TECPR1 conjugates LC3 to damaged endomembranes upon detection of sphingomyelin exposure

Abstract

Invasive bacteria enter the cytosol of host cells through initial uptake into bacteria-containing vacuoles (BCVs) and subsequent rupture of the BCV membrane, thereby exposing to the cytosol intraluminal, otherwise shielded danger signals such as glycans and sphingomyelin. The detection of glycans by galectin-8 triggers anti-bacterial autophagy, but how cells sense and respond to cytosolically exposed sphingomyelin remains unknown. Here, we identify TECPR1 (tectonin beta-propeller repeat containing 1) as a receptor for cytosolically exposed sphingomyelin, which recruits ATG5 into an E3 ligase complex that mediates lipid conjugation of LC3 independently of ATG16L1. TECPR1 binds sphingomyelin through its N-terminal DysF domain (N'DysF), a feature not shared by other mammalian DysF domains. Solving the crystal structure of N'DysF, we identified key residues required for the interaction, including a solvent-exposed tryptophan (W154) essential for binding to sphingomyelin-positive membranes and the conjugation of LC3 to lipids. Specificity of the ATG5/ATG12-E3 ligase responsible for the conjugation of LC3 is therefore conferred by interchangeable receptor subunits, that is, the canonical ATG16L1 and the sphingomyelin-specific TECPR1, in an arrangement reminiscent of certain multi-subunit ubiquitin E3 ligases.

Keywords: ATG5-ATG12 E3 ligase; DysF; autophagy; membrane damage; sphingomyelin.

Figure 1. TECPR1 detects sphingomyelin on the cytosolic face of bacteria‐containing vacuoles

1. A

   Schematic of liposome‐binding assay.

2. B

   Venn diagram of liposome‐binding proteins identified in the cell lines shown with an enrichment value of > 1.4 in sphingomyelin versus control liposomes. Those proteins enriched in all three cell lines are indicated.

3. C–H

   (C, D, H) Confocal micrographs of HeLa cells expressing GFP‐TECPR1 either alone or together with FLAG‐tagged neutral sphingomyelinase 2 (nSMase2 [H]) and treated with osmotic shock assay or LLOMe (C), or infected with the bacteria indicated (D) and fixed at 1 h (S. Typhimurium and L. monocytogenes) or 0.5 h (S. flexneri) postinfection. Cells were stained for Galectin‐8 and DNA (DAPI). Scale bar, 20 μm. (E) Selected frames from live imaging of HeLa cells expressing GFP‐TECPR1 and mCherry‐Galectin 8 infected with BFP‐expressing S. Typhimurium. Yellow arrowhead indicates a bacterium to which TECPR1 has been recruited prior to Galectin 8. Scale bar, 10 μm (F) Percentage of bacteria positive for mCherry‐TECPR1 in ATG5 knockout (KO) or ATG5‐complemented MEF cells at 1 h (S. Typhimurium) or 0.5 h (S. flexneri) postinfection. Mean ± SD of four (S. Typhimurium) or two (S. flexneri) independent experiments. n > 100 bacteria per coverslip. (G) Percentage of S. Typhimurium positive for GFP‐TECPR1 at 1 h postinfection in HeLa cells pretreated or not with 100 nM Wortmannin. Mean ± SD of two independent experiments. n > 100 bacteria per coverslip.

4. I

   Percentage of S. Typhimurium positive for GFP‐TECPR1 or endogenous Galectin‐8 at 1 h postinfection in HeLa cells expressing FLAG‐nSMase2 or not. Mean ± SD of two (TECPR1) or three (Gal8) independent experiments. n > 100 bacteria per coverslip.

Source data are available online for this figure.

Figure EV1. (corresponding to Fig 1). TECPR1 detects sphingomyelin on the cytosolic face of bacteria‐containing vacuoles

1. A

   Histograms displaying the diameter of liposomes composed of phosphatidylcholine:cholesterol (60:40; top) or sphingomyelin:phosphatidylcholine: cholesterol (50:10:40; bottom), as measured by dynamic light scattering.

2. B

   Confocal micrograph of HeLa cells expressing GFP‐TECPR1 infected with S. Typhimurium ΔprgH + inv, fixed at 1 h postinfection and stained with DAPI. Scale bar, 10 μm.

3. C

   Percentage of S. Typhimurium WT or ΔprgH + inv positive for GFP‐TECPR1 in Hela cells fixed at 1 h postinfection. Mean ± SD of two independent experiments performed in duplicate. n > 100 bacteria per coverslip.

4. D

   ATG5‐deficient mouse embryonic fibroblasts were retrovirally transduced with an AU1‐tagged ATG5 construct or not, lysates prepared and analysed by SDS–PAGE and Western blotted with the indicated antibodies. Upper band in anti‐ATG5 blot is conjugated ATG5‐ATG12 while the lower is monomeric ATG5. PCNA serves as a loading control.

