Co‐chaperone BAG3 enters autophagic pathway via its interaction with microtubule associated protein 1 light chain 3 beta

The co‐chaperone BAG3 is a hub for a variety of cellular pathways via its multiple domains and its interaction with chaperones of the HSP70 family or small HSPs. During aging and under cellular stress conditions in particular, BAG3, together with molecular chaperones, ensures the sequestration of aggregated or aggregation‐prone ubiquitinated proteins to the autophagic‐lysosomal system via ubiquitin receptors. Accumulating evidence for BAG3‐mediated selective autophagy independent of cargo ubiquitination led to analyses predicting a direct interaction of BAG3 with LC3 proteins. Phylogenetically, BAG3 comprises several highly conserved potential LIRs, LC3‐interacting regions, which might allow for the direct targeting of BAG3 including its cargo to autophagosomes and drive their autophagic degradation. Based on pull‐down experiments, peptide arrays and proximity ligation assays, our results provide evidence of an interaction of BAG3 with LC3B. In addition, we could demonstrate that disabling all predicted LIRs abolished the inducibility of a colocalization of BAG3 with LC3B‐positive structures and resulted in a substantial decrease of BAG3 levels within purified native autophagic vesicles compared with wild‐type BAG3. These results suggest an autophagic targeting of BAG3 via interaction with LC3B. Therefore, we conclude that, in addition to being a key co‐chaperone to HSP70, BAG3 may also act as a cargo receptor for client proteins, which would significantly extend the role of BAG3 in selective macroautophagy and protein quality control.

of unfolded proteins. 1,27][8] Together with HSP70, its multidomain structure-consisting of the WW, PxxP and BAG domains, as well as its IPV and 14-3-3 motifs-enable BAG3 to serve as a hub for proteotoxicity-induced signaling. 9The cargo of small HSPs and HSP70s attaches to BAG3 via the IPV motifs and the BAG domain, respectively. 7,10,11nonically, BAG3-mediated selective macroautophagy (referred to as autophagy), requires the recognition of the cargo's ubiquitination status. 12[15] In most mammals, six genes encode for members of ATG8-family proteins.Three genes each encode for variants of either microtubule-associated proteins 1A/1B light chain 3 (LC3A, LC3B and LC3C) or γ-aminobutyric acid receptor-associated proteins (GABARAP, GABARAPL1 and GABARPL2).Conjugated to phosphatidylethanolamine, these are the only membrane-anchored representatives of ubiquitin-like proteins. 16st ATG8-interacting proteins comprise an ATG8-interacting motif (AIM, LC3 interacting region [LIR] or GABARAP interacting motif [GIM]), which is characterized by an aromatic residue and a branched hydrophobic sidechain ([F/W/Y] 0 XX[L/I/V] 3 ) separated by two amino acids. 17In many cases, this core motif is flanked by acidic or polar residues, predominantly N-terminally. 183][24][25] In the past years, few studies addressed the binding preferences and specificity of ATG8s. 21,26,27Thus far, some ligand preferences have been characterized. 28e different binding modes and the varying preferences of ATG8 proteins illustrate the complexity in regulating the sequestration of autophagic cargo.Apart from the bona fide autophagy receptor p62, which links ubiquitinated cargo with ATG8 proteins, emerging data also indicate ubiquitin-independent autophagic degradation of cargo sequestrated by BAG3. 8,29A direct interaction of BAG3 with ATG8 proteins might explain this degradation route.Indeed, our previous in silico analyses have identified five potential LIR motifs in BAG3 that are conserved within the craniotes. 30In human BAG3, these include Y 86 PQL 89 , Y 93 IPI 96 , Y 205 ISI 208 , Y 247 HKI 250 and Y 451 LMI 454 .
Here, we further investigated BAG3 for its interaction with LC3 proteins.Our results strongly suggest a specificity of these in silico identified LIRs for LC3B.Inactivation of these LIRs significantly reduced BAG3 levels in autophagic vesicles.These findings suggest the potential function of BAG3 as an adapter for non-ubiquitinated autophagic cargo in addition to its co-chaperone activity, significantly expanding the role of BAG3 in selective autophagy.

