Recruitment of autophagy initiator TAX1BP1 advances aggrephagy from cargo collection to sequestration

Since TAX1BP1 is a crucial factor for the initiation of autophagosome biogenesis at the p62 condensates, we wanted to follow its dynamics in relation to p62 and NBR1 in cells. To this end, we endogenously tagged p62 with GFP, NBR1 with mScarlet and TAX1BP1 with iRFP. Comparison of the autophagic flux between the engineered and parental cell line revealed no significant difference in the stabilization of TAX1BP1, NBR1, p62 as well as the activated form of p62 phosphorylated at S349 by Bafilomycin (Baf) treatment (Fig. EV1A,B). Under basal conditions p62 and NBR1 colocalized extensively when imaged by spinning disk live-cell microscopy (Fig. 1B, left panel). By contrast, TAX1BP1 showed an even distribution throughout the cell and a low level of colocalization with p62 and NBR1 (Fig. 1B, left panel). However, treatment with the class III phosphatidylinositol 3-phosphate kinase (PI3KC3) inhibitor VPS34 IN1 (Bago et al, 2014), which blocks autophagosome biogenesis, induced foci formation for TAX1BP1 (Fig. 1B, right panel). These foci colocalized with a fraction of the p62 and NBR1 condensates (Fig. 1C,D). Consistently, TAX1BP1 also accumulated at p62 condensates in ATG14 KO, ATG7 KO and FIP200 KO cells (Fig. EV1C). The finding that TAX1BP1 is not a constitutive component of the p62 condensates and that we needed to arrest autophagosome biogenesis to detect robust colocalization of TAX1BP1 with these condensates was surprising because in a fully reconstituted system in which condensate formation is induced by mixing GST-4xUB, p62, and NBR1, TAX1BP1 is recruited to almost all condensates in a NBR1 dependent manner (Fig. 1E,F) (Turco et al, 2021; Zaffagnini et al, 2018). These results suggested that there is an unknown level of regulation for the recruitment of TAX1BP1 to the p62 condensates in cells. Our data further suggested that the recruitment of TAX1BP1 may coincide with induction of autophagosome biogenesis and thus the switch from cargo collection to its sequestration by autophagosomes.

To confirm the role of TAX1BP1 as a regulator of autophagic flux of p62 condensates in our cell system, we generated a TAX1BP1 knockout (KO) in the background of a GFP-p62 expressing cell line. Employing live-cell imaging, we found that the overall size and total number of p62 condensates was significantly increased in the TAX1BP1 KO cells compared to the parental cell line (Fig. 2A,B). The addition of the autophagy inhibitor VPS34 IN1 did not lead to a significant further increase in the number of p62 condensates in the TAX1BP1 KO cells, unlike in the parental cell line (Fig. 2B). Consistently, NBR1 levels were significantly increased in the TAX1BP1 KO cell line (Fig. 2C,D). Furthermore, TAX1BP1 KO cells showed a robust increase in the number of p62 condensates compared to the parental cell line upon inhibition of the proteasome by MG132 (Appendix Fig. S1A). Taken together, our data show that autophagic flux of p62 condensates is regulated by TAX1BP1 in these cells. Next, we probed for colocalization between p62 and the autophagy machinery by immunofluorescence. This revealed a significant reduction in colocalization between p62 condensates and the autophagy core components ATG9A and FIP200 in the absence of TAX1BP1 (Fig. 2E,F; Appendix Fig. S1B) (Turco et al, 2021), supporting the hypothesis that TAX1BP1 is the main recruiter of the autophagy machinery.

To assess whether TAX1BP1 also mediates the recruitment of other components of the selective autophagy machinery to the condensates, we focused on TBK1. The TBK1 kinase has emerged as a major pro-autophagic factor in selective autophagy and binds to TAX1BP1 via the NAP1/SINTBAD adapters (Fu et al, 2018). In addition, it has recently been shown that the depletion of TAX1BP1 results in reduced TBK1 activation under basal conditions (Yamano et al, 2024). Indeed, we found that phosphorylation of p62 at S403, which is mediated by TBK1 (Matsumoto et al, 2011; Pilli et al, 2012), was abolished in the TAX1BP1 KO cells. This was particularly evident upon inhibition of autophagy with VPS34 IN1, which resulted in the accumulation of phosphorylated p62 (Fig. 3A,B). We also observed a significant reduction in phosphorylation of the TBK1 adapter protein NAP1 (Fig. 3A,C). Thus, we speculated that TAX1BP1 might promote the recruitment of TBK1 to p62 condensates. Consistently, we found that NAP1 showed a significant reduction in colocalization with p62 condensates in the TAX1BP1 KO cells compared to the parental cell line (Fig. EV2A,B). In addition, the colocalization of p62 condensates with active TBK1, phosphorylated at S172, was also reduced in the TAX1BP1 KO cells (Figs. 3D,E and EV2C). Since the TBK1 adapters NAP1 and SINTBAD are also required for the recruitment of TBK1 to p62 condensates (Fig. EV2D), we speculated that TAX1BP1 recruits TBK1 to p62 condensates via these adapter proteins. To test this more directly, we reconstituted these steps in vitro. We incubated p62, NBR1, TAX1BP1, the TBK1 adapters SINTBAD or NAP1 as well as TBK1. The addition of GST-4xUB induced p62-driven condensation and revealed that TAX1BP1 is both necessary and sufficient to recruit TBK1 to p62 condensates via the NAP1/SINTBAD adapters (Figs. 3F–J and EV2E–H). SINTBAD and TBK1 appeared to be equally distributed within the p62 condensates, in contrast to TAX1BP1, which showed enrichment at the center of the condensates in vitro (Fig. EV2F). Taken together, our data showed that TAX1BP1 plays a major role in mediating TBK1 recruitment to p62 condensates and thus potentially promotes the local phosphorylation of p62 and other components in these structures (Fig. 3K).

