Eukaryotes have evolved various quality control mechanisms to promote proteostasis in the endoplasmic reticulum (ER). Selective removal of certain ER domains via autophagy (termed as ER-phagy) has emerged as a major quality control mechanism. However, the degree to which ER-phagy is employed by other branches of ER-quality control remains largely elusive. Here, we identify a cytosolic protein, C53, that is specifically recruited to autophagosomes during ER-stress, in both plant and mammalian cells. C53 interacts with ATG8 via a distinct binding epitope, featuring a shuffled ATG8 interacting motif (sAIM). C53 senses proteotoxic stress in the ER lumen by forming a tripartite receptor complex with the ER-associated ufmylation ligase UFL1 and its membrane adaptor DDRGK1. The C53/UFL1/DDRGK1 receptor complex is activated by stalled ribosomes and induces the degradation of internal or passenger proteins in the ER. Consistently, the C53 receptor complex and ufmylation mutants are highly susceptible to ER stress. Thus, C53 forms an ancient quality control pathway that bridges selective autophagy with ribosome-associated quality control in the ER.
For cells to survive they need to be able to remove faulty or damaged components. The ability to recycle faulty parts is so crucial that some of the molecular machinery responsible is the same across the plant and animal kingdoms. One of the major recycling pathways cells use is autophagy, which labels damaged proteins with molecular tags that say 'eat-me'. Proteins called receptors then recognize these tags and move the faulty component into vesicles that transport the cargo to a specialized compartment that recycles broken parts.
Cells make and fold around 40% of their proteins at a site called the endoplasmic reticulum, or ER for short. However, the process of folding and synthesizing proteins is prone to errors. For example, when a cell is under stress this can cause a ‘stall’ in production, creating a build-up of faulty, partially constructed proteins that are toxic to the cell. There are several quality control systems which help recognize and correct these errors in production. Yet, it remained unclear how autophagy and these quality control mechanisms are linked together.
Here, Stephani, Picchianti et al. screened for receptors that regulate the recycling of faulty proteins by binding to the ‘eat-me’ tags. This led to the identification of a protein called C53, which is found in both plant and animal cells. Microscopy and protein-protein interaction tests showed that C53 moves into transport vesicles when the ER is under stress and faulty proteins start to build-up. Once there, C53 interacts with two proteins embedded in the wall of the endoplasmic reticulum. These proteins form part of the quality control system that senses stalled protein production, labelling the stuck proteins with ‘eat-me’ tags. Together with C53, they identify and remove half-finished proteins before they can harm the cell.
The fact that C53 works in the same way in both plant and human cells suggests that many species might use this receptor to recycle stalled proteins. This has implications for a wide range of research areas, from agriculture to human health. A better understanding of C53 could be beneficial for developing stress-resilient crops. It could also aid research into human diseases, such as cancer and viral infections, that have been linked to C53 and its associated proteins.
Autophagy is an intracellular degradation process where eukaryotic cells remove harmful or unwanted cytoplasmic contents to maintain cellular homeostasis (Dikic and Elazar, 2018; Klionsky and Deretic, 2011; Marshall and Vierstra, 2018). Recent studies have shown that autophagy is highly selective (Johansen and Lamark, 2020; Stolz et al., 2014) and is mediated by receptors that recruit specific cargo, such as damaged organelles or protein aggregates. Autophagy receptors and their cargo are incorporated into the growing phagophore through interaction with ATG8, a ubiquitin-like protein that is conjugated to the phagophore upon activation of autophagy (Stolz et al., 2014; Zaffagnini and Martens, 2016). The phagophore grows and eventually forms a double-membrane vesicle termed the autophagosome. Autophagosomes then carry the autophagic cargo to lytic compartments for degradation and recycling. Selective autophagy receptors interact with ATG8 via conserved motifs called the ATG8 interacting motif (AIM) or LC3-interacting region (LIR) (Birgisdottir et al., 2013). In contrast to mammals and yeast, cargo receptors that mediate organelle recycling remains mostly elusive in plants (Stephani and Dagdas, 2020).
The endoplasmic reticulum (ER) is a highly dynamic heterogeneous cellular network that mediates folding and maturation of ~40% of the proteome (Walter and Ron, 2011; Sun and Brodsky, 2019). Proteins that pass through the ER include all secreted and plasma membrane proteins and majority of the organellar proteins. This implies, ER could handle up to a million client proteins in a cell every minute (Karagöz et al., 2019). Unfortunately, the folding process is inherently error prone and misfolded proteins are toxic to the cell (Sun and Brodsky, 2019; Karagöz et al., 2019; Fregno and Molinari, 2019). To maintain the proteostasis in the ER, eukaryotes have evolved dedicated quality control mechanisms that closely monitor, and if necessary, trigger the removal of terminally misfolded proteins. Degradation of the faulty proteins is mediated by proteasomal and vacuolar degradation pathways (Fregno and Molinari, 2019).
One of the main vacuolar/lysosomal degradation processes is ER-phagy. It has emerged as a major quality control pathway, and defects in ER-phagy is linked to various diseases (Chino and Mizushima, 2020; Hübner and Dikic, 2020; Stolz and Grumati, 2019; Wilkinson, 2020). ER-phagy involves cargo receptors that mediate removal of certain regions of the ER via autophagy. Several ER-resident ER-phagy receptors have been identified. These include Fam134B, RTN3L, ATL3, Sec62, CCPG1, and TEX264 in mammals and ATG39 and ATG40 in yeast (Khaminets et al., 2015; Grumati et al., 2017; Chen et al., 2019; Fumagalli et al., 2016; Smith et al., 2018; An et al., 2019; Chino et al., 2019; Mochida et al., 2015). A recent study showed reticulon proteins could also function as ER-phagy receptors in plants (Zhang et al., 2020). In addition, CALCOCO1 and Epr1 have been recently identified as cytosolic ER-phagy receptors that associate with ER-resident VAP proteins to recycle ER tubules (Nthiga et al., 2020; Zhao et al., 2020). Altogether, these receptors are activated during starvation or stress conditions and work together to remodel the highly heterogeneous and dynamic ER network to maintain proteostasis. Despite the emerging links, how ER-phagy cross-talks with the core ER quality control pathways remains largely unknown (Chino and Mizushima, 2020; Dikic, 2018).
Here, using a peptide-competition coupled affinity proteomics screen, we identified a highly conserved cytosolic protein, C53, that is specifically recruited into autophagosomes during ER stress. C53 interacts with plant and mammalian ATG8 isoforms via a non-canonical ATG8 interacting motif (AIM), termed shuffled AIM (sAIM). C53 is recruited to the ER by forming a ternary receptor complex with the UFL1, the E3 ligase that mediates ufmylation, and its ER membrane adaptor DDRGK1 (Gerakis et al., 2019). C53-mediated autophagy is activated upon ribosome stalling during co-translational protein translocation and results in the degradation of specific ER proteins.
To identify specific cargo receptors that mediate selective removal of ER compartments during proteotoxic stress, we performed an immunoprecipitation coupled to mass spectrometry (IP-MS) screen to identify AIM-dependent ATG8 interactions triggered by ER stress. We hypothesized that a synthetic AIM peptide that has higher affinity for ATG8 can outcompete, and thus reveal, AIM-dependent ATG8 interactors. To identify this synthetic peptide, we performed a peptide array analysis that revealed the AIM wt peptide (Figure 1—figure supplement 1A,B; Supplementary file 1). Using isothermal titration calorimetry (ITC), we showed that the AIM wt binds ATG8 with nanomolar affinity (KD = ~ 700 nM), in contrast to the AIM mutant peptide (AIM mut), which does not show any binding (Figure 1—figure supplement 1C–D) or the low micromolar-range affinities measured for most cargo receptors (Zaffagnini and Martens, 2016). As plants have an expanded set of ATG8 proteins, we first tested if any of the ATG8 isoforms specifically responded to ER stress induced by tunicamycin (Kellner et al., 2017). Tunicamycin inhibits glycosylation and leads to proteotoxic stress at the ER (Bernales et al., 2006). Quantification of ATG8 puncta in transgenic seedlings expressing GFP-ATG8A-I revealed that tunicamycin treatment significantly induced all nine ATG8 isoforms (Figure 1—figure supplement 2). Since all ATG8 isoforms were induced and ATG8E has a broad expression pattern, we chose ATG8E, and performed peptide competition coupled IP-MS analysis (See Materials and methods for detailed description). In addition to well-known AIM dependent ATG8 interactors such as ATG4 (Autophagy related gene 4) and NBR1 (Neighbour of BRCA1) (Wild et al., 2014), our analyses revealed that the highly conserved cytosolic protein C53 (aliases: CDK5RAP3, LZAP, IC53, HSF-27) is an AIM-dependent ATG8 interactor (Figure 1A, Supplementary file 2, Figure 1—figure supplement 3).
To confirm our IP-MS results, we performed in vitro pull-down experiments. Arabidopsis thaliana (At) C53 specifically interacted with GST-ATG8A, and this interaction was outcompeted with the AIM wt, but not AIM mut peptide. Consistently, ATG8 receptor accommodating site mutations (LDS − LIR Docking Site) prevented C53 binding (Figure 1B). We extended our analysis to all Arabidopsis ATG8 isoforms and showed that AtC53 interacts with eight of nine Arabidopsis isoforms (Figure 1C). To probe for evolutionary conservation of C53-ATG8 interaction, we tested the orthologous proteins from the basal land plant Marchantia polymorpha (Mp) and showed that MpC53 interacts with one of two Marchantia ATG8 isoforms (Figure 1—figure supplement 4). As C53 is highly conserved in multicellular eukaryotes and has not been characterized as an ATG8 interactor in mammals, we tested whether human C53 (HsC53) interacts with human ATG8 isoforms (LC3A-C, GABARAP, -L1, -L2). HsC53 interacted with GABARAP and GABARAPL1 in an AIM-dependent manner via the LIR docking site, similar to plant C53 homologs (Figure 1D,E). Of note, we have also tested other modes of ATG8 binding such as the recently identified UDS or the hydrophobic pocket accommodating the atypical LIR motif found in ufmylation enzyme UBA5 (Marshall et al., 2019; Huber et al., 2019). The UDS mutation rendered ATG8A unstable (Figure 1B), whereas mutating the atypical LIR accommodating site did not affect C53 binding (Figure 1F). Altogether, these data suggest that C53-ATG8 interaction is conserved across kingdoms and mediated via the LIR Docking Site.
