HEATR3 recognizes membrane rupture and facilitates xenophagy in response to Salmonella invasion

Abstract

Bacterial invasion into the cytoplasm of epithelial cells triggers the activation of the cellular autophagic machinery as a defense mechanism, a process known as xenophagy. In this study, we identified HEATR3, an LC3-interacting region (LIR)-containing protein, as a factor involved in this defense mechanism using quantitative mass spectrometry analysis. HEATR3 localizes intracellularly invading Salmonella, and HEATR3 deficiency promotes Salmonella proliferation in the cytoplasm. HEATR3 also localizes to lysosomes damaged by chemical treatment, suggesting that Salmonella recognition is facilitated by damage to the host cell membrane. HEATR3 deficiency impairs LC3 recruitment to damaged membranes and blocks the delivery of the target to the lysosome. These phenotypes were rescued by exogenous expression of wild-type HEATR3 but not by the LIR mutant, indicating the crucial role of the HEATR3-LC3 interaction in the receptor for selective autophagy. HEATR3 is delivered to lysosomes in an autophagy-dependent manner. Although HEATR3 recruitment to the damaged membrane was unaffected by ATG5 or FIP200 deficiency, it was markedly impaired by treatment with a calcium chelator, suggesting involvement upstream of the autophagic pathway. These findings suggest that HEATR3 serves as a receptor for selective autophagy and is able to identify damaged membranes, facilitate the removal of damaged lysosomes, and target invading bacteria within cells.

Keywords: HEATR3; NOD2 signaling; autophagy; salmonella infection; xenophagy receptor.

Fig. 1.

Identification of HEATR3 as an LIR protein that directly associates with LC3 family proteins. (A) Schematic presentation of stable isotope labeling by amino acids in cell culture (SILAC)—mass spectrometry performed in this study. HEK293T cells expressing OSF-LC3B^WT and OSF-LC3B^K51A/L53A were labeled with C¹²N¹⁴ (Light) medium or with C¹³N¹⁵ (Heavy) medium, respectively. After mixing these cells, lysed cells and LC3B proteins were purified with Strep-Tactin Sepharose beads. Copurified proteins were analyzed using mass spectrometry, and the amounts of proteins from each culture were quantified. (B) The ratio of the amounts of proteins labeled with C¹³N¹⁵ (Heavy) and proteins labeled with C¹²N¹⁴ (Light). Each dot indicates each identified factor by mass spectrometry. A total of 601 proteins were identified and quantified in relative amounts. The top 20 proteins identified in the Heavy fraction rather than the Light fraction were magnified, including known LIR-containing proteins. HEATR3 and some known LIR-containing proteins included in the top 20 were indicated in the red dot. (C) Strep-tag pull-down assay. HEK293T cells were cotransfected with pCAG-OSF-empty, OSF-LC3B^WT, or OSF-LC3B^K51A with pCAG-Myc-HEATR3. After 48 h of transfection, lysed cells, and pulled down OSF-LC3B with Strep-Tactin Sepharose beads. Myc-HEATR3 in the beads-bound fraction (Top) and in whole cell lysate (Bottom), and OSF-LC3B in the beads-bound fraction (Middle) were detected by Western blotting using the anti-Myc antibody and anti-FLAG antibody, respectively. The ratio of HEATR3 levels based on the protein signal on Western blotting between the beads-bound fraction and whole cell lysate was quantified using Image J (Bottom graph) (one‐way ANOVA, Dunnett’s multiple comparison tests, **P < 0.01 versus OSF-LC3B^WT). (D and E) GST pull-down assay using purified proteins. Purified OSF-HEATR3 were mixed with purified GST-LC3 or GABARAP family proteins (D), or tetra-ubiquitin (E). After pulled-down using Glutathione beads, samples were separated by SDS-PAGE and detected GST-fused proteins by CBB (Bottom panels) or HEATR3 by western blotting using anti-FLAG antibody (Top panels).

Fig. 2.