How TAX1BP1 is recruited to the p62 condensates is not well understood. While it was shown that the interaction with NBR1 requires a central part of TAX1BP1, referred to as N-domain (aa420-506) (Ohnstad et al, 2020), the mechanistic basis of this interaction remains unexplored. We previously showed that the FW domain of NBR1 is sufficient for the interaction with TAX1BP1 (Turco et al, 2021). However, the FW domain is not essential because its deletion did not abolish the binding to TAX1BP1. We therefore set out to systematically define the interaction between TAX1BP1 and NBR1 with the aim to reveal the basis of a potential regulatory mechanism. We started by testing if the TAX1BP1 N-domain is not only required but also sufficient for the interaction with NBR1. Employing a microscopy-based protein–protein interaction assay we found that the N-domain is indeed sufficient to recruit NBR1 (Figs. 4A,B and EV3A). Next, we determined the corresponding binding region in NBR1. Multiple attempts to define this interaction site using AlphaFold 2 gave inconclusive results. We therefore used a peptide array where we spotted NBR1 peptides on a membrane. Each peptide consisted of 12 amino acids with two residues overlap between peptides. We then probed for interacting peptides with recombinant TAX1BP1. We identified two regions capable of binding to TAX1BP1: parts of the FW domain, previously implicated to bind TAX1BP1 weakly (Turco et al, 2021) and the LC3-interacting region (LIR) 1 (Kirkin et al, 2009) (designated as binding region 1 and 2, respectively; aa347–aa365 and aa731–aa749) (Fig. 4C). Next, we tested if the deletion of the identified binding regions would abrogate the binding between NBR1 and TAX1BP1. Transfection of HeLa PentaKO cells (lacking p62, NBR1, TAX1BP1, NDP52, and OPTN (Lazarou et al, 2015)) with corresponding deletions of NBR1 and subsequent pulldowns from the cell lysates using GST-TAX1BP1 immobilized on GSH beads, demonstrated that binding region 2, encompassing the LIR1 of NBR1, is required for the interaction while binding region 1 is not (Fig. 4D). We also confirmed that a small fragment of NBR1 (referred to as CC2-LIR1; aa685– aa757), which contains the binding region 2, alone is sufficient for the interaction with TAX1BP1 in vitro (Fig. EV3B,C). To obtain further insights into the interaction of the NBR1 LIR1 region with the N-domain of TAX1BP1, we performed AlphaFold 2 modeling with sequences of TAX1BP1 and NBR1 encompassing the identified binding sites (Figs. 4E and EV3D)(Jumper et al, 2021). Based on the highest-ranking model, we generated point mutants for both TAX1BP1 and NBR1 and assessed their impact on the interaction of the two proteins. TAX1BP1 point mutants were validated in a microscopy-based interaction assay (Figs. 4F,G and EV3E). NBR1 point mutants were validated by performing a pulldown with cell lysates (Fig. 4H). Residues Y449 and L442 in TAX1BP1 and F740 of NBR1 were identified to be essential for the interaction between the two proteins (Figs. 4F–H and EV3E), consistent with our AF2 model. Given the proximity of residue F740 to the LIR1 motif, we tested whether it is also required for binding to GABARAP/LC3 proteins. To this end, we performed a pulldown with GST-LC3B and detected only a mild reduction in LC3B binding for this mutant (Fig. EV3F).

Based on the discovery that the LIR1 region of NBR1 is essential for the binding to TAX1BP1, we speculated that the interaction of NBR1 with TAX1BP1 and ATG8 proteins such as GABARAP might be mutually exclusive. To test this, we co-incubated GFP-NBR1 coated beads simultaneously with mCherry-TAX1BP1 and increasing amounts of GABARAP. Binding of TAX1BP1 to NBR1 was gradually reduced by the addition of increasing concentrations of GABARAP (Fig. 5A,B; Appendix Fig. S2A). The addition of GABARAP also outcompeted an already established TAX1BP1–NBR1 interaction in vitro (Fig. 5C,D). Next, we asked if GABARAP would also reduce the recruitment of TAX1BP1 to p62 condensates. Indeed, the presence of GABARAP notably reduced the fraction of TAX1BP1-positive p62 condensates in our condensation assay (Fig. 5E,F; Appendix Fig. S2B).

Thus, our findings delineate the interaction between NBR1 and TAX1BP1, revealing that it involves the LIR1 region of NBR1 and that the interaction can be outcompeted by high concentrations of LC3/GABARAP proteins.