In order to examine the in vivo link between C53 and ATG8 function, we generated transgenic AtC53-mCherry Arabidopsis lines and measured autophagic flux during ER stress. Without stress, AtC53 displayed a diffuse pattern in the cell, partially overlapping with the ER marker GFP-HDEL (Figure 2—figure supplement 1A). Similarly, upon carbon starvation (-C, Figure 2A), which is commonly used to activate bulk autophagy, AtC53-mCherry remained mostly diffuse (Marshall and Vierstra, 2018). However, tunicamycin (Tm) treatment led to a rapid increase in AtC53 puncta as observed in both native promoter driven and ubiquitin 10 promoter driven transgenic lines. The C53 puncta did not colocalize with HDEL-GFP puncta formed during ER stress, suggesting C53 puncta are highly specific (Figure 2—figure supplement 1A,B). The number of puncta was further increased upon concanamycin A (ConA) treatment that inhibits vacuolar degradation, suggesting that AtC53 puncta are destined for vacuoles (Figure 2A). The AtC53 puncta disappeared when AtC53-mCherry lines were crossed into core autophagy mutants atg5 and atg2, confirming that formation of AtC53 puncta is dependent on macroautophagy (Figure 2A). Consistent with this, other ER-stressors such as phosphate starvation, cyclopiazonic acid (CPA), and dithiothreitol (DTT) treatments also induced AtC53 puncta (Figure 2—figure supplement 1C; Fumagalli et al., 2016; Smith et al., 2018; Naumann et al., 2019). The AtC53 puncta co-localized with GFP-ATG8A and GFP-ATG11, indicating that they are autophagosomes (Figure 2B, Figure 2—figure supplement 2A). Moreover, as recently shown for other selective autophagy receptors, AtC53 and HsC53 directly interacted with the mammalian ATG11 homolog FIP200 (PTK2/FAK family-interacting protein of 200 kDa) (Figure 2—figure supplement 2B; Lahiri and Klionsky, 2018; Turco et al., 2019; Ravenhill et al., 2019; Vargas et al., 2019). Ultrastructural analysis using immunogold labelling showed that C53 is associated with ER under non-stress conditions, consistent with previous findings showing C53 associates with ER membrane proteins (Yang et al., 2019). Electron micrographs also showed that AtC53 is recruited to autophagosomes during ER stress, consistent with our live cell imaging results (Figure 2C, Figure 2—figure supplement 3). Similar to plant proteins, transfected HsC53-GFP co-localized with mCherry-GABARAP upon tunicamycin treatment in HeLa cells. The number of HsC53 puncta increased upon bafilomycin (BAF) treatment, which inhibits lysosomal degradation; suggesting that HsC53 puncta eventually fuse with lysosomes (Figure 2D). To support our imaging-based autophagic flux assays, we also performed western blot based autophagic flux analyses, using antibodies raised against AtC53 and HsC53. These autophagic flux assays further demonstrated ER-stress-specific autophagic degradation of AtC53 and HsC53 (Figure 3).
Having validated C53 as an autophagy substrate, we next sought to identify its ATG8-interacting motif (AIM). For this purpose, we reconstituted the binary complex in vitro and determined the stoichiometry of the C53-ATG8 interaction by native mass spectrometry (nMS). Both HsC53 and AtC53 formed 1:1 and 1:2 complexes with GABARAP and ATG8A, respectively; pointing to the existence of multiple binding epitopes (Figure 4A). Initially, we tested all predicted canonical AIMs in AtC53. However, even the pentuple AIM mutant bound at similar levels to ATG8, suggesting non-canonical AIMs mediate the C53-ATG8 interaction (Figure 4—figure supplement 1). To narrow down the ATG8-binding region of C53, we performed in vitro pull downs using truncated proteins. C53 contains an intrinsically disordered region (IDR) that bridges two α-helical domains located at the N and C termini. In vitro pull downs revealed that the IDR is necessary and sufficient to mediate ATG8 binding, as also confirmed with ITC and nMS experiments (Figure 4B–D, Figure 4—figure supplement 2). Multiple sequence alignment of the C53-IDR uncovered three highly conserved sites with the consensus sequence ‘IDWG’, representing a shuffled version of the canonical AIM sequence (W/F/Y-X-X-L/I/V) (Figure 4C, Figure 4—figure supplement 3). Mutational analysis of the three shuffled AIM sites in HsC53 and AtC53 revealed the importance of the sAIM epitopes for binding to GABARAP and ATG8, respectively; though in AtC53, an additional canonical AIM had to be mutated to fully abrogate the binding (Figure 5A). ITC experiments with the purified IDRs, as well as nMS and surface plasmon resonance (SPR) experiments with full-length proteins, also supported sAIM-mediated ATG8-binding for both HsC53 and AtC53 (Figure 5B,C, Figure 5—figure supplement 1). Circular dichroism spectroscopy showed that sAIM mutants had very similar secondary structures to the wild-type proteins, suggesting that lack of ATG8 binding is not due to misfolding (Figure 5—figure supplement 1C). To verify our in vitro results in vivo, we analyzed the subcellular distribution of sAIM mutants in transgenic Arabidopsis lines and transfected HeLa cells. Confocal microscopy analyses showed that C53sAIM mutants were not recruited into autophagosomes and had diffuse localization patterns upon ER stress induction (Figure 5B,C). Altogether these biochemical and cell biological analyses show that C53 is recruited to the autophagosomes by interacting with ATG8 via the non-canonical sAIMs.
Next, we looked for client proteins subject to C53-mediated autophagy. Quantitative proteomics analyses of wild type and AtC53 mutant lines revealed that AtC53 mediates degradation of ER resident proteins as well as proteins passaging the ER to the cell wall, apoplast, and lipid droplets (Figure 6, Supplementary file 3, 4). These data are consistent with a recent study, showing that ER-resident proteins accumulate in a conditional mutant of mouse C53 (Yang et al., 2019). Since C53 is a cytosolic protein, we then explored how it senses proteotoxic stress in the ER lumen, considering four likely scenarios: C53 may collaborate with (i) a sensor of the unfolded protein response (UPR) (Karagöz et al., 2019) or (ii) a component of the ER-associated degradation pathway (ERAD) (Sun and Brodsky, 2019). Alternatively, it may sense clogged translocons caused by (iii) ribosome stalling triggered during co-translational protein translocation (Wang et al., 2020) or (iv) aberrant signal recognition particle (SRP) independent post-translational protein translocation events (Ast et al., 2016; Figure 7A). In plants, there are two major UPR branches: the Ire1 pathway and bZIP17/28 pathway (Pastor-Cantizano et al., 2020). To test the connection with the UPR system, we performed autophagic flux assays. AtC53 flux was already higher than wild type in Arabidopsis UPR sensor mutants ire1a/b and bzip17/28, consistent with elevated ER stress levels in these mutants (Figure 7B,C; Koizumi et al., 2001; Kim et al., 2018). Furthermore, C53 puncta induced by tunicamycin treatment did not colocalize with Ire1b-YFP oligomers (Figure 7D). Finally, inhibition of Ire1 activity in HeLa cells using chemical inhibitors 4μ8c or KIRA6 also increased HsC53 puncta (Figure 7E). Together these data indicate that recruitment of C53 to the autophagosomes does not depend on UPR sensors (Maly and Papa, 2014). Next, we performed colocalization analyses using model ERAD substrates. In transgenic plant lines expressing model ERAD substrates, the client proteins did not colocalize with AtC53 puncta (Figure 7—figure supplement 1A; Shin et al., 2018). Likewise, the model mammalian ERAD substrates GFP-CFTRΔF508 (ERAD-C), A1ATNHK-GFP (ERAD-L), and INSIG1-GFP (ERAD-M) only partially colocalized with HsC53 puncta in HeLa cells (Figure 7—figure supplement 1B), suggesting C53-mediated autophagy may cross-talk with the ERAD pathway (Leto et al., 2019).
Next, we tested the effect of clogged translocons on C53 function. Remarkably, HsC53 significantly colocalized with the ER-targeted poly-lysine construct ER-K20 that leads to ribosome stalling (Wang et al., 2020), but not with an SRP-independent translocon clogger (Ast et al., 2016), despite both leading to a blockage at the Sec61 translocon (Figure 7—figure supplement 2A). To further corroborate these findings, we tested a suite of translation inhibitors that block different steps in translation. Consistent with C53 responding to ribosome stalling, elongation inhibitors such as Anisomycin, Emetine or Puromycin induced AtC53 puncta, whereas initiation inhibitors Harringtonine or Hygromycin B did not have any effect. All inhibitors triggered mCherry-ATG8A puncta formation, suggesting the effect caused by elongation inhibitors is specific to C53 (Figure 7—figure supplement 2B). HsC53 puncta were also induced by anisomycin treatment (Figure 7—figure supplement 2C). Consistently, silencing of HsC53 using shRNA significantly reduced lysosomal trafficking of ER-K20 (Figure 7—figure supplement 2D; Wamsley et al., 2017). These data suggest that C53 is activated upon ribosome stalling during co-translational protein translocation and mediates autophagic degradation of the stalled nascent chain.
How is C53 recruited to the ER during ribosome stalling? Notably, C53 has been previously linked to UFL1, an E3 ligase that mediates ufmylation of stalled, ER-bound ribosomes, modifying ribosomal protein RPL26 (Wang et al., 2020; Walczak et al., 2019). To test if C53 is a part of a higher order receptor complex, we analysed the interaction of C53 with UFL1 and its ER membrane adaptor DDRGK1 (Gerakis et al., 2019). We were able to observe both DDRGK1 and HsC53 in a single UFL1 pull down experiment (Figure 8—figure supplement 1A). Further in vitro pull-down assays and yeast two hybrid analyses with the plant proteins showed that AtUFL1 directly interacts with AtC53 and AtDDRGK1 (Figure 8A, Figure 8—figure supplement 1B). Consistently, AtC53 associates with DDRGK1 and UFL1 in in vivo coimmunoprecipitations and affinity purification mass spectrometry experiments (Figure 8B, Supplementary file 5). Furthermore, co-localization of UFL1 and DDRGK1 with AtC53 in punctate structures increases upon ER stress and these puncta are delivered to the vacuole (Figure 8C,D, Figure 8—figure supplement 1C,D). Strikingly, AtC53 autophagic flux requires functional UFL1 and DDRGK1, as the number of AtC53 puncta was significantly lower in ufl1 and ddrgk1 mutants (Figure 8E, Figure 8—figure supplement 1E). Ultimately, autophagic flux assays using the ufmylation machinery mutants confirmed that AtC53 flux requires a functional ufmylation machinery (Figure 8E, Figure 8—figure supplement 1F,G). Taken together, our data indicate that C53 is recruited to the ER by forming a heteromeric receptor complex with UFL1 and DDRGK1.