WEEL tetra peptide motif on HEATR3 has a role in direct association with LC3. (A) The model of HEATR3 structure. AlphaFold2 conformation model prediction was performed and predicted structure is shown in ribbon representation. Yellow and blue regions correspond to N-terminal fragment (HEATR3^N: 1 to 326) and C-terminal fragment (HEATR3^C: 327 to 680), respectively. HEAT domain was indicated in purple. Four putative LIR sequences (W/F/Y)-X-X-(L/I/V) were highlighted in light blue. (B) Schematic structure of human HEATR3 and its truncated mutants used in this study. The amino acid positions were shown on the Left of the scheme. (C) Strep-tag pull-down assay. HEK293T cells were cotransfected with pCAG-OSF-empty, OSF-LC3B or OSF-GABARAP with pCAG-Myc-HEATR3 and its derivatives. After 48 h of transfection, lysed cells, and pulled down OSF-LC3B or OSF-GABARAP with Strep-Tactin Sepharose beads. Myc-HEATR3 in the beads-bound fraction (Left, Top) and in whole cell lysate (Left, Bottom), and OSF-LC3B in the beads-bound fraction (Left, Middle) were detected by Western blotting using the anti-Myc antibody and anti-FLAG antibody, respectively. (D and E) Yeast two-hybrid assay. GAL4 DNA binding domain (DBD) fused proteins and GAL4 activation domain (AD) fused proteins were transiently coexpressed in yeast with the represented combination. Yeast was cultured in medium lacking the essential amino acids, which are indicated at the Bottom of the panels. (F) GST pull-down assay using purified proteins. Purified WT or LIR2A mutant of OSF-HEATR3 were mixed with purified GST-LC3B WT, K51A mutant, or LC3 family proteins. After pulled-down using Glutathione beads, samples were separated by SDS-PAGE and detected GST-fused proteins by CBB (Bottom panels) or HEATR3 by western blotting using anti-FLAG antibody (Top panels).

Fig. 3.