| Antibodies
See Table 1 for the list of antibodies used.
T A B L E 1 List of all antibodies used.
Cells were transfected with three different plasmids each encoding a unique U6 driven gRNA sequence and a CMV driven Cas9 nuclease.
Forty-eight hours posttransfection, the cells were singularized into 96-well plates, and the knock-out was confirmed by sequencing and immunoblot analysis (Figure 1A).

| Autophagic vesicle purification
Autophagic vesicles were purified based on a novel purification method described previously. 32At least 1 Â 10 7 cells were collected using Trypsin/ethylenediaminetetraacetic acid (EDTA) and centrifuged at 306 Â g for 4 min.After resuspension in PBS supplemented with cOmplete™ EDTA-free (Roche), cell disruption was performed using a UP50H ultrasonic processor (Hielscher) for 3 Â 2 s with an amplitude of 60%.Samples were then centrifuged at 3000 Â g for 10 min at 4 C, supernatants were collected and centrifuged at 18 620 Â g for
BAG3-EGFP (enhanced green fluorescent protein) was defined as EGFP-positive (488 nm, BP 530/30) events and positivity was conducted according to the background given by a negative control transfected with an empty vector as well as stained with an APCconjugated GABARAP/GABARAPL1/GABARAPL2 antibody.Analyses were performed using the LSRFortessa SORP (BD Biosciences) flow cytometer and subsequently analyzed with FlowJo v10.6.1 (BD Biosciences).] with due regard to the background given by a negative control transfected with an empty vector.

| Immunoprecipitation
The quantitative levels of BAG3 within LC3B-positive structures were defined as the mean area [px 2 ] of up to 70 000 single colocalizing BAG3 spots.All additional Plot profiles were created using FIJI v2.1.0 33showing the gray value of the different fluorescence channels depending on the distance of a given region of interest.

| Peptide array
Spot synthesis for peptide array of human BAG3 (15mer peptides, three residues shift per spot, N-terminally acetylated) on cellulose membrane was acquired from Intavis Peptides Services (Tübingen, Germany).Protein binding was performed according to the manufacturer's specification using heterologously expressed and glutathione S-transferase (GST)-tagged LC3s (LC3A, Novusbio, H00084557-P01; LC3B, Novoprolabs, 509 467; LC3C, Abnova, H00440738-P01) at 2 μg/mL.Binding of the proteins was evaluated using an anti-GST-horseradish peroxidase conjugate (GE Healthcare, RPN1236).For comparative evaluation of affinities, the signal intensities of the respective LIR motif-containing peptides were averaged and normalized to the averaged intensity of the most intense motif.In the case of peptides containing two LIR motifs, the signals from overlapping peptides were excluded from evaluation.according to the manufacturer's instructions (Table 2).Mutagenesis was confirmed via sequencing.

| Proximity ligation assay
In situ protein-protein interaction was evidenced using Duolink ® proximity ligation assay (PLA; Merck, DUO920082, DUO92004, DUO92008).Cells were fixed with formaldehyde and permeabilized using ice-cold 90% methanol.The sample preparation was conducted according to the manufacturer's protocol with an amplification time of 150 min.Stained cells were imaged with a Zeiss LSM710 confocal microscope.Puncta per cell of at least three biological replicates were counted using ImageJ for three randomly selected areas with both EGFP-positive and -negative cells and statistically analyzed as indicated.

| Immunoblotting
Samples were mixed with reducing SDS-loading buffer, incubated for 5 min at 98 C and subjected to SDS-PAGE.Gels were then blotted onto nitrocellulose membranes and subsequently incubated in/blocked with 5% nonfat dried milk powder dissolved in Tris-buffered saline (TBS)-T (0.05% Tween-20) for 1 h at room temperature.Membranes were probed with the required primary (1:X00 in TBS-T) and peroxidase-conjugated secondary antibodies (1:1000 in TBST) as indicated.Detection was performed with an Amersham Imager 600 system (GE Scientific).