Since, DDRGK1 is an ER-membrane protein and physically linked to C53, we analyzed the degradation of DDRGK1 during ER stress. Transgenic lines expressing DDRGK1-GFP in c53 and atg5 mutant revealed that recruitment of DDRGK1 from ER membrane to punctate structures during ER stress required both C53 and ATG5 (Figure 8—figure supplement 2A). Furthermore, DDRGK1 puncta colocalized with mCherry-ATG8A in a C53-dependent manner (Figure 8—figure supplement 2B). Western-blot-based autophagic flux assays further confirmed AtC53-dependent degradation of DDRGK1 (Figure 8—figure supplement 2C). Interestingly, abundant ER proteins such as the Calnexin, BIP or SMT1 are not degraded by AtC53-dependent ER-phagy. Likewise, small and large ribosomal subunits are not degraded by AtC53 (Figure 8—figure supplement 2C). These results are consistent with the C53 cargo clientele defined by quantitative proteomics, and point toward a highly selective, yet unknown cargo selection mechanism of C53.
We then explored how C53 is kept inactive under normal conditions. We hypothesized that the Ubiquitin like modifier UFM1 may safeguard the C53 receptor complex under normal conditions and keep ATG8 at bay. Upon ER stress, UFM1 would be transferred to RPL26, exposing sAIMs on C53. To test this, we first analyzed the UFM1-C53 interaction by in vitro pull-down assays and could show that AtC53 can interact with AtUFM1 (Figure 9A). Furthermore, in vitro competition experiments revealed a competition between UFM1 and ATG8 for C53 binding (Figure 9A). This result is reminiscent of the mutually exclusive UFM1 and GABARAP binding of UBA5, the E1 enzyme in the ufmylation cascade (Huber et al., 2019). We then performed in vivo co-immunoprecipitation experiments during ER stress. Consistent with our hypothesis and in vitro data, ER stress led to depletion of UFM1 and enhanced AtC53-ATG8 interaction (Figure 9B,C, supplement 1). Altogether, these data suggest that the two ubiquitin-like proteins UFM1 and ATG8 compete with each other for association with the C53 receptor complex (Figure 9D).
Finally, we examined if C53 is physiologically important for ER stress tolerance. First, we tested if C53 plays a general role in autophagy using carbon and nitrogen starvation assays. Carbon and nitrogen starvation are typically used to characterize defects in bulk autophagy responses (Marshall and Vierstra, 2018). In contrast to the core autophagy mutants atg5 and atg2, CRISPR-generated Atc53 mutants did not show any phenotype under carbon or nitrogen starvation conditions (Figure 10A,B). However, consistent with increased flux, Atc53 mutants were highly sensitive to phosphate starvation, which has been shown to trigger an ER stress response (Naumann et al., 2019; Figure 10C, Figure 10—figure supplement 1A). Similarly, in both root length and survival assays, Atc53 mutants were sensitive to tunicamycin treatment (Figure 10D, Figure 10—figure supplement 1B,C). In addition, ufmylation machinery mutants (Figure 10E), including ufl1 and ddrgk1, were also sensitive to tunicamycin treatment but insensitive to carbon and nitrogen starvation (Figure 10F, Figure 10—figure supplement 1D,E). Lastly, the Marchantia polymorpha c53 mutant was also sensitive to tunicamycin, suggesting C53 function is conserved across the plant kingdom (Figure 10—figure supplement 1F). We then performed complementation assays using wild-type AtC53 and the AtC53sAIM mutant. AtC53 expressing lines behaved like wild-type plants in tunicamycin supplemented plates (Figure 10G). However, AtC53sAIM mutant did not complement the tunicamycin sensitivity phenotype, and had significantly shorter roots (Figure 10G, Figure 10—figure supplement 1G). Parallel to analyzing C53-mediated ER homeostasis in plants, stress tolerance assays in HeLa cells showed that silencing of HsC53 led to an induction of Bip3 chaperone protein levels (Figure 10—figure supplement 1H), indicating increased ER stress. Complementation of Hsc53 silenced lines with HsC53-GFP dampened Bip3 expression (Figure 10—figure supplement 1H). Altogether, these results demonstrate that C53 coordinated ER-phagy is crucial for ER stress tolerance in plant and mammalian cells.
The endoplasmic reticulum is a highly heterogeneous and dynamic network that handles folding and maturation of up to a million proteins per minute in a cell (Karagöz et al., 2019). It constantly tailors the proteome in a cell-type and physiological state dependent manner. Unfortunately, protein synthesis, folding, and maturation events are all error prone, and faulty proteins have to be efficiently recycled to prevent accumulation of toxic by-products. Since, selective autophagy is a highly efficient quality control pathway that could very quickly recycle large amounts of proteins and membranous compartments, it is not surprising to have various ER-phagy receptors focusing on re-shaping the ER during stress (Chino and Mizushima, 2020; Wilkinson, 2020). With C53, eight ER-phagy receptors have now been identified in metazoans, working together to maintain ER homeostasis under changing cellular conditions. However, since most of ER-phagy pathways were studied during nutrient starvation, which supposedly activates bulk recycling mechanisms, selective cargo recruitment triggered upon quality control defects is still poorly understood. It is thus a major challenge to elucidate the coordination of different ER-phagy receptors and their cross-talk with the core ER-quality control pathways (Chino and Mizushima, 2020).
Our findings reveal C53-mediated ER-phagy to be a central mechanism operating at the interface of key quality control pathways, controlling ER homeostasis across different kingdoms of life. Using various model systems including divergent model plant species and human cell lines, we show that C53 forms an ancient autophagy receptor complex that is closely connected to the ER quality control system via the ufmylation pathway. Unlike other ER-phagy receptors studied so far, C53 seem to be highly specific in resolving ribosome stalling triggered during SRP-dependent co-translational protein translocation. However, it remains to be shown how C53 recruit specific cargo into the autophagosomes.
Interestingly, recent genome wide CRISPR screens identified ufmylation as a major regulator of ER-phagy, the ERAD pathway, and viral infection (Leto et al., 2019; Liang et al., 2020; Kulsuptrakul et al., 2019). Using fluorescent reporter lines and genome wide CRISPRi screens, Liang et al., showed that ufmylation plays a major role in regulating starvation induced ER-phagy. They showed that both DDRGK1 and UFL1 are critical for starvation-induced ER-phagy, whereas C53 mutants did not show any ER-phagy defects (Liang et al., 2020). Our results using stable transgenic organisms show that C53-mediated autophagy is not activated by carbon or nitrogen starvation that are typically used to activate bulk autophagy (Figure 2 and Figure 2—figure supplement 1). C53 is activated by ER stress caused by phosphate depletion (Naumann et al., 2019). Consistently, phenotyping experiments revealed that C53 and the ufmylation machinery mutants are asymptomatic during carbon or nitrogen starvation but are highly sensitive to ER-stress treatments (Figure 10, Figure 10—figure supplement 1). Together, the two complementary studies indicate that the ufmylation machinery is tightly associated with ER-phagy in multicellular eukaryotes and plays a crucial role in ER stress tolerance.
It should be noted that C53 and ufmylation proteins are essential for mammalian development (Gerakis et al., 2019). Defects in C53 receptor complex have been associated with various diseases including liver cancer, pancreatitis, and cardiomyopathy (Gerakis et al., 2019). Our results suggest C53 and ufmylation is also critical for stress tolerance in plants, but they are not essential for development; suggesting plants have evolved compensatory mechanisms during adaptation to sessile life. Future comparative studies could reveal these mechanisms and help us develop sustainable strategies for promoting ER proteostasis during stress in mammals and plants.
Constructs for Arabidopsis thaliana and E. coli transformation were generated using the GreenGate (GG) cloning method (Lampropoulos et al., 2013). Plasmids used are listed in materials section. The coding sequence of genes of interest were either ordered from Twist Biosciences or Genewiz or amplified from Col-0 or HeLa cDNA using the primers listed in the materials section. The internal BsaI sites were mutated by site-directed-mutagenesis without affecting the amino acid sequence.
For Marchantia polymorpha Gateway Cloning (Ishizaki et al., 2015) was used to generate all constructs.
For HeLa expression experiments, plasmids used are listed in the materials section. The constructs were made by conventional restriction enzyme-based cloning.
The CRISPR/Cas9 constructs for mutating c53, DDRGK1 and UFM1 in Arabidopsis thaliana were prepared according to the protocol described by Xing et al., 2014 and Ma et al., 2015. The pHEE401E and pCBC-DT1T2 vectors for expressing two sgRNAs were provided by Youssef Belkhadir and Jixiang Kong, GMI Vienna. sgRNA target sites were chosen using the website http://crispr.hzau.edu.cn/CRISPR2/. Each gene was targeted by two sgRNAs to remove a fragment of the gene. The CRISPR cassettes of each gene were generated by PCR amplification using pCBC‐DT1T2 as template with the primer pairs BsF/F0 and BsR/R0, using adaptors containing the BsaI-restriction sites, respectively (see materials section). The PCR products were digested with BsaI, ligated into the pHEE401E plasmid, and transformed into DH5α E. coli. Floral dipping proceeded as described previously (Clough and Bent, 1998). Genotyping primers P1 5′‐xxx‐3′ and P2 5′‐xxx‐3′ flanking each target site were used to select T1 plants that carried deletions. Sanger sequencing was performed to define the deletion. Through backcrossing with Col-0 plants and genotyping, Cas9-free plants were achieved.
In Marchantia polymorpha, CRISPR/Cas9 constructs were generated by selecting two target sequences in c53 and ire1. Synthetic oligonucleotides were annealed and inserted at the BsaI site of the entry vector pMpGE_En03 flanked by attL1 and attL2 sequences (Sugano et al., 2018). The resultant cassettes were inserted to the destination vector pMpGE011 by LR Clonase II Enzyme Mix. The vectors were introduced into thalli of TAK1 via A. tumefaciens GV3101+pSoup, and the transformants were selected with 0.5 µM chlorsulfuron (Kubota et al., 2013). Genomic DNA from transformants was amplified by PCR and sent for sequencing to verify mutations.
All Arabidopsis thaliana lines used originate from the Columbia (Col-0) ecotype background. Mutant lines used in this study are listed in the materials section. All transgenic plants were generated by the floral dipping method (Clough and Bent, 1998) for which the plasmid constructs were prepared using the green gate cloning method (Lampropoulos et al., 2013).
