on LGG-1 by generating recombinant C. elegans ATG-4.2 and in a biochemically-reconstituted system examining either: 1) ATG-4.2 priming activity on the C-terminal of LGG-1, using a construct with an extended C-terminus (LGG-1-His); or 2) delipidation activity on LGG-1 conjugated onto liposomes, as described (Kauffman et al., 2018). We observed that after 8 minutes, 89% of liposomal LGG-1 was cleaved, but only 18% of LGG-1-His was cleaved (average of three independent experiments; Figure 6B), indicating that ATG-4.2 preferentially delipidates (or removes) LGG-1 from autophagosomal membranes. Our biochemical results are consistent with our cell biological and genetic data, and together suggest a model whereby delipidation of LGG-1, specifically by ATG-4.2, is a necessary step for correct maturation of autophagosomes.

To gain further insights into the effects of ATG-4.2 in autophagic progression we examined if there were defects in autolysosomes of atg-4.2 mutant animals. Formation of autolysosomes, which occurs through the fusion of autophagosomes and lysosomes, is a necessary step in autophagosome maturation, and can be assayed by probing the co-localization of lysosomal proteins, such as LAAT-1 (Liu et al., 2012) and the autophagy protein LGG-1 (Hyttinen et al., 2013; Maday et al., 2012; Zhang et al., 2015). We observed that in wild-type animals 91% of the LGG-1 puncta in the cell body co-localized with LAAT-1 (Figure S7I-S7P)), consistent with a progression towards autophagosomes-lysosome fusion and acidification as autophagic structures are transported towards the neuronal cell body. In atg-4.2 mutant animals, while there was no change in LGG-1/LAAT-1 colocalization in the synaptic region, nor in the total number of LGG-1 and/or LAAT-1 structures in the cell body (Figure S7I and S7Q), there was a significant reduction in colocalization of LGG-1 and LAAT-1 (Figure 6C–6E). These findings are consistent with a role for atg-4.2 in promoting autolysosome formation in the neuronal cell body.

Next, we examined the maturation state of the autolysosomes by using the LGG-1 tandem marker while simultaneously visualizing LAAT-1:BFP. Consistent with our previous findings, we observed that in wild-type animals 92% of LAAT-1 and LGG-1 colocalized structures were mature (red positive, green negative), while in atg-4.2(ola316) mutant animals only 33% of LAAT-1 and LGG-1 colocalized structures were mature (Figure 6F–6H). This effect was, at least in part, due to a failure to delipidate LGG-1 from the cytoplasmic face of the autolysosomes, but might also reflect a defect in internal acidification. Our observations are consistent with atg-4.2 mutant animals having a defect in the maturation of autophagic structures, and suggest that ATG-4.2 is necessary for the formation of autolysosomes.

To examine how atg-4.2 affects the acidification of autophagic cargo, we visualized the co-localization of synaptic proteins with LGG-1 in the neuronal cell body. Consistent with the role of autophagy in degrading synaptic material, and with the role of ATG-4.2 in the clearance of autophagic structures, we observed that atg-4.2 mutant animals displayed an increased percentage of autophagosomes co-localized with synaptic proteins in the cell soma (Figure 6I–6K and Figure S7R). Moreover generation of a synaptic tandem marker, which enabled visualization of the internal acidification state of compartments containing synaptic material, revealed that in atg-4.2 mutant animals fewer compartments in which synaptic proteins were acidified (Figure 6L–6O). Together, our findings indicate that atg-4.2 mutant animals are defective in acidification, resulting in an accumulation of immature autophagosomes with undigested cargo, including synaptic material.

CONTACT FOR REAGENT AND RESOURCE SHARING

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Daniel A. Colón-Ramos (daniel.colon-ramos@yale.edu).

EXPERIMENTAL MODEL AND SUBJECT DETAILS

METHOD DETAILS

QUANTIFICATION and STATISTICAL ANALYSIS