BCL2 regulates antibacterial autophagy in the intestinal epithelium

Abstract

Autophagy is a key innate immune defense mechanism in intestinal epithelial cells. Bacterial invasion of epithelial cells activates antibacterial autophagy through a process that requires the innate immune adaptor protein MYD88, yet how MYD88 signaling connects to the autophagy machinery is unknown. Here, we show that the mouse intestinal pathogen Salmonella enterica Serovar Typhimurium (Salmonella Typhimurium) triggers MYD88 signaling that regulates binding of the anti-autophagy factor B cell lymphoma 2 (BCL2) to the essential autophagy protein Beclin1 (BECN1) in small intestinal enterocytes, a key epithelial cell lineage. Salmonella infection activated the kinase c-Jun N-terminal protein kinase 1 (JNK1) downstream of MYD88. JNK1 induced enterocyte BCL2 phosphorylation, promoting dissociation of the inhibitory BCL2-BECN1 complex and releasing BECN1 to initiate autophagy. Mice with BCL2 phosphorylation site mutations that prevent BCL2-BECN1 dissociation showed increased Salmonella invasion of enterocytes and dissemination to extraintestinal sites. These findings reveal that BCL2 links MYD88 signaling to enterocyte autophagy initiation, providing mechanistic insight into how invading bacteria trigger autophagy in the intestinal epithelium.

Keywords: autophagy; enterocyte; innate immunity; intestinal epithelium; pathogenic bacteria.

Fig. 1.

Bacterial infection promotes phosphorylation of BCL2 and induces dissociation of the BCL2–BECN1 complex in small intestinal epithelial cells. (A) Schematic of the BCL2–BECN1 interaction and its role in activating autophagy. BECN1 initiates autophagosome formation (9). BCL2 complexes with BECN1 to inhibit autophagy, and metabolic signals arising from starvation or exercise trigger dissociation of the BCL2–BECN1 complex as a prerequisite to autophagy (12, 13). The mechanism requires phosphorylation of BCL2, which induces dissociation from BECN1 and promotes autophagy (15). (B) Immunofluorescence microscopy of LC3 in the small intestines of uninfected and S. Typhimurium–infected conventional wild-type mice. Mice were infected intragastrically with 5 × 10⁹ colony-forming units (CFU) of S. Typhimurium (SL1344) and killed 24 h later. Sections of the distal small intestine were stained with rabbit anti-LC3 and Cy3-conjugated anti-rabbit IgG (red) and counterstained with DAPI (blue) to detect nuclei. [Scale bars, 50 μm (Left); 10 μm (Right)]. (C) Immunofluorescence microscopy of LC3 in the small intestinal villus tips of uninfected and S. Typhimurium–infected germ-free wild-type mice. Mice were infected and LC3⁺ structures were detected as in B. Examples of LC3⁺ puncta are indicated with arrowheads. (Scale bars, 10 μm.) Results are representative of images from n = 3 mice per group and five independent experiments. Immunofluorescence controls are shown in SI Appendix, Fig. S1. (D) Immunoblot analysis of small intestinal epithelial cells from uninfected or S. Typhimurium–infected germ-free wild-type mice. Mice were infected intragastrically as in B. Epithelial cell lysates were blotted and probed with antibodies against BCL2, p-BCL2, LC3, and Mito70 (loading control). LC3-I and LC3-II denote nonlipidated and lipidated LC3, respectively. Each lane represents one mouse. Results are representative of three independent experiments. (E) p-BCL2 and BCL2 band densities in D were measured by scanning densitometry and the ratio of the densities was calculated. Each data point represents one mouse. (F) Immunofluorescence microscopy of organoids derived from the small intestines of wild-type C57Bl/6 mice. Organoids were infected with 10⁶ CFU of S. Typhimurium (SL1344) for 4 h, then fixed, embedded, and sectioned. Sections were detected with rabbit anti-LC3 and Cy3-conjugated anti-rabbit IgG (red) and counterstained with DAPI (blue) to detect nuclei. Arrowheads indicate examples of LC3⁺ puncta. (Scale bars, 10 μm.) (G) LC3⁺ structures in images from E were counted. Each data point is the average number of LC3⁺ puncta in each cell of organoids from an individual mouse; at least 10 individual organoids were counted from each mouse (n = 3 mice per group). Results are representative of three independent experiments. (H) Immunoblot analysis of wild-type small intestinal organoids with and without S. Typhimurium infection. Organoids cultured from wild-type mice were infected with 10⁶ CFU of S. Typhimurium for 4 h. Organoid lysates were immunoblotted with antibodies against BCL2, p-BCL2, LC3, and Mito70 (loading control). LC3-I and LC3-II denote nonlipidated and lipidated LC3, respectively. Results are representative of three independent experiments. (I) Coimmunoprecipitation of BECN1 with BCL2 in small intestinal epithelial cells from uninfected or S. Typhimurium–infected germ-free mice. Epithelial cell lysates were immunoprecipitated with mouse anti-BECN1, blotted, and probed with mouse anti-BCL2-HRP plus rabbit anti-BECN1 (Novus Biologicals) and goat anti-rabbit IgG-HRP. The presence of two BECN1 isoforms in the immunoprecipitate has been observed in published reports (–18). Each lane represents one mouse. Results are representative of three independent experiments. (J) Band intensities from the BECN1 immunoprecipitations in I were measured by scanning densitometry and the ratios of the BCL2 and BECN1 densities were calculated. Each data point represents one mouse. uninf., uninfected; S. Tm, Salmonella Typhimurium; sm. int. epith. cells, small intestinal epithelial cells; p-BCL2, phosphorylated BCL2; IP, immunoprecipitation. Means ± SEM are plotted; *P < 0.05; ***P < 0.001; ****P < 0.0001 by Student’s t test.