Seeds were then spread on plates or liquid culture with half-strength MS media (Murashige and Skoog salt + Gamborg B5 vitamin mixture) with 1% sucrose, 0.5 g/L MES and 1% plant agar. pH was adjusted to 5.7 with NaOH. Seeds were imbibed for 4 days at 4°C in darkness. Plants were grown at 21°C at 60% humidity under LEDs with 50 µM/m2s and 12 hr:12 hr photoperiod.
For the autophagy flux assay, TMT and in vivo immunoprecipitation, seedlings were grown in liquid culture under continuous light.
Male Marchantia polymorpha accession Takaragaike-1 (Tak-1) was maintained asexually and cultured through gemma using half-strength Gamborg’s B5 medium containing 1% agar under 50–60 mmol photons m−2s−1 continuous white light at 22° C unless otherwise defined (Kubota et al., 2013).
Seedlings were grown for 9 days on media supplemented with the indicated drug concentration. Plates were scanned on day 0 and then quantified daily starting from day 2 to day 9. Large-scale root-length quantification was conducted using the automated plant imaging analysis software BRAT (Buschlab Root Analysis Toolchain) (Slovak et al., 2014) with the inhouse high-performance computer cluster MENDEL. Before analysis, collected data was passed through software quality control.
Carbon starvation: Seedlings were grown on half-strength MS media with 1% sucrose for 7 days. They were then transferred to media without sucrose, followed by wrapping the plates in aluminium foil and placing them under the same growth conditions as before for 9 days.
Nitrogen starvation: Seedlings grew on half-strength MS media with 0.5% sucrose for 7 days. They were then transferred to media without nitrogen and put under the same growth conditions as before for 14 days.
Seedlings were arranged in a similar fashion to Jia et al., 2019.
Phosphate starvation: The method was previously described by Naumann et al., 2019. Seeds were surface-sterilized and germinated 5 days on +Pi medium prior to transfer to 1% (w/v) Phyto-Agar (Duchefa) containing 2.5 mM KH2PO4, pH 5.6 (high Pi or +Pi medium) or no Pi supplement (low Pi or –Pi medium), 5 mM KNO3, 0.025 mM Fe-EDTA, 2 mM MgSO4, 2 mM Ca(NO3)2, 2.5 mM MES-KOH, 0.07 mM H3BO3, 0.014 mM MnCl2, 0.01 mM NaCl, 0.5 µM CuSO4, 1 µM ZnSO4, 0.2 µM Na2MoO4, 0.01 µM CoCl2, 5 g/L sucrose. The agar was routinely purified by repeated washing in deionized water and underwent subsequent dialysis using DOWEX G-55 anion exchanger (Ticconi et al., 2009). ICP-MS analysis of the treated agar (7.3 µg/g Fe and 5.9 µg/g P) indicated a contribution of 1.25 µM Fe and 1.875 µM P to the solid 1% agar medium. Images were analyzed using ImageJ software.
Seedlings were grown on 9 cm round plates supplemented with the indicated drug at the indicated concentration. Seedling survival was quantified after 14 days. Differentiation between live and dead seedlings was carried out similar to Yang et al., 2016. Surviving seedlings were defined as seedlings which had two green cotyledons and two green true leaves. Plants with yellow leaves or cotyledons were defined as dead.
Tunicamycin sensitivity: 14 days old plants were transformed to half-strength Gamborg’s B5 medium containing indicated concentration of tunicamycin and grown in continues light at 22°C to determine survival rates.
20–30 seedlings for western blot or 0.5–1 g seedlings for immunoprecipitation and mass spectrometry were grown in liquid culture for 5 days under continuous light with shaking at 80 rpm. Media was supplemented with different drugs (3 µM Torin, 10 µg/ml Tunicamycin or other drugs dissolved in DMSO) as indicated. 1 µM of concanamycin was added, if indicated in figures, to track the contribution of vacuolar degradation. For nutrient starvation, seedlings were transferred to phosphate, nitrogen- or sucrose-depleted media (–C, –P, -N). The plants were kept in the dark to reduce sucrose production by photosynthesis or to provide drug stability. Pure DMSO was added to control samples. For analyzing total protein degradation such as TMT, seedlings were flash frozen in liquid nitrogen after 24 hr treatment. For interaction analysis such as Co-immunoprecipitation, seedling treatment was stopped after 8 hr of treatment.
Samples were homogenized in a bead mill (RetschMM300, Haan, Germany; 30 Hz, 90 s) at 4°C with zirconium oxide grinding beads or ground by mortar and pestle for bigger sample volumes. For western blotting, SDS loading buffer was added and the sample boiled at 95°C for 10 min. Lysates were cleared by centrifugation at 16,000 g for 10 min and protein concentration was normalized by Amidoblack staining (Sigma). Western blotting was performed following standard protocols as described below. 5 µg of lysate was loaded per lane.
HeLa-Kyoto and HEK293T cells maintained in Dulbecco’s modified Eagle’s Medium (DMEM) with 10% FBS, 1% L-Glutamine and 1% Penicillin/Streptomycin. Transfection was performed with GeneJuice transfection reagent according to manufacturer’s instructions. 100 µl of empty media was mixed with 3 µL of GeneJuice and after 5 min of incubation a total of 1 µg of DNA mixture per transfection was added. After 20 min of incubation, transfection mixture was added dropwise to the cells. Cells were incubated with DNA for 24 hr. DNA containing media was removed and replaced with media. Both cell lines were authenticated using STR profiling and repeatedly tested negative for mycoplasma contamination. Testing and authentication were performed using the in house core facilities.
Lentiviral transduced shRNA-mediated knockdown of c53 in HeLa cells:
The knockdown was performed in S2 conditions. HEK293T cells were seeded 24 hr prior to transfection in DMEM without antibiotics. At 50–60% confluency, cells were transfected with 1 µg shRNA, 750 ng psPAX2 and 250 ng pMD2.G utilizing 6 µL of GeneJuice in 250 µL of empty DMEM. After 48 hr of incubation, the virus containing media was harvested and mixed 1:1 with full media. This mixture was applied to HeLa cells that were seeded 24 hr prior. Polybrene was added to a final concentration of 4 µg/ml.
After 24 hr of incubation, the medium on target cells was exchanged with full media. After 24 hr, selection with 2 µg/ml Puromycin was started. No living cells were observed in a control plate after 24 hr. After splitting cells in S2 conditions, cells were transferred into S1 conditions.
Cells were seeded 24 hr prior to treatment. At 50–60% confluency treatments were started by replacing media containing the indicated drugs or full media (untreated). Tunicamycin was added with a final concentration of 2.5 µg/ml and Torin with a final concentration of 3 µM. The treatments were stopped after 16 hr by removing the media and washing the cells with 1xPBS. A 2 hr recovery period was started by adding either media containing 100 nM Bafilomycin A1 or full media. Cells were put on ice and lysed with 100 µL of Lysis buffer (50 mM HEPES, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 25 mM NaF, 10 µM ZnCl2, 1% Triton X-100% and 10% Glycerol) per well. After centrifugation, supernatant was mixed 1:1 with 2x Laemmli Buffer and denatured by heating to 95°C for 5 min.
Each sample was loaded onto a 4–20% SDS-PAGE gradient gel (BioRad) and electrophoresis was run at 100V for 1.5 hr.
SDS-PAGE was performed using gradient 4–20% Mini-PROTEAN TGX Precast Protein Gels (BioRad). Blotting on nitrocellulose membranes was performed using a semi-dry Turbo transfer blot system (BioRad). For images of human LC3B, a wet transfer to PVDF membranes was performed at 200 mA for 70 min. Membranes were blocked with 5% skimmed milk or BSA in TBS and 0.1% Tween 20 (TBS-T) for 1 hr at room temperature or at 4°C overnight. This was followed by incubation with primary and subsequent secondary antibody conjugated to horseradish peroxidase. After three times 10 min washes with TBS-T, the immune-reaction was developed using ECL Super-Pico Plus (Thermo) and detected with ChemiDoc Touch Imaging System (BioRad).
Protein bands intensities were quantified with Image Lab 6 (BioRad). Equal rectangles were drawn around the total protein gel lane and the band of interest. The lane profile was obtained by subtracting the mean intensity of the background. The adjusted volume of the peak in the profile was taken as a measure of the band intensity. The protein band of interest was normalized for the total protein level of the whole lane. Average relative intensities and a standard error of at least three independent experiments were calculated.
To generate AtC53 antibody, purified protein was sent to Eurogentec for immunization of rabbits via their 28 day program. The final bleed was purified on column conjugated with the purified protein.
For pulldown experiments, 10 µl of glutathione magnetic agarose beads (Pierce Glutathione Magnetic Agarose Beads, Thermo Scientific) were equilibrated by washing them two times with wash buffer (100 mM Sodium Phosphate pH 7.2, 300 mM NaCl, 1 mM DTT, 0.01% (v/v) IGEPAL). Normalized E. coli clarified lysates or purified proteins were mixed, according to the experiment, added to the washed beads and incubated on an end-over-end rotator for 1 hr at 4°C. Beads were washed five times in 1 ml wash buffer. Bound proteins were eluted by adding 100 µl Laemmli buffer. Samples were analysed by western blotting or Coomassie staining.
Yeast two hybrid assay (Y2H) was performed according to the Mathmaker GAL4 Two hybrid system (Clonetech) following the protocol from the manufacture. Different genes were fused in frame to GAL4 activation domain of the prey vector pGADT7 and GAL4 binding domain from the bait vector pGBKT7. Split-GFP was used as positive control. Combinations of pGADT7 and pGBKT7 vectors carrying the different genes were transformed in the yeast strains Y187 (MAT α) and AH109 (MAT a), respectively. After mating between bait and prey strains, the diploid yeast was selected for growth on (SD)-Leu /- Trp, (SD)-Leu /- Trp /- His and (SD)-Leu /- Trp /- His/-Ade plates at 28°C for 2 to 4 days.
0.5–1 g seedlings were grown in liquid and treated as described under section Autophagy Flux Assay. After homogenization of frozen samples by bead-mill, G-TEN buffer (10% Glycerole, 50 mM Tris/HCl pH 7.5, 1 mM EDTA, 300 mM NaCl, 1 mM DTT, 0.1% [v/v] Nonidet P-40/Igepal, Complete protease inhibitor tablet) was added, vortexed, and lysates were cleared by centrifugation at 16,000 g for 10 min at 4°C. Protein concentration was equally adjusted using Bradford protein assay (Sigma).
