Reduced endosomal microautophagy activity in aging associates with enhanced exocyst‐mediated protein secretion

Abstract Autophagy is essential for protein quality control and regulation of the functional proteome. Failure of autophagy pathways with age contributes to loss of proteostasis in aged organisms and accelerates the progression of age‐related diseases. In this work, we show that activity of endosomal microautophagy (eMI), a selective type of autophagy occurring in late endosomes, declines with age and identify the sub‐proteome affected by this loss of function. Proteomics of late endosomes from old mice revealed an aberrant glycation signature for Hsc70, the chaperone responsible for substrate targeting to eMI. Age‐related Hsc70 glycation reduces its stability in late endosomes by favoring its organization into high molecular weight protein complexes and promoting its internalization/degradation inside late endosomes. Reduction of eMI with age associates with an increase in protein secretion, as late endosomes can release protein‐loaded exosomes upon plasma membrane fusion. Our search for molecular mediators of the eMI/secretion switch identified the exocyst‐RalA complex, known for its role in exocytosis, as a novel physiological eMI inhibitor that interacts with Hsc70 and acts directly at the late endosome membrane. This inhibitory function along with the higher exocyst‐RalA complex levels detected in late endosomes from old mice could explain, at least in part, reduced eMI activity with age. Interaction of Hsc70 with components of the exocyst‐RalA complex places this chaperone in the switch from eMI to secretion. Reduced intracellular degradation in favor of extracellular release of undegraded material with age may be relevant to the spreading of proteotoxicity associated with aging and progression of proteinopathies.


| INTRODUC TI ON
Cells constantly renew their proteome to ensure proteostasis and to adjust protein levels to cellular needs. Proteostasis is accomplished through coordinated function of chaperones and proteolytic systems, which work together to identify and eliminate damaged proteins (Sala et al., 2017). Loss of proteostasis is a hallmark of aging and contributes to the pathogenesis of age-related pathologies such as neurodegenerative diseases, metabolic conditions, or cancer (Kaushik & Cuervo, 2015;Lopez-Otin et al., 2013). Autophagy, the degradation of intracellular contents in lysosomes or late endosome/multivesicular bodies (LE/MVBs), is an essential component of the proteostasis network. Several forms of autophagy co-exist in most mammalian cells including macroautophagy, which mediates degradation of proteins and whole organelles entrapped in double membrane vesicles that then fuse with lysosomes (Galluzzi et al., 2017), and chaperone-mediated autophagy (CMA), dedicated to selective degradation of individual cytosolic proteins that directly cross the lysosomal membrane (Kaushik & Cuervo, 2018). A third type of mammalian autophagy is endosomal microautophagy (eMI), which allows for the selective uptake of cytosolic proteins in LE/ MVBs (Sahu et al., 2011). Like CMA, eMI is initiated by the recognition of a pentapeptide KFERQ-like motif (Kirchner et al., 2019) in the amino acid sequence of substrate proteins by the heat shock cognate protein of 70 kDa (Hsc70) (Sahu et al., 2011). Hsc70 then binds to phosphatidylserine residues on the LE/MVB membrane (Morozova et al., 2016), and through interaction with the chaperone Bag6 (Krause et al., 2022) triggers internalization of the substrate protein via invaginations of the LE/MVB limiting membrane.
These invaginations form in a manner dependent on the endosomal sorting complexes required for transport (ESCRT) (Sahu et al., 2011). In addition to protein internalization through eMI and other microautophagy-related pathways (Schuck, 2020), the LE/ MVB receives extracellular cargo through heterophagy pathways (pinocytosis, phagocytosis, endocytosis) (Johannes, 2021;Müller et al., 2015;Piper & Katzmann, 2007) and contributes to the turnover of ubiquitinated plasma membrane receptors via the ESCRT pathway (Shields & Piper, 2011), further emphasizing the importance of this compartment in protein degradation.
Because eMI was only recently discovered, its physiological relevance is still poorly understood. Activation of eMI as part of the starvation response in Drosophila has been described (Mukherjee et al., 2016). The effect of nutrient deprivation on eMI in mammalian cells is less clear, with upregulation of a variant of eMI very early in starvation (Mejlvang et al., 2018) and gradual decline in eMI activity as starvation is sustained (Krause et al., 2022). eMI contributes to clearance of prone-to-aggregate proteins such as Tau; however, pathogenic Tau variants found in Tauopathies fail to be degraded by eMI and instead hinder this type of autophagy by promoting docking of LE/MVBs in the plasma membrane and subsequent release of the pathogenic Tau variants (Caballero et al., 2018;Caballero et al., 2021). These findings suggest that eMI dysfunction could be involved in the pathogenesis and progression of Tauopathies through propagation of proteotoxicity. Extracellular release of undegraded eMI cargo may represent an alternative way for cells to rid themselves of toxic products, particularly when degradation is impaired, as is the case in aging (Buratta et al., 2020). LE/MVBs isolated from old mice have been found to contain high levels of oxidized proteins, suggesting an age-related defect in the degradative capacity of LE/MVBs that could result in defective eMI in aging (Cannizzo et al., 2012). However, possible changes in eMI with age and the molecular defect(s) behind those changes have not been directly studied.
In this work, we demonstrate a decline in eMI activity with age and place it at the level of internalization and degradation of substrate proteins. We next identify changes with age in the stability and dynamics of LE/MVB-associated Hsc70 that we attribute to age-dependent glycation of this chaperone. In the second part of this study, we focus on the potential relationship between the eMI degradative function of LE/MVBs and their known role in exosome biogenesis and release (Hessvik & Llorente, 2018). We demonstrate an increase in protein secretion in aging that we attribute to reduced intracellular protein degradation and identify an additional previously unknown function of the exocyst complex and the GTPase RalA (Exocyst-RalA complex), known for their involvement in vesicle secretion (Moskalenko et al., 2003;Wang et al., 2004), as negative regulators of eMI. Abnormally elevated levels of the exocyst/ RalA complex in LE/MVBs with age provides a potential mechanism linking reduced eMI degradation and enhanced exocytosis of undegraded material in aging.