| In silico modeling
In silico modeling of ΔLIR-BAG3 based on wild-type BAG3 (AlphaFold model) was performed using SWISS-MODEL. 34The superposition of the two models was created using matchmaker tool implemented in UCSF Chimera. 35I G U R E 1 Analysis of BAG3-LC3B interaction.3 | RESULTS

| BAG3 interacts with LC3B in situ and favors the Y 86 PQL 89 motif in vitro
We generated a BAG3-deficient HeLa cell line using a CRISPR/Cas9 approach to provide a cell line with a defined background.The cells were characterized both at the genetic and protein level (Figure 1A).
First, we investigated the potential interaction of BAG3 with the bona fide ATG8 protein LC3B in cellulo by proximity ligation.This assay employs immunodetection coupled with an amplification of DNA probes to detect the proximity (max.40 nm distance) of two immunolabeled proteins.In approximation, this method only uncovers direct interactions of two different proteins in situ.We were thus able to identify an interaction between BAG3 and LC3B using BAG3-EGFP overexpression in a BAG3-deficient background (Figure 1B,E).We treated the cells with MG-132 to inhibit proteasomal degradation and induce the BAG3-mediated selective autophagy pathway, and additionally treated with BafA1 to raise the total quantity of autophagic vesicles.Furthermore, we incubated cells with PYR-41 36 to diminish ubiquitination.Negative controls with the single antibodies against BAG3 and LC3B only (Figure S1A), and the positive controls with well-described interaction partners-such as HSP70 for BAG3 and NBR1 for LC3B (Figure 1C,D)-validated the method for detecting the binding of BAG3 to LC3B in situ.The evidence of a BAG3-HSP70 interaction is restricted only to cells displaying BAG3-EGFP expression, whereas the LC3B-NBR1 interaction is independent of BAG3 expression (Figure 1C,D).We also summarized the analysis of puncta per cell of the proximity ligation and its statistical evaluation (Figure 1E).Accordingly, detection of the BAG3-LC3B complex, which also occurred in BAG3-positive cells only, is proof of principle for an in cellulo interaction.
Subsequently, before we evaluated which of the putative LIR motifs is functional, we performed an initial immunoprecipitation assay to verify which of the human ATG8 paralogs binds to wild-type BAG3.To this end, cells were treated with BafA1 and PYR-41 to induce BAG3-mediated selective autophagy and enhance the likelihood of binding.Interestingly, we found BAG3 to interact with all LC3 proteins, but with none of the GABARAP proteins (Figure S1C).We furthermore verified that the PYR-41 treatment did not disturb the autophagic flux in any relevant way (Figure S1D).Based on these data, we excluded the GABARAP proteins in subsequent experiments.
Next, we performed a BAG3 peptide array to assign binding to the five putative LIRs.Due to the lack of structural information on BAG3, a restriction to particular motifs in advance did not seem reasonable.We screened the amino acid sequence for binding regions using synthetic 15mer peptides by shifting each peptide by three residues (Figure 1F).specificity for LC3 binding, we decided to employ a ΔLIR-BAG3 with disabled functionality of all putative LIR motifs.Within this ΔLIR-BAG3 mutant the tyrosine residues of the LIR motifs were changed to serine to retain the free hydroxyl group of the side chain.In the case of Y 451 LMI 454 , the isoleucine residue was mutated to a threonine, to reduce possible effects on the integrity and function of the BAGdomain.In silico modelling analyses also did not reveal any effect of these mutations on the structure of BAG3 (Figure S2).
Subsequently, we reassessed the interaction of BAG3 with the bona fide autophagy marker LC3B via immunoprecipitation.Overexpressed wild-type BAG3 and a ΔLIR-BAG3 mutant (Figure S1B), were immunoprecipitated under baseline and proteostasis stress conditions as indicated (Figure 1I).Under basal conditions, we only observed a weak, nonspecific coprecipitation of LC3B.Upon induction of BAG3-mediated autophagy via MG-132, PYR-41 treatment 36 to block ubiquitination, and BafA1 supplementation, we observed a strong increase of coprecipitating LC3B (Figure1I,J).This induction of interaction was completely lost in the ΔLIR-BAG3 mutant (Figure 1J).Indeed, calculating a Δ of coimmunoprecipitated LC3B between treated and untreated cells resulted in significantly less coimmunoprecipitated LC3B in the ΔLIR-BAG3 mutant (Figure 1K).Both results indicate the functionality of one or more LIR motifs in the interaction of BAG3 and LC3B.