25 µl of RFP or GFP-Trap_A beads (Chromotek) were equilibrated and added to each lysate and incubated for 2 hr at 4°C on a turning wheel. Beads were washed three times with 1 mL G-TEN buffer.
For western Blot analysis, beads were resuspended in 30 µl SDS-loading buffer (116 mM Tris-HCl pH 6.8, 4.9% glycerol, 10 mM DTT, 8% SDS). On-bead bound proteins were eluted by boiling the beads for 10 min at 70°C and analysed by western blotting with indicated antibodies.
For mass spectrometry experiments, the beads were further washed five times with mass spectrometry compatible buffer (50 mM Tris/HCl pH 7.5, 1 mM EDTA). Buffer resuspended beads were then submitted for trypsin digestion.
Bead-bound bait proteins were incubated with fluorescently labelled prey protein as described previously by Turco et al., 2019.
10 µl of Glutathione Sepharose 4B beads (GE Healthcare, average diameter 90 mm) were incubated for 30 min at 4°C (16 rpm horizontal rotation) with GST-tagged bait proteins (4 mg/mL for GST and GST-FIP200 CTR). The beads were washed two times in 10x bead volume of washing buffer (25 mM HEPES pH 7.5, 150 mM NaCl, 1 mM DTT). The buffer was removed, and the beads were resuspended 1:1 in washing buffer. 10 µL of a 2–5 µM dilution of fluorescently labeled binding partners (GFP, C53-GFP and GFP-p62) were added to the bead suspension and incubated for 30–60 min at room temperature before imaging with a Zeiss LSM700 confocal microscope with 20 X magnification. For quantification, the maximum gray value along the diameter of each bead (n ≥ 15) was measured.
MS/MS Data analysis: Raw files were processed with Proteome Discoverer (version 2.3, Thermo Fisher Scientific, Bremen, Germany). Database searches were performed using MS Amanda (version 188.8.131.5214) (Dorfer et al., 2014) against the TAIR10 database (32785 sequences). The raw files were loaded as fractions into the processing workflow. Carbamidomethylation of cysteine and TMT on peptide N-termini were specified as fixed modifications, phosphorylation on serine, threonine and tyrosine, oxidation of methionine, deamidation of asparagine and glutamine, TMT on lysine, carbamylation on peptide N-termini and acetylation on protein N-termini were set as dynamic modifications. Trypsin was defined as the proteolytic enzyme, cleaving after lysine or arginine. Up to two missed cleavages were allowed. Precursor and fragment ion tolerance were set to 5 ppm and 15 ppm, respectively. Identified spectra were rescored using Percolator (Käll et al., 2007), and filtered to 0.5% FDR at the peptide spectrum match level. Protein grouping was performed in Proteome Discoverer applying strict parsimony principle. Proteins were subsequently filtered to a false discovery rate of 1% at protein level. Phosphorylation sites were localized using IMP-ptmRS implemented in Proteome Discoverer using a probability cut-off of >75% for unambiguous site localization.
TMT-quantification: TMT reporter ion S/N values were extracted from the most confident centroid mass within an integration tolerance of 20 ppm. PSMs with average TMT reporter S/N values below 10 as well as PSMs showing more than 50% co-isolation were removed. Protein quantification was determined based on unique peptides only. Samples were sum normalized and missing values were imputed by the 5% quantile of the reporter intensity in the respective sample. Statistical significance of differentially abundant proteins was determined using limma (Smyth, 2004). Gene Ontology (Ashburner et al., 2000) enrichment was determined using DAVID (Dennis et al., 2003) (version 6.8). Cross species comparison of regulated proteins was performed by mapping proteins to ortholog clusters available in eggnog (Huerta-Cepas et al., 2016). Proteins containing signal peptides were predicted using SignalP 5.0 (Almagro Armenteros et al., 2019).
High-density peptide array analysis was performed commercially by PEPperPRINT. This comprised a full substitution scan of wild-type peptide GVSEWDPILEELQEM, with exchange of all amino acid positions with 23 amino acids including citrulline (Z), methyl-alanine (O) and D-alanine (U). The analysis also included an N- and C-terminal deletion series of wild-type peptide GVSEWDPILEELQEM; an additional 32 spots of custom control peptide KPLDFDWEIVLEEQ, and acidic variants of this control peptide involving exchanges of selected amino acid positions with glutamic acid €. The resulting peptide microarrays contained 416 different linear peptides printed at least in triplicate (1412 peptide spots; wild-type peptides were printed with a higher frequency), and were framed by HA (YPYDVPDYAG, 88 spots) control peptides (See Supplementary file 1 for the array map).
Peptide microarrays were pre-stained with rabbit anti-GST Dylight680 at a dilution of 1:2000 to investigate background interactions with the variants of wild-type peptides GVSEWDPILEELQEM and KPLDFDWEIVLEEQ that could interfere with the main assays. Subsequent incubation of other peptide microarrays with proteins GST-ATG8A and GST at a concentration of 10 µg/ml in incubation buffer was followed by staining with secondary antibody rabbit anti-GST Dylight680 and read-out at a scanning intensity of 7 (red). The control staining of the HA epitopes with control antibody mouse monoclonal anti-HA (12CA5) DyLight800 was finally done as an internal quality control to confirm the assay quality and the peptide microarray integrity. Read-out of the control staining was performed at a scanning intensity of 7/7 (red/green).
Quantification of spot intensities and peptide annotation were based on the 16-bit grey scale tiff files at a scanning intensity of 7 that exhibit a higher dynamic range than the 24-bit colorized tiff files; microarray image analysis was done with PepSlide Analyzer. A software algorithm breaks down fluorescence intensities of each spot into raw, foreground and background signal (see ‘Raw Data’ tabs), and calculates averaged median foreground intensities and spot-to-spot deviations of spot duplicates.
Four-day-old seedlings were treated as indicated under the autophagy flux assay section. Seedlings were imaged between 3 hr - 6 hr of drug incubation. Roots were placed on a microscope slide with indicated treatment buffer and closed with coverslip. Imaging was performed in the root differentiation zone where root hair growth starts.
Transfected and treated cells were grown on coverslips and fixed utilizing 0.4% Paraformaldehyde solution in PBS for 30 min. Fixed cells were mounted in VectaShield mounting medium without DAPI and sealed using clear nail polish.
Samples were imaged at an upright ZEISS LSM800 or LSM 780 confocal microscope (Zeiss) with an Apochromat 40x or 63x objective lens at 1x magnification.
Excitation/detection parameters for GFP and mCherry were 488 nm/463 nm and 510 nm and 561 nm/569 to 635 nm, respectively, and sequential scanning mode was used for colocalization of both fluorophores. Identical settings, including an optical section thickness of 2 µm per z-stack, were used during the acquisition for sample comparison, and the images processed using identical parameters. Confocal images were processed with ZEN (version 2011) and ImageJ (version 1.48 v) software.
Autophagic puncta were counted using ImageJ. Several (at least five) z-stack merged images were manually background subtracted, thresholded and the same threshold value was applied to all the images and replicates of the same experiment. The image was converted to eight-bit grayscale and then counted for ATG8 puncta either manually or by the Particle Analyzer function of ImageJ. The average number of autophagosomes per z-stack was averaged between 10 or more different roots.
Colocalization analysis was performed by calculating Pearson’s correlation coefficient as previously described using ImageJ software with the plug-in JACoP (Bolte and Cordelières, 2006). Values near one represent almost perfect correlation, whereas values near 0 reflect no correlation. The average Pearson’s correlation coefficient was determined in five or more different roots.
For high-pressure freezing, 5-day-old Arabidopsis seedling roots expressing AtC53-mCherry were cut and high-pressure frozen (EM PACT2, Leica, Germany), prior to subsequent freeze substitution in acetone containing 0.4% uranyl acetate at −85°C in an AFS freeze-substitution unit (Leica, Wetzlar, Germany). After gradient infiltration with increasing concentration of HM20, root samples were embedded and ultraviolet polymerized for ultra-thin sectioning and imaging. TEM images were captured by an 80 kV Hitachi H-7650 transmission electron microscope (Hitachi High-Technologies Corporation, Japan) with a charge-coupled devise camera. IEM analysis were performed as previously described (Zhuang et al., 2017).
Arabidopsis roots were fixed in 2% paraformaldehyde and 0.2% glutaraldehyde (both EM-grade, EMS, USA) in 0.1 M PHEM buffer (pH 7) for 2 hr at RT, then overnight at 4°C. The fixed roots were embedded in 12% gelatin and cut into 1 mm3 blocks which were immersed in 2.3 M sucrose overnight at 4°C. These blocks were mounted onto a Leica specimen carrier (Leica Microsystems, Austria) and frozen in liquid nitrogen. With a Leica UCT/FCS cryo-ultramicrotome (Leica Microsystems, Austria) the frozen blocks were cut into ultra-thin sections at a nominal thickness of 60 nm at −120°C. A mixture of 2% methylcellulose (25 centipoises) and 2.3 M sucrose in a ratio of 1:1 was used as a pick-up solution. Sections were picked up onto 200 mesh Ni grids (Gilder Grids, UK) with a carbon coated formvar film (Agar Scientific, UK). Fixation, embedding and cryo-sectioning was conducted as described by Tokuyasu, 1973.
Prior to immunolabeling, grids were placed on plates with solidified 2% gelatine and warmed up to 37°C for 20 min to remove the pick-up solution. After quenching of free aldehyde-groups with glycine (0.1% for 15 min), a blocking step with 1% BSA (fraction V) in 0.1 M Sörensen phosphate buffer (pH 7.4) was performed for 40 min. The grids were incubated in primary antibody, rabbit polyclonal to GFP (ab6556, Abcam, UK), diluted 1:125 in 0.1 M Sörensen phosphate buffer over night at 4°C, followed by a 2 hr incubation in the secondary antibody, a goat-anti-rabbit antibody coupled with 6 nm gold (GAR 6 nm, Aurion, The Netherlands), diluted 1:20 in 0.1 M Sörensen phosphate buffer, performed at RT. The sections were stained with 4% uranyl acetate (Merck, Germany) and 2% methylcellulose at a ratio of 1:9 (on ice). All labeling steps were conducted in a wet chamber. The sections were inspected using a FEI Morgagni 268D TEM (FEI, The Netherlands) operated at 80kV. Electron micrographs were acquired using an 11-megapixel Morada CCD camera from Olympus-SIS (Germany).