To test the relevance of the interaction between TAX1BP1 and NBR1 for the autophagic degradation of p62 condensates in cells, we re-introduced fluorescently tagged wild-type (WT), ∆N (lacking the N-domain) and L442A TAX1BP1 into the TAX1BP1 KO cell line by viral transduction (Fig. EV4A). After doxycycline-induced expression, we observed that in contrast to TAX1BP1 WT (R-TAX1BP1) and consistent with our in vitro data, TAX1BP1 ∆N and L442A (R-TAX1BP1∆N, R-TAX1BP1 L442A) failed to colocalize with p62 condensates upon treatment with VPS34 IN1 (Fig. 6A,B). Intriguingly, even upon overexpression, TAX1BP1 did not colocalize with p62 puncta and the number of TAX1BP1 foci upon VPS34 IN1 remained lower than for p62 (Figs. 6A and EV4B). Furthermore, re-expression of TAX1BP1 WT rescued the phenotype of the TAX1BP1 KO reducing the number of p62 condensates, while the mutants failed to do so (Fig. 6C). In addition, the TAX1BP1 mutants were unable to restore TBK1-mediated phosphorylation of p62 and the adapter protein NAP1 (Fig. 6D–F). Doxycycline-induced re-expression of TAX1BP1 WT also reduced the NBR1 protein levels, in contrast to the TAX1BP1 mutants (Fig. 6D,G). Moreover, re-expression of TAX1BP1 rescued the recruitment of FIP200 and ATG9 to p62 condensates (Fig. EV4C).

To test the impact of the F740A mutant of NBR1 on TAX1BP1 recruitment, we first generated a GFP-p62 expressing NBR1 KO cell line. Consistent with previous reports, there was a major loss of p62 condensates in the NBR1 KO (Fig. EV4D,E) (Turco et al, 2021). Re-expression of mScarlet-NBR1 (R-NBR1) rescued the phenotype and enabled p62 condensate formation in the cells (Fig. EV4D,E). Similar to the NBR1-binding deficient TAX1BP1 mutants described above, introduction of the F740A point mutant into NBR1 (R-NBR1 F740A) increased the number of p62 condensates significantly in comparison to the NBR1 WT (Fig. 6H,I). The NBR1 F740A mutant also showed no further increase in the number of p62 condensates upon treatment with the VPS34 inhibitor (Fig. 6H,I). In contrast to NBR1 WT, the point mutant also failed to recruit TAX1BP1 to the p62 condensates (Fig. 6H,J).

Taken together, the interaction between TAX1BP1 and NBR1 is crucial for TBK1 recruitment and to promote autophagy flux of p62 condensates in cells.

Based on the insights we obtained regarding the TAX1BP1–NBR1 interaction, we aimed to understand how the TAX1BP1 recruitment and by implication the initiation autophagosome biogenesis is regulated. p62 condensates harbor two known binding cues for TAX1BP1: ubiquitin, which is bound by TAX1BP1 via its C-terminal ZnF domain (EV5A–C) (Sarraf et al, 2020; Tumbarello et al, 2015) and the region around the LIR1 motif of NBR1, which we identified above. To identify the respective contribution of each individual binding cue for the recruitment of TAX1BP1 to p62 condensates, we reconstituted this step in vitro (Fig. 7A). Deletion of the N-domain completely abolished the recruitment of TAX1BP1 to p62 condensates, while the N-domain alone was still recruited, albeit to a lower level (Figs. 7B,C and EV5D). Conversely, deletion of the ZnF domain resulted in a loss of ubiquitin binding (Fig. EV5A–C) and a decreased TAX1BP1 recruitment to p62 condensates in vitro (Fig. 7B,C), consistent with recent findings (Ferrari et al, 2024). The levels of recruitment of TAX1BP1ΔZnF and of the isolated N-domain were similar, suggesting that the residual recruitment of TAX1BP1ΔZnF is mediated by the N-domain (Figs. 7B,C and EV5D). To further corroborate our in vitro results, we re-introduced TAX1BP1∆ZnF (labeled as R-TAX1BP1∆ZnF) into our TAX1BP1 KO cell line in a doxycycline-inducible manner. In line with our in vitro result, we observed a mild reduction in TAX1BP1 puncta overlapping with p62 and a weaker signal for phosphorylated NAP1 and phosphorylated p62 in comparison to the TAX1BP1 WT (Fig. 7D–H). These results suggest that under the conditions tested, the N-domain-mediated interaction with NBR1 is essential for TAX1BP1 recruitment, while the ZnF-mediated recruitment via ubiquitin merely enhances the recruitment or stabilizes the interaction.