5. E

   Confocal micrographs of ATG5‐deficient MEF cells, complemented with AU1‐ATG5 or not and expressing mCherry‐TECPR1, were infected with S. Typhimurium, fixed at 1 h postinfection and stained with DAPI to label bacteria. Scale bar, 20 μm.

6. F, G

   Confocal micrographs of HeLa cells expressing GFP‐TECPR1 alone (G) or together with mCherry‐tagged FYVE domain from WDFY1 (F) were pretreated with 100 nM Wortmannin or mock control (DMSO) infected with S. Typhimurium for 1 h, fixed and stained with anti‐WIPI2 (G) and DAPI to label bacteria. Scale bar, 20 μm.

Figure 2. The N‐terminal DysF domain of TECPR1 detects sphingomyelin

1. A

   Schematic of TECPR1 domain structure. AIR, ATG5‐interacting region; PH, Plekstrin homology domain. Tectonin repeat indicated.

2. B

   Confocal micrographs of HeLa cells expressing the indicated GFP‐tagged TECPR1 constructs, infected with S. Typhimurium, fixed at 1 h postinfection and DNA stained with DAPI. Scale bar, 20 μm.

3. C

   Percentage of S. Typhimurium positive for the indicated constructs in HeLa cells, fixed at the indicated timepoints postinfection. Mean ± SEM of 3–4 independent experiments performed in duplicate. n > 100 bacteria per coverslip.

4. D

   Confocal micrographs of HeLa cells expressing TECPR1 N‐terminal (N′) DysF‐GFP infected with mCherry‐expressing S. Typhimurium, fixed at 30 min postinfection. Scale bar, 20 μm.

5. E

   Confocal micrographs of HeLa cells expressing TECPR1 N′ DysF‐GFP with or without FLAG‐nSMase2, infected with S. Typhimurium, fixed at 30 min postinfection and stained with anti‐Galectin 8 antibody. DNA stained with DAPI. Scale bar, 20 μm.

6. F

   Percentage of S. Typhimurium positive for N′ DysF‐GFP in Hela cells, expressing FLAG‐nSMase 2 or not, fixed at 30 min postinfection. Mean ± SD of two independent experiments performed in duplicate.

7. G, H

   Confocal micrographs of HeLa cells expressing the indicated GFP‐tagged DysF domains, infected with mCherry‐S. Typhimurium and fixed at 30 min postinfection (G) or treated with osmotic shock assay, fixed and stained with anti‐Galectin 8 antibody (H). The bottom row in H depicts an enlarged version of the boxed region shown above. Scale bar, 20 μm.

Source data are available online for this figure.

Figure EV2. (corresponding to Fig 2). The N‐terminal DysF domain of TECPR1 detects sphingomyelin

1. A

   Confocal micrographs of Hela cells expressing N‐terminal DysF domain of TECPR1 (N'DysF) as a C‐terminal GFP fusion were infected with S. Typhimurium, fixed at the indicated times postinfection, stained with anti‐Galectin 8 and DAPI. Arrowhead indicates bacteria to which DysF‐GFP but not Galectin 8 is recruited, whereas arrow indicates those to which both proteins are recruited. Scale bar, 20 μm.

2. B

   Percentage of S. Typhimurium positive for N‐terminal DysF‐GFP and/or anti‐Galectin 8 in HeLa cells fixed at the indicated times postinfection. Mean ± SD of two independent experiments performed in duplicate.

3. C

   Hela cells stably expressing the indicated GFP‐tagged DysF constructs were prepared for SDS–PAGE and Western blotted with anti‐GFP antibody.

Figure 3. Identification of the sphingomyelin‐binding site in the N‐terminal DysF domain

1. A

   Ribbon diagram of the crystal structure of the N‐terminal DysF domain with amino acid positions corresponding to full‐length TECPR1 indicated.

2. B, C

   Percentage of S. Typhimurium positive for the indicated GFP‐tagged N‐terminal DysF constructs at 30 min postinfection. Mean ± SD of two (EK134,135AA) or three independent experiments. n > 100 bacteria per coverslip. P‐value from one‐way ANOVA with Dunnett's multiple comparison test.

3. D

   Ribbon diagram of the distal tip of N‐terminal DysF domain structure (boxed area in [A]) showing those residues required for recruitment to S. Typhimurium depicted with side chains in magenta and those not in grey (top). The two exposed tryptophan residues are highlighted. The comparable region from the structures of the inner DysF domains of Myoferlin (PDB:2K2O; middle) and Dysferlin (PDB:4CAH; bottom) are depicted.