Statistical analyses were performed with GraphPad Prism eight software. For all the quantifications described above, statistical analysis was performed. Statistical significance of differences between two experimental groups was assessed wherever applicable by either a two-tailed Student’s t-test if the variances were not significantly different according to the F test, or using a non-parametric test (Mann-Whitney or Kruskal-Wallis with Dunn’s post-hoc test for multiple comparisons) if the variances were significantly different (p<0.05). Differences between two data sets were considered significant at p<0.05 (*); p<0.01 (**); p<0.001 (***); p<0.0001 (****); n.s., not significant.
Recombinant proteins were produced using E. coli strain Rosetta2(DE3)pLysS grown in 2x TY media at 37°C to an A600 of 0.4–0.6 followed by induction with 300 µM IPTG and overnight incubation at 18°C. Pelleted cells were resuspended in lysis buffer (100 mM Sodium Phosphate pH 7.0, 300 mM NaCl) containing protease inhibitors (Complete, Roche) and sonicated. The clarified lysate was first purified by affinity, by using HisTrap FF (GE HealthCare) columns. The proteins were eluted with lysis buffer containing 500 mM Imidazole. The eluted fraction was buffer exchanged to 10 mM Sodium Phosphate pH 7.0, 100 mM NaCl and loaded either on Cation Exchange, Resource S, or Anion Exchange, Resource Q, chromatography columns. The proteins were eluted by NaCl gradient (50% in 20 CV). Finally, the proteins were separated by Size Exclusion Chromatography with HiLoad 16/600 Superdex 200 pg or HiLoad 16/600 Superdex 75 pg, which were previously equilibrated in 50 mM Sodium Phosphate pH 7.0, 100 mM NaCl. The proteins were concentrated using Vivaspin concentrators (3000, 5000, 10000 or 30000 MWCO). Protein concentration was calculated from the UV absorption at 280 nm by DS-11 FX+ Spectrophotometer (DeNovix).
Binding of AIM wt (EPLDFDWEIVLEEEM) and AIM mutant (EPLDFDAEIALEEEM) peptide to GST-GABARAP and GST-ATG8A, respectively, was investigated by surface plasmon resonance analysis using a Biacore T200 instrument (GE Healthcare) operated at 25°C. In additione, AIM-dependent binding of HsC53 and AtC53 to GST-GABARAP and GST-ATG8A were studied. The running buffer used for all experiments was 50 mM sodium phosphate pH 7.0 supplemented with 100 mM NaCl, 0.05% (v/v) Tween-20% and 0.1% (w/v) bovine serum albumin.
Polyclonal anti-GST antibodies (GST Capture Kit, GE Healthcare) were amine coupled on to a Series S CM5 sensor chip (GE Healthcare) using two adjacent flow cells (i.e. the reference and active cell) according to the manufacturer’s instructions.
To determine specific binding, GST-GABARAP or GST-ATG8A were captured on the active cell (concentration: 5 µg/ml; contact time: 30 s; flow rate: 10 µl/min) and GST was captured on the reference cell (concentration: 10 µg/ml; contact time: 30 s; flow rate: 10 µl/min) to perform background subtraction.
To qualitatively show whether the analytes, HsC53, HsC53 123A (i.e. HsC53W269A, W294A, W312A), AtC53 and AtC53 1234A (i.e. AtC53W276A, W287A, Y304A, W335A), interact or do not interact in an AIM-dependent manner with GST-GABARAP or GST-ATG8A, the two flow cells were exposed to four sets of double consecutive injections (1st set: 10 µM analyte, running buffer; 2nd set: 10 µM analyte, 10 µM analyte; 3rd set: 10 µM analyte, 10 µM analyte + 6.4 µM AIM wt peptide; 4th set: 10 µM analyte, 10 µM analyte + 6.4 µM AIM mutant peptide. Contact time 1st injection: 30 s; contact time 2nd injection: 30 s; dissociation time: 60 s; flow rate: 30 µl/min).
To quantify the binding affinities of the AIM wt peptide to GST-ATG8 or GST-GABARAP, multi-cycle kinetic experiments with increasing concentrations of the AIM wt peptide (25, 50, 100, 200, 400, 800, 1600, 3200 nM and 400 nM as internal replicates) were performed (contact time: 60 s; dissociation time: 60 s; flow rate: 30 µl/min). As a negative control, the chip was exposed to 3200 nM of the AIM mutant peptide (contact time: 60 s; dissociation time: 60 s; flow rate: 30 µl/min).
To quantify the apparent binding affinity of the AIM wt peptide to GST-GABARAP in presence of HsC53, multi-cycle kinetic experiments with increasing concentrations of the AIM peptide (0, 25, 50, 100, 200, 400, 800, 1600, 3200 nM and 400 nM as internal replicates), containing 10 µM of HsC53, in running buffer (contact time: 60 s; dissociation time: 60 s; flow rate: 30 µl/min). For negative controls, the chip was exposed to 3200 nM of the AIM mutant peptide, containing 10 µM of HsC53 or 10 µM of HsC53 123A, and to 3200 nM of the AIM wt peptide, containing 10 µM of HsC53 123A (contact time: 60 s; dissociation time: 60 s; flow rate: 30 µl/min).
To quantify the apparent binding affinity of the AIM wt peptide to GST-ATG8A in the presence of AtC53, multi-cycle kinetic experiments with increasing concentrations of the AIM peptide (0, 50, 100, 200, 400, 800, 1600, 3200, 6400, 12800 nM and 400 nM as internal replicate) containing 10 µM of AtC53 were performed (contact time: 60 s; dissociation time: 60 s; flow rate: 30 µl/min). As negative controls, the chip was exposed to 6400 nM of the AIM mutant peptide, containing 10 µM of AtC53 or 10 µM of AtC53 1234A, and to 6400 nM of the AIM wt peptide, containing 10 µM of AtC53 1234A (contact time: 60 s; dissociation time: 60 s; flow rate: 30 µl/min).
After each cycle, regeneration was performed with 2 injections of 10 mM glycine-HCl pH 2.1 for 120 s at a flow rate of 10 µL/min.
The sensograms obtained were analyzed with Biacore T200 Evaluation software (version 3.1) by global fitting of the data to a 1:1 steady-state affinity model.
Molecular weights and sources of the proteins for the SPR experiments are reported in the following table:
|Sample name||Source||MW (Da)|
|AIM wt peptide||Synthetized in house||1894.08|
|AIM mutant peptide||1750.89|
|HsC53||Escherichia coli recombinant expression||59191.15|
Calculation for the apparent () of the AIMwt was done by using the following formula (Nelson, David L. Lehninger Principles Of Biochemistry. New York: W.H. Freeman, 2008):
Where, , and
All experiments were carried out at 25°C in 50 mM sodium phosphate buffer pH 7.0, 100 mM NaCl, using the PEAQ-ITC Automated (Malvern Panalytical Ltd). For protein-protein interactions, the calorimetric cell was filled with 40 μM GABARAP or ATG8A and titrated with 250 μM HsC53 or AtC53 IDRs, respectively. A single injection of 0.4 μl of HsC53 or AtC53 IDRs (not taken into account) was followed by 18 injections of 2 μl each. Injections were made at 150 s intervals with a duration of 4 s and a stirring speed of 750 rpm. The reference power was set to 10 µcal/s, the feedback mode was set to high. For protein-peptide interactions, the calorimetric cell was filled with 40 μM GABARAP or ATG8A and titrated with 600 μM peptide from the syringe. The titrations were performed as described above. For the control experiments, equivalent volumes of the IDRs, or the peptides, were titrated to buffer, equivalent volumes of buffer were titrated to GABARAP or ATG8A and equivalent volumes of buffer were titrated to buffer, using the parameters above. The raw titration data were integrated, corrected for the controls and fitted to a one-set-of-sites binding model using the PEAQ-ITC analysis software (Version 1.22).
Proteins were buffer exchanged into ammonium acetate using BioRad Micro Bio-Spin 6 Columns and concentrations were measured using DS-11 FX+ Spectrophotometer (DeNovix).
Native mass spectrometry experiments were carried out on a Synapt G2Si instrument (Waters, Manchester, UK) with a nanoelectrospray ionisation source. Mass calibration was performed by a separate infusion of NaI cluster ions. Solutions were ionised through a positive potential applied to metal-coated borosilicate capillaries (Thermo Scientific). The following instrument parameters were used; capillary voltage 1.3 kV, sample cone voltage 40 V, extractor source offset 30 V, source temperature 40°C, trap gas 3 mL/min. A higher capillary voltage (1.9 kV) was required for ionization of the 1:2 AtC53-AtG8A complex. Data were processed using Masslynx V4.1 and spectra were plotted using R. Peaks were matched to protein complexes by comparing measured m/z values with expected m/z values calculated from the mass of individual proteins which are given in table below.
|Protein||Expected mass from sequence/Da||Measured Mass/Da|
|AtC53||64 399.6||64 401.3|
|AtC53-1234A||63 962.1||63 976.7|
|HsC53||59 191.1||59 193.0|
|AtG8A||15 965.3||15 964.0|
|GABARAP||15 968.3||15 968.2|
CD spectroscopy experiments were performed using a Chirascan-Plus CD spectrophotometer (Applied Photophysics). Purified proteins in 50mM sodium phosphate pH 7.0, 100mM NaCl were diluted to approximately 0.2 mg/ml and spin-filtered with an 0.1µm filter. CD measurements were carried out in a quartz glass cuvette with 0.5 mm path length. To obtain overall CD spectra, wavelength scans between 180 nm and 260 nm were collected at 25°C using a 1.0 nm bandwidth, 0.5 nm step size, and time per point of 0.5 s. Both CD and absorbance data were collected at the same time over three accumulations and averaged. CD data at wavelengths where the absorptivity was above 2.5 are not shown (data below 194nm). The raw data in millidegree units were corrected for background and drift (). Subsequently, the differential molar extinction coefficient per peptide bond (Δε) was calculated, taking into account the absorptivity measured at 205 nm (A205) and the calculated protein extinction coefficient at 205 nm (ε205) using the equation
SignalP 5.0 improves signal peptide predictions using deep neural networksNature Biotechnology 37:420–423.https://doi.org/10.1038/s41587-019-0036-z
Gene ontology: tool for the unification of biologyNature Genetics 25:25–29.https://doi.org/10.1038/75556
A mass-spectrometry-based framework to define the extent of disorder in proteinsAnalytical Chemistry 86:10979–10991.https://doi.org/10.1021/ac5027435
A guided tour into subcellular colocalization analysis in light microscopyJournal of Microscopy 224:213–232.https://doi.org/10.1111/j.1365-2818.2006.01706.x
ER-Phagy: quality control and turnover of endoplasmic reticulumTrends in Cell Biology 30:384–398.https://doi.org/10.1016/j.tcb.2020.02.001
Mechanism and medical implications of mammalian autophagyNature Reviews Molecular Cell Biology 19:349–364.https://doi.org/10.1038/s41580-018-0003-4
MS Amanda, a universal identification algorithm optimized for high accuracy tandem mass spectraJournal of Proteome Research 13:3679–3684.https://doi.org/10.1021/pr500202e
Proteasomal and lysosomal clearance of faulty secretory proteins: er-associated degradation (ERAD) and ER-to-lysosome-associated degradation (ERLAD) pathwaysCritical Reviews in Biochemistry and Molecular Biology 54:153–163.https://doi.org/10.1080/10409238.2019.1610351
Translocon component Sec62 acts in endoplasmic reticulum turnover during stress recoveryNature Cell Biology 18:1173–1184.https://doi.org/10.1038/ncb3423
Selective autophagy: atg8 family proteins, LIR motifs and cargo receptorsJournal of Molecular Biology 432:80–103.https://doi.org/10.1016/j.jmb.2019.07.016
The unfolded protein response: detecting and responding to fluctuations in the Protein-Folding capacity of the endoplasmic reticulumCold Spring Harbor Perspectives in Biology 11:a033886.https://doi.org/10.1101/cshperspect.a033886
ATG8 expansion: a driver of selective autophagy diversification?Trends in Plant Science 22:204–214.https://doi.org/10.1016/j.tplants.2016.11.015
ER-Anchored transcription factors bZIP17 and bZIP28 regulate root elongationPlant Physiology 176:2221–2230.https://doi.org/10.1104/pp.17.01414
Autophagy : molecular mechanisms and disease outcomesAccessed March 1, 2017.
