The Autophagy Receptor TAX1BP1 and the Molecular Motor Myosin VI Are Required for Clearance of Salmonella Typhimurium by Autophagy

Abstract

Autophagy plays a key role during Salmonella infection, by eliminating these pathogens following escape into the cytosol. In this process, selective autophagy receptors, including the myosin VI adaptor proteins optineurin and NDP52, have been shown to recognize cytosolic pathogens. Here, we demonstrate that myosin VI and TAX1BP1 are recruited to ubiquitylated Salmonella and play a key role in xenophagy. The absence of TAX1BP1 causes an accumulation of ubiquitin-positive Salmonella, whereas loss of myosin VI leads to an increase in ubiquitylated and LC3-positive bacteria. Our structural studies demonstrate that the ubiquitin-binding site of TAX1BP1 overlaps with the myosin VI binding site and point mutations in the TAX1BP1 zinc finger domains that affect ubiquitin binding also ablate binding to myosin VI. This mutually exclusive binding and the association of TAX1BP1 with LC3 on the outer limiting membrane of autophagosomes may suggest a molecular mechanism for recruitment of this motor to autophagosomes. The predominant role of TAX1BP1, a paralogue of NDP52, in xenophagy is supported by our evolutionary analysis, which demonstrates that functionally intact NDP52 is missing in Xenopus and mice, whereas TAX1BP1 is expressed in all vertebrates analysed. In summary, this work highlights the importance of TAX1BP1 as a novel autophagy receptor in myosin VI-mediated xenophagy. Our study identifies essential new machinery for the autophagy-dependent clearance of Salmonella typhimurium and suggests modulation of myosin VI motor activity as a potential therapeutic target in cellular immunity.

Fig 1. Evolutionary analysis of myosin VI associated adaptors and autophagy-related proteins.

(A) Coulson plot [30] showing the evolutionary distribution within the holozoa and metazoa of selected myosin VI partners implicated in autophagy. Genes or gene families are represented as rows and taxa are columns. A filled circle indicates the presence of a homolog, empty circles denote instances where no homolog was identified. Grey circles denote identified sequences that could not be reliably placed through phylogenetic reconstruction. Columns are coloured according to the phylogenetic distribution of taxa, indicated with a schematic tree below the table. Unless indicated all homology assignments are supported through phylogenetic reconstruction. Small circles denote paralagous expansion of gene families and where these paralogs have been classified they are stated to the right. Large oval in M. brevicollis represents the single ATG8 gene within this taxa reflecting the ancestral configuration[31]. Note the expansion of the CALCOCO, OPTN/NEMO and LCA/B gene families occurring at the base of the vertebrates. (B) Phylogenetic analysis of the expansion of the CALCOCO gene family. Branch support is shown for major clades. Domain organisation of metazoan CALCOCO gene products corresponding to those in. (C) Western blot of lysates harvested from HeLa cells following TAX1BP1 alone, NDP52 alone, or TAX1BP1 and NDP52 double siRNA transfection. HeLa cells transfected with siRNA targeted to TAX1BP1, NDP52, or both together were subjected to an infection with mCherry expressing Salmonella for 8 hours followed by processing for confocal microscopy and quantitation of the % of infected cells with ubiquitin (+) Salmonella. Experiments were performed in triplicate and error bars represent s.d.

Fig 2. TAX1BP1 localises to Salmonella that have escaped into the cytosol of infected cells.

(A) HeLa cells transfected with GFP-TAX1BP1 were infected with mCherry expressing Salmonella for 1 hour, processed for confocal immunofluorescence microscopy, and immunostained for ubiquitin. (B) HeLa cells infected for 1 hour with mCherry expressing Salmonella were immunostained for TAX1BP1 and p62 or (C) TAX1BP1 and LC3. Scale bar, 20 μm. (D) Mouse embryonic fibroblasts were infected with mCherry Salmonella for 1 hour and were immunostained for TAX1BP1 and processed for confocal microscopy. Nuclei are labelled with Hoechst (blue) (E) Still frames taken from RPE cells expressing GFP-TAX1BP1 or GFP-NDP52 infected with mCherry Salmonella imaged on a spinning disk live cell microscope. Elapsed time in min:sec is displayed.

Fig 3. Myosin VI localises to autophagosomes containing Salmonella.

RPE cells stably expressing GFP-myosin VI were infected with blue fluorescent protein (BFP) expressing (A, B) or mCherry expressing (C) Salmonella for 1 hour. Cells were immunostained for ubiquitin (A, red), p62 (B, red), or LC3 (C, blue) and processed for confocal microscopy. (D) Mouse embryonic fibroblasts (MEFs) were infected with mCherry-expressing Salmonella and processed for confocal microscopy by immunostaining for myosin VI (green), LC3 (blue) and nuclei were labelled with Hoechst (cyan). (E) HeLa cells expressing GFP-myosin VI were infected with mCherry expressing Salmonella (red) for 1 hour, processed for confocal microscopy and immunostained for GFP (green) and TAX1BP1 (blue). Scale bar, 20 μm.

Fig 4. Overlapping binding of TAX1BP1 zinc finger domains with ubiquitin and myosin VI.

(A) Domain organisation of TAX1BP1 and designated point mutations. LIR, LC3-interaction region; ZF, zinc-finger; SKICH, SKIP carboxyl homology. (B) Ribbon presentation of the structures of the TAX1BP1 zinc fingers together with two molecules of ubiquitin. The residues are colored according to the extent of the chemical shift changes observed where blue represents small or no chemical shift changes observed, and red represents the maximum chemical shift changes observed for the backbone amides. The side chains of the residues in TAX1BP1 that when mutated abolish ubiquitin binding are displayed and labelled. (C) RPE cells were transfected with GFP empty vector, with GFP-TAX1BP1 wild-type, GFP-TAX1BP1 ZF1 mutant (Q743A/E747K), GFP-TAX1BP1 ZF2 mutant (Q770A/E774K) or the Q743A/E747K/Q770A/E774K double zinc finger mutant of GFP-TAX1BP1 followed by GFP immunoprecipitation and pull-down with K63-linked polyubiquitin (2–7). Western blot analysis was performed and immunoblotting against the indicated proteins. (D) Mammalian 2-hybrid assay in CHOK.1 cells with myosin VI tail as bait and TAX1BP1 full-length wild-type and mutants as prey. Results are represented as the normalised luciferase activity against a bait only control.

Fig 5. Binding of TAX1BP1, NDP52 and optineurin to different ubiquitin chain types.

GFP immunoprecipitation followed by either linear tetra-ubiquitin (A), K48-linked polyubiquitin (B), or K63-linked polyubiquitin (C) pull-downs (PD) from RPE cells transfected with GFP alone, GFP-TAX1BP1, GFP-NDP52, or GFP-optineurin wild-type and mutants. Western blot analysis performed using antibodies specific for ubiquitin and GFP.

Fig 6. Binding of TAX1BP1 zinc finger domains to ubiquitin and myosin VI.

Dependence of the chemical shift changes of resonances of residues in TAX1BP1 as a function of ubiquitin concentration: (A) and (B) overlay of sections of the ¹H/¹⁵N HSQC spectra of ¹⁵N labeled TAX1BP1 zinc fingers at different concentrations of unlabeled ubiquitin showing the concentration dependence of the peak positions of residues in each of the zinc fingers. (C) Graph showing the concentration dependence of these peaks together with two other peaks from each of the zinc fingers. The chemical shift change of the backbone amide group (Δδ_HN−N) is computed with Δδ_HN−N = √ _(δ_H ²+ (δ_N/10)²), where δ_H and δ_N are the changes in ¹H and ¹⁵N chemical shift, respectively. If changes in the spectra are observed and the signals are in fast exchange, the dissociation constant (K_D) is calculated by fitting the NMR titration data to the formula: $\Delta {\delta }_{\text{HN-N}}=\frac{{\delta }_b-{\delta }_f}{2\left[P_0\right]}\left(\left[P_0\right]+\left[L_0\right]+K_D-\sqrt{{\left(\left[P_0\right]+\left[L_0\right]+K_D\right)}^2-4\left[P_0\right]\left[L_0\right]}\right)$ where y is Δδ, δ_b-δ_f is the difference in chemical shift between bound and free state (= Δδ_max), P₀ is the total protein concentration and L₀ is the total ligand concentration. A simple two-state binding model is assumed. Protein concentration is treated as a constant (measured before starting the titration). For our experiments P₀ was 150 μM. (D) The direct interaction of myosin VI CBD with TAX1BP1 was measured as changes in thermophoretic mobility of 75 nM fluorescently labeled myosin VI CBD in the presence of various TAX1BP1 concentrations. The binding affinity of the complex was estimated by a fit to a Hill function, as a K _D of about 5–10 μM. The experiment was repeated in the presence of 1 mM ubiquitin. Even the presence of a large excess of free ubiquitin does not change the binding affinity between myosin VI and TAX1BP1. Experiments were performed in triplicate (no ubiquitin) or duplicate (1mM ubiquitin) and error bars represent s.d. (E) Pull-down assay of myosin VI CBD and TAX1BP1 in the presence of ubiquitin. 15 μM TAX1BP1 was incubated for 30 min with increasing amounts of ubiquitin (0–1 mM) before adding 10 μM GST-myosin VI CBD. After pull-down with Glutathione-Sepharose beads, the amount of TAX1BP1 binding to GST-myosin VI CBD was visualised by SDS-PAGE. Upper gel shows the total input and lower panel the pull-down.

Fig 7. TAX1BP1 interacts with LC3B as well as LC3C and requires its zinc finger domains to localise to ubiquitylated Salmonella.

(A) Mammalian 2-hybrid assay in CHOK.1 cells with LC3A, B, or C and GABARAP, GABARAPL1 or L2 as bait and TAX1BP1 full-length wild-type as prey. Results are represented as the normalised luciferase activity against a bait only control. (B) Mammalian 2-hybrid assay in CHOK.1 cells using LC3B or LC3C as bait and full-length TAX1BP1, NDP52, optineurin wild-type and LIR mutants as prey. Results are represented as the normalised luciferase activity against a bait only control. (C) Mammalian 2-hybrid assay in CHOK.1 cells with LC3B as bait and TAX1BP1 full-length wild-type and mutants as prey. Results are represented as the normalised luciferase activity against a bait only control. (D) HeLa cells transfected with GFP-TAX1BP1 wild-type or V144S LIR mutant, Q743A/E747K/Q770A/E774K double zinc finger mutant, and V144S/Q743A/E747K/Q770A/E774K LIR and double zinc finger mutant were infected with mCherry expressing Salmonella for 1 hour prior to saponin extraction and fixation. Cells were immunostained for GFP and ubiquitin and processed for confocal microscopy. Nuclei are labelled with Hoechst (blue). Scale bar, 20 μm.

Fig 8. Suppression of myosin VI expression leads to a hyper-proliferation of Salmonella and an accumulation of ubiquitylated Salmonella within LC3-positive autophagosomes.

(A) HeLa cells following myosin VI siRNA or TAX1BP1, NDP52, OPTN (TNO) siRNA transfection were subjected to Salmonella infection for the indicated time points. Gentamicin protection assays were performed and colonies were counted at each time point. Results are represented as the fold replication from 2 hours, which represents the amount of bacterial proliferation. Results are from >3 independent experiments and error bars represent s.d. Western blot analysis was performed on whole cell lysates and immunoblotting was performed to the indicated proteins. (B) HeLa cells mock, myosin VI siRNA, or TNO siRNA treated were infected with mCherry expressing Salmonella for 8 hours, followed by processing for confocal microscopy. Cells were immunostained for ubiquitin (green). Nuclei were labelled with Hoechst (blue). The % of infected cells with ubiquitin (+) Salmonella was quantified. The results depict >3 independent experiments and the error bars represent the s.d. (C) HeLa cells transfected with myosin VI or TNO siRNA were infected with BFP-expressing Salmonella for 8 hours. Cells were processed for confocal microscopy and immunostained for LC3 (green) and ubiquitin (red). Scale bar, 20 μm. Quantitation at 4 and 8 hours post-infection, with BFP-expressing Salmonella, of the % of ubiquitin (+) Salmonella which are LC3 (+). Results represent 3 independent experiments and the error bars indicate the s.d.