HEATR3 accumulates invaded Salmonella and has an inhibitory role in the propagation of intracellular Salmonella. (A) Localization of HEATR3 in Salmonella infected HeLa cells. HeLa cells were transfected with pCAG-OSF-HEATR3. After 24 h of transfection, cells were infected with YFP-Salmonella (MOI = 100). Salmonella infected cells were fixed and stained with anti-FLAG antibody. Salmonella (green) and OSF-HEATR3 (magenta) were detected by Structured Illumination Microscopy (SIM) (Top). Magnified images of each channel and merged channel are shown at below the main image. The dashed line indicates the outline of a cell. (Scale bar, 5 µm.) (B) Localization of HEATR3 C-terminal fragment and Galectin3 (Gal3) in Salmonella infected HeLa cells. HeLa cells transiently expressed OSF-HEATR3^C and YFP-Gal3, and infected with mCherry-Salmonella (MOI = 500). At 5 h postinfection (h.p.i.), infected cells were fixed and stained with anti-FLAG antibody. Salmonella (magenta) and each protein (OSF-HAETR3^C: cyan, YFP-Gal3: green) in cells were detected using a confocal laser microscope. The dashed line indicates the outline of a cell. Red arrowheads indicate Gal3-positive Salmonella and yellow arrowheads indicate Gal3-negative Salmonella. (Scale bar, 10 µm.) (C) Colocalization rate of HEATR3 with infected Salmonella. Colocalization rate of HEATR3 with Gal3-positive and -negative Salmonella was calculated, each dot representing a separate image (n = 12). Red asterisks indicate statistically significant differences compared to Gal3-positive Salmonella (**P < 0.01, unpaired two-sample Student’s t test). (D) Establishment of HEATR3 knock-out (KO) and Hexa KO HeLa cells. WT, HEATR3 KO, Penta KO, Hexa KO HeLa cells lysate were separated by SDS-PAGE and HEATR3 (Top), p62 (Middle), and α-Tubulin (Bottom) were detected by western blotting using antibodies recognizing each endogenous protein. kDa, kilodalton. (E) LC3 recruitment to the invaded Salmonella. At 1 h.p.i, mCherry-Salmonella (MOI = 100)-infected WT or KO HeLa cells were fixed and stained with anti-LC3 antibody. Both LC3 signal-positive and -negative Salmonella were counted, and the ratio of LC3-positive Salmonella was calculated. Each dot indicates each cell (n = 10). Red asterisks indicate statistically significant differences versus WT (*P < 0.05, **P < 0.01) (one‐way ANOVA and Dunnett’s multiple comparison test) NS, not significant. (F) Flow cytometry-based intracellular Salmonella growth assay. A schematic presentation of this experiment was shown (Top). The Bottom panels show the results of flow cytometry analysis for the WT HeLa cells infected YFP-Salmonella. Representative dot plot shows Salmonella infected cell cultures at indicated time (Bottom). (G) HEATR3 negatively regulate intracellular Salmonella growth. YFP-Salmonella were infected to WT, HEATR3 KO, or ATG5 KO HeLa cells (MOI = 50). After incubation at the indicated time, cells were analyzed by a flow cytometer. The number of cells with YFP signal high (YFP^High) and YFP signal low (YFP^Low) were counted and the ratio of YFP^High/YFP-positive (YFP^Low+YFP^High) were calculated. The graph displays the average results obtained from three distinct experiments conducted independently. Red asterisks indicate statistically significant differences versus WT (**P < 0.01) (two‐way ANOVA and Dunnett’s multiple comparison test). (H) Gentamicin protection assay. Salmonella was infected to WT, HEATR3 KO, or ATG5 KO HeLa cells (MOI = 50). After incubation with indicated time with gentamicin, cells were lysed and the number of Salmonella was counted by colony formation assay. The fold increase of colony formation units (CFU)/mL was shown (Bottom) with bar chart indicates the raw value of CFU/mL (Top). The graph displays the average results obtained from three distinct experiments conducted independently. Red asterisks indicate statistically significant differences versus WT (**P < 0.01, two‐way ANOVA and Dunnett’s multiple comparison test). (I) LC3 interaction is required for HEATR3 function in Salmonella growth inhibition. WT cells, HEATR3 KO cells, HEATR3 KO cells stably expressing HEATR3 WT or LIR2A mutant HeLa cells were infected with salmonella and performed gentamicin protection assay. Fold change of internal salmonella (8 h/1 h) were shown in the upper graph. The graph displays the average results obtained from three distinct experiments conducted independently. Red asterisks indicate statistically significant differences versus WT (**P < 0.01) (one‐way ANOVA and Dunnett’s multiple comparison test). NS, not significant (Bottom panel). The protein expression of HEATR3 and α-Tubulin was confirmed by western blotting using anti-HEATR3 antibody, anti-α-Tubulin, respectively (Top). (J) HEATR3 is required for NOD2 mediated NF-kB signaling. HEK293T WT, HEK293T HEATR3 KO, HEATR3 WT reexpressed HEATR3 KO or HEATR3 LIR2A-reexpressed HEATR3 KO cells were transfected with pCAG-OSF-NOD2 and p5xNF-kB-luciferase reporter. After 24 h of transfection, cells were treated with or without muramyl dipeptide (MDP) for 6 h. Luciferase activity in the cell lysates were measured. The graph displays the average results obtained from three distinct experiments conducted independently. Red asterisks indicate statistically significant differences versus WT (**P < 0.01) (two‐way ANOVA and Tukey’s multiple comparison test). NS, not significant.

Fig. 4.

HEATR3 is recruited to the damaged lysosome and degraded by lysophagy. (A) Subcellular localization of HEATR3, Gal3, and LAMP-1 in HeLa cells treated with LLOMe. HeLa cells were transfected with a plasmid expressing YFP-Gal3 and Myc-fused full length of HEATR3. The cells were treated with 1 mM of LLOMe for 15 min and then washed. After washout of LLOMe, cells were incubated for 2 h. The cells were fixed and detected YFP-Gal3 (green), Myc-fused HEATR3 (magenta), and LAMP-1 using anti-Myc and anti-LAMP-1 antibodies, respectively. Magnified images of each channel (Bottom) and merged images of YFP-Gal3 and HEATR3 (Right) were shown. (Scale bar, 10 µm.) (B) Subcellular localization of endogenous HEATR3 and Gal3 in HeLa cells treated with LLOMe. HeLa cells transiently transfected with YFP-Gal3 (green) expressing vector were treated with 1 mM of LLOMe for 15 min and then washed. After washout of LLOMe, cells were incubated for 2 h. The cells were fixed and stained with anti-HEATR3 (magenta) and ant-LAMP1 antibodies, respectively. (Scale bar, 10 µm.) Magnified images of each channel (Bottom) and merged images (Right) were shown. (Scale bar, 10 µm.) (C) Stimulated emission depletion (STED) superresolution microscopy analysis of HEATR3 localized on the damaged membrane. HeLa cells were transfected with a plasmid expressing YFP-Gal3 and Myc-HEATR3^C. After 24 h post transfection, the cells were treated with 1 mM of LLOMe for 15 min and then washed. After washing out of LLOMe, the cells were further incubated for 2 h. The cells were fixed and YFP-Gal3 (green) and HEATR3^C (magenta), which were stained with anti-Myc, were observed STED microscopy. Scale bars in magnified panels represent 5 µm (Left), 2 µm (Middle), and 500 nm (Right), respectively. (D) A scheme of lysosome delivery assay using tandem fluorescent (tf) tag fusion proteins. (E) HEATR3 is delivered to lysosomes in an autophagy-dependent manner. HeLa WT, ATG5 KO and FIP200 KO cells expressing mRFP-EGFP-HEATR3^C were treated with 1 mM of LLOMe 15 min. After washout LLOMe, cells were further incubated 2 or 24 h. The cells were fixed and observed RFP and GFP signals by fluorescent microscopy. (Scale bar, 10 µm.) GFP- and RFP-positive dots and RFP-positive dots were counted, and the ratio of GFP⁻ RFP+/GFP⁺ RFP⁺ (Right graph) was calculated. W/O indicates time after LLOMe washout. Dots indicate each cell (WT n = 24, ATG5 KO n = 27, FIP200 KO n = 25) (^∗∗P < 0.01) (one‐way ANOVA and Dunnett’s multiple comparison test).