| eMI activity decreases with age
Previous data showed that LE/MVBs from old animals accumulate undegraded oxidized proteins (Cannizzo et al., 2012), suggesting a possible age-related decline in eMI function. To investigate this possibility, we isolated LE/MVBs from young (4-6 m) and old (22-24 m) mouse livers and reconstituted the different steps of eMI in vitro using a previously developed assay (Krause & Cuervo, 2021;Sahu et al., 2011) (Figure 1a; comparable purity of fractions isolated from young and old mice was verified by immunoblot for LE/MVBs markers and enzymatic assays for the endolysosomal enzyme β-hexosaminidase [ Figure S1a-c]). LE/MVBs were incubated with two KFERQ-motif containing proteins (Tau and αsynuclein), previously shown to undergo degradation by selective eMI (Caballero et al., 2018;Krause et al., 2022) or with Cyclophilin A, which lacks a KFERQ motif and can be internalized and degraded in LE/MVBs through non-selective eMI (Sahu et al., 2011) ( Figure 1b). Incubation in the presence or absence of protease inhibitors allows for the differentiation of substrate binding (substrate recovered in LE/MVBs not treated with protease inhibitors) and substrate internalization/degradation (calculated as the difference between the substrate recovered in LE/MVBs pre-treated with protease inhibitors and those untreated) (Figure 1a). We found reduced binding for the three proteins in LE/MVBs from old mice (Figure 1b,c), and significantly reduced internalization/degradation for both selective substrates, Tau and α-synuclein, that was not observed for Cyclophilin A (Figure 1b,d), suggesting that the functional decrease with age preferentially affects KFERQselective eMI.
Defective eMI with age could be a consequence of changes in activity per LE/MVB or in the fraction of cellular LE/MVBs that can perform eMI. To differentiate between these possibilities, we incubated LE/MVBs with the eMI substrate Tau and performed immunofluorescence of the glass-spotted samples for Tau and the endolysosomal marker LAMP-1 to measure the percentage of LE/ MVBs competent for eMI (LAMP-1 + /Tau + ) (Figure 1e). We observed no difference between the two ages in the percentage of LE/MVBs (LAMP-1 + vesicles) that colocalized with the eMI substrate Tau (Figure 1f), suggesting that the fraction of LE/MVBs capable of performing eMI is preserved with age and that it is the absolute eMI activity per LE/MVB that decreases in the old group.
To determine if the reduction in eMI activity with age is observed in vivo and to explore possible changes in substrate targeting to LE/MVBs with age (a step not reconstituted in the in vitro system), we injected young and old mice with saline or the endolysosomal protease inhibitor leupeptin to prevent degradation in LE/ MVBs (Figure 1g,h; efficacy of leupeptin injection was confirmed in total homogenate with proteins known to undergo lysosomal degradation [ Figure S1d]). Immunoblot for three KFERQ-bearing proteins (GAPDH, Aldolase and LRRK2) did not reveal reduction with age in the levels of endogenous eMI substrates associated with LE/MVBs (Figure 1i), suggesting that Hsc70-mediated targeting to these organelles was comparable in young and old mice.
However, once again, we found clear trends of reduced internalization/degradation for two of the selective endogenous substrates (Aldolase and LRRK2) (Figure 1j), which match their higher steady-state levels in this compartment ( Figure 1i) (likely bound to the membrane but not efficiently internalized/degraded). This decrease in eMI was not observed for one of the other endogenous substrates (GAPDH), suggesting possible substrate-specific differences in their ability to undergo eMI with age and likely competition for LE/MVB internalization/degradation based on the cellular abundance of these proteins.
We next used a modified in vitro assay to analyze substrate internalization separately from degradation by measuring the proteolysis of a pool of radiolabeled proteins incubated with isolated LE/MVBs with intact membranes or upon disruption of their membranes with detergent to allow direct access of the LE/MVB enzymes to the radiolabeled substrates, and thus, eliminate the internalization step.
We found a marked reduction in the degradation of radiolabeled proteins in both cases (Figure 1k), suggesting that at least part of the age-related reduction in eMI is due to less efficient proteolysis by luminal enzymes.
Despite lower proteolytic activity, immunoblot of isolated LE/ MVBs from old mice did not reveal major changes with age in levels of the luminal enzymes GBA and Cathepsin D ( Figure S1a,b).
We next considered the possibility of defective acidification.
However, measurement of the pH of isolated LE/MVBs using the radiometric pH probe LysoSensor Yellow/Blue DND-160 revealed higher acidification in old mice LE/MVBs (pH shift from 5.75 to 5.36) (Figure 1l). Direct measurement of the activity of other pHdependent enzymes in LE/MVBs such as β-hexosaminidase, which hydrolyzes amino sugars, also confirmed higher activity in LE/ MVBs from old mice ( Figure S1c), further supporting their proper acidification.
Overall, these findings unveil a decline in selective eMI with age mostly due to defective degradation of internalized cargo. Reduced substrate degradation does not seem to be a consequence of overall changes in LE/MVB properties such as acidification but appears to be instead limited to defective protein break down.