Fig. 2.

BCL2 phosphorylation promotes antibacterial autophagy in small intestinal epithelial cells. (A) Schematic depicting how the BCL2^AAA mutation impacts the BCL2–BECN1 interaction and autophagy activation. Activating signals promote BCL2 phosphorylation at three conserved amino acid residues: T69, S70, and S87 (15). Bcl2^AAA mice harbor a knock-in Bcl2 allele in which these residues are mutated, rendering BCL2 nonphosphorylatable and unable to release BECN1 to initiate autophagy. These mice are thus deficient in stimulus-induced but not basal autophagy (13). (B) Immunoblot of small intestinal organoids from wild-type (WT) and Bcl2^AAA mice. Organoids were infected with S. Typhimurium–GFP (10⁶ CFU) for 4 h or were left uninfected. Immunoblots were detected with antibodies against BCL2, p-BCL2, LC3, and Mito70 (loading control). LC3-I and LC3-II denote nonlipidated and lipidated LC3, respectively. Results are representative of three independent experiments. (C) Immunofluorescence microscopy of small intestines from infected and uninfected conventionally raised wild-type and Bcl2^AAA littermates. Mice were infected intragastrically with S. Typhimurium–GFP (5 × 10⁹ CFU per mouse) and were killed 24 h later. Sections of the distal small intestine were stained with rabbit anti-LC3 and Cy3-conjugated anti-rabbit IgG (red), and goat anti-GFP and DyLight 488-conjugated anti-goat IgG (green). Intrinsic GFP fluorescence is destroyed by tissue fixation, necessitating antibody detection of GFP. Sections were counterstained with DAPI (blue) to detect nuclei. (Scale bars, 50 μm.) (D) Immunofluorescence microscopy of small intestinal villus tips from uninfected and S. Typhimurium–infected mice. Mice were infected and tissues stained as in C. Examples of LC3⁺ structures are indicated with arrowheads, and examples of intracellular S. Typhimurium–GFP are indicated with arrows. (Scale bars, 10 μm.) Images are representative n = 6 mice per group and two independent experiments. Immunofluorescence controls are shown in SI Appendix, Fig. S1. (E) LC3⁺ puncta in images from D were counted. Each data point is the average number of LC3⁺ puncta per cell from one mouse (n = 6 mice per group). At least 200 crypt-villus units were counted per mouse, and all epithelial cells in each crypt-villus unit were counted (note that enterocytes at the villus tips typically have more LC3⁺ puncta than enterocytes in the lower regions of the crypt-villus unit) (5). (F) Cell-associated S. Typhimurium–GFP in images from D were counted. Each data point is the average number of S. Typhimurium per intestinal epithelial cell (IEC) from one mouse (n = 6 mice per group); at least 200 crypt-villus units were counted per mouse. (G) Immunofluorescence microscopy of small intestinal organoids grown from wild-type and Bcl2^AAA littermates. Organoids were infected with 10⁶ CFU of S. Typhimurium–GFP for 4 h, then fixed, embedded, and sectioned. Sections were detected with rabbit anti-LC3 and Cy3-conjugated anti-rabbit IgG (red), and goat anti-GFP, DyLight 488-conjugated anti-goat IgG (green), and DAPI (blue) to detect nuclei. Arrowheads indicate examples of LC3⁺ puncta. (Scale bars, 5 μm.) Images are representative of organoids from n = 5 mice per group and two independent experiments. (H) LC3⁺ puncta in images from G were counted. Each point is the average number of LC3⁺ puncta in each organoid cell from one mouse (n = 5 mice per group); at least 10 organoids were counted from each mouse. (I) Intracellular S. Typhimurium–GFP in images from G were counted. Each point represents organoids from one mouse (n = 8 mice per group and three independent experiments). (J) Numbers of intracellular S. Typhimurium–GFP in organoids grown from wild-type and Bcl2^AAA littermates as determined by dilution plating. Each point represents organoids from one mouse (n = 6 mice per group). (K) Bacterial burdens CFU in the small intestine (sm. int.), mesenteric lymph node (MLN), spleen, and liver of wild-type and Bcl2^AAA littermates 24 h after intragastric infection with 5 × 10⁹ CFU S. Typhimurium–GFP. Bacterial counts were determined by dilution plating. Each point represents an individual mouse, and data are from three independent experiments. Geometric means ± SEM are plotted. WT, wild-type; S. Tm, Salmonella Typhimurium; GFP, green fluorescent protein; IEC, intestinal epithelial cell; p-BCL2, phosphorylated BCL2; CFU, colony-forming units. Means ± SEM are plotted except where noted; *P < 0.05; **P < 0.01; ****P < 0.0001; ns, not significant by Student’s t test.