| Inactivation of predicted LIR-motifs within BAG3 affects BAG3-LC3B interplay in cellulo
Next, we analyzed the colocalization dynamics for both wild-type BAG3 and ΔLIR-BAG3 under proteostasis stress conditions as this appeared to be necessary for the induction of the interaction with LC3B (Figure 1I).Here, we included LC3A and LC3C in addition to LC3B as the peptide array displayed a potential, although weak, interaction with Y 86 PQL 89 or Y 205 ISI 208 peptides (Figure 1G).Considering that other members of the LC3 protein family may compensate a potential loss of LC3B in the autophagic system, as we have previously reported, 37 this was of particular interest.

Our analysis of the relative colocalization area of wild-type BAG3
with LC3B under MG-132 treatment clearly demonstrated a superposition compared with the empty vector control.Compared with LC3A (Figure S3A,C) colocalization appeared to be elevated for LC3B (Figure 2A,C).For LC3C, we only observed interference at the level of the background given by the empty vector control, which was associated with a massive variation in the relative signal (Figure S3B,D).
These results again suggest a certain selectivity for LC3B.Additional treatment with BafA1 slightly, but not significantly, elevated the relative area of colocalization with LC3A-or LC3B-positive structures (Figures 2C and 3C).The ΔLIR-BAG3 mutant exhibited an almost identical relative area of colocalization with LC3B as the wild type (Figure 2C).Here, we would like to mention that the colocalizing area itself did not provide any functional conclusion.For instance, merely the BafA1 treatment-induced increase in LC3-positive structures might intrinsically elevate the area of colocalization with ΔLIR-BAG3 without any functional significance.For assessing the functionality, solely the difference in the mean BAG3 concentration/intensity within LC3-positive structures is of significance.Accordingly, by considering parameters such as the mean BAG3 concentration (wild-type vs. ΔLIR), for example, the intensity within the colocalizing LC3 structures, we verified a functional component of the LIRs.In the first step, we compared the fluorescence intensity profiles of designated cell regions displaying a correlation of BAG3 and LC3B within all tested paradigms.However, the analyzed wild-type BAG3 fluorescence intensity closely correlated with the intensity peaks of LC3B (Figure 2B).This correlation was hardly evident for ΔLIR-BAG3.
Subsequently, we analyzed the relative amount of wild-type BAG3 or ΔLIR-BAG3, respectively, within the LC3B-positive structures (Figure 2D).For wild-type BAG3, we observed a significant increase in the mean abundance of BAG3 colocalizing with LC3B-positive puncta following BafA1 treatment.This specific induction did not occur for ΔLIR-BAG3 (Figure 2D).Accordingly, since the BafA1 treatment in general did not provoke an accumulation, this enrichment of BAG3 within LC3B-positive structures has to depend on the specificity of one or more predicted LIR motifs.Next, we addressed the ensuing questions about the localization of BAG3 in autophagosomes and a LIR-dependent targeting into these structures by analyzing the content of isolated autophagic vesicles.

| BAG3 accumulates in isolated autophagic vesicles in a LIR-dependent manner
The recent establishment of a purification protocol in our lab 32 enabled us to extract and enrich intact native autophagic vesicles from cultured cells by means of a combination of successive centrifugations, an antibody-based fluorescence tagging of ATG8-positive structures and a subsequent sorting via flow cytometry (Figures 3A   and S4).The different centrifugation/washing steps combined with fluorescent labeling of established autophagosomal marker proteins such as LC3B or GABARAP and, eventually, sorting via fluorescenceactivated cell sorter effectively reduced the unlipidated form of LC3B accompanied by enriched levels of lipidated LC3B (Figure 3B).In accordance with Schmitt et al. 32 the specific detection and sorting of ATG8-positive structures eventually resulted in purified vesicles containing prominent autophagic markers such as LC3B and SQSTM1/ p62 in the absence of cytosolic proteins (e.g., LC3B-I) or other contaminants like fragments of the cytoskeleton such as tubulin (Figure 3B).This enabled us to confirm a localization of BAG3 inside autophagic vesicles via flow cytometry.Using the specified protocol, we were able to confirm the presence of BAG3-EGFP in autophagosomes of transfected BAG3-deficient cells (Figure 3C).To test the LIR dependency of this autophagosome-localized BAG3 population, we transfected BAG3-deficient HeLa cells either with an empty vector (pN1), wild-type BAG3, or ΔLIR-BAG3.