Fig. 5.

HEATR3 is required for elimination of damaged lysosome. (A) Lysosomal translocation of Gal3 was impaired in HEATR3 KO cells. HeLa WT, HEATR3 KO, Hexa KO, or Penta KO cells expressing mRFP-EGFP-Gal3 were treated with 1 mM of LLOMe 15 min. After washout LLOMe, cells were further incubated for 24 h. The cells were fixed and observed RFP and GFP signals by fluorescent microscopy. Magnified images were shown (Bottom). (Scale bar, 10 µm.) W/O indicates time after LLOMe washout. GFP- and RFP-positive dots and RFP-positive dots were counted and calculated the ratio of GFP⁻ RFP+/GFP⁺ RFP⁺ (Right graph). The graph was extracted from SI Appendix, Fig. S4C W/O 24 h data. The number of the cells analyzed was shown in parenthesis (^**P < 0.01) (two-way ANOVA, and Tukey’s multiple comparison test). NS, not significant. (B) LIR of HEATR3 is required for Gal3 translocation to the lysosome. WT HeLa cells, HEATR3 KO cells, HEATR3 KO cells stably expressing HEATR3 WT or LIR2A mutant were expressed with mRFP-EGFP-Gal3. Cells were treated with 1 mM of LLOMe 15 min. After washout LLOMe, cells were further incubated 24 h. This experimental method was the same as that in (A) (WT n = 25, HEATR3 KO n = 23, HEATR3 KO+WT n = 27, HEATR3 KO+LIR2A n = 21) (^**P < 0.01 versus HEATR3KO cells) (one-way ANOVA and Dunnett’s multiple comparison test). NS, not significant. (C) Colocalization of Gal3 and LC3 was impaired in HEATR3 KO cells. WT HeLa cells (Upper panels), HEATR3 KO cells (Middle panels), Penta KO cells, and Hexa KO cells (Lower panels) expressing YFP-Gal3 were treated with 1 mM of LLOMe 15 min. After washout LLOMe, cells were further incubated for 24 h. The cells were fixed, YFP-Gal3 (green) and LC3 (magenta), which were stained with anti-LC3 antibody, were observed by fluorescent microscopy. (Scale bar, 10 µm.) Magnified images were shown in Right panels. (Scale bar, 5 µm.) The area of overlap between Gal3 and LC3 signals was calculated by means of image analysis. Each dot indicates each cell (WT n = 49, HEATR3 KO n = 60, Penta KO n = 61, Hexa KO n = 62) (^∗∗P < 0.01) (one-way ANOVA and Dunnett’s multiple comparison test).

Fig. 6.