Since we found above that the binding of TAX1BP1 to NBR1 is weakened in the presence of GABARAP (Fig. 5), we tested whether the ZnF-mediated interaction with ubiquitin aids in overcoming the competition with GABARAP for efficient TAX1BP1 recruitment. We performed reconstitution experiments using cargo-mimetic beads coated with GST-4xUB and compared the recruitment of GFP-TAX1BP1 full-length and ∆ZnF to the beads in the presence of p62, NBR1 and increasing amounts of GABARAP. To ensure that ubiquitin remains a limiting factor and that TAX1BP1 recruitment depends also on NBR1 (as observed in cells (Fig. 6A–C)), we employed low amounts of ubiquitin and pre-coated the beads with mCherry-p62 and NBR1. Indeed, under the conditions tested, the full-length TAX1BP1 protein was robustly recruited to the beads even in the presence of high concentrations of GABARAP while TAX1BP1∆ZnF was barely detectable (Figs. 7I and EV5E). We also performed condensation assays with longer ubiquitin chains (GST-8xUB vs GST-4xUB +/− GABARAP) to mimic a higher ubiquitin load on the cargo and found that more ubiquitin on the cargo rescued TAX1BP1 recruitment to p62 condensates in the presence of GABARAP in vitro (Fig. 7J,K). This suggests that due to their competition with TAX1BP1 for NBR1 binding, the GABARAP/LC3 proteins render the ubiquitin interaction more important for TAX1BP1 recruitment. Hence, we propose that TAX1BP1 is recruited to p62 condensates once a certain local ubiquitin threshold is surpassed, to trigger the switch from cargo collection to autophagosome biogenesis when the cargo load reaches a sufficient level (Fig. 8).

All mammalian cell lines were cultured at 37 °C in a humidified 5% CO₂ atmosphere. HeLa PentaKO cells were obtained from Michael Lazarou (Lazarou et al, 2015). HEK293T cells were purchased from the American Type Culture Collection (ATCC). HAP1 cells were originally purchased from Horizon Discovery. GFP-p62 and GFP-p62 mSc-NBR1 were previously generated and used as parental cell lines in this study (Turco et al, 2021; Zaffagnini et al, 2018). HeLa and HEK293T cells were grown in Dulbecco Modified Eagle Medium (DMEM, 41966-029, Thermo Fisher) supplemented with 10% (v/v) fetal bovine serum (FBS, Thermo Fisher), 25 mM HEPES (15630080, Thermo Fisher), 1% (v/v) non-essential amino (NEAA, 11140050, Thermo Fisher), and 1% (v/v) penicillin–streptomycin (15140122, Thermo Fisher). HAP1 cells were cultured in Iscove’s Modified Dulbecco’s Medium (IMDM, 31980-022, Thermo Fisher), supplemented with 10% (v/v) fetal bovine serum (FBS, Thermo Fisher) and 1% (v/v) penicillin–streptomycin (15140122, Thermo Fisher). Cell lines were regularly tested for mycoplasma contaminations.

Sf9 insect cells (12659017, Thermo Fisher) were cultured as a shaking liquid culture at 27 °C in insect culture medium (96-001-01, ESF 921 Expression System LLC) supplemented with 1% penicillin–streptomycin (15140122, Thermo Fisher).

The following chemical compounds have been used in this study: Bafilomycin A1 at 400 nM for 3 h (sc-201550, Santa Cruz Biotech), Doxycycline at 50 ng/ml for 16 h (D9891, Sigma), DMSO (D2438, Sigma), Puromycin at 5 µg/ml (A11138-03, Thermo Fisher), VPS34 IN1 at 2 µM for 3 h (APE-B6179, ApexBio). siRNAs were purchased from Dharmacon as a pool of 4 individual siRNAs and used at a final concentration of 20 nM.

The sequences for all the inserts were generated by amplifying existing plasmids or through gene synthesis (GenScript). Plasmids were generated either by Gibson assembly or gene synthesis. For Gibson assembly, the vector and inserts were generated by restriction digestion or PCR. DNA was cleaned up with Promega Wizard SV gel and PCR Cleanup System (A2891, Promega). The inserts and the backbone were mixed in a 3:1 ratio and incubated with the 2x NEBuilder HiFi DNA assembly enzyme mix (M5520A, New England Biolabs) for 1 h at 50 °C. Competent DH5a E.coli were transformed with the Gibson reaction and incubated overnight at 37 °C. The next day single colonies were picked, grown overnight, and plasmids were isolated using the GeneJet Plasmid Miniprep Kit (K0503, Thermo Fisher). For sequence verification, the isolated plasmids were sent for Sanger sequencing (MicroSynth AG).

sgRNAs for generating HAP1 GFP-p62 mSc-NBR1–TAX1BP1-iRFP cell line were designed by CRISPOR (http://crispor.tefor.net). Both sgRNAs were cloned into an all-in-one vector encoding Cas9D10A nickase (Addgene) with puromycin as a selection marker. The plasmid bearing the repair template was generated by gene synthesis (GenScript). All-in-one vector and repair template were transfected into GFP-p62 mSc-NBR1 (previously validated in (Turco et al, 2021)) using Fugene 6 (E2691, Promega). After selection with puromycin, the remaining cells were single-cell sorted by FACS and validated by PCR, western blot and Sanger sequencing. The selected clone was also validated for protein levels and autophagy flux.

The same sgRNAs for generating the TAX1BP1 KO and NBR1 KO cell line has been used as previously reported (Lazarou et al, 2015; Sarraf et al, 2020) and cloned into pSpCas9(BB)-2A-Puro vector backbone. The plasmids were transfected into HAP1 GFP-p62 cells (previously described in (Zaffagnini et al, 2018)). After selection with puromycin, cells were single-cell sorted by FACS. Individual clones were validated using PCR, western blots and Sanger sequencing.