Efficient Agrobacterium-mediated transformation of the liverwort Marchantia polymorpha using regenerating thalliBioscience, Biotechnology, and Biochemistry 77:167–172.https://doi.org/10.1271/bbb.120700
The EMBL-EBI search and sequence analysis tools APIs in 2019Nucleic Acids Research 47:W636–W641.https://doi.org/10.1093/nar/gkz268
Druggable sensors of the unfolded protein responseNature Chemical Biology 10:892–901.https://doi.org/10.1038/nchembio.1664
Autophagy: the master of bulk and selective recyclingAnnual Review of Plant Biology 69:173–208.https://doi.org/10.1146/annurev-arplant-042817-040606
CALCOCO1 acts with VAMP-associated proteins to mediate ER-phagyThe EMBO Journal 39:e103649.https://doi.org/10.15252/embj.2019103649
Functional diversification of ER stress responses in ArabidopsisTrends in Biochemical Sciences 45:123–136.https://doi.org/10.1016/j.tibs.2019.10.008
Linear models and empirical bayes methods for assessing differential expression in microarray experimentsStatistical Applications in Genetics and Molecular Biology 3:1–25.https://doi.org/10.2202/1544-6115.1027
Plant selective Autophagy-Still an uncharted territory with a lot of hidden gemsJournal of Molecular Biology 432:63–79.https://doi.org/10.1016/j.jmb.2019.06.028
Cargo recognition and trafficking in selective autophagyNature Cell Biology 16:495–501.https://doi.org/10.1038/ncb2979
A technique for ultracryotomy of cell suspensions and tissuesThe Journal of Cell Biology 57:551–565.https://doi.org/10.1083/jcb.57.2.551
A UPR-Induced soluble ER-Phagy receptor acts with VAPs to confer ER stress resistanceMolecular Cell S1097-2765:30510–30514.https://doi.org/10.1016/j.molcel.2020.07.019
Michael RapeReviewing Editor; University of California, Berkeley, United States
Suzanne R PfefferSenior Editor; Stanford University School of Medicine, United States
In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.
While having been intensely studied in yeast or mammalian cells, autophagy receptors remain less well understood in plants. In this study, the authors identify C53 as a conserved autophagy receptor from plants to human cells. C53 recognizes a shuffled ATG8-interaction motif to bind ATG8. While C53 does not play a role in general autophagy, it is recruited to clogged ER translocons during times of ER stress. C53 accumulates at the ER via DDRGK1 and UFL1, thus coordinating UFM1-modification of RPS26 in stalled ribosomes with formation of an autophagophore. Consistent with these observations, mutants in C53 or the plant UFMylation machinery were specifically sensitive to ER stress. This paper therefore provides important insight into conserved mechanisms underlying selective ER-phagy from plants to humans.
Decision letter after peer review:
Thank you for submitting your article "A cross-kingdom conserved ER-phagy receptor maintains endoplasmic reticulum homeostasis during stress" for consideration by eLife. Your article has been reviewed by three peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by Suzanne Pfeffer as the Senior Editor. The reviewers have opted to remain anonymous.
The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.
As the editors have judged that your manuscript is of interest, but as described below that additional experiments are required before it is published, we would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). First, because many researchers have temporarily lost access to the labs, we will give authors as much time as they need to submit revised manuscripts. We are also offering, if you choose, to post the manuscript to bioRxiv (if it is not already there) along with this decision letter and a formal designation that the manuscript is "in revision at eLife". Please let us know if you would like to pursue this option. (If your work is more suitable for medRxiv, you will need to post the preprint yourself, as the mechanisms for us to do so are still in development.)
The endoplasmic reticulum (ER) orchestrates the folding and secretion of a large number of proteins. Protein misfolding in the ER can be corrected by the ubiquitin-proteasome system, as well as by autophagy of parts of the ER, a process referred to as ER-phagy. While several adaptors that link ER to the autophagy machinery have been described, how this process is being regulated is only beginning to be understood.
In this manuscript, Stephani and co-workers identify a protein, C53, that is recruited to ATG8 proteins upon ER stress. C53 recognizes ATG8, a hallmark modification of autophagy, through a novel "shuffled ATG8-interacting motif". C53 localization changes in cells expressing a reporter for stalled ribosomes, and cells lacking C53 are particularly sensitive to ER-stress that is elicited by aberrant co-translational protein synthesis and ER targeting. The experiments describing the interaction between C53 and ATG8, both in plants and human cells, have been performed well and establish C53 as a bona fide ATG8-binding protein. Also, the change in localization of C53 upon specific ER stresses is very interesting and indicative of a potential function of this protein in ER-phagy. However, there are several major issues that would need to be addressed before this manuscript could be accepted for publication in eLife. We recognize that these issues require significant experimentation, yet due to Covid-19, we will not impose a time limit to the authors on when to resubmit to eLife.
Essential, major issues that need to be addressed with experiments prior to publication are:
1) The authors do not show that C53 localizes to ER upon stress, a prerequisite for a specific ER-phagy receptor. They also do not demonstrate that C53 is required for specific ER-phagy. These questions would need to be addressed through better localization analyses (i.e. co-localization of C53 and ER markers) and flux studies (to show that bulk ER proteins are turned over in a C53- and autophagy-dependent manner). Moreover, the authors should include a more careful investigation of the role of the UPR in these processes.
2) The authors would need to improve their investigation of C53's role in turning over stalled ribosomes at the ER. The current data is not sufficient and would need to be supported by flux assays that show reporter turnover in a C53- and autophagy-dependent manner in a pulse-chase setting.
3) Given their focus and previous publications, the authors would also need to show that the UFM1 machinery is required for C53's role in ER-phagy. They also should perform sequential IPs to show that C53 is really part of a ternary complex with UFL1 and DDRGK1, as explicitly stated by their model.
4) In addition to these experimental revisions, this manuscript requires substantial editing to allow readers understand the experiments and their significance. The reasoning between experiments needs to be explained in more detail. For example, the authors need to explicitly discuss concentrations (i.e. are protein levels used for competition experiments comparable to cellular concentrations?); moreover, they should describe how many AIMs are outside of the IDR region, how were they identified, and how many were tested; finally, the authors should describe better what they mean by a "shuffled AIM" – is this a new ATG8-interaction motif or do they believe it is just a minor variation of the standard recognition element? These are examples – the whole manuscript would greatly benefit from more explicit explanations of experiments to make it more accessible to a larger audience.https://doi.org/10.7554/eLife.58396.sa1
Essential, major issues that need to be addressed with experiments prior to publication are:
1) The authors do not show that C53 localizes to ER upon stress, a prerequisite for a specific ER-phagy receptor.
In the previous version of the manuscript, we had already analyzed the localization of C53 in plants by immunogold labeling with two different antibodies and live cell imaging using confocal microscopy under normal and stress conditions.
Our electron micrographs obtained from non-stressed cells already suggested C53 is in close proximity to the ER (original Figure 2—figure supplement 3). We have now performed a more detailed electron microscopy analysis demonstrating that C53 localizes to the ER and delivered to the vacuole, but does not localize to other subcellular compartments such as multivesicular bodies, Golgi apparatus or mitochondria. (updated Figure 2—figure supplement 3). In addition, we carried out colocalization assays with (i) a canonical ER marker (HDEL-GFP) (now presented in Figure 2—figure supplement 1A), (ii) DDRGK1, a transmembrane domain containing protein that localize uniformly at the ER (Liang et al., 2020) (original Figure 8 and new Figure 8—figure supplement 2), and (iii) UFL1, the ufmylation E3 ligase residing in the ER (original Figure 8). This confocal microscopy analysis also shows that upon ER stress, C53 specifically co-localizes with both DDRGK1 and UFL1 (presented in Figure 8C and the new Figure 8—figure supplement 2).
In addition, we have already showed in the previous version of the manuscript that C53 puncta co-localize with the ER targeted ribosome stalling substrate (presented in Figure 7—figure supplement 2) in HeLa cells. This substrate has been shown to be trapped in Sec61 translocons that are on the ER (Wang et al., 2020).
Altogether the new additional electron microscopy and confocal microscopy data and the previously presented imaging data suggest C53 is recruited to ER during stress.
They also do not demonstrate that C53 is required for specific ER-phagy. These questions would need to be addressed through better localization analyses (i.e. co-localization of C53 and ER markers) and flux studies (to show that bulk ER proteins are turned over in a C53- and autophagy-dependent manner).