| DISCUSSION
We have recently described the potential of BAG3 to serve as an autophagy receptor, since BAG3 features a surprisingly strong conservation of putative LIR motifs within the craniotes. 30Therefore, we investigated an interaction of BAG3 with LC3 proteins, especially signal may likewise partly be the result of unspecific binding due to the high frequency of hydrophobic residues 36 like in the PxxP domain within BAG3, which are also most frequent among the LC3s in LC3B, therefore, this was not further investigated.Whether these hydrophobic residues are accessible at all in the native protein is currently not evident due to the lack of structural information on BAG3.The effects of the ΔLIR-BAG3 mutant (Figure 1I-K) likewise indicate that such interactions, if they indeed exist within the intact BAG3 protein at all, play a minor role in comparison to the predicted LIRs.
Additionally, LC3C indicated a high affinity towards a Y 247 HKI 250 containing peptide in the array (Figure 1G,H).Since we only detected the peptide containing this motif at the free N-terminus of the spotted peptide, the absence of interaction for the other three peptides containing Y 247 HKI 250 might be due to a folding-related inaccessibility of the motif as well as due to unspecific binding.Therefore, this result of the peptide array is not conclusive.To circumvent ambiguities of the array, we decided to inactivate the potential LIR motifs altogether and to study them collectively.This allowed us to work out the selectivity for LC3B in HeLa cells (Figures 2 and S1C).Although LC3B is often considered the bona fide autophagy marker linking ubiquitin receptors and their cargo to the autophagosomal membrane, it is not strictly necessary for autophagosome formation or cargo recognition. 37Indeed, other LC3s may very well compensate for a loss of LC3B. 37Furthermore, the phagophore is not exclusively equipped with a single type of LC3.Still, LC3B and LC3C appear to work at opposing ends of the autophagic spectrum. 38,39Notably, HeLa cells in general have a reportedly low expression of LC3C with 0.1 nTPM according to the human protein atlas (proteinatlas.org)project. 40This might have compromised the traceability of an interaction with LC3C by the methods used.Further support for an LC3B selectivity might be provided by the peptide array, which likewise suggested the most intense binding for LC3B (Figure 1G).Collectively, our colocalization analysis regarding selectivity of a LIR-dependent interaction of BAG3 with LC3 proteins upon proteostasis stress suggests a preference for LC3B.A specific interference to LC3C was not ascertained (Figure S3B,D).Therefore, the ability of BAG3 to interact with at least LC3B underlines a potential function as an autophagy receptor.
We did not detect a significant LIR or treatment-specific effect on the proportional area of LC3 colocalizing BAG3 (Figure 2C).As discussed above, for example, a BafA1-induced increase in LC3B area or Previous data suggested that BAG3 may partake in autophagic processes independent of ubiquitin and may potentially also deal with non-ubiquitinated cargo. 8Therefore, we also looked at the amount of BAG3 in autophagic vesicles in presence of PYR-41 and a disabled E1 ubiquitin ligase activity.Inhibition of ubiquitination by PYR-41 also resulted in a similar reduction of BAG3 as the ΔLIR mutant.Although the reduced levels post-PYR-41 treatment indicate preferential functionality for ubiquitinated cargo, a function for nonubiquitinated cargo cannot be excluded.Since we could not detect such an effect for ΔLIR-BAG3 (Figure 3G), the effects are likely not just additive.Ubiquitination events have very complex effects on, for instance, autophagosome maturation 41 and, as recently demonstrated, proteasome inhibition promote alternative pathways beyond autophagy towards lysosomal degradation. 42Therefore a more detailed analysis beyond pharmacological inhibition of ubiquitination is required to reveal the functional aspects of ubiquitination-dependent amounts of BAG3 in autophagic vesicles.
In conclusion, our data strongly suggest that LIRs are required for efficient translocation of BAG3 into autophagosomes (Figure 3E,F).
Here, we were able to identify an on demand interaction of BAG3 with LC3B in the context of upregulated BAG3-mediated selective macroautophagy pathway.Therefore, we hypothesize that a LIRdependent translocation of BAG3 to autophagosomes may also affect sequestrated HSP70 cargo and drive its autophagic degradation.
chosen optical fields from three independent biological replicates.BAG3 and LC3 spots were detected using the Harmony High-Content Imaging and Analysis software (Perkin-Elmer) within the cytoplasm excluding positive signals in the nucleus.Colocalization was calculated for up to 200 000 single BAG3 spots for each condition and cell line for three biological replicates as colocalizing BAG3 spot area [px 2 ] in relation to total BAG3 spot area [px 2