| LE/MVB-associated Hsc70 is glycated with age
We next sought to understand the molecular defects behind the functional decline in eMI by analyzing possible changes in the levels of eMI-related components with age. Immunoblot analysis revealed that total cellular levels ( Figure S2a protein), consistent with an age-related sugar modification as previously detected for other proteins in LE/MVBs from old organisms (Cannizzo et al., 2012). While this modification is detectable in liver homogenates from old mice (27% of the total cellular Hsc70), it is highly enriched in LE/MVBs where it represents 53% of LE/MVBassociated Hsc70 (Figure 2c,d). It is unlikely that the glycosylated form of hsc70 in LE/MVBs is a cytosolic contaminant, since only after long exposure a small fraction of cytosolic hsc70 (<10%) displayed the molecular weight shift ( Figure S2c). Mass spectrometry analysis identified that Hsc70 in LE/MVBs from old mice had sugar moieties in five asparagine residues not present or barely detectable in Hsc70 from young LE/MVBs (Figure 2e). The modified residues were distributed (i) along the nucleotide binding region (with N 35 located in the protein/protein interface of the interaction of Hsc70 with co-chaperones of the Bag family and N 151 in close proximity to the nucleotide binding pocket), (ii) in the substrate binding region, and (iii) in the lid domain (with N 584 adjacent to K 583 , one of the lysine residues known to mediate interaction of Hsc70 with phosphatidylserine on the LE/MVB membrane during eMI (Morozova et al., 2016)) ( Figure S2d). We propose that glycation of these residues may affect the ability of Hsc70 to hydrolyze ATP, required for releasing bound substrates, its interaction with the co-chaperone Bag6, recently shown to co-operate with Hsc70 in the substrate internalization step of eMI (Krause et al., 2022), or interfere with the docking of Hsc70 at the LE/MVB (Morozova et al., 2016).
Since glycation can have a toxic effect on protein function (Simm, 2013), we analyzed changes in the behavior of Hsc70 in the LE/MVB compartment. To investigate the topology of glycated Hsc70, we incubated isolated LE/MVBs at room temperature with increasing concentrations of trypsin, which only degrades surface proteins (mTOR shown as control in Figure S2e). A larger fraction of the glycated form of Hsc70 was still present after trypsinization, suggesting that it may be more readily internalized into LE/MVBs than unmodified Hsc70 ( Figure S2e). Although persistence of glycated Hsc70 upon trypsinization could reflect intrinsic resistance to degradation by the protease, we discarded this possibility because we found that both unmodified and glycated Hsc70 displayed comparable kinetics of degradation in LE/MVBs. Incubation of LE/MVBs at physiological temperature (37°C) revealed a sharp reduction in levels of both the glycated and unmodified form of Hsc70 in old LE/ MVB that could be prevented by inhibition of proteolysis, in support of the proposed higher internalization and degradation of both forms of Hsc70 with age ( Figure 2f,g). In fact, Hsc70 behavior in old LE/MVBs resembled that of the co-chaperone Bag6 (Figure 2f,g), previously shown to be internalized, with a fraction of it degraded along with internalized substrates as part of the eMI process (Krause et al., 2022). Stability of another eMI component, the ESCRT protein Vps4, was also reduced in old LE/MVBs, albeit to a lower extent (Figure 2f,g). Whether Vps4 instability is related to the change in Hsc70 with age or is an independent event will require further investigation.
To further investigate possible aberrant protein-protein interactions and/or overall changes in the organization of Hsc70 in LE/ MVBs with age, we performed blue-native electrophoresis of isolated organelles. We detected a high molecular weight protein complex (~970 kDa) immunoreactive for Hsc70 only present in old LE/MVBs (Figure 2h), and two additional Hsc70 complexes (~930 and 600 kDa), also detected in young LE/MVBs but in significantly lower abundance than in old LE/MVBs (Figure 2h,i). The increase in F I G U R E 1 eMI activity decreases with age. (a-d) in vitro eMI reconstitution assay with isolated mouse liver LE/MVBs. Experimental design (a) and representative immunoblots (b) of LE/MVBs isolated from livers of 4 m and 22 m old mice incubated with (+) or without (−) protease inhibitors (PI) and the indicated substrate proteins. GBA and Tsg101 are shown as LE/MVB markers. Substrate binding (c) and internalization/degradation (int./deg.) (d) quantified in n = 12 (Tau) or 8 (Cyclophilin A, α-synuclein) mice. (e,f) Representative confocal microscopy images (e) and quantification (f) from LE/MVBs isolated from 4 m and 22 m old mice immunostained for the indicated proteins following incubation with Tau (e). Inset: Higher magnification of the boxed area with colocalization mask (in white). N ≥ 15 fields from 6 (4 m) and 3 (22 m) mice. (g-j) Experimental design (g), representative immunoblots (h) and quantification of steady-state levels (i) and degradation (j) of the indicated endogenous eMI substrates in LE/MVBs isolated from 4 m and 22 m old mice injected without or with leupeptin to block endolysosomal degradation. Data is expressed relative to average value of saline injected 4 m old animals. N = 5 mice. (k) Proteolytic activity of intact (left) and detergent disrupted (broken, right) LE/MVBs from 4 m and 22 m old mice incubated with a pool of radiolabeled cytosolic proteins. N = 9 (intact), 6 (broken) mice. (l) pH measurement from isolated LE/MVBs of 4 m and 22 m old mice n = 3 mice. All data are mean ± SEM and individual values. Ponceau staining is shown as loading control in the immunoblots. Paired (c,d) and unpaired (rest) twotailed t-test were used. Differences were significant for *p < 0.05 and **p < 0.01. ns, not significant was less evident in old LE/MVBs, where higher amounts of chaperones were internalized per molecule of substrate (slope = 0.7303, R 2 = 0.14, p = 0.037). It is possible that chaperone internalization with age occurs in part in a substrate-independent manner and/or that the observed clustering of Hsc70 in the old LE/MVBs promotes internalization of multi-chaperone complexes per substrate in these conditions. In fact, we also noticed some level of dissociation with age in the LE/MVB dynamics of Bag6 relative to Hsc70. We recently found that the ratio of Bag6 internalized for every Hsc70 molecule changes with eMI activity (Krause et al., 2022), with a higher amount of Bag6 to Hsc70 internalized when eMI is activated, as shown here for young LE/MVBs ( Figure 3i). However, similar analysis in the old LE/MVBs revealed a loss of the coordinated substrate-dependent increase in Bag6 internalization with Hsc70 (Figure 3i), in support of altered Bag6/Hsc70 dynamics.
Overall, our findings suggest that reduced eMI with age may be a consequence of altered Hsc70 dynamics at the LE/MVB membrane that lead to its higher internalization and subsequent augmented degradation in this compartment. We attribute the disrupted dynamics to age-related post-translational modifications (j) Proposed model of the effect of Hsc70 changes with age on eMI activity. Data are mean ± SEM and individual values. Ponceau staining is shown as loading control in the immunoblots. One sample multiple t-tests (a,i), unpaired two-sided t-test (d), and two-way ANOVA with Bonferroni multiple comparisons post hoc test (g) were used. Differences were significant for *p < 0.05, **p < 0.01, ***p < 0.001, ns, not significant F I G U R E 3 Age-related changes in the LE/MVB dynamics of Hsc70 and Bag6. (a-c) Representative τ-STED microscopy images of isolated LE/MVBs from 4 m (as previously reported (Krause et al., 2022)) and 22 m old animals incubated with (+) or without (−) the eMI substrate Tau and immunostained for LAMP-1, Hsc70, and Bag6. Single channel black and white original images (a, left) and pseudocolored 3 channels merged image (a, right), 2.5-D density plots (b) and fluorophore images (a), 2.5-D density plots (b) and fluorophore intensity plotting along the LE/MVB membrane (c) are shown. Data are mean ± SEM and individual values. Two-way ANOVA with Bonferroni's multiple comparisons post hoc test (d,g), unpaired two-sided t-test (f), and simple linear correlation (h,i) were used. Differences were significant for *p < 0.05, ***p < 0.001. ns, not significant on Hsc70, such as the glycation identified here, that affect the normal interaction of this chaperone with substrates and other eMI components.