For stable cell line generation, lentiviral transductions were performed. HEK293T cells were transfected with VSV-G, GAG-POL and lentiviral plasmids carrying the transgene using Lipofectamine 3000 (L3000008, Thermo Fisher). After 24 h, the medium of the transfected cells was exchanged to the appropriate medium for the target cells. HAP1 GFP-p62 TAX1BP1 KO and HAP1 GFP-p62 NBR1 KO cells were seeded on a 6 well plate at 400k cells/well. 24 h later the medium containing the lentivirus was collected and filtered with a 0.45-µm filter (729025, Macherey-Nagel). In total, 8 mg/ml polybrene (Sigma) was added to the viral supernatant. The target cells were incubated for 24 h with the virus containing medium. The cells were washed with PBS on the next day and virus-free medium was added back to the cells. After selection with antibiotics, the expression of the transgene was validated by western blot.

For western blot analysis, cells were harvested using trypsin and washed 2x with PBS. Cells were resuspended in an appropriate volume of lysis buffer (50 mM Tris-HCl pH 7.4, 1 mM EGTA, 1 mM EDTA, 1% Triton X-100, 0.27 M sucrose, 1 mM DTT, cOmplete EDTA-free protease inhibitor (11836170001, Roche)) and incubated for 30 min on ice. Lysates were cleared by centrifugation and protein concentration was determined by Bradford (Biorad). Overall, 10 µg of protein was boiled for 5 min at 95 °C and then loaded on a 4–15% SDS gel (4561086, Biorad).

Proteins were transferred on a nitrocellulose membrane (10600001, Cytiva) by wet blotting (Biorad) at 120 V for 60 min. Membranes were blocked with 3% non-fat dry milk in TBS + 0.1% Tween-20 (blocking buffer) for 1 h at RT. Primary antibodies were added for overnight incubation at an appropriate dilution (see below). The next day membranes were washed 3× with TBS + 0.1% Tween-20 (TBST) and incubated with the secondary antibody diluted in blocking buffer. After three washes with TBST, membranes were developed with Super Signal West femto chemiluminescence substrate (34096, Thermo Fisher). Images were taken with a ChemiDoc MP Imaging system (Bio-Rad). Antibodies dilutions used for western blot: rabbit anti-AZI2/NAP1 (Abcam 1:1000), mouse anti-NBR1 (Abnova 1:1000), mouse anti-GAPDH (Sigma, 1:25,000), mouse anti-p62 (BD Bioscience, 1:1000), rabbit anti-p62 p-S349 (Cell Signaling, 1:1000), rabbit anti-p62 p-S403 (Cell Signaling, 1:1000), rabbit anti-TAX1BP1 (Cell Signaling, 1:1000). Secondary antibodies: goat anti-rabbit HRP (Jackson Immunoresearch, 1:10,000), goat anti-mouse HRP (Jackson Immunoresearch, 1:10,000).

For quantification, ImageJ was used. A rectangle was drawn around the corresponding protein band and the intensity was plotted using the “Plot Lanes” function. Measured values were adjusted and corrected by GAPDH as a loading control.

Cells were seeded in 10-well glass-bottom imaging chambers (543079, Greiner Bio-One) at 6000 cells/ well. 48 h later, cells were treated and imaged using a Visitron Live Spinning Disk microscope (Plan-Apochromat 63×/1.4 Oil DIC objective and an EM-CCD camera; including a temperature & CO₂-controlled environment). Standard imaging settings were a gain of 500 for all channels and 100 ms exposure for GFP & mScarlet and 200 ms for iRFP. Laser settings were the following: 5% laser power for GFP, 10% laser power for mScarlet and 10% laser power for iRFP. For dox-inducible TAX1BP1 constructs, iRFP settings were a gain of 500, 150 ms exposure and 15% laser power. Settings were kept consistent across replicates. Multiple areas within the wells were imaged to obtain comparable numbers of cells (50–150 cells). No blinding was performed during image acquisition.

For quantifications a macro for ImageJ was used as previously described (Turco et al, 2021). In short, binary pictures were generated by thresholding. Analyze punctae with a cut off of “>3 pixel units” was used to count puncta and to save coordinates as ROIs. ROIs from one channel were overlayed with the other channel(s) to determine the degree of colocalization. Threshold was determined manually and was kept consistent across replicates. Cells with a significant amount of background noise were excluded from the analysis.

Immunofluorescence was performed similar to previous studies (Turco et al, 2021). In short, cells were seeded in six-well plates on coverslips (0117520, Marienfeld Superior). After 24 h the cells were treated, washed 3× with PBS and fixed with 4% Paraformaldehyde in PBS for 20 min at RT. After fixation, the cells were washed 3× with PBS and permeabilized with 0.1% Triton X-100 in PBS for 5 min at RT except for experiments that included FIP200. For FIP200, cells were permeabilized with 0.25% Triton X-100 in PBS for 15 min at RT. Cells were washed 2× with PBS, incubated with blocking buffer (1% BSA in PBS) at RT for 1 h and incubated with the primary antibodies for 1 h at RT in a humid chamber except for FIP200. For FIP200, coverslips were incubated immediately with the primary antibody in a humid chamber for 1 h at 37 °C without a previous blocking step. The cells were washed 3x with PBS and incubated with the secondary antibody for 1 h at RT in a humid chamber. Cells stained with FIP200 were incubated for 1 h at 37 °C in a humid chamber with the secondary antibody. All cells were washed 3× with PBS and mounted on glass slides (AA00000112E01MNZ10, epredia) by transferring them onto a droplet of the mounting medium DAPI-Fluoromont-GTM (0100-200, Southern Biotech).