Fig. 3.

JNK1 phosphorylates BCL2 and regulates antibacterial autophagy in small intestinal enterocytes. (A) Immunoblot of small intestinal epithelial cells from uninfected or S. Typhimurium–infected germ-free wild-type mice. Mice were infected intragastrically with 5 × 10⁹ CFU of S. Typhimurium–GFP and were killed 24 h later. Small intestinal epithelial cell lysates were blotted and probed with antibodies against JNK1, p-JNK, LC3, and Mito70 as a loading control. LC3-I and LC3-II denote nonlipidated and lipidated LC3, respectively. Each lane represents one mouse. Results are representative of three independent experiments. (B) The p-JNK and JNK1 band intensities in A were measured by scanning densitometry and the ratio of the densities was calculated. Each data point represents one mouse. (C) Immunoblot of wild-type small intestinal organoids with and without S. Typhimurium infection. Organoids were infected with 10⁶ CFU of S. Typhimurium–GFP for 4 h. Organoid lysates were immunoblotted with antibodies against JNK1, p-JNK, LC3, and Mito70 (loading control). Results are representative of three independent experiments. (D) Immunoblot of small intestinal epithelial cells from conventional wild-type and Jnk1^−/− mice. Mice were infected intragastrically with S. Typhimurium–GFP (5 × 10⁹ CFU per mouse) and were killed 24 h later. JNK1, BCL2, p-BCL2, and Mito70 (loading control) were detected. Each lane represents one mouse. Results are representative of three independent experiments. (E) The p-BCL2 and BCL2 band intensities in D were measured by scanning densitometry and the ratio of the densities was calculated. Each data point represents one mouse (n = 4 mice per group). (F) Immunofluorescence microscopy of small intestines from S. Typhimurium–infected conventional wild-type and Jnk1^−/− mice. Wild-type and Jnk1^−/− littermates were orally infected with S. Typhimurium–GFP (5 × 10⁹ CFU per mouse) and were killed 24 h later. Sections of the distal small intestine were stained with rabbit anti-LC3 and Cy3-conjugated anti-rabbit IgG (red), and goat anti-GFP and DyLight 488-conjugated anti-goat IgG (green). Sections were counterstained with DAPI (blue) to detect nuclei. Examples of LC3⁺ structures are indicated with arrowheads, and examples of intracellular S. Typhimurium–GFP are indicated with arrows. (Scale bars, 5 μm.) Images are representative of n = 5 mice per group and two independent experiments. (G) LC3⁺ puncta from images in F were counted. Each data point represents the average number of LC3⁺ puncta per cell from one mouse; at least 200 villi were counted per mouse (n = 5 mice per group). (H) Epithelial cell-associated S. Typhimurium–GFP from images in F were counted. Each data point represents the average number of S. Typhimurium per IEC from one mouse; at least 200 villi were counted per mouse (n = 5 mice per group). (I) Immunoblot of small intestinal organoids from wild-type and Jnk1^−/− mice. Organoids were infected with 10⁶ CFU of S. Typhimurium–GFP for 4 h, then embedded and sectioned. BCL2, p-BCL2, JNK1, LC3, and Mito70 (loading control) were detected with antibodies as in A. Results are representative of three independent experiments. (J) Immunofluorescence microscopy of small intestinal organoids from wild-type and Jnk1^−/− mice. Organoids were fixed, embedded, and sectioned. Sections were detected with rabbit anti-LC3 and Cy3-conjugated anti-rabbit IgG (red), and goat anti-GFP, DyLight 488-conjugated anti-goat IgG (green), and DAPI (blue) to detect nuclei. Arrowheads indicate examples of LC3⁺ puncta. (Scale bars, 5 μm.) Images are representative of organoids from n = 3 mice per group and two independent experiments. (K) LC3⁺ puncta in images from J were counted. Each data point is the average number of LC3⁺ puncta in each cell; at least 10 organoids were counted from each mouse. (L) Intracellular S. Typhimurium–GFP in images from J were counted. Each data point represents organoids from one mouse (n = 5 mice per group). (M) Numbers of intracellular S. Typhimurium–GFP in organoids grown from wild-type and Jnk1^−/− mice as determined by dilution plating. Each data point represents organoids from one mouse (n = 6 mice per group). sm. int. epith. cells, small intestinal epithelial cells; uninf., uninfected; S. Tm, Salmonella Typhimurium; p-JNK, phosphorylated JNK; p-BCL2, phosphorylated BCL2; IEC, intestinal epithelial cell; GFP, green fluorescent protein; WT, wild-type. Means ± SEM are plotted; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns, not significant by Student’s t test.

