Ubr1-induced selective endophagy/autophagy protects against the endosomal and Ca2+-induced proteostasis disease stress

The cellular defense mechanisms against cumulative endo-lysosomal stress remain incompletely understood. Here, we identify Ubr1 as a protein quality control (QC) E3 ubiquitin-ligase that counteracts proteostasis stresses by facilitating endosomal cargo-selective autophagy for lysosomal degradation. Astrocyte regulatory cluster membrane protein MLC1 mutations cause endosomal compartment stress by fusion and enlargement. Partial lysosomal clearance of mutant endosomal MLC1 is accomplished by the endosomal QC ubiquitin ligases, CHIP and Ubr1 via ESCRT-dependent route. As a consequence of the endosomal stress, a supportive QC mechanism, dependent on both Ubr1 and SQSTM1/p62 activities, targets ubiquitinated and arginylated MLC1 mutants for selective endosomal autophagy (endophagy). This QC pathway is also activated for arginylated Ubr1-SQSTM1/p62 autophagy cargoes during cytosolic Ca2+-assault. Conversely, the loss of Ubr1 and/or arginylation elicited endosomal compartment stress. These findings underscore the critical housekeeping role of Ubr1 and arginylation-dependent endophagy/autophagy during endo-lysosomal proteostasis perturbations and suggest a link of Ubr1 to Ca2+ homeostasis and proteins implicated in various diseases including cancers and brain disorders. Supplementary Information The online version contains supplementary material available at 10.1007/s00018-022-04191-8.


Introduction
Disease-causing mutations, as well as perturbations in intracellular Ca 2+ homeostasis, may result in protein misfolding and/or accumulation in the cytosol or membrane organelles, mitochondrial oxidative damage 1 , the endoplasmic reticulum (ER) stress or lysosomal swelling and exocytosis 2 that have been linked to numerous Mendelian disorders, neurodegenerative diseases, and cancers. To protect protein homeostasis (proteostasis), cells have developed spatial or organelle-specific protein QC mechanisms 3 . At the ER, QC mechanisms recognize various degradation signals (degrons) in luminal, transmembrane or cytosolic segments of misfolded membrane proteins as part of the ER-associated degradation (ERAD) and ERphagy 4,5 . The cytosolic PQC preferentially relies on the coordinated actions of molecular chaperones, ubiquitin (Ub) conjugation machinery, proteasomes (Ub-proteasome systems) and chaperone-mediated autophagy 3 .
Conformationally defective cell surface membrane proteins generated either in situ or escaped from the ER QC are recognized by the peripheral PQC machinery. Primarily, increased post-translational conjugation of ubiquitin (ubiquitination) in non-native plasma membrane (PM) proteins is a signal for ESCRT (endosomal sorting complex required for transport)dependent targeting to the multivesicular body (MVB) and lysosomal degradation [6][7][8][9][10] . The role of peripheral PQC is to maintain native protein composition and intra-and inter-organelle homeostasis during membrane dynamics between Golgi, the plasma membrane (PM)-and endo-lysosomal compartments. The peripheral PQC is less studied, and only three QC E3 Ubligases have been implicated in the clearance of non-native PM and endosomal membrane proteins to lysosomal proteolysis. The C-terminus of Hsc70-Interacting Protein (CHIP), Nedd-4 (and its yeast homologues Rsp5), as well as the RFFL, represent both chaperone-dependent and -independent PQC mechanisms 6,7,9,11,12 .
From these, arginylation has a role outside QC as N-terminal degron for n-recognin type proteins in fast proteasomal degradation 17 . Previous research also found that N-terminal arginylation of BiP (HSPA5, GRP78) released from the ER and cytosolic misfolded soluble proteins can be cleared by autophagy upon proteasomal inhibition or cytokine-induced oxidative stress [18][19][20][21] . In the current study, we found new roles for n-recognins in PQC during Ca 2+ -assault, endosomes and cargo selective autophagy, different from their previous roles at the N-degron pathway.
The formation of phagophores, the newly formed membranes that engulf autophagy cargo, has been demonstrated at Rab11+ early endosomes 22 . Instead, endosomal autophagy cargo recognition and clearance mechanisms for misfolded endosomal membrane proteins or in endosomal stress QC remain undefined. To address these questions, we start the investigation by examining the endosomal PQC activity using disease-associated regulatory membrane protein MLC1 (megalencephalic leukoencephalopathy with subcortical cyst) and end the study addressing separately endosomal compartment PQC during Ca 2+ -stress.
MLC1 is a unique part of the astrocyte PM macromolecular regulatory signalling cluster, pertinent for astrocytes homeostasis, motility/morphology [23][24][25] , inflammatory responses 26 , signalling and ion homeostasis regulation of the brain extracellular space [27][28][29] . The cluster encompasses integral membrane proteins including Na⁺/K⁺-ATPase, TRPV4 cation and ClC-2 chloride channels, EGF receptors and GlialCAM adhesion molecules [30][31][32][33] . GlialCAM is required for the correct MLC1 cell surface entry and tetherin 34 . At the cell surface, MLC1/GlialCAM regulates the diffusional partitioning and endosomal dynamics of the cluster 34 , as well as the activity of ClC-2 32 , gap junctions 25,28 and Na + -K + -ATPase 35,36 . Disease mutations in MLC1 or GlialCAM and/or altered expression or compromised PM tethering of the cluster result in MLC1-dependent loss of endo-lysosomal organellar identity and impaired cargo sorting 34 . Thus, MLC1 has far-reaching consequences on astrocytic and neuronal proteostasis with the incompletely understood molecular mechanism.
Here we show that Ubr1 is a key player in the biological stress PQC to protect the endosomal compartment proteostasis. We found that disease-causing MLC1 variants cause endo-lysosomal and intracellular Ca 2+ -stress are partly offset by E3 Ub-ligases CHIP and Ubr1 to facilitate mutants ESCRT-mediated lysosomal proteolysis. Remarkably, as an alternative triage mechanism, Ubr1-and SQSTM1/p62-dependent selective endosomal autophagy (endophagy) cleared ubiquitinated and arginylated MLC1 clients. Further evidence showed Ubr1/SQSTM1/p62-axis to be a wider QC mechanism during Ca 2+ -load involved in cargo selective dual ubiquitin/arginine-client autophagy to maintain proteostasis. Our study connects Ubr1 and ubiquitin/arginine targets to a previously unidentified form of PQC and potentially to many human proteostasis diseases.