| Impact of reduced eMI activity with age on the intracellular proteome
Since our analysis of endogenous eMI substrates indicated that the impact of aging on eMI may be, to some extent, substrate-dependent ( Figure 1i,j), we performed comparative quantitative proteomics of isolated LE/MVBs from young and old mice to determine changes in the profile of eMI substrates with age. We injected half of the mice with leupeptin, as in Figure 1g, to inhibit proteolysis in the LE/ MVB lumen and thus cause the accumulation of endogenous substrates (proteins with an increase of ≥20% upon leupeptin injection) (Schneider et al., 2014). When proteolysis was not inhibited, more proteins displayed higher levels in LE/MVBs from old mice when To differentiate substrate proteins reaching LE/MVBs through eMI (from the cytosol), from those degraded in LE/MVBs through heterophagy (pinocytosis, phagocytosis, endocytosis), we used as before (Krause et al., 2022) Gene Ontology (GO) analysis to eliminate proteins associated with the terms "extracellular space" and "plasma membranes", which included 34% and 25% of the degraded Interestingly, in addition to the 676 proteins no longer degraded in the old LE/MVBs, there was a subset of proteins (197) not previously detected to undergo degradation in this compartment in young mice that were degraded in old mice ( Figure S3f). This group included proteins normally present in LE/MVBs as constitutive components that become unstable in this compartment with age (184 proteins), as well as 13 cellular proteins only routed to LE/MVBs for degradation in the old group ( Figure S3g). We also identified proteins with defective eMI targeting to LE/MVBs with age as those degraded in young LE/MVBs and no longer detected in the LE/MVBs from old mice injected or not with leupeptin ( Figure S3g). Only 7 proteins fulfilled these criteria, further confirming that LE/MVB targeting of eMI substrates is preserved with age and that the primary defect in this autophagic pathway is in internalization/degradation of cargo.