Fig. 4.

Constitutively active BECN1 rescues the autophagy block in MYD88-deficient enterocytes. (A) BECN1^F121A harbors a Phe-to-Ala point mutation that disrupts the interaction of BECN1 with BCL2, yielding BECN1 that is constitutively active (12, 26, 27). Becn1^F121A knock-in mice thus have increased basal levels of autophagy (12). (B) Immunoblot analysis of BCL2 phosophorylation in wild-type and Becn1^F121A small intestinal organoids with and without S. Typhimurium infection. Organoids were infected with 10⁶ CFU of S. Typhimurium–GFP for 4 h. Organoid lysates were immunoblotted and detected with antibodies against BCL2, p-BCL2, LC3, and Mito70 (loading control). Results are representative of three independent experiments. (C) Immunofluorescence microscopy of small intestinal villus in Myd88^fl/fl, Myd88^ΔIEC, MyD88^fl/fl/Becn1^F121A and MyD88^ΔIEC/Becn1^F121A mice. Mice were infected intragastrically with S. Typhimurium–GFP (5 × 10⁹ CFU per mouse) and killed 24 h later. Sections of the distal small intestine were stained with rabbit anti-LC3 and Cy3-conjugated anti-rabbit IgG (red). Sections were counterstained with DAPI (blue) to detect nuclei. Examples of LC3⁺ puncta are indicated with arrowheads. (Scale bars, 10 μm.) Images are representative of n = 3 mice per group and two independent experiments. (D) LC3⁺ puncta from images in C were counted. Each data point represents the average number of LC3⁺ puncta per cell from one mouse; at least 200 crypt-villus units were counted per mouse (n = 6 mice per group). S. Tm, Salmonella Typhimurium; WT, wild-type; p-BCL2, phosphorylated BCL2; sm. int. epith. cells, small intestinal epithelial cells. Means ± SEM are plotted; ***P < 0.001; ****P < 0.0001; ns, not significant by Student’s t test.

Fig. 5.