Mutant MLC1s are targeted for ubiquitination
GlialCAM functions as a chaperone-like QC factor to MLC1 in the ER ensuring its cell surface expression 34 . Disease-causing mutations in MLC1 or GlialCAM have reduced cellular and PM turnover, and the same enlarged endosomal compartment phenotype 34 . The expression of regulatory MLC1 appears to be critical in the PM cluster. Therefore, we set out to examine characterized 34 mutations possessing gradual decline (P92S:mild, S280L:moderate, C326R:severe) in their PM turnovers and expression ( 34 , Fig. 1A 37 ) using doxycyclineinducible HeLa and astrocytic U251N cells, immunoprecipitations (IP) and live-cell surface (cs)-ELISA for the PM-endosomal kinetics ( 31,32,34 , Methods). The expressed MLC1-wt was comparable to that of the endogenous MLC1 in U251N cells (Fig. S1A). Proteasomal inhibition (MG132, 2h) unmasked increased total poly-Ub of the IPed P92S relative to the wt (Fig. S1B), suggesting that misfolding increased Ub and reduced steady-state expression. To examine specifically the post-Golgi turnover, we expressed MLC1-P92S in ts20 cells harbouring the thermolabile E1 Ub-activating enzyme. Inactivation of the E1 enzyme at 40°C (Fig. S1C) not only eliminated the MLC1-P92S ubiquitination ( Fig. S1D) but also decreased four-fold its PM turnover from a half-life (T1/2) of ~1.5h to ~4h, approaching that of the wt (Fig.1B). This suggested that ubiquitination of the ER QC escaped MLC1-P92S accelerated the PM turnover.
As support, P92S endosomal internalization was inhibited by >50% in ts20 cells than in control E36-wt cells (Fig. S1E). The MLC1-C326R had pronounced PM destabilization (T1/2 ~0.5h) and steady-state expression defect (Fig.1B) 31 . These results indicate that the significantly reduced steady-state expression of mutants is attributed to their accelerated turnover in post-Golgi compartments.
MLC1 lacks N-linked complex glycosylation, which could have been used to separate the immature ER forms from the processed post-Golgi ones. Instead, we used the association of GlialCAM with MLC1 31,34 to selectively isolate the cell surface ubiquitinated MLC1.
MLC1/GlialCAM was cell surface immunoprecipitated (cs-IP) with anti-GlialCAM antibody (Ab) bound to the cell surface. The affinity-purification followed by the denaturation and second IP for MLC1 was done to ensure examination of direct Ub-conjugation in MLC1 with anti-Ub Ab. Immunoblotting revealed increasing ubiquitination of mild P92S (~5-fold) and severe C326R (~40-fold) relatively to natively folded MLC1-wt at the cell surface (Fig. S1F).
MLC1 mutations perturb cytosolic Ca 2+ homeostasis and stimulate lysosomal exocytosis Intracellular Ca 2+ levels are potent regulators of proteostasis. MLC1 mutations cause endosomal compartment fusions without lysosomal permeabilization of protons 34 , which could relate to perturbations in the cytosolic Ca 2+ . The cytosolic Ca 2+ -transients upon activation of the store-operated Ca 2+ influx from the extracellular space were examined using Fura-2 Ca 2+sensitive dye in MLC1-wt and S280L U251N cells 39 . In comparison to MLC1-wt, cells expressing MLC1-S280L exhibited diminished Ca 2+ -entry through the PM store-operated Ca 2+ -channels in response to intracellular Ca 2+ -stores depletion using thapsigargin (Tg) ( Fig.   2A). Measurements of basal cytosolic Ca 2+ levels using Calbryte520 and fluorescence reading showed a significant ~2.5-fold increase in resting levels of cytosolic Ca 2+ in S280L cells relative to MLC1-wt cells (Fig. 2B). The chronic intracellular Ca 2+ increase implies dysregulation in the Ca 2+ homeostasis, which can cause observed proteostasis stress to endosomal compartments. As a lysosomal Ca 2+ -response readout, we monitored the lysosomal Lamp1 exocytosis 40 and hexosaminidase secretion 41 . The constitutive secretion of lysosomes in the absence of ionomycin and the inability of ionomycin (5µM) to stimulate lysosomal hexosaminidase exocytosis were consistent with the elevated cytosolic Ca 2+ level imposing proteostasis stress to MLC1-P92S and S280L cells (Fig. 2C-D).