| Protein secretion is elevated with age
As the LE/MVB is also the site of exosome biogenesis (Stoorvogel et al., 2002), one of the potential consequences of reduced degradation of cargo internalized through eMI in aging could be increased exosome secretion as a way to remove toxic protein products from the cell. In fact, previous studies from our group have shown that this is the case for pathogenic forms of Tau proteins, which have inhibitory effects on different autophagic pathways. These proteins reach the extracellular space in large part through their eMIdependent delivery to LE/MVBs and subsequent fusion of these compartments with the plasma membrane (Caballero et al., 2021).
To study the relationship between the impairment of intracellular proteolysis and changes in protein secretion with age, we used primary ear fibroblasts from young and old mice. To avoid the confounding effect of elevated secretion in cellular senescence (Coppé et al., 2008), we only used early passage cultures, which we previously confirmed contained less than 1% of senescent cells (positive for β-galactosidase) (Bejarano et al., 2018). We used a pulse of metabolic labeling with 3 H-leucine and then measured secretion of radiolabeled proteins in the extracellular media of cultured fibroblasts ( Figure 4a). We found a significant 73% increase in protein secretion by fibroblasts from old mice compared with young mice (Figure 4b).
Interestingly, this increase in secretion can be reproduced in young per cell a reliable measurement of eMI. We used a library of short guide RNAs (sgRNAs) targeting proteostasis-related genes (including genes involved in protein folding, degradation and secretion and in endolysosomal and autophagy pathways [Horlbeck et al., 2016]) and used fluorescent activated cell sorting to separate cells with high (top 33%) and low (bottom 33%) Split Venus signal to determine the frequency of individual sgRNAs in each population by nextgeneration sequencing ( Figure S5b).
Among the top hits of the screen, we noticed that transcriptional repression of 3 components of the exocyst complex, previously linked to vesicle secretion (Guo et al., 1999;TerBush et al., 1996), led to significant changes in eMI activity ( Figure S5c,d). The exocyst is an eight-member protein complex organized into two different subcomplexes (SCI and SCII) (Figure 5a) (Ahmed et al., 2018), whose assembly and cellular localization are often regulated by the GTPases RalA and RalB (Moskalenko et al., 2003). The exocyst mediates tethering and localization of secretory vesicles to the plasma membrane, but some exocyst components have been reported to be associated with LE/MVBs (Monteiro et al., 2013). We confirmed that members of the two exocyst subcomplexes, as well as RalA and RalB, associate (d-f) Representative images (d) of NIH3T3 mouse fibroblasts stably expressing the KFERQ-split Venus reporter control (ctrl) or singly knocked-down for the indicated components of the exocyst complex or RalA. Cells were untreated (none) or treated with ammonium chloride and leupeptin (+N/L) to block endolysosomal proteolysis and maintained in the presence (serum+) or absence (serum-) of serum. Nuclei are highlighted with Hoechst. Insets: Higher magnification of the boxed regions in the fields of NIH3T3 serum-supplemented with N/L. Quantification of eMI binding (left) and degradation (right) in serum+ (e) and serum-(f) conditions. N ≥ 2500 cells from 2 independent experiments (ie). Data are mean ± SEM and individual values. One sample multiple t-test relative to the control cells was used (e,f). Differences were significant for *p < 0.05, **p < 0.01, ***p < 0.001 and ****p < 0.0001 Although limited by the availability of antibodies suitable for reverse co-immunoprecipitation, we were able to at least confirm that pull-down with anti-Exoc4 co-immunoprecipitated Hsc70 ( Figure S6e). Interaction of Exoc4 or Exoc7 with Hsc70 and Bag6 was not detected in cytosol (Figure 6e)