MYD88 controls JNK1 and BCL2 phosphorylation and dissociation of the BCL2–BECN1 complex in small intestinal epithelial cells. (A) Immunofluorescence microscopy of small intestinal villus tips from uninfected or S. Typhimurium–infected germ-free wild-type or Myd88^−/− mice. Mice were infected intragastrically with S. Typhimurium–GFP (5 × 10⁹ CFU per mouse) and were killed 24 h later. Sections of the distal small intestine were stained with rabbit anti-LC3 and Cy3-conjugated anti-rabbit IgG (red), and goat anti-GFP and DyLight 488-conjugated anti-goat IgG (green). Sections were counterstained with DAPI (blue) to detect nuclei. Examples of LC3⁺ puncta are indicated with arrowheads, and examples of intracellular S. Typhimurium–GFP are indicated with arrows. (Scale bars, 10 μm.) Images are representative of n = 3 mice per group and two independent experiments. (B) LC3⁺ puncta in images from A were counted. Each data point is the average number of LC3⁺ puncta per cell from one mouse (n = 6 mice per group). At least 200 crypt-villus units were counted per mouse, and all cells in each crypt-villus unit were counted. (C) Epithelial cell-associated S. Typhimurium–GFP in images from A were counted. Each data point is the average number of S. Typhimurium per IEC from one mouse (n = 6 mice per group); at least 200 crypt-villus units were counted per mouse, and all cells in each crypt-villus unit were counted. (D) Immunoblot analysis of small intestinal epithelial cells from uninfected or S. Typhimurium–infected germ-free wild-type and Myd88^−/− mice. Mice were infected intragastrically with S. Typhimurium–GFP and were killed 24 h later. Small intestinal IEC lysates were blotted and probed with antibodies against JNK1, p-JNK, BCL2, p-BCL2, LC3, and Mito70 (loading control). LC3-I and LC3-II denote nonlipidated and lipidated LC3, respectively. Each lane represents one mouse. Results are representative of three independent experiments. (E) The band intensities in D were measured by scanning densitometry and the ratios of the p-JNK and JNK1 densities (Top) and p-BCL2 and BCL2 densities (Bottom) were calculated. Each data point represents one mouse. (F) Immunoblot analysis of wild-type and Myd88^−/− small intestinal organoids with and without S. Typhimurium infection. Organoids were infected with 10⁶ CFU of S. Typhimurium–GFP for 4 h. Organoid lysates were immunoblotted with antibodies against JNK1, p-JNK, BCL2, p-BCL2, LC3, and Mito70 (loading control). Results are representative of three independent experiments. (G) Immunofluorescence microscopy of wild-type and Myd88^−/− organoids. Organoids were infected with 10⁶ CFU of S. Typhimurium–GFP for 4 h, then fixed, embedded, and sectioned. Sections were detected with rabbit anti-LC3 and Cy3-conjugated anti-rabbit IgG (red), and goat anti-GFP and DyLight 488-conjugated anti-goat IgG (green). Sections were counterstained with DAPI (blue) to detect nuclei. Arrowheads indicate examples of LC3⁺ puncta, and arrows indicate examples of S. Typhimurium–GFP. (Scale bars, 5 μm.) Results are representative of n = 9 mice and three independent experiments. (H) LC3⁺ puncta in images from G were counted. Each data point is the average number of LC3⁺ puncta in each cell of organoids from an individual mouse; at least 10 organoids were counted from each mouse (n = 5 mice per group). Results are representative of two independent experiments. (I) Cell-associated S. Typhimurium–GFP in images from G were counted. Each data point is the average number of S. Typhimurium per IEC from one mouse (n = 5 mice per group). At least 200 crypt-villus units were counted per mouse, and all cells in each crypt-villus unit were counted. (J) Numbers of intracellular S. Typhimurium in organoids grown from wild-type and Myd88^−/− littermates as determined by dilution plating. Each data point represents organoids from one mouse (n = 6 mice per group). (K) Coimmunoprecipitation of BECN1 with BCL2 in small intestinal epithelial cells from uninfected or S. Typhimurium–GFP–infected germ-free wild-type and Myd88^−/− mice. Epithelial cell lysates were immunoprecipitated with mouse anti-BECN1, blotted, and probed with mouse anti-BCL2-HRP plus rabbit anti-BECN1 (Santa Cruz) and goat anti-rabbit IgG HRP. Each lane represents one mouse. Results are representative of three independent experiments. (L) The band intensities in K were measured by scanning densitometry and the ratios of the BCL2 and BECN1 densities were calculated. Each data point represents one mouse. (M) Model: Bacterial infection of intestinal enterocytes induces phosphorylation of JNK1 downstream of MYD88. JNK1 phosphorylates BCL2, which dissociates from BECN1 and releases BECN1 to initiate autophagy. BCL2 thus regulates antibacterial autophagy in enterocytes to limit bacterial infection of the intestinal epithelial barrier. S. Tm, Salmonella Typhimurium; WT, wild-type; sm. int. epith. cells, small intestinal epithelial cells; IEC, intestinal epithelial cell; p-JNK, phosphorylated JNK; p-BCL2, phosphorylated BCL2; IP, immunoprecipitation. Means ± SEM are plotted; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns, not significant by Student’s t test.