Identification of QC E3-ligases in post-Golgi compartments
The mechanism(s) of how endosomal compartments release proteostasis stress is not understood. To identify Ub E3-ligases that dispose of non-native endosomal membrane proteins and/or protect endosomal QC during Ca 2+ -perturbations, we performed a cell-based phenotypic E3-ligase siRNA sub-library screen against E3-ligases that may be involved in PQC. The objective was to select E3-ligases that impede the accelerated PM-endosomal turnover of MLC1-S280L that caused Ca 2+ changes.
The PM turnover of MLC1-S280L was measured after one hour chase at 37°C using cs-ELISA in siRNA-transfected HeLa cells lacking endogenous MLC1. Targets that delayed the turnover more than three standard errors from the non-target siRNA (siNT) controls were considered hits (Fig. 2E). siRNA against TSG101, which is indispensable for ESCRTdependent lysosomal degradation of ubiquitinated misfolded PM proteins, was used as a positive control 6,7,9,42 .
Measuring the PM-endosomal kinetics with cs-ELISA showed that depletion of Ubr1 (siUbr1) and CHIP (siCHIP) decreased the PM turnover of six transmembrane/cytosolic MLC1 disease mutations without an additive effect with siCHIP/siUbr1 or influence on the MLC1-wt or the native-like N141K mutation in the extracellular part of MLC1 (Fig. 2F, S2B). The specificity of siUbr1 to diminish the fast PM turnover of P92S and S280L was confirmed with three different siRNA targets (Fig.2G). As further evidence toward misfolded proteins, Ubr1 depletion did not influence constitute recycling of TfR or the lysosomal targeting of the Ubdependent dependent (CD4t-Ubn, CD4cc-UbRG4) and Ub-independent (CD4t-LAMP) model transmembrane cargoes [48][49][50] (Fig. S2C-D). These results suggest that CHIP and Ubr1 have PQC specificity toward misfolded MLC1 without targeting constitutive Ub-dependent cargo sorting by the ESCRT-machinery.
CHIP typically functions with chaperones/co-chaperones that enhance recognition of non-native conformations [51][52][53] . The cell surface IP (cs-IP) of MLC1/GlialCAM and immunoblotting revealed the highest ~40-80-fold association of Hsc70, Hsp90, and Hsp40 (DNAJB1) with the severe C326R and half of that with milder P92S paralleling the severity in their PM defects. DNAJB1 type Hsp40 recruits misfolded substrates to Hsp70 as well as targets a range of human Hsp70-based disaggregase clients 54 . CHIP association increased ~2-5-fold in mutants relative to that of MLC1-wt (Fig. 3A). The cs-ELISA assay showed that depletion of CHIP attenuated ~66% of the PM turnover and ~50% of endosomal internalization of the misfolded MLC1-P92S, which was reversed by CHIP-wt overexpression (

Ubr1-dependent ubiquitination of MLC1 variants
Lack of CHIP caused only partial inhibition of the endo-lysosomal transfer of MLC1 variants (Fig. 3B). The found second E3-ligase Ubr1 has been suggested to have PQC functions outside its role as n-recognin, mostly in yeast, toward a limited number of cytosolic, misfolded ER, mitochondrial and cytosolic PQC clients [45][46][47] . Ubr1 in the human ER or endosomal PQC is not well understood. We selected MLC1-S280L with moderate expression and turnover defect ( Fig. 1A, 1E) for further QC studies.
First, to address the cellular and the ER role of Ubr1, the disappearance kinetics of MLC1-S280L was measured upon translational inhibition by cycloheximide (CHX) and immunoblotting in siUbr1 and siNT treated cells. The fast degradation rate of MLC1-S280L during the first 0.5 h chase was inhibited by ~50% in siUbr1 treated cells (Fig. 3C), suggesting that Ubr1 contributes to the ER clearance of misfolded MLC1. To address the ER effect of Ubr1, non-native MLC-wt was accumulated in the ER by preventing its ER exit with Brefeldin A (BFA) and ERAD with proteasomal inhibitor Bortezomib. The ubiquitination of non-native MLC-wt was reduced by >75% in siUbr1 treated cells (Fig. 3D). These results showed that a fraction of unassembled or incompletely folded MLC1 is targeted for Ubr1-dependent ERAD.
Next, we assessed Ubr1 capacity for Ub-conjugation at the PM-endosomes. MLC1/GlialCAM was isolated using cs-IP followed by a second IP for MLC1 after denaturation to reveal direct MLC1 ubiquitination using immunoblotting and anti-Ub Ab (Fig.   3E, as in Fig. 1C, S1F). Ubiquitination was normalized to the MLC1 amount. Ubr1-depletion decreased the ubiquitination of misfolded MLC1-S280L by ~75% at the PM, while its PM amount increased ~3-fold (Fig. 3E, lower panel) without affecting natively folded MLC1-wt at the PM.
To demonstrate the interaction of Ubr1 with MLC1 at the PM-endosomes, misfolded MLC1-S280L was expressed with Flag-Ubr1-wt or catalytically inactive (Ubr1-CI). After mild in vivo cross-linking, S280L/GlialCAM was isolated using cs-IP without denaturation. Ubr1wt reduced and Ubr1-CI elevated S280L amount at the PM, however, Ubr1 detection was low.
As a more sensitive in vivo approach, we measured the MLC1/Ubr1 interaction with the proximity biotinylation technique using MLC1-Bir* as the bait and cs-IP 59 . The endogenous Ubr1 displayed ~8-fold higher biotinylation susceptibility by the MLC1-S280L-Bir* than MLC1-wt (Fig. 3F). These data showed that the non-native astrocytic regulatory cluster membrane protein MLC1 variant represents previously unrecognized PQC clients for Ubr1 in the ER and the PM-endosomes.

Ubr1 functions as an endosomal E3 ubiquitin QC-ligase
Ubr1 subcellular location in human cells is not well known and was evaluated using indirect immunostaining and fluorescence microscopy for endogenous-Ubr1, Flag-Ubr1 and mCherry-Ubr1 without the presence of misfolded protein or Ca 2+ -stress. The majority of Ubr1 was associated with membrane-puncta and partly with the PM (Fig. S4A), phalloidin stained F-actin at endosomal invaginations, as well as the clathrin adaptor (AP2) and the clathrin light chain (CLC) (Fig. S4B) being consistent with its role at the PM-endosomal compartment.
Additionally, Ubr1 was found partly in the nuclei (Fig. S4A) and with the ER-markers KDEL and ERp57, but not with the lysosome marker Lamp1 or autophagosome marker LC3b (Fig.   S4B).
Next, the Flag-Ubr1 colocalization was assessed in the presence of endocytoses misfolded MLC1-S280L and native MLC1-wt. Extracellular MLC1 HA-epitope was selectively labelled at the PM and endocytosed 30 min for immunofluorescence or proximity ligation assay (PLA) 31,32 to visualize protein-protein interactions. Both Ubr1 and MLC1-S280L colocalized with early endosomal marker EEA1 ( Fig. 4A-B). We confirmed the preferential interaction of endosomal misfolded MLC1-S280L with the Ubr1-wt and Ubr1-CI, as well as EEA1+ using the PLA (Fig. 4C-D). The immunolocalization and proximity interaction were substantially enhanced between Ubr1-CI and S280L over the MLC1-wt at the PM proximity (Fig. S4C), reinforcing the notion that Ubr1 interacts with misfolded MLC1 specifically at PM-endosomal compartments (Fig. 3E).
Next, the significance of Ubr1 on the endosomal sorting of MLC1 was interrogated by monitoring misfolded MLC1-S280L endo-lysosomal trafficking with pHv-analysis (as in