4. E

   Percentage of S. Typhimurium positive for indicated GFP‐TECPR1 constructs in HeLa cells at 30 min postinfection. Mean ± SEM of two (ΔPH) or six independent experiments performed in duplicate. n > 100 bacteria per coverslip. P‐value from one‐way ANOVA with Tukey's multiple comparison test.

5. F

   Confocal micrographs of HeLa cells expressing either GFP‐TECPR1 WT or W154A and treated with LLOMe or DMSO vehicle control for 10 min and fixed. Scale bar, 20 μm.

6. G

   Number of GFP‐positive puncta per cell was enumerated in samples depicted in (F). Data from 50 cells from a single experiment, representative of two. P‐value from one‐way ANOVA with Tukey's multiple comparison test.

Source data are available online for this figure.

Figure EV3. (corresponding to Fig 3). Identification of the sphingomyelin‐binding site in the N‐terminal DysF domain

1. A

   Coomassie stain of SDS–PAGE gel of recombinant TECPR1 N‐terminal DysF domain purified from E. coli.

2. B

   Mass spectrometry of recombinant TECPR1 N‐terminal DysF protein before (top) and after (bottom) reductive methylation. Note the shift in mass of major species from 14,173 to 14,398 Da (225 Da) corresponding to eight dimethyl groups attached to seven lysine and N‐terminal methionine residues.

3. C

   Overlay of the crystal structure of TECPR1 N‐terminal DysF domain and that of the NMR structure of the inner DysF domain of myoferlin (PDB: 2K2O) and the crystal structure of the inner DysF domain of Dysferlin (PDB: 4CAH).

4. D

   Confocal micrographs of Hela cells expressing N'DysF‐GFP infected with mCherry‐Salmonella for 30 min and fixed. Micrograph is same as that shown in Fig 2G. Insets depict SCVs that are scored as being either positive (left) or negative for DysF‐GFP (right). Scale bar, 20 μm.

5. E

   Coomassie stain of SDS–PAGE gels of recombinant N'DysF WT or W154A protein before and after labelling with streptavidin^AlexaFluor46.

Figure 4. TECPR1 in complex with ATG5/ATG12 catalyses the conjugation of LC3 to lipids in sphingomyelin‐displaying membranes

1. A, B

   Binding of the indicated streptavidin^Alexa546‐conjugated N‐terminal DysF proteins to Hela cells in the presence or absence of a 10‐fold molar excess of indicated Lysenin^CTD‐GFP proteins. Representative histograms of DysF protein binding (left) and mean fluorescence intensity ± SD of three independent experiments (right). P‐value from one‐way ANOVA with Tukey's multiple comparison test.

2. C–E

   Liposome sedimentation assay. The indicated proteins were incubated with either liposomes (containing sphingomyelin or not) or buffer only, pelleted by centrifugation and equal portions of supernatant and pellet fractions analysed by SDS–PAGE and Coomassie staining. Representative gels depicted (C). Fraction of TECPR1 (D) or DysF WT or W154A (E) proteins present in pellet under the indicated conditions, expressed as % binding. Note pelleting in the absence of liposomes likely due to the affinity of proteins for tube walls. Mean ± SD of three independent experiments. P‐value from one‐way ANOVA with Tukey's multiple comparison test.

3. F

   In vitro LC3 lipid conjugation assay. Liposomes containing sphingomyelin or not were incubated with LC3 lipid conjugation machinery: ATG7 (E1), ATG3 (E2), ATG5‐ATG12 (E3), LC3B, MgCl₂, ATP and the indicated TECPR1 proteins. Control reactions without ATG3 or ATP were included. Mixes were incubated at 37°C for 3 h, centrifuged and equal fractions of both supernatant and pellet analysed by SDS–PAGE and Coomassie staining. Lipidated LC3B‐II appears in pellet fraction whereas both soluble LC3B‐I and AMPylated LC3B are in the supernatant, the latter of the same apparent molecular weight as LC3B‐II.

4. G

   Fraction of LC3B protein present in pellet from complete in vitro LC3 lipid conjugation assay with the indicated TECPR1 proteins in either absence or presence of sphingomyelin. Mean ± SD from three independent experiments. P‐value from one‐way ANOVA with Tukey's multiple comparison test.

Source data are available online for this figure.

Figure 5. TECPR1 recruits ATG5 to Salmonella‐containing vacuoles for ATG16L1‐independent LC3 conjugation

1. A–D

   (A, C) Confocal micrographs of indicated MEF cell lines infected with mCherry‐expressing S. Typhimurium, fixed at 30 min postinfection and stained with anti‐ATG5 antibody (A) or anti‐LC3 antibody (C). Scale bar, 20 μm. (B, D) Percentage of S. Typhimurium positive for endogenous ATG5 (B) or LC3 (D) in indicated MEF cells at 30 min postinfection. Mean ± SEM of 3 (B) or 3–4 (D) independent experiments performed in duplicate. n > 200 bacteria per coverslip. P‐value from one‐way ANOVA with Tukey's multiple comparison test.

2. E

   Percentage of S. Typhimurium positive for endogenous LC3 in indicated MEF cells treated with 100 nM Wortmannin or DMSO vehicle control, fixed at 1 h postinfection. Mean ± SEM of three independent experiments performed in duplicate. n > 100 bacteria per coverslip. P‐value from one‐way ANOVA with Tukey's multiple comparison test.

3. F

   Confocal micrographs of either control of ATG16L1‐deficient MEF cells, transfected with either control or TECPR1 siRNA (siRNA no. 66 depicted), infected with mCherry‐expressing S. Typhimurium for 30 min, fixed and stained with anti‐LC3 antibody. Scale bar, 20 μm.

4. G

   Percentage of S. Typhimurium positive for LC3 in either control or ATG16L1‐deficient MEF cells transfected with indicated siRNAs and infected with mCherry‐expressing S. Typhimurium for 30 min. Mean ± SEM of three independent experiments performed in at least duplicate and enumerated after blind labelling of coverslips. P‐value from one‐way ANOVA with Dunnett's multiple comparison test.

5. H

   Fold intracellular proliferation of S. Typhimurium in indicated MEF cells as assessed by colony forming unit assay, expressed as a ratio of intracellular bacteria present at 6 h versus 1 h. Mean ± SEM of three (ATG16L1 KO, ATG5 KO), four (TECPR1 KO) or five (Control) independent experiments. P‐value from one‐way ANOVA with Tukey's multiple comparison test.

Source data are available online for this figure.

Figure EV4. (corresponding to Fig 5). TECPR1 recruits ATG5 to Salmonella‐containing vacuoles for ATG16L1‐independent LC3 conjugation

1. A

   Western blot of lysates from MEF cells firstly lentivirally transduced to stably express Cas9 followed by stable lentiviral expression of indicated gRNAs for 7 days and blotted with antibodies shown. Hatched line in (F) denotes where intervening, irrelevant lanes of the blot were excised.

2. B

   Percentage of S. Typhimurium positive for anti‐Galectin 8 in TECPR1 WT MEF cells, KO or KO complemented with TECPR1 construct fixed at 1 h postinfection. Mean + SEM of three independent experiments performed in duplicate. Statistical significance, based on a one‐way ANOVA with the Tukey's multiple comparison test, was not reached between any of the groups.

3. C

   Relative expression level of GFP in HeLa cells stably transduced with H6Pneo lentivirus, selected with G418 and measured by flow cytometry. H6Pneo harbours GFP under the control of one of five different, progressively shorter portions of the spleen focus‐forming virus (SFFV) promoter (A–E) as described in Materials and Methods.

4. D

   Western blot of TECPR1 WT or KO MEFs or KO complemented with TECPR1 or GFP control from indicated H6P‐based lentiviruses harbouring SFFV promoters corresponding to those in (B). Arrow indicates specific band corresponding to TECPR1 at 135 kDa, and star denotes nonspecific band. Actin serves as loading control.

Figure 6. TECPR1 can damage sphingomyelin‐positive membranes

1. A

   Percentage of S. Typhimurium positive for Galectin‐8 and/or GFP‐TECPR1 in either control Hela cells or those expressing GFP‐TECPR1 at 1 h postinfection. Mean ± SEM of three independent experiments performed in duplicate. n > 100 bacteria per coverslip.

2. B

   Confocal micrographs of HeLa cells expressing the indicated GFP‐TECPR1 constructs or not (control), infected with S. Typhimurium for 30 min, fixed and stained with anti‐Galectin‐8 and DAPI. Scale bar, 20 μm. Arrowheads indicate bacteria positive for Galectin 8, of which more are present in cells expressing GFP‐TECPR1 WT but not W154A. Arrow indicates bacterium to which GFP‐TECPR1 but not Galectin 8 has been recruited.

3. C

   Percentage of S. Typhimurium positive for Galectin‐8 in Hela cells expressing the indicated GFP‐TECPR1 constructs or not (control) and infected with S. Typhimurium for 1 h. Mean ± SEM of five independent experiments performed in triplicate. P‐value from one‐way ANOVA with Tukey's multiple comparison test.

4. D

   Fold intracellular proliferation of S. Typhimurium in HeLa cells expressing the indicated GFP‐TECPR1 constructs or not (control) as assessed by colony forming unit assay. Mean ± SEM of four (ΔPH) or five (control, WT, W154A and ΔAIR) independent experiments. P‐value from one‐way ANOVA with Dunnett's multiple comparison test.

Source data are available online for this figure.