BAG3- deficient
HeLa cells were transiently transfected via calcium phosphate precipitation.Transfected cells were washed 24 h posttransfection and the medium was refreshed.Treatment was administered in an appropriate manner to allow harvesting the cells 48 h after transfection.Expression constructs of wild-type human BAG3 and C-terminal EGFP-tagged BAG3 were used as previously published. 6Single point LIR mutants and eventually ΔLIR-BAG3 (Y 86 S, Y 93 S, Y 205 S, Y 247 S, I 454 T) were generated using, the Quik-Change Lightning Site-Directed Mutagenesis Kit (Agilent, 210 518) (A) BAG3 deficiency on protein level in HeLa cells confirmed by immunoblot.(B-D) Representative images of proximity ligation assay (PLA) of BAG3-LC3B interaction (cyan) (B), BAG3-HSP70 interaction (C) and LC3B-NBR1 interaction (D).All PLA were performed on BAG3-deficient cells transiently transfected with human BAG3-EGFP and treated with MG-132 (10 h, 25 μM), Bafilomycin A1 (4 h, 2 μM) and PYR-41 (4 h, 12.5 μM).Displayed are representative images obtained from three independent experiments.Magnification: 100Â.Scale bar: 10 μm.(E) Analysis of proximity ligation puncta per cell of (B-D) and of Figure S1A, Statistical significances were reported after two-way ANOVAs followed by a Benjamini-Hochberg post hoc tests."****" indicates p ≤ 0.0001.(F) Exemplary of the first and last three oligopeptides part of the BAG3 peptide array.With each peptide the primary structure of the oligopeptide shifts by three amino acids towards the C-terminus.(G) Membranes of the BAG3 peptide arrays incubated with the indicated recombinant human glutathione S-transferase (GST)-tagged LC3 protein.LC3 binding was detected using an anti-GST antibody.Oligopeptides containing the predicted LIR motif are highlighted with a unique color.(H) Heatmap comparing mean signal intensity of all peptides containing each LIR of the arrays displayed in G.The motif with the strongest mean signal intensity was defined as 100%.(I) Immunoprecipitations of BAG3-deficient cells transfected with empty vector (pN1), wild type (WT)-BAG3 or ΔLIR-BAG3.Cells were treated for 8 h with MG-132 (25 μM), PYR-41 (12.5 μM) and Bafilomycin A1 (2 μM) (M-P-B) or dimethyl sulfoxide (DMSO); 48 h posttransfection whole cell lysates were obtained and immunoprecipitated using an anti-BAG3 antibody.Red arrows indicate which signals were quantified.(J) LC3B/BAG3 ratio for WT-BAG3 and ΔLIR-BAG3, with and without treatment obtained from I (four independent experiments).(K) Accumulation of co-precipitated LC3B by M-P-B treatment in cells transfected with WT-BAG3 or ΔLIR-BAG3 obtained from I (four independent experiments).