Antibody dilutions (in blocking buffer): rabbit anti-ATG9 (1:100, Cell Signaling), rabbit anti-FIP200 (1:200, Cell Signaling), mouse anti-p62 (1:100, BD Bioscience), rabbit anti-AZI2/NAP1 (1:100, Abcam), rabbit anti-TAX1BP1 (1:100, Cell Signaling), anti-TBK1 p-S172 (1:100, Cell Signaling), goat anti-rabbit Alexa Fluor 546 (1:1000, Invitrogen), goat anti-mouse Alexa Fluor 488 (1:1000, Invitrogen), goat anti-rabbit Alexa Fluor 647(1:500, Jackson ImmunoResearch).

Images were taken with a Zeiss LSM 700 confocal microscope equipped with Plan-Apochromat 63×/1.4 Oil DIC objective. Multiple areas within samples were imaged to obtain comparable numbers of cells (50–150 cells). No blinding was performed during image acquisition. To avoid cross-contamination between fluorochromes, each channel was imaged sequentially using the multitrack recording module before merging. Images from fluorescence and confocal acquisitions were processed and analyzed with ImageJ software.

For quantification, a macro for ImageJ was used as described before (Turco et al, 2021). In short, binary pictures were generated by thresholding. Analyze puncta with a cut off of “>3 pixel units” was used to count the number of puncta and to save coordinates as ROIs. ROIs from one channel were overlayed with the other channel(s) to determine the degree of colocalization. The threshold was determined manually and was kept consistent across replicates. Cells with a significant amount of background noise were excluded from the analysis.

N-domain (aa420-506 of TAX1BP1), GFP-N-domain, Tetra-ubiquitin (4xUB), GFP and GFP-CC2-LIR1 were cloned into a pGEX -4T1 plasmid vector backbone. mCherry-p62 S403E was cloned into the pET vector backbone. Octa-ubiquitin (8xUB) was generated by gene synthesis and cloned into the pGEX-4T1 vector backbone by GenScript. After transformation into competent E. coli Rosetta pLysS cells (70956-4, Novagen), a single colony was picked and incubated overnight in 100 ml of LB including the appropriate selection marker. The next day, the cultures were scaled up to the range of liters. The cultures were incubated at 37 °C until an OD₆₀₀ of 0.6 was reached and then the expression was induced by addition of 0.2 mM isopropylthiogalactoside (IPTG). The cells were harvested after 18–20 h at 18 °C and pellets were flash-frozen in liquid nitrogen and stored at −70 °C. For mCherry-p62 S403E, the cells were incubated until they reached the OD 0.8. The expression was induced with 0.2 mM IPTG and the cells were harvested after 5 h shaking at 25 °C, flash-frozen and stored at −70 °C.

10x-His-GFP-TAX1BP1, 10x-His-GFP-TAX1BP1∆N, 10x-His-GFP-TAX1BP1∆ZnF, 10x-His-mCherry-TAX1BP1, 10x-His-mCherry-TAX1BP1∆N, GST-TAX1BP1, GST-TAX1BP1∆ZnF, GST-TAX1BP1∆N, GST-TAX1BP1 L442A, GST-TAX1BP1 Y449A were cloned into the pLIB vector backbone for insect cell culture expression. Codon-optimized GFP-TBK1, TBK1, SINTBAD and SINTBAD-GFP were generated by gene synthesis (GenScript) for insect cell expression. The plasmids were transformed into electrocompetent DH10BacY cells (Vijayachandran et al, 2011). For the transformation, 1 µl of the plasmids at a concentration of 25 ng/µl were mixed with bacteria and electroporated using a Biorad Micropulser at the appropriate settings for bacteria (2.5 kV potential difference). For recovery, the electroporated cells were mixed with 5 ml of LB and incubated for at least 5 h at 37 °C. Afterward, the cells were plated on LB agar plates containing 50 µg/ml Kanamycin, 7 µg/ml Gentamicin, 10 µg/ml Tetracycline, 200 µg/ml X-Gal and 165 µM IPTG. White colonies were screened on the next day by colony PCRs. Positive colonies were picked and incubated with 10 ml of LB and appropriate antibiotics overnight at 37 °C. The next day the bacmids were isolated and the DNA measured. The bacmids were used freshly for insect cell transfections. 1 million of Sf9 insect cells were seeded on 6 well plates. 5 µg of DNA and 5 µl of FuGene HD (E2311, Promega) was used for transfection. The expression of the reporter YFP was tracked over several days by microscopy. Once most cells showed YFP expression (usually after 5 days), the cell suspension (V0) was added to 30 ml of insect cells at 1 mio/ml. Viability of cells and YFP expression was monitored for several days. Once, the cells showed viability drop to below 90%, the cells were spun down and the supernatant was filtered with a 0.45-µm filter (729025, Macherey-Nagel) and stored at 4 °C. For expression, 1 ml of the V1 was added to 900 ml of Sf9 insect cells at 1 mio/ml. Once the viability dropped below 90%, the cells were harvested, washed once with PBS, flash-frozen and stored at −70 °C.