Ubr1 is activated by proteostasis stress in endosomal compartments
The role of Ubr1 in a cytosolic and ER QC has been demonstrated upon cellular stress in yeast 46,60 . We sought evidence for Ubr1 connection to endosomal proteostasis stress in human diseases using MLC1 mutant cells with endosomal fusions 34  Inhibiting the autophagosome-lysosome fusion by bafilomycin A1 (BafA1) resulted in cytosolic dispersion of SQSTM1/p62 and the reduced colocalization of the SQSTM1/p62 with MLC1-S280L, consistent with the endosomal mutant degradation by auto-lysosomal pathway ( Fig. 5E-F). Furthermore, Western blot analysis revealed that the cellular SQSTM1/p62 and the autophagy membrane LC3b-II/I ratio were increased upon siRNA-mediated Ubr1 depletion in MLC1-S280L cells as well as in control cells ( Fig. 6A-B). This suggests that Ubr1 has a substantial role in general proteostasis QC. Consistently, the lack of Ubr1 resulted in the accumulation of pre-autophagic isolation membranes with compromised maturation to autophagosomes pinpointing its importance for selective autophagy (Fig. 5C-F, 6E).
Collectively, the data so far ( Fig.1-6D) indicated that misfolded MLC1 changes intracellular Ca 2+ homeostasis and renders proteostasis stress to the endosomal pathway culminating in the activation of Ubr1-and SQSTM1/p62-dependent cargo selective endophagy.
Arginylation can occur on exposed N-terminal residues or internally located side-chains of proteins 64 . To assess whether SQSTM1/p62 can recognize misfolded MLC1 via arginylation, arginyltransferase was inhibited by tannic acid in MLC1-S280L cells and analyzed using immunofluorescence microscopy. Tannic acid induced the SQSTM1/p62 dispersion prohibitin autophagosome formation in S280L cells similar to lack of Ubr1 ( Fig. 6F-G, S5A). Inhibition of arginylation and Ubr1 depletion prevented the formation of autophagosomes and enhanced S280L and Lamp1 accumulation at the PM ( Fig.6F-G, S5A).
The above data indicates that both arginylation and Ubr1-induced ubiquitination of misfolded MLC1 is required for the SQSTM1/p62-driven endo-phagosome formation. To test this, we monitored the arginylation and ubiquitination of the affinity-purified MLC1-wt and S280L by tandem mass spectrometry (Fig. 7C, peptides S5C). Arginylation was found D31 and R42 residues in the N-terminal tail and E168-169 residues in the 2 nd cytosolic loop of S280L, adjacent to the Ub-acceptor sites (K33, K173 and 175) forming structural proximity clusters (Fig.7D, red circles). We also observed folding-independent arginylation in the wt and A recent study showed that proteostasis stress increased the cytosolic accumulation of R-BiP 19 . Immunofluorescence microscopy confirmed that the R-BiP level was increased in the cytosol of S280L cells and R-Bip was also confined with LC3b, Lamp1 and S280L to autolysosomes ( Fig. S6A-D). Considering that the recruitment of Ubr1 by arginylation to misfolded MLC1 was limited ( Fig. 6E-F, S6A-D), we tested the effect of the N-terminal UBRbox deletion (ΔN), responsible for the recognition of N-terminal arginylation by Ubr1. The PM turnover of MLC2-S280L, measured by cs-ELISA, showed that ΔN-Ubr1 overexpression accelerated the PM turnover of S280L more than Ubr1-wt, suggesting that Ubr1 is recruited via an alternative mechanism (Fig. 7E). This is consistent with the Hsp70-dependent and Ndegron-independent PQC function of Ubr1 in the cytosol 65 .
To evaluate the involvement of molecular chaperones in MLC1-S280L recognition by Ubr1, the PM resident MLC1 was isolated with cs-IP (as in Fig.3A) or cellular pool using IP from siUbr1-depleted, and Flag-Ubr1-wt or Ubr1-CI expressing cells followed by immunoblotting of interacting chaperones. Hsp90β (83.2kDa) preferentially interacted at the PM, while both Hsp90α (90kDa) and Hsp90β were interacting with the cellular MLC1-S280L ( Fig. 7F, S6E). Lack of Ubr1 decreased Hsc70/Hsp40/Hsp90β association with the mutant in both compartments.
Ubr1/ SQSTM1/p62 clear ubiquitin/arginine PQC clients during Ca 2+ -induced stress As a finale, we asked whether Ubr1 regulates a wider biological endosomal pathway stress QC. Because of the lack of markers for the endo-lysosomal stress, we used a morphological readout of Lamp1+ lysosomes (Fig. 5) in Hela cells. Ubr1 is important for endo-phagy stress QC caused by MLC1 mutants, which also elevated cytosolic Ca 2+ . Next, we tested the Ubr1 link to Ca 2+ -stress and whether it can clear a variety of clients via cargo selective SQSTM1/p62 autophagy (Fig.8C condition). Intracellular Ca 2+ was elevated by Ca 2+ ionophore, ionomycin, known to trigger proteostasis stress via mitochondrial oxidative stress 66 , the release of lysosomal enzymes (Fig. 2D), by reducing PtdIns(4,5)P2 level critical for the PM protein stability and endocytosis 67 and imposing ERstress 68 . Ionomycin-induced (ION) enlargement of Lamp1+ lysosomes, the appearance of Lamp1/SQSTM1/p62 positive autolysosomes and arginylation signal to lysosomes (Lamp1/rArg) with SQSTM1/p62 (Fig.8D, E). Lack of Ubr1 during Ca 2+ -stress abrogated SQSTM1/p62 expression, autolysosomes and decreased significantly lysosomal diameter (Fig.8D, F). Inhibition of arginylation abrogated selective autolysosome formation and partly the lysosomal enlargement being full with the removal of Ubr1 in Ca 2+ -stress (Fig.8D, F-G).

Discussion
Here, we describe previously unrecognized actions of Ubr1 and SQSTM1/p62 in cargo selective endo-phagy/autophagy clearing specifically ubiquitinated/arginylated clients during proteostasis stress. We found Ubr1 important for attenuating proteostasis stress at the fused endosomal compartments with altered cargo sorting that were initiated either by conformationally challenged astrocyte signalling cluster MLC1 variants possessing proximity clustering of ubiquitin-arginine posttranslational modifications and/or cytoplasmic Ca 2+overload. In turn, the lack of Ubr1 alone and especially with lack of arginylation resulted in endo-lysosomal compartments stress. Importantly, we provided several lines of evidence indicating that the most crucial role of Ubr1 in autophagy is to guide ubiquitinated/arginylatedcargo to SQSTM1/p62 and ensure SQSTM1/p62 oligomerization-dependent selective autophagosome/endo-phagosome maturation. These mechanisms represent a previously unrecognized Ub E3-ligase-dependent Stress QC route targeting damaged proteins for Autophagy QC. Notably, we demonstrate that damaged integral PM proteins can be substrates to Ubr1 at the ER QC and Endo-phagy QC. Ubr1 likely supports during proteostasis stress the constitutive PM-endosome PQC pathway using CHIP and ubiquitin-dependent ESCRTmachinery described here and previously 6,7,9 . Collectively, these findings link Ubr1 stress QC to a major biological disease pathway of Ca 2+ -signalling and selective autophagy with implications to a variety of human diseases (Summary Fig. 8H and S6F).
Multiple Ca 2+ -binding proteins link Ca 2+ -changes to ubiquitination, arginylation and biogenesis of autophagosomes and lysosomes 69  Only a few Ub QC-ligases have been identified at the PM-endosomes or for proteostasis stress. CHIP relies on the Hsc70/Hsp90-chaperone-dependent mechanism, while the endosomal RFFL recognizes conformationally defective substrates independently of chaperones 6,7,9,11 . The yeast Rsp5 12 limits the PM accumulation of heat-stressed proteins in concert with arrestin-related adaptors 81 . Ubr1 has been invoked in chaperone-dependent andindependent 45,65,82 , and stress-induced cytosolic-and ER QC functions in yeast 60 . Unlike CHIP, Ubr1 has not been associated with membrane protein QC at the PM-endosomes or Stress QC in mammalians. We demonstrated that Ubr1 is a QC-ligase that has a pivotal role in the ERAD of conformationally defective MLC1s similar to a previously found role for yeast CFTR 46 and endo-phagy QC during Ca 2+ proteostasis stress in human cells.
The dynamics of molecular chaperones and proteotoxic stress may play a role for both CHIP and Ubr1 in QC. CHIP loss-of-function increases sensitivity to senescence, stress and ageing [83][84][85] . Thus, stress-activated Ubr1 can constitute an important backup QC mechanism toward certain clients. Where the Hsc70-CHIP complex recognizes substrates with exposed short hydrophobic sequences, Ubr1 as n-recognin targets destabilizing amino acids 86 . Our data and a recent report 60 suggest alternative mechanisms for the stress PQC. Upon acute stress, CHIP can sense Hsp70 deficiency and relocate to support membrane organelles QC independently of chaperones 87 . Osmotic stress activates Ubr1 independently of the UBR-box to degrade some misfolded cytosolic and ER membrane proteins in yeast 60 consistent with its n-recognin-independent function 65,82 . We found that the lack of UBR-box in Ubr1 enhances QC, Ubr1 is recruited with Hsp90-complex and is a requirement for SQSTM1/p62 oligomerization, underscoring the role of Ubr1 stress QC in human cells. Diversification of Ubr1 recognition capacity during proteostasis stress may expand the limited substrate specificity and capacity of the mammalian QC ligases.
The plethora of genetic mutations, including in risk genes for cancers and neurodegenerative diseases, underscores the importance of proper protein/membrane flow at the endosomal pathway. The identified Ubr1 QC-mechanism advances our understanding of these diseases by demonstrating the capacity of cells to alleviate proteostasis Ca 2+ -stress clients toward selective auto/endo-phagy. Our results highlight the role of ubiquitination and arginylation in the stress QC and offer novel therapeutic targets in diseases afflicting membrane proteins and organelle proteostasis.

Experimental Models
Parental or inducible Lenti-X Tet-On 6 HeLa and U251N cells with and without 2HA-MLC1 (GeneID: 23209) expression were used in experiments and cultured under standard conditions 31,32,34 . Lentivirus production, transduction and doxycycline induction 6 were done as described before 31 as well as transfection of GlialCAM-Flag 31, 34 . Transfection of plasmid DNA was performed using Lipofectamine 2000 (Thermo Fisher Scientific) and siRNA using RNAiMAX or Oligofectamine transfection reagent (Thermo Fisher Scientific). Thermolabile E1 mutant CHO cells ts20 and control E36 cells were preincubated at 40°C for 3 hr to inactivate E1 7 . Flag-Ubr1-wt in pCMV-Tag2B was a gift from Y.Yamaguchi (GeneID: 499877). C1011 was mutated to alanine to generate inactive ligase (CI). To create a mCherry vector, Ubr1-wt and Ubr1-CI were transferred to pmCherry-C1 (Clonetech).

Immunoprecipitation and protein analyses
In total cell IP, the Ab was added directly to cell lysates. Selective isolation of MLC1complex from the PM was achieved by cs-IP using anti-GlialCAM Ab (1:2000, R&D Systems).
Ab was bound on ice to live cells for 45min, after which the unbound Ab was washed off. Cells were lysed in Triton X-100 lysis buffer (1% Triton X-100, 25 mM Tris-Cl, 150 mM NaCl, pH 8.0, 10 μM MG132 containing 20 μM PR-619, 10 μg/ml pepstatin + leupeptin, 1 mM phenylmethylsulfonyl fluoride, and 5 mM N-ethylmaleimide) on ice or in co-IP assays, a milder lysis buffer was used by changing Triton X-100 to 0.4% NP-40. For detecting direct ubiquitination of MLC1, lysates were denatured using 1% SDS for 5min, after which the SDS concentration was adjusted to 0.1%. The second IP step was performed using anti-HA (Biolegend, for MLC1). Treatment with cycloheximide (100 μg/ml) was carried out in full medium at 37°C for indicated times. BFA treatment (5 μg/ml) was done in the full medium at