2. 12 |
Statistical methodsGraphs and statistics were assembled using GraphPad Prism 9 (Graph-Pad Software, Inc.)All statistical analyses were performed according to normal distribution and variance differences by one-way or twoway analysis of variance (ANOVA) followed by original Benjamini and Hochberg post hoc tests.The results display mean ± standard deviation (SD) or standard error of the mean (SEM) as indicated.Statistical significance was accepted at a level of p < 0.05 ( p-value ≤ 0.05 = *, ≤0.01 = **, ≤0.001 = ***, ≤0.0001 = ****).
Figure 3D displays the total cell lysate (TL) applied for the autophagosomal isolation protocol.Despite minor variations, wild-type BAG3 and ΔLIR-BAG3 displayed similar expression levels and stability (Figure S1B).The specific gating strategy of the flow cytometry for the final separation of vesicles is illustrated in Figure S4.To define whether BAG3 accumulates in a LIR-dependent manner in autophagosomes, we comparatively analyzed the amount of wild-type BAG3 as well as ΔLIR-BAG3 in these autophagic vesicles.The results in Figure 3E clearly confirmed a high concentration of wild-type BAG3.Regarding an LIR-dependent targeting of BAG3 to autophagosomes, in a comparative analysis, we indeed observed that mutations of the LIR motifs resulted in a massive reduction of ΔLIR-BAG3 (Figure 3E,F) in isolated vesicles.For the wild-type, additional treatment with PYR-41 effected a similar reduction of BAG3 in autophagic vesicles.Interestingly, we did not observe a reduction in ΔLIR-BAG3 levels in autophagic vesicles following PYR-41 treatment (Figure 3G).Since PYR-41 disables cellular ubiquitination capabilities, the reduction of BAG3 observed in autophagosomes, at first glance, may indicate at least a partial dependency on ubiquitination of cargo in order for BAG3 to act as an autophagy receptor.Likewise, the static behavior of ΔLIR-BAG3 (with its loss of function in all LIRs) when treated with PYR-41 further solidifies BAG3's potential function as an F I G U R E 2 Colocalization analysis of wild type (WT) and ΔLIR-BAG3 with LC3B.(A) Representative images of BAG3-deficient cells overexpressing WT-BAG3 or ΔLIR-BAG3 upon MG-132 treatment (8 h, 10 μM) as well as dimethyl sulfoxide (DMSO; control) or Bafilomycin A1 (8 h, 2 μM) as indicated.Shown images are each representative for 15 slices of three independent experiments.BAG3 shown in green, the indicated LC3 in red.Nuclei were stained by DAPI.Magnification: 40Â.Scale bar: 20 μm.(B) Fluorescence intensity profiles of (A) showing the gray values of the given channels within designated cell regions defined by distance [μm].BAG3 shown in green, the indicated LC3 in red (C) Quantification of BAG3 colocalizing with indicated LC3 within the cells displayed in A. Shown is the relative area of BAG3 puncta colocalizing with LC3 normalized to the total BAG3 area.Statistics are depicted as mean ± SD of three independent experiments comprising up to 200 000 single BAG3 puncta for each condition and cell line; one-way ANOVA; no significant alterations were observed.(D) Relative mean area of colocalizing BAG3 puncta in LC3B-positive structures within the cells described in A. Statistics are depicted as mean ± SEM of up to 70 000 individual BAG3 puncta out of three independent experiments for each condition and cell line; one-way ANOVA followed by a Benjamini-Hochberg post hoc test; "*" indicates p ≤ 0.05 and "***" p ≤ 0.001.autophagy receptor.In summary, this approach confirmed a LIRdependent accumulation of BAG3 in autophagic vesicles, with only minor amounts of ΔLIR-BAG3 present, likely representing BAG3 as cargo and not as a receptor.