| Age-related changes in exocyst complex organization in LE/MVBs
Given the inhibitory effect of the exocyst complex on eMI, we investigated if changes in this complex with age may underlie some of the reduction in eMI with aging. While overall levels of the tested exocyst components (Exoc2, Exoc3, Exoc4, Exoc5, and Exoc7) did  Ponceau staining is shown as loading control in the immunoblots. One sample multiple t-tests (a-c), unpaired two-sided t-test (l) and two-way ANOVA with Bonferroni's multiple comparisons post hoc test (n) were used. Differences were significant for *p < 0.05, **p < 0.01, ****p < 0.0001. ns, not significant levels observed in young mice. Daily oral administration of CA77.1, a novel, specific CMA activator suitable for in vivo use  in old mice significantly reduced the age-related accumulation of RalA without affecting levels of Exoc2 (SCI) or Exoc7 (SCII) (Figure 7n).
Overall, our findings are consistent with an inhibitory role of the exocyst complex in eMI activity and with RalA-dependent changes with age in exocyst complex dynamics in LE/MVBs, likely as a result of its interaction with glycosylated Hsc70, that result in an inhibitory effect on eMI activity ( Figure S8). Defective degradation of RalA by CMA provides a basis for the age-related accumulation of RalA and places CMA as a key regulator of eMI activity. The age-related increase in the organization of exocyst proteins into complexes may reflect both a mechanism for the reduced eMI activity observed in aging and a link between eMI and protein secretion when substrates can no longer be effectively degraded by the pathway (Figure S8).

| DISCUSS ION
This work provides evidence for reduced eMI activity with age and identifies the causes and possible consequences of the failure of this component of the proteostasis network. We propose that the age- Our work supports that the reduction in eMI activity with age occurs at the level of substrate internalization and degradation, whereas targeting of eMI substrate proteins to LE/MVBs remains, for the most part, unchanged. These findings resemble those described for CMA, the type of autophagy most closely related to eMI, where proper targeting of substrate proteins by Hsc70 to lysosomes was still observed in old organisms (Cuervo & Dice, 2000). However, while in CMA, the decrease with age in levels of the lysosomal receptor for this pathway makes substrate binding to lysosomes the first affected step in aging (Cuervo & Dice, 2000), in the case of eMI we found that substrate internalization and mostly degradation  in protein aggregates reduced the availability of this chaperone for removal of clathrin from early endocytic vesicles (Yu et al., 2014).
In the case of LE/MVBs, we observed instead an increase in Hsc70 levels with age (Figure 2a), suggesting that either different endocytic process are differentially affected by age or that the chaperone depletion observed in the context of neurodegeneration may not occur until more advanced ages than tested here or only in instances with high levels of cytosolic protein aggregation. Future studies will determine the possible contribution of the failure of endocytosis in those instances to eMI function.
A growing number of examples support that if proteins are unable to be degraded due to a blockage in the intracellular proteolytic systems, a last resort to rid the cell of toxic products could be to secrete them outside of the cell (Caballero et al., 2021;Ferreira et al., 2019;Solvik et al., 2022). Different mechanisms contribute to this release of undegraded materials, likely dependent on the origin of the proteolytic blockage. Failure to acidify lysosomes has been shown to increase secretion of proteins in extracellular vesicles in a macroautophagy-dependent manner (secretory autophagy) (Solvik et al., 2022). In contrast, selective inhibition of CMA leads to rerouting of substrate proteins toward LE/MVBs via eMI and subsequent extracellular release in exosomes upon docking of LE/MVBs with the plasma membrane (Caballero et al., 2021). Here, we demonstrated that protein secretion is elevated during aging using primary fibroblasts from young and old mice and that this secretion occurs in a manner independent of conventional secretion mechanisms based on the lack of sensitivity to Brefeldin A treatment (Figure 4).
The studies in the first part of this work were all done in isolated liver LE/MVBs because the amount and purity of the recovered organelles allows for the reconstitution of eMI in vitro to directly analyze changes with age in this pathway separately from other agerelated changes. In contrast, the studies on protein secretion were performed in primary fibroblasts in culture given their suitability for collection of ECVs from a single cell type. However, the fact that elevation in secretion was reproduced by acute blockage of proteolysis in young primary mouse fibroblasts helped link the observations in the two different experimental systems and suggests that the inability to degrade proteins may trigger their extracellular release rather than risk their intracellular accumulation. Future studies comparing degradation and secretion in the same model system are needed to elucidate the molecular mechanisms behind the crosstalk between both processes.
The absence of the previously reported increase in autophagy receptors in the secreted material when released through secretory autophagy (Solvik et al., 2022), along with the marked increase of RalA and other exocytosis-related components in LE/MVB with age, make us propose exocytosis as one of the major contributors to the observed increase in protein secretion with age. Interestingly, a recent study has shown that inhibition of proteolysis with chloroquine in several mammary cancer cell lines led to changes in cargo released in EVs but not in the total amount of secreted protein in these vesicles (Xu et al., 2022). It is possible that consequences of endo/ lysosomal proteolysis inhibition are different between primary, untransformed cells and cancer cells, as the rapid division of the latter makes protein quality control less of a priority. Alternatively, failure to increase protein content in EVs in that study could be due to the previously reported inhibitory effect of chloroquine on protein synthesis (Ciak & Hahn, 1966), not observed with the endolysosomal proteolysis inhibitors used in our work (leupeptin and ammonium chloride).
We have uncovered a previously unknown role for the exocyst complex and the GTPase RalA in regulating eMI activity ( Figures 5   and 6). The exocyst complex inhibits eMI substrate internalization and degradation directly at the LE/MVB membrane. This inhibition may be mediated through an interaction with the eMI chaperones Hsc70 and Bag6 based on co-precipitation of Exoc4 and 7 with these chaperones in the LE/MVB compartment. Elucidating the exact mechanism of exocyst inhibition of eMI activity will require future studies, but based on our findings we favor a model where for RalA (Figure 7). Future studies are needed to dissect the protein composition of the high molecular weight complexes of SCI and SCII that accumulate in old LE/MVB ( Figure 7); however, the fact that similar size complexes, albeit less abundant, are also visible in young LE/MVBs supports that they may be a result of enhanced assembly or reduced disassembly, rather than non-specific interactions with other proteins at the old LE/MVB membrane. Since glycosylated Hsc70 can be precipitated with SCI components and RalA, we propose that this post-translational modification may be the culprit of the disrupted kinetics of SCI and RalA in LE/MVBs with age.
RalA-mediated exocyst assembly may be one mechanism through which LE/MVBs can be transitioned from a degradative to a secretory compartment and explain the relationship between reduced degradation and increased secretion observed in this work ( Figures 6 and 7). The exocyst complex has been shown to work both independently (Monteiro et al., 2013)

or in conjunction with
RalA (Wang et al., 2004) to facilitate vesicle fusion with the plasma membrane. It is unclear which of these mechanisms is primarily responsible for the elevated protein secretion observed in aging, although the above-mentioned coincidence of higher RalA levels and enhanced exocyst complex assembly in old LE/MVBs makes it likely that the exocyst and RalA act coordinately in the switch from eMI degradation to secretion. Future studies are needed to determine if the high molecular weight complex containing subunits of SCI and SCII but free of Hsc70 is enough to promote secretion. Alternatively, the Hsc70-SCI-RalA complex detected in higher abundance in old LE/MVBs, could be the one targeting these organelles to SCII containing regions of the plasma membrane to promote full exocyst assembly and vesicular content release.
We found it interesting that the observed increase with age in cellular levels of RalA can be in large part explained by a decrease in its degradation through CMA (Figure 7), providing an additional point of interplay between these two selective types of autophagy.
CMA-dependent degradation of RalA may be a way to titrate eMI activity, as we found that small changes in RalA levels can have a dramatic effect on this pathway ( Figure 5) and to coordinate CMA and eMI activities. Furthermore, this previously unknown degradation of RalA by CMA highlights a possible impact of changes in CMA activity on the many cellular processes regulated by RalA (Yan & Theodorescu, 2018). Finally, the efficient normalization in RalA levels upon chemical activation of CMA provides a potential mechanism to reduce release of undegraded or toxic protein products that fail to undergo degradation through eMI. This may have therapeutic potential in the context of neurodegenerative diseases as a way to limit the extracellular release of pathogenic proteins like Tau and αsynuclein in favor of their intracellular degradation, which may help limit the propagation of disease (Miranda & Di Paolo, 2018).
Overall, this work establishes that eMI, like other proteostasis pathways, declines in function with age and provides possible mechanisms and consequences for this decline. The characterization of the exocyst complex and RalA as endogenous inhibitory modulators of eMI further adds to our still limited knowledge of how this pathway is regulated and establishes a link between eMI and protein secretion, which is used by aging cells to ameliorate proteotoxicity.

| Animal models and cell culture
This study used adult (3-6 months) male Wistar rats (Charles River Laboratories) and male young (4-6 months) and old ( Table S1. Details on animal maintenance and treatments and on cell culture conditions and in vitro treatments are provided under Extended Experimental Procedures (Appendix S1).

| Chemicals and antibodies
Sources of chemicals and dilutions and sources of antibodies used in this study are detailed under Extended Experimental Procedures (Appendix S1).

| Studies on isolated organelles
Rodent liver lysosomes and LE/MVBs were isolated through differential centrifugation and separation into discontinuous density gradients. Cytosol was prepared by centrifugation at 100,000×g for 1 h of the post-nuclear supernatant. Extracellular vesicles (ECVs) were isolated from the culture media by sequential centrifugation and concentration through molecular exclusion columns. LysoSensor Yellow/Blue DND-160 was used for pH determination, PNGase F and Endo H for deglycosylation, trypsinization to determine protein topology and incubation at physiological temperature and immunoblot to determine protein stability as described in detail under Extended Experimental Procedures (Appendix S1).

| eMI measurements
In vitro analysis of eMI was performed using isolated LE/MVBs and purified substrate proteins as described before (Krause & Cuervo, 2021). Binding was calculated as the amount of protein associated with LE/MVBs untreated with protease inhibitors and internalization/degradation as the difference between proteins present in organelles treated with protease inhibitors after subtracting for the amount bound. Proteolysis in isolated LE/MVBs was assayed using a pool of radiolabeled proteins as before (Sahu et al., 2011).

Measurement of eMI in cultured cells was done using the KFERQ-
Split Venus reporter (Caballero et al., 2018)

described in detail under
Extended Experimental Procedures (Appendix S1).

| Protein analysis procedures
Protein secretion was measure upon 3 H-leucine metabolic labeling of cultured cells by analysis of radioactivity on the acid precipitable fraction collected from the culture media. Co-immunoprecipitation was performed upon solubilization in a mild detergent/low salt buffer using primary antibodies and Protein A/G Plus agarose beads. Protein electrophoresis and immunoblot, quantitative proteomics, and protein pathway analysis and glycoproteomics were performed using standard procedures described in detail under Extended Experimental Procedures (Appendix S1).

| Image-based procedures
Immunofluorescence was performed in glass spotted organelles, fixed, and incubated with primary and secondary antibodies following standard procedures. STED microscopy of isolated organelles was performed in a Leica TCS SP8 STED 3X outfitted with a τ-STED module and analysis of membrane/lumen distribution and organization of proteins in hot spots based on protein abundance were performed as described before (Krause et al., 2022) and detailed under Extended Experimental Procedures (Appendix S1).

| CRISPRi screen and RNA quantification
CRISPR interference (CRISPRi) screen with an sgRNA library targeting 1176 genes related to proteostasis pathways was performed using cells expressing nuclease-dead form of Cas9 fused to the transcriptional repressor KRAB (dCas9-KRAB) and the KFERQ-Split Venus fluorescent eMI reporter as described in detail under Extended Experimental Procedures (Appendix S1). Quantitative RT-PCR was performed after reverse transcription of RNA into cDNA using the primers shown in Table S2 following isolation of total RNA using the RNeasy Plus kit

ACK N OWLED G M ENTS
We thank Dr. Shilpa Dilipkumar for technical assistance with τ-STED microscopy and Dr. Susmita Kaushik for proofreading the manuscript and for helpful feedback.

FU N D I N G I N FO R M ATI O N
This work was supported by grants from the National Institutes of Health/National institute on Aging AG021904, AG054108 (to AMC), AG031782 (to AMC and JJBC) and AG062359 (to MK), National in-

DATA AVA I L A B I L I T Y S TAT E M E N T
There are no restrictions on data availability in this manuscript.