To demonstrate that C53 is required for ER-phagy, we performed the suggested confocal microscopy and autophagic flux assays. Using confocal microscopy and western blot-based flux assays, we show that ER resident protein DDRGK1, but not the abundant ER resident proteins SMT1, BIPs or Calnexins are degraded via C53-mediated autophagy (Presented in Figure 8—figure supplement 2). In addition, DDRGK1 form punctate structures that colocalize with ATG8A in a C53 dependent manner (Presented in Figure 8—figure supplement 2). We also have presented data in the previous version of the manuscript that showed that C53 mediates lysosomal delivery of the ER targeted ribosome stalling construct in HeLa cells (Figure 7—figure supplement 2). Together, our data confirm that C53 mediates ER-phagy, however functions differently than the known ER-phagy receptors. Rather than shaping the ER upon starvation or ER-stress, it is specifically involved in recycling stalled nascent chains at the ER. To clarify this point, we added a paragraph in the Discussion, comparing C53 with other ER-phagy receptors.
With regards to its mechanism, our data suggest that C53 is not involved in recycling of ER tubules or sheets per se but rather enables a quality control pathway to recycle arrested polypeptides of ER-stalled ribosomes. Moreover, the quality-control activity of C53 seems to be highly specific: even though both the Clogger and ER-K20 constructs are localized at Sec61 translocons, only ER-K20, a co-translational ribosome stalling construct is capable to activate C53 and get degraded by C53 autophagy.
To test specific induction of C53 upon ribosome stalling, we tested its activation by different translation inhibitors. Our new data show that C53 is activated by various translation elongation inhibitors, but not by drugs that inhibit initiation of translation. This phenotype contrasts to the occurrence of ATG8 puncta, which are activated with all translation inhibitors (new Figure 7—figure supplement 2). Together, these data indicate that C53 is not involved in reticulophagy or recycling of ER sheets, but in recycling stalled translation products. Consistently, quantitative proteomics data presented in the original manuscript (Figure 6) revealed that proteins accumulating in C53 deficient cells are mostly proteins that are synthesized in the ER-bound ribosomes that could trigger ribosome stalling during translational elongation.
Moreover, the authors should include a more careful investigation of the role of the UPR in these processes.
Plants utilize two major UPR branches, the Ire1 and bZIP17/28 (homologous to ATF6), however lack PERK homologs (Pastor-Cantizano, et al., 2020). To address the connection with the plant UPR system, we thus focused our analysis on the Ire1 and bZIP17/28 ER stress sensors. In the original version of the manuscript, we already presented data showing that C53 protein levels are elevated in bZIP17/28 mutants and that C53 autophagic flux is not affected in neither ire1a/b double mutant or bzip17/28 double mutants. Furthermore, we presented data showing that C53 still forms puncta in response to ER stress. We also corroborated these findings using Ire1 inhibitors in HeLa cells. To further confirm these data, we carried out further experiments showing that C53 puncta are still induced upon ER stress in bZIP28/bZIP60 mutants. Moreover, C53 puncta do not colocalize with IRE1 oligomers that form upon ER stress (updated Figure 7). As C53 is highly conserved and plants only employ two UPR pathways, our data conclusively show that C53 mediated autophagy is not activated by UPR sensors.
We also would like to note that all plant experiments are performed with stable transgenic lines. As there has been several examples, where cell culture studies were not transferrable to organisms, the use of stable transgenic organisms – as in the present study – is highly beneficial for obtaining robust and reproducible results.
2) The authors would need to improve their investigation of C53's role in turning over stalled ribosomes at the ER. The current data is not sufficient and would need to be supported by flux assays that show reporter turnover in a C53- and autophagy-dependent manner in a pulse-chase setting.
As described in the original manuscript, our quantitative proteomics data suggest that C53 does not degrade stalled ribosomes. New data presented in Figure 8—figure supplement 2B confirm our notion. In contrast to DDRGK1, neither 40S nor 60S ribosomal subunits are degraded via C53 (new Figure 8—figure supplement 2). Moreover, we show that abundant ER proteins, such as calnexin or the BIP chaperones, are also not degraded via the C53 pathway. However, a comprehensive analysis of cargo selection and recruitment within the ER is beyond the scope of this manuscript.
3) Given their focus and previous publications, the authors would also need to show that the UFM1 machinery is required for C53's role in ER-phagy.
In the original version of our manuscript, we have presented data illustrating that upon ER stress, UFL1 forms puncta that colocalize with C53 and that these puncta are delivered to the vacuole. We have also shown that in stable ufl1 mutant plants, C53 puncta do not form upon ER stress (Figure 8—figure supplement 1). We also presented data showing that all ufmylation pathway mutants are sensitive to ER stress but not starvation (Figure 10—figure supplement 1). To provide further experimental evidence for the proposed mechanism, we carried out an autophagic flux experiment showing that C53 flux is affected in all ufmylation mutants (Figure 8—figure supplement 1F). Together, our data confirm the connection of UFM1 machinery and C53-mediated ER-phagy.
They also should perform sequential IPs to show that C53 is really part of a ternary complex with UFL1 and DDRGK1, as explicitly stated by their model.
We based our conclusion that C53-UFL1-DDRGK1 forms a tripartite receptor complex (now modified to heteromeric receptor complex) based on in vitro pull downs, in vivo pull downs, yeast two hybrid assays and colocalization studies. To further support our model, we have performed additional pull-down experiments, where we were able to pull down both DDRGK1 and C53 in a single UFL1 pull down (Figure 8—figure supplement 1A). Additionally, we have included DDRGK1, UFL1 and C53 affinity purification mass spectrometry results under normal and stress conditions that showed that all three proteins associate with each other, and this interaction becomes stronger upon ER stress treatments (Supplementary file 5).
4) In addition to these experimental revisions, this manuscript requires substantial editing to allow readers understand the experiments and their significance. The reasoning between experiments needs to be explained in more detail. For example, the authors need to explicitly discuss concentrations (i.e. are protein levels used for competition experiments comparable to cellular concentrations?).
We have now significantly edited the manuscript to explain the experiments more clearly. We have extended the text to better explain the rationale behind the experimental design and details, and our conclusions. We have also extended the Discussion section to compare and contrast C53 mediated autophagy to other known ER-phagy receptors. We believe the revised text allows readers to follow and judge our findings much better.
Regarding the protein levels, for the in vivo competition experiments, we used native antibodies against ATG8 and UFM1. For C53 flux analyses, we performed assays using native C53 antibodies, native promoter driven C53 transgenic lines and Ubiquitin10 promoter driven transgenic lines. In all cases, we have obtained comparable results. Nevertheless, we have added new confocal microscopy data that shows that “native” promoter driven C53 forms similar levels of puncta when compared to “Ubiquitin10” promoter driven C53 lines (Presented in Figure 2—figure supplement 1).
Moreover, they should describe how many AIMs are outside of the IDR region, how were they identified, and how many were tested.
We are absolutely confident of our assessment that C53 interacts with ATG8 via the sAIMs. We have reached to this conclusion by (i) truncation experiments and ITC assays that showed that the IDR of both human and Arabidopsis C53 is necessary and sufficient to bind to ATG8 (Figure 4), (ii) native mass spectrometry experiments that showed that C53sAIM mutants are unable to form a complex with ATG8 (now Figure 4—figure supplement 2), (iii) SPR experiments with full length C53 proteins that showed that C53sAIM mutants are unable to bind ATG8 (Figure 5), and (iv) complementation experiments in Arabidopsis and HeLa cells that showed that C53sAIM mutants are unable to form puncta that co-localize with ATG8 (Figure 5). CD spectroscopy and western blot experiments showed that sAIM mutants fold similar to the wild type proteins and are equally stable. So, the lack of puncta is not due to protein instability but loss of ATG8 interaction. Nevertheless, to address the raised point, we mutagenized all potential canonical AIM sites in the Arabidopsis C53 protein one by one and in a combinatorial manner. These experiments confirm that the canonical AIMs of C53 do not mediate ATG8 binding (Presented in Figure 4—figure supplement 1).
Finally, the authors should describe better what they mean by a "shuffled AIM" – is this a new ATG8-interaction motif or do they believe it is just a minor variation of the standard recognition element? These are examples – the whole manuscript would greatly benefit from more explicit explanations of experiments to make it more accessible to a larger audience.
As clearly indicated in the original manuscript, our in vitro data suggest sAIM binds ATG8 via the canonical LIR-docking site. However, while the critical AIM residues (IDWG) are maintained, their order is mixed up relative to the standard motif (should be W-D,G-I) and could not be reverted. Therefore, we decided to use the term “shuffled AIM” rather than defining it as a new ATG8 interacting motif. We modified the text to clarify this.https://doi.org/10.7554/eLife.58396.sa2
- Madlen Stephani
- Lorenzo Picchianti
- Tim Clausen
- Yasin Dagdas
- Yasin Dagdas
- Sascha Martens
- Armin Djamei
- Adriana Savova
- Alexander Gajic
- Emilio Skarwan
- Victor Sanchez de Medina Hernandez
- Azadeh Mohseni
- Marion Clavel
- Christian Loefke
- Alibek Abdrakhmanov
- Yasin Dagdas
- Sascha Martens
- Mateusz Matuszkiewicz
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
We thank R Strasser, M Schuldiner, R Kopito, J Christianson, K Mukhtar, S Howell, D Hofius, R Vierstra, Y Ye, W Yarbrough and T Ueda for sharing plasmids or plant lines. We acknowledge the Vienna BioCenter Core Facilities GmbH (VBCF) facilities Plant Sciences (J Jez), Electron Microscopy (T Heuser, S Jakob, M Brandstetter), and Protein Technologies (P Stolt-Bergner, A Sedivy, J Neuhold, A Lehner) as well as the GMI/IMBA/IMP Protein Chemistry Core facility (M Madalinski, E Roitinger, K Stejskal, R Imre) and A Schleiffer for their help with the experiments. The SPR and ITC equipment were kindly provided by the EQ-BOKU VIBT GmbH and the BOKU Core Facility Biomolecular and Cellular Analysis. We thank members of the Vienna Biocenter Ubiquitin Club for fruitful discussions. This work has been funded by the Vienna Science and Technology Fund (WWTF) through project LS17-047 (YD, TC), the Austrian Science Fund (FWF): P32355 (YD), P30401-B21 (SM), I3033-B22 (AD), Unidocs Fellowship (AS), ERC grant No.646653 (SM), and the Austrian Academy of Sciences. MM is financially supported by The Financial Supports for Young Scientists (WULS-SGGW) International Research Scholarship Fund No. BWM 315/2018. We thank Claudine Kraft, Elif Karagoz, James Watson and Youssef Belkhadir for critical evaluation of the manuscript. We also acknowledge Sabbi Lall, Life Science Editors for editing assistance.
- Suzanne R Pfeffer, Stanford University School of Medicine, United States
- Michael Rape, University of California, Berkeley, United States
© 2020, Stephani et al.
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.