LC3B as a bona fide representative of ATG8 proteins, in vitro and in cellulo.Our data suggest a direct interaction of BAG3 especially with LC3B but barely with LC3C.All assays performed point to a LIR dependency of this interaction.The results of the peptide array only provided evidence of an interaction with LC3A and LC3B for Y 86 PQL 89 and Y 205 ISI 208 containing peptides.The concomitant strong conservation of the other predicted LIRs, Y 93 IPI 96 and Y 451 LMI 454 throughout vertebrate evolution may also be explained by these LIR's being part of the first IPV motif essential for interaction with small HSPs or being part of the BAG domain, respectively.Y 205 ISI 208 is equally a part of the second IPV motif.Interaction with an LC3 protein would presumably influence binding of small HSPs to BAG3.Possibly, this might represent a potential switch for interactions from HSPB6/8 to LC3 proteins.In particular, for LC3B, the peptide array exhibits presumably nonspecific non-LIR-containing interaction sites.Here, based on the recognized sequences, there are currently no data to elucidate whether these are noncanonical binding motifs or yet unknown interaction mechanisms.However, since these regions are not phylogenetically conserved, apart from conserved areas important for the PxxP or BAG domain, little data argues in favor of this avenue.Rather, the F I G U R E 3 BAG3 in isolated native autophagic vesicles.(A) Schematic illustration of autophagic vesicle isolation.TL, total lysate; P1-2, pellet fractions; S1-2, supernatants; AV, autophagic vesicles.Gating strategy is shown in Figure S4.(B) Analysis of purification progress by immunoblot; fractions as indicated in (A); (C) Flow cytometric analysis of autophagic vesicles within BAG3-deficient cells overexpressing BAG3-EGFP in comparison to an unstained negative control as well as an empty vector (pN1) control.Cells were treated with MG-132 (8 h, 10 μM) and Bafilomycin A1 (8 h, 2 μM).Shown images are representative for three independent experiments.Gating strategy was set as indicated in Figure S4.(D) Immunoblot analysis of BAG3 in BAG3-deficient cells overexpressing WT-BAG3 or ΔLIR-BAG3 under MG-132 (8 h, 10 μM), Bafilomycin A1 (8 h, 2 μM) and PYR-41 (8 h, 12.5 μM) treatment as indicated.(E) BAG3 detection in autophagic vesicles isolated from BAG3-deficient cells described in (D).(F,G) Statistical analysis of (E).BAG3 levels were normalized to LC3B as a loading control as well as BAG3 levels in total lysates shown in (C) to compensate differences in transfection efficiency.Statistics are depicted as mean ± SD of three independent experiments; one-way ANOVA followed by a Benjamini-Hochberg post hoc test; "**" indicates p ≤ 0.01 and "***" p ≤ 0.001.
cargo-mediated translocation to autophagosomes without a direct BAG3-LC3 interaction might have impacts on changes in the relative proportion of BAG3 colocalizing with LC3.Comparison of the fluorescence intensity profiles of wild type versus ΔLIR-BAG3 within designated cell region indicated-although both paradigms showed a correlation-an elevated specificity of wild-type BAG3 towards LC3Bpositive structures especially after BafA1-treatment (Figure2B).To identify the specific effects, we looked at the enrichment of BAG3 (mean area of BAG3-positive puncta) within the LC3B-positive structures (Figure2D).Here, a BafA1-inducible LIR dependency of the specific BAG3 enrichment became apparent.Whereas the concentration of wild-type BAG3 in LC3B-positive structures was significantly induced after BafA1 treatment, the absence of this effect for the ΔLIR-BAG3 clearly indicated a LIR-dependent translocation to these structures (Figure2D).The ΔLIR-BAG3 not only caused a collapse of the inducible BAG3 amount colocalizing in BAG3/LC3B-positive puncta, but furthermore massively decreased the quantity of BAG3 in autophagic vesicles (Figure3E,F).Our data suggest that about half of the BAG3 present in autophagic vesicles in the context of the selected treatment could be LIR-dependent.The minor amounts of ΔLIR-BAG3 in autophagic vesicles might also be explained via a residual translocation by interactions beyond the LIRs as suggested by the peptide array or as part of the ternary complex, for example, with HSP70 including its cargo, targeted via classical autophagy receptors.
List of primers used for mutagenesis.
30tative canonic LIR motifs we have identified previously, these regions are either not conserved at all, or part of the WW or BAG domain, which likely dictate conservation of these regions.30Therefore,weassumed that these interactions are nonphysiological and may be caused by the high hydrophobicity of the spotted peptides.As this array did not provide a clear preference for a single LIR orT A B L E 2