The recruitment of HEATR3 to the damaged membrane is dependent on Ca²⁺ influx, but not autophagy induction. (A) Colocalization of HEATR3 and LC3 on the damaged membrane. HeLa cells were transfected with a plasmid expressing YFP-Gal3 and Myc-HEATR3^C. After 24 h post transfection, the cells were treated with 1 mM of LLOMe for 15 min and then washed-out. After washing out of LLOMe, the cells were further incubated for 2 h. The cells were fixed and stained with anti-Myc and anti-LC3 antibodies. YFP-Gal3 (green), Myc-HEATR3^C (magenta), and LC3 (cyan) were detected using confocal laser microscopy. (Scale bar, 10 µm.) Magnified images of each channel and merged image also shown. (B) Dynamics of HEATR3 and Galectin3 dots formation after LLOMe treatment. Twenty-four hours after transfection of YFP-Gal3 or mCherry-HEATR3^C expressing vector, the cells were treated with LLOMe and live-cell imaging analysis was performed. The Y-axis indicates the number of dots per cell. The X-axis indicates time after LLOMe treatment (minutes). The graph displays the average of dots number in each cell (n = 10). (C) Recruitment of HEATR3 to the damaged membrane is independent of its LC3 interaction. Hela cells were transiently expressing CFP-Gal3 and mCherry-HEATR3^WT or HEATR3^LIR2A were treated with (Bottom) or without (Top) 1 mM LLOMe for 15 min. After washing-out LLOMe, cells were further incubated 2 h. Then, cells were fixed and CFP (green) and mCherry (magenta) signals were observed by fluorescent microscopy. Magnified images were shown in Right panels. (Scale bar, 10 µm.) (D) Recruitment of HEATR3 to the damaged membrane in ATG5 or FIP200 KO cells. ATG5 or FIP200 KO HeLa cells were transfected with a plasmid expressing YFP-Gal3 and Myc-HEATR3^C. The experimental method was the same as (A). YFP-Gal3 (green) and LAMP1 (magenta) and HEATR3 (cyan), which were detected using anti-LAMP1 and anti-Myc antibodies. (Scale bar, 10 µm.) Magnified images were shown in Right panels. (E) Ca²⁺ influx is important for HEATR3 recruitment to the damaged membrane. HeLa cells transiently expressing YFP-Gal3 and mScarlet-HEATR3^C were treated for 30 min with (Bottom) or without (Top) 15 µM BAPTA-AM, then treated with 1 mM LLOMe for 15 min. After washing out of LLOMe, cells were further cultured for 2 h. After fixation, the mScarlet (magenta) and YFP (green) signals were observed. (Scale bar, 10 µm.) Magnified images were shown in Right panels. Colocalization ratio (Pearson’s R) of HEATR3 and Gal3 was calculated using image J (Right graph). Each dot indicates each cell (BAPTA-: n = 29, BAPTA+: n = 23) (unpaired two-sample Student’s t test). (F) LLOMe treatment induces Ca²⁺ flux from lysosomes into the cytoplasm. A schematic presentation of the LAMP1-GCaMP Ca²⁺ sensor is shown (Left panel). The results of LAMP1-GCaMP activation after LLOMe treatment (Right panel). LAMP1-GCaMP transiently expressed in HeLa cells was treated with 1 mM LLOMe, and live-cell imaging analysis was performed. The Y-axis represents the change in intensity of LAMP1-GCaMP signals, normalized to the average LAMP1-GCaMP signal intensity before LLOMe addition (F0). The X-axis represents time (min) (n = 13 cells). (G) Local Ca²⁺ influx at the lysosomal membrane damage site following LLOMe treatment. LAMP1-GCaMP and mCherry-Gal3 were transiently expressed in HeLa cells, which were treated with 1 mM LLOMe for 15 min, washed, and incubated for an additional 15 min before fixation. LAMP1-GCaMP (green) and mCherry-Gal3 (magenta) signals were detected using confocal laser microscopy. (Scale bar, 10 µm.) Magnified images of each channel and the merged image are shown. (H) HEATR3 is recruited to the site of Ca²⁺ influx on the lysosomal membrane. HeLa cells transfected with a plasmid expressing LAMP1-GCaMP and mScarlet-HEATR3^C were subjected to the same treatment as in (G). LAMP1-GCaMP (green) and mScarlet-HEATR3^C (magenta) were detected using confocal laser microscopy. (Scale bar, 10 µm.) Magnified images of each channel and the merged image are shown. (I) Model of HEATR3 activation during Salmonella infection. HEATR3 recognizes and accumulates at membrane damage sites, triggering NOD2 signaling and inducing xenophagy by recruiting LC3 through its LIR.