GFP-NBR1 and NBR1 were cloned into the pCAGGSsH-C vector and expressed in HEK293 via the VBCF ProTech facility (https://www.viennabiocenter.org/vbcf/protein-technologies/). Pellets were stored at −70 °C.

mCherry-p62 S403E, GST-4xUB, GFP-NBR1, NBR1, GFP-TAX1BP1 FL, mCherry-TAX1BP1 FL were purified as described previously (Ferrari et al, 2024; Turco et al, 2021; Zaffagnini et al, 2018).

Truncations and point mutants of TAX1BP1 were purified following the same procedure as for the full-length TAX1BP1 (FL) protein.

GFP-TBK1, TBK1, SINTBAD, and SINTBAD-GFP were also purified as previously described (Adriaenssens et al, 2024).

GST-N-domain, GFP-N-domain, GST-GFP-CC2-LIR1, and GST-8xUB were purified from bacterial pellets. Pellets corresponding to 2 L culture were resuspended in lysis buffer (50 mM HEPES, pH 7.5, 300 mM NaCl, 1 mM DTT, 2 mM MgCl₂, cOmplete EDTA-free protease inhibitor (Roche), DNase (Sigma),) and lysed by sonication. To clear the lysate, the suspension was centrifugated with 18,000 rpm at 4 °C. The supernatant was filtered through a 0.45-µm filter (6880-2504, Cytiva) and incubated with 3 ml pre-equilibrated GSH beads (17075605, GE Healthcare). After 1 h incubation rolling at 4 °C, the beads were washed 5× with Wash Buffer I (50 mM HEPES, pH 7.5, 300 mM NaCl, 1 mM DTT), once with Wash Buffer II (50 mM HEPES, pH 7.5, 700 mM NaCl, 1 mM DTT) and again 2× with Wash Buffer I. For elution, beads were either resuspended in 3 ml Wash Buffer I + TEV protease for cleavage overnight or by incubation with elution buffer (Wash Buffer I + 50 mM glutathione (Sigma) - pH adjusted to 7.5). The next day, the eluate was filtered through a 0.45-µm filter, concentrated to 500 µl using an appropriate protein concentrator (Merck Millipore), and further purified by size exclusion chromatography on a Superdex 200 Increase 10/300 column (Cytiva) in SEC Buffer (25 mM HEPES pH 7.5, 150 mM NaCl, 1 mM DTT).

Microscopy-based interaction assay was performed as previously described (Sawa-Makarska et al, 2020; Turco et al, 2021; Zaffagnini et al, 2018). In short, GSH (17075601, GE Healthcare), GFP-trap or RFP-trap beads (gta & rta, Chromotek) were washed first 2x with H₂O and then equilibrated 2× with SEC buffer (25 mM HEPES, pH 7.5, 150 mM NaCl, 1 mM DTT). Beads were resuspended in 40 µl of SEC buffer, and the bait protein was added. For experiments with GST-TAX1BP1 (FL, domain, truncations, point mutants) in Fig. 4A, 2.5 µM of bait protein was added to the beads. The beads were incubated for 1 h at 4 °C and washed afterwards 3× with SEC Buffer. The beads were pipetted into a 384-well glass-bottom plate (781892, Greiner Bio-One) containing 1 µM GFP-NBR1 as prey in a total volume of 20 µl. For experiments in Fig. EV5A, 5 µM of GST-4xUB as bait protein was added and treated as described above. Beads were pipetted into a 1 µM GFP-TAX1BP1 (FL, truncation) containing prey solution. For experiments in Fig. 5A,C, GFP-trap beads were incubated with 2.5 µM GFP-NBR1 and treated as described above. Beads were added to 500 nM mCherry-TAX1BP1 and increasing amounts of GABARAP protein (0 nM–500 nM–1 µM–5 µM). For experiments in Fig. 7I, GSH beads were first incubated with a mix of 4.5 µM GST + 0.5 µM GST-4xUB as a bait for 1 h. After the incubation, the beads were washed to remove the unbound proteins and further incubated with 1 µM mCh-p62 and 1 µM NBR. After 30 min of incubation, 500 nM GFP-TAX1BP1 FL or GFP-TAX1BP1∆ZnF, together with 5 µM GABARAP, was added to the beads. Prey & bait were incubated for 15 min to 30 min at RT and imaged with a Zeiss LSM 700 confocal microscope equipped with Plan-Apochromat 20X/0.8 WD 0.55 mm objective. For each experimental condition, three biological replicates were performed. For the quantification, we employed an artificial intelligence (AI) script that automatically quantifies signal intensities from microscopy images by drawing line profiles across beads and recording the difference between the minimum and maximum gray values along the lines. The AI was trained to recognize beads employing cellpose (Stringer et al, 2021). Source code available at https://www.maxperutzlabs.ac.at/research/facilities/biooptics-light-microscopy. Background-corrected values were plotted and subjected for statistical analysis. For the GABARAP competition assay, we normalized to the control (0 nM GABARAP) to minimize the inter-experiment variability.

For pulldowns with cell lysates, HeLa PentaKO cells were seeded at 500k/well. Twenty-four hours later, the cells were transfected with plasmids encoding GFP-NBR1 (FL, truncations, point mutants). Forty-eight hours after transfections, cells were harvested and lysed using lysis buffer (50 mM Tris-HCl pH 7.4, 1 mM EGTA, 1 mM EDTA, 1% Triton X-100, 0.27 M sucrose, 1 mM DTT, cOmplete EDTA-free protease inhibitor (Roche)). GST-TAX1BP1 coated beads were prepared as described above. In total, 100–200 µg of total protein lysates was mixed with beads and incubated for 1 h at 4 °C. After the incubation, beads were washed 3× with IP wash buffer (20 mM Tris-HCl pH 7.4, 150 mM NaCl, 1 mM EDTA, 0.05% Triton X-100, 5% glycerol) to remove unbound material. Overall, 40 µl of 2× loading dye was added to the beads. Samples were boiled for 10 min at 95 °C. SDS PAGE and western blotting for detection were performed as described above.

In vitro condensation assays have been performed as previously described (Turco et al, 2021; Turco et al, 2019; Zaffagnini et al, 2018). In short, proteins were spun down with max speed for 10 min at 4 °C. For experiments in Figs. 1E and 7B, 2 µM mCherry-p62 S403E, 1 µM NBR1, 2 µM GFP-TAX1BP1 (FL, ∆N or ∆ZnF) were mixed in SEC buffer (25 mM HEPES, pH 7.5, 150 mM NaCl, 1 mM DTT) in a 384-well glass-bottom plate (781892, Greiner Bio-One). Buffer without NaCl was used to keep the salt level at 150 mM NaCl. 5 µM GST-4xUB was added at the microscope to trigger condensation of the proteins. Images were taken as a time course over 2 h with a Spinning disk microscope equipped with LD Achroplan 20X/0.4 Corr objective and EM-CCD camera.

For the reconstitution of the recruitment of TBK1 & SINTBAD to p62 condensates in Fig. 3G,H and EV3E, 500 nM mCherry-p62 S403E, 250 nM NBR1, 250 nM mCherry-TAX1BP1, 250 nM SINTBAD or NAP1, 500 nM GFP-TBK1, 500 nM SINTBAD-GFP were mixed in SEC buffer (25 mM HEPES, pH 7.5, 150 mM NaCl, 1 mM DTT) in a 384-well plate (781892, Greiner Bio-One). In all, 1.5 µM GST-4xUB was added to the wells at the microscope to trigger condensation of the proteins.

For the competition assay with the GABARAP proteins in Fig. 5E, 500 nM mCherry-p62 S403E, 250 nM NBR1, 250 nM GFP-TAX1BP1 and increasing amounts of GABARAP (0 nM–500 nM–1 µM–5 µM) were mixed in SEC buffer (25 mM HEPES, pH 7.5, 150 mM NaCl, 1 mM DTT) in a 384-well plate. 1.5 µM GST-4xUB was added at the microscope to trigger condensation of the proteins.

For the condensation assay in Fig. 7J, 500 nM mCherry-p62 S403E, 250 nM NBR1, 250 nM GFP-TAX1BP1 and either 0 µM or 5 µM GABARAP were mixed in SEC buffer (25 mM HEPES, pH 7.5, 150 mM NaCl, 1 mM DTT) in a 384-well plate. 1.5 µM GST-4xUB or GST-8xUB were added to the wells at the microscope to trigger condensation of the proteins.

Images depicted in Figs. 1E, 2G, 2H, 5E, and 7B show indicated timepoints after subtracting the background using rolling ball background reduction (rolling ball radius = 5 pixels). For quantifications, the same macro was used as previously described (Turco et al, 2021; Zaffagnini et al, 2018). In short, rolling ball background subtraction of ImageJ was performed, followed by thresholding and puncta counting across all channels and timepoints using the Analyze puncta function.

The peptide array was synthesized by the Peptide Synthesis facility of the IMP (https://cores.imp.ac.at/peptide-synthesis/). The entire NBR1 protein sequence was spotted as 12 aa/spot in a 2 aa window. The membrane was first soaked in methanol and then washed 3× with TBS. The membrane was incubated with blocking buffer (3% BSA dissolved in 1× TBS) overnight at 4 °C.The next day the membrane was washed 1× with TBST (TBS + 0.1% Tween-20) and incubated with our protein (GFP-TAX1BP1) at a concentration of 50 µg/ml in 15 ml of SEC Buffer (25 mM HEPES, pH 7.5, 150 mM NaCl, 1 mM DTT) for 2 h at RT. After the incubation, the membrane was washed 3x with TBST.

For detection, a western blot using a PVDF membrane (88520, Thermo) was performed. After the transfer by wet blotting, the membranes were treated as described above.

For the statistical analysis, Prism GraphPad was used. For all microscopy-based interaction assays a one-way ANOVA was performed. For immunofluorescence images Student’s t test (unpaired) was performed. For live-cell imaging data, two-way ANOVA or Student’s t test was performed. For western blot quantifications, one-way ANOVAs or Student’s t tests (unpaired) were performed. For condensation assays, either one-way ANOVA or Student’s t test (unpaired) was performed depending on the number of conditions. For comparing the number of foci between TAX1BP1 WT and ∆ZnF we performed a one-sample t and Wilcoxon test. P values were corrected for multiple testing using the suggested method. All values comparing data points in the figures refer to P values. Error bars are reported as mean ± standard deviation.