Post-translational modification identification by mass spectrometry
Affinity purification of MLC1 was done using anti-HA in a lysis buffer as described above and final washing twice in 50 mM NH4HCO3. Beads were resuspended to 20mM Tris-HCl (pH 8.0) and 750ng of trypsin was added for 24h, and an additional 250 ng for 3h, and incubated at 37°C. Peptides were lyophilized and formic acid was added in 2%. LC-MS/MS was done on the Orbitrap Fusion Tribrid Instrument connected to an UltiMate 3000 UHPLC liquid chromatography system (Thermo-Fisher Scientific). The acquired raw files of mass spectra were searched against the Uniprot human database and analyzed using Byonic or MaxQuant with dynamic modifications for ubiquitination (+ 114.04293 Da) and arginylation (+156.10111). The false discovery rate (FDR) was set to 1%.

Live-cell ELISA
The PM/endosomal protein expression and turnover were measured using the PM epitope labelling and cs-ELISA in live cells 6,7,32 . Endosomal internalization was measured for 5 min and turnover for indicated times at 37°C. TfR was measured using horseradish peroxidase (HRP)-conjugated transferrin (Thermo Fisher Scientific) and CD4 with anti-CD4 (BD Pharmigen). The transferrin-HRP or HRP-conjugated secondary Abs were measured either by luminescence using HRP-Substrate (SuperSignal West Pico, Thermo Fisher Scientific) or Ampilite (ATT Bioquest) or Amplex Red assay (Thermo Fisher Scientific).
Differences in cytosolic Ca 2+ -fold levels were measured using Calbryte 520 AM (5μM) (ATT Bioquest). The dye was loaded into cells for 45min at 37°C and 30min at RT after which 1mM Probenecid was added to each well before the fluorescence plate reading.

Lysosomes exocytosis and enzyme assays
The lysosomal exocytosis to the PM or lysosomal beta-hexosaminidase secretion was

Microscope imaging
The pH of endocytic vesicles containing indicated cargo molecules was measured using live-cell single vesicle fluorescence microscopy and image analysis as described previously 6,7,31,34,56,57 . Membrane protein cargo was labelled sequentially with appropriate primary Ab against extracellular epitope and with FITC-goat anti-mouse secondary Fab (Jackson Immunoresearch). Internalization was initiated at 37°C and allowed to continue for indicated times. Recycling endosomes were labelled with 5 µg/ml FITC-Tf (Jackson Immunoresearch) for 1 h. At least >250 vesicles from 25-50 cells per experiment were analyzed, and the average weighted mean was calculated for at least three or more independent experiments. The analysis was performed on an inverted fluorescence microscope Nikon TI-E equipped with Lumencor Spectra X light source and electron-multiplying charge-coupled device (Photometrics) equipped with an Evolve 512 electron-multiplying charge-coupled device (EM CCD) camera (Photometrics Technology) and a 63×/1.4 numerical aperture (NA) Plan Apochromat oilimmersion objective. The acquisition was performed at 490 ± 5-and 440 ± 10-nm excitation wavelengths using a 535 ± 25-nm emission filter and was analyzed with NIS-Elements (Nikon). and EEA1 were detected from fixed samples. Anti-mouse plus and anti-rabbit minus probes were used to crosslink the desired two epitopes and visualized them as described above.
Cells on glass coverslips were loaded with Fura2-AM (5 µM, Thermo Fisher Scientific) for 30min, washed and incubated in Krebs-Ringer-Hepes solution for 10min at 37 °C. The signal was measured using a Nikon TE300 Eclipse microscope equipped with a Sutter DG-4/OF wavelength switcher, Omega XF04 filter set for Fura-2, Photonic Science ISIS-3 intensified CCD camera and MetaFluor software. Images were obtained every 20s using a 20X objective. Ratio values (340/380 nm) were transformed to cytosolic [Ca 2+ ] using the equation derived by 89 .

Statistics
Paired or unpaired two-tailed Student's t-test was used for p-values as indicated in the figure legends. Statistical significance was set to p< 0.05. All data in curves and bar plots represent the means average of at least three or more independent experiments. Data are expressed as means ± SEM.    34 were measured using cs-ELISA (as in Fig.1B). G) Ubr1 depletion decreases the PM turnover of misfolded MLC1. Cells were depleted using three different Ubr1 siRNA target sequences. siNT served as a negative and siTsg101 as a positive control. Cs-ELISA was used to measure the PM turnover after the 1h chase (as in Fig.1B).