A Dual Role for Receptor-interacting Protein Kinase 2 (RIP2) Kinase Activity in Nucleotide-binding Oligomerization Domain 2 (NOD2)-dependent Autophagy

Background: Autophagy is triggered by NOD2 as an anti-bacterial response.

Results: NOD2-stimulated autophagy requires RIP2-dependent activation of p38 MAPK and repression of the PP2A phosphatase in intestinal epithelial cell lines.

Conclusion: RIP2 kinase activity is necessary for anti-bacterial autophagy induction by NOD2.

Significance: These findings provide novel molecular targets for modulation of autophagy as an anti-bacterial response.

Abstract

Autophagy is triggered by the intracellular bacterial sensor NOD2 (nucleotide-binding, oligomerization domain 2) as an anti-bacterial response. Defects in autophagy have been implicated in Crohn's disease susceptibility. The molecular mechanisms of activation and regulation of this process by NOD2 are not well understood, with recent studies reporting conflicting requirements for RIP2 (receptor-interacting protein kinase 2) in autophagy induction. We examined the requirement of NOD2 signaling mediated by RIP2 for anti-bacterial autophagy induction and clearance of Salmonella typhimurium in the intestinal epithelial cell line HCT116. Our data demonstrate that NOD2 stimulates autophagy in a process dependent on RIP2 tyrosine kinase activity. Autophagy induction requires the activity of the mitogen-activated protein kinases MEKK4 and p38 but is independent of NFκB signaling. Activation of autophagy was inhibited by a PP2A phosphatase complex, which interacts with both NOD2 and RIP2. PP2A phosphatase activity inhibited NOD2-dependent autophagy but not activation of NFκB or p38. Upon stimulation of NOD2, the phosphatase activity of the PP2A complex is inhibited through tyrosine phosphorylation of the catalytic subunit in a process dependent on RIP2 activity. These findings demonstrate that RIP2 tyrosine kinase activity is not only required for NOD2-dependent autophagy but plays a dual role in this process. RIP2 both sends a positive autophagy signal through activation of p38 MAPK and relieves repression of autophagy mediated by the phosphatase PP2A.

Introduction

Autophagy is a cell stress response that causes the encapsulation of cellular contents in multilamellar vesicles for subsequent degradation and recycling (1). It is best known as a starvation response; however, autophagy is also essential for the capture and removal of intracellular bacteria, such as Salmonella typhimurium, Listeria monocytogenes, and Mycobacterium tuberculosis (2). Genetic variants in several autophagy genes are associated with Crohn's disease (CD),³ a debilitating and chronic inflammatory bowel disease (3–6). There is a strong link between bacteria and CD pathogenesis; therefore, it has been proposed that ineffective bacterial clearance due to impaired anti-bacterial autophagy is an important contributor to the pathogenesis of this chronic inflammatory disease (2, 6).

The first identified CD risk gene is NOD2 (nucleotide-binding oligomerization domain 2), which encodes an intracellular bacteria sensor involved the innate immune response to bacteria (7). NOD2 detects a conserved component of bacterial peptidoglycan consisting of muramyl dipeptide (MDP). MDP is released from bacteria when the cell wall is fragmented as a part of bacterial killing, as well as during bacterial division, or is co-injected into cells with pathogen effector proteins by type III or IV secretion systems. Upon stimulation by MDP, NOD2 oligomerizes and recruits the receptor-interacting protein 2 kinase (RIP2/RICK/CARDIAK). Activation of RIP2 recruits ubiquitin-modifying enzymes and stimulates protein kinase cascades, resulting in the activation of NFκB and the mitogen-activated kinases (MAPKs) p38, JNK, and ERK1/2. These pathways coordinately regulate inflammatory cytokine production, anti-bacterial killing, and the recruitment of other professional immune response cells. Functional analyses of CD-associated NOD2 variants demonstrate defects in both inflammatory signaling and bactericidal activity in response to MDP (8).

Recent reports have demonstrated that NOD2 stimulates autophagy as an anti-bacterial response and that this process is impaired by CD-associated variants in either NOD2 or the autophagy gene, ATG16L1 (9–11). The exact mechanism behind how NOD2 directs autophagosome formation has not yet been determined. There remains some discrepancy regarding whether this process requires NOD2 signaling (9, 10, 12) or is mediated solely by direct recruitment of autophagic machinery to sites of bacterial invasion (11). In this report, we determine that NOD2-dependent signaling is required for autophagy induction in intestinal epithelial cells and examine which signaling components are required for this process. Our studies illustrate a dual role for RIP2 tyrosine kinase activity in NOD2-dependent autophagy through activation of p38 MAPK and repression of PP2A phosphatase activity.

EXPERIMENTAL PROCEDURES

Reagents and Antibodies

MDP was purchased from Bachem (Torrance, CA). Erlotinib (LC Laboratories, Woodburn, MA) was a gift of Derek Abbott (Case Western Reserve University, Cleveland, OH). Antibodies against FLAG (M2), MEKK4 (clone MEKK4-338), Atg1/ULK1 (A7481), or tubulin (clone DM1A) were obtained from Sigma. Antibodies to GAPDH (clone 14C10), phosphorylated NFκB p65 (Ser-536) (clone 93H1), p38 MAPK (catalog no. 9212), phosphorylated p38 (Thr-180/Tyr-182) (catalog no. 9211), and phosphorylated ULK1 (Ser-555) (clone D1H4) were purchased from Cell Signaling (Boston, MA). Anti-LC3B antibody (catalog no. NB100-2220) was purchased from Novus Biologicals (Littleton, CO). Antibodies to RIP2/RICK (H-300), PP2A (1D6), phosphory lated PP2A (Tyr-307) (F-8), PPP2R1A (A-5), and the epitope tags Omni (M-21) and HA (Y-11) were purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Anti-HA (HA.11) antibody was purchased from Covance (Emeryville, CA). A rat monoclonal antibody to RIP2/RICK (Nick-1) was purchased from Assay Designs (Farmingdale, NY). HRP-conjugated donkey secondary antibodies were purchased from Jackson ImmunoResearch Laboratories (West Grove, PA).

RNA Interference (RNAi) Reagents

MISSION short hairpin RNA (shRNA) constructs targeting NOD2 (NM_022162.1-2959s1c1), RIP2 (NM_003821.5-2364s21c1), p38 (NM_139012.x-503s1c1), JNK (NM_002752.x-356s1c1), ERK1 (NM_002746.1-876s1c1), ULK1 (NM_003565.x-5154s1c1), p65 NFκB (NM_021975.1-888s1c1), MEKK4 (NM_011948.1-1792s1c1), ATG16L1 (NM_030803.5-2181s1c1), and nonspecific control (SHC002) were purchased from Sigma. Control siRNA (D-001210-01) and siGENOME SMARTpool siRNA to PPP2R1A (M-101259-02-0005) were obtained from Thermo Scientific (Waltham, MA).

Plasmids

NOD2 expression plasmids as well as the NFκB and CMV β-gal reporters have been described previously (13). The kinase-dead (KD) mutant of MEKK4 (pCMV5-MEKK4KD), dominant negative mutants of IKKα and IKKβ (pFLAG-IKKα DN and pEBG-IKKβ DN), and the wild type and T95M point mutant of RIP2 (pcDNA3-Omni-RIP2 and pcDNA3-Omni-RIP2 T95M) were kind gifts of Derek Abbott (14). Kinase-dead mutants of p38 (pCMV-p38 KD), JNK (pCE4-JNK KD), and ERK1 (pCE4-ERKKD) were gifts of Melanie Cobb (University of Texas Southwestern Medical Center, Dallas, TX). The IκBα superrepressor mutant (S32A/S36A) (pBabe-IκBα-SR, Addgene plasmid 15291) was created by William Hahn (Harvard Medical School, Boston, MA) (15) and purchased from Addgene (Cambridge, MA). Wild type PP2Ac (pEF4C-HA-PP2A), a phosphatase-dead mutant of PP2Ac (pEF4C-HA-PP2Amut), wild type PPP2R1A (pEF4C-FLAG-PPP2R1A), and an amino-terminal deletion mutant of PPP2R1A that lacks binding to PP2Ac (pEF4C-FLAG-PPP2R1AΔN) were kind gifts of Daniel Krappmann (Institute of Toxicology, Nuremberg, Germany) (16).

Cell Culture and Transfection

HCT116 and HEK293T cell lines were maintained in Dulbecco's modified Eagle's medium (Invitrogen) with 10% fetal bovine serum (FBS; Lonza, Allendale, NJ). HCT116 cell lines stably expressing shRNAs were generated by lentiviral infection, followed by selection with 0.5 μg/ml puromycin for 2 weeks. Transient transfection of HCT116 cells was performed by Polyfect (Qiagen, Valencia, CA) or nucleofection with Kit V (Lonza) according to the manufacturers' instructions and assayed 48 h post-transfection. HEK293T cells were transfected using Polyfect (Qiagen) or calcium phosphate as described previously (10, 13) and assayed 24–48 h post-transfection.

Immunofluorescence

HCT116 shRNA cell lines were plated onto glass coverslips for 48 h. Cells were stimulated with 20 μg/ml MDP for 4 h, fixed in 4% paraformaldehyde/phosphate-buffered saline (PBS), and permeabilized in 0.4% Triton X-100, PBS. Cells were washed with PBS, blocked in 2% FBS, PBS, and incubated with anti-LC3B antibody (1:200; catalog no. 2775, Cell Signaling Technology) in 2% FBS, PBS overnight at 4 °C. Cells were subsequently stained with goat anti-rabbit-Alexa488 antibody (1:1,000; catalog no. A11034, Invitrogen) and mounted on slides with Vectashield plus DAPI (Vector Laboratories, Burlingame, CA). Samples were visualized by confocal microscopy using a ×40 objective lens on a Leica TCS-SP spectral laser-scanning confocal microscope equipped with a Q-Imaging Retiga EXi cooled CCD camera and Image-Pro Plus Capture and Analysis software (Media Cybernetics, Silver Spring, MD). LC3⁺ vesicles were scored in z-stack overlays from at least four separate fields and quantitated in an automated fashion using a customized visual basic Image-Pro Plus macro.

Immunoblots

For analysis of phosphosignaling events and protein expression levels, cells were washed twice in PBS and then lysed in Ginger buffer (310 mm Tris, pH 6.8, 25% glycerol, 5% SDS, 715 mm β-mercaptoethanol, 125 mg/ml bromphenol blue) on ice. The resulting lysate was separated by SDS-PAGE, and proteins were transferred to a PVDF membrane. Membranes were blocked in 5% milk, Tris-buffered saline, 0.1% Tween 20 and then probed with primary antibodies overnight at 4 °C. Blots were developed by enhanced chemiluminescence (Millipore, Billerica, MA).

Gentamycin Protection Assays

An overnight culture of Salmonella enterica serovar typhimurium SL1344 was diluted 1:7 and grown at 30 °C for 1 h and then added to cells at a multiplicity of infection of 10 for 30 min. Cells were washed twice with PBS, and then DMEM supplemented with 10% FBS and 50 μg/ml gentamycin (Sigma) was added for 1 h. Cells were lysed in 50 μl of lysis buffer (PBS, 0.1% Triton X-100), and dilutions were plated in duplicate onto Luria broth plates. After growth overnight at 30 °C, colonies recovered were counted, and colony-forming units/well were calculated (10). Statistical significance between groups was determined by using a two-tailed t test. Results were considered significant when p was ≤0.05.

Immunoprecipitation-coupled Mass Spectral Screen

Identification of NOD2-interacting proteins by immunoprecipitation of NOD2 complexes from MDP-stimulated HEK293T cells stably expressing low levels of FLAG-NOD2 followed by liquid chromatography-coupled tandem mass spectrometry has been described in detail previously (17).

NFκB and p38 MAPK Reporter Gene Assays

Luciferase reporter gene assays to measure NFκB or p38 MAPK activity were performed in HCT116 or HEK293T cells as described previously (17).

Cytokine Secretion Assays

Cells were plated in triplicate and stimulated for 18 h with MDP (100 ng/ml). Interleukin-8 (IL-8) levels secreted into the cell culture medium were determined by the Quantikine human IL-8 immunoassay (R&D Systems, Minneapolis, MN) according to the manufacturer's instructions.

RESULTS

The Kinase Activity of RIP2 Is Required for MDP-stimulated Autophagy Induction

The autophagic process is highly conserved and can be broken down into discrete steps (1). First, an isolation membrane nucleates in a process initiated by the ULK1 (uncoordinated-51-like kinase) complex, which subsequently activates a type III phosphatidylinositol 3-kinase (PI3K) complex. The isolation membrane is then elongated under the control of two ubiquitin-like conjugation systems composed of autophagy-related gene (ATG) proteins. The first system is composed of ATG7 and ATG10, which assist formation of an elongation complex containing ATG5, ATG12, and ATG16L1. The ATG5-ATG12-ATG16L1 complex localizes to the isolation membrane and aids in the recruitment of microtubule-associated protein light chain 3 protein (LC3), which is modified from a cytosolic form (LC3-I) to a lipid-conjugated form (LC3-II) and inserted into the forming autophagosomal membrane. The autophagosome then fuses with a lysosome, resulting in the degradation of autophagosomal contents by lysosomal enzymes.

The molecular requirements for NOD2-dependent autophagy induction have not been clearly elucidated. S. typhimurium is an intracellular bacteria targeted by autophagy, and the clearance of this bacterium is often used to monitor autophagy (18). We have previously demonstrated that MDP stimulation during infection enhances Salmonella killing in a NOD2- and ATG16L1-dependent manner (10). To further dissect the essential components of NOD2-dependent autophagy, we examined the requirement for RIP2 (a kinase required for NOD2 signaling), ULK1 (a kinase involved in autophagy initiation), and Beclin-1 (a component of the PI3K complex) in this process by RNAi knockdown in the human intestinal epithelial cell line HCT116 (Fig. 1, A, B, and J, and supplemental Fig. 1). Our data demonstrate that expression of ULK1, Beclin-1, or ATG16L1 is required for enhancement of Salmonella killing by MDP in gentamycin protection assays (Fig. 1A and supplemental Fig. 1A). MDP-enhanced bacterial killing was also ablated when expression of NOD2 or RIP2 was inhibited by RNAi (Fig. 1, A and B). The role of these proteins in MDP-stimulated autophagy was confirmed in additional assays. The formation of LC3⁺ vesicles in response to MDP was dependent on both NOD2 and RIP2 expression as determined by quantification of confocal micrographs (Fig. 1, D and E). Likewise, an MDP-stimulated increase in LC3-II levels was dependent on NOD2 and RIP2 expression (Fig. 1, G and H). These findings suggest that NOD2-dependent autophagy is dependent on the intersection of NOD2 signaling with an autophagic cascade initiated by ULK1.

FIGURE 1.

MDP-enhanced bacterial killing is mediated by NOD2, RIP2 kinase activity, and autophagy. A, gentamycin protection assay performed in HCT116 cells stably expressing shRNA targeting the indicated proteins. Cells were treated with medium or MDP (10 μg/ml) during the course of Salmonella infection. The assay was performed in triplicate, with averages ± S.D. (error bars) graphed. **, p < 0.01. B, gentamycin protection assay performed in HCT116 cells stably expressing RIP2 shRNA. The assay was performed as in A. *, p < 0.05. C, gentamycin protection assay performed in HCT116 cells pretreated with DMSO (Mock) or 100 nm erlotinib for 1 h. Cells were treated with medium or MDP (10 μg/ml) during the course of Salmonella infection. The assay was performed in triplicate, with averages ± S.D. graphed. *, p < 0.05. D, quantification of confocal micrographs of anti-LC3-immunostained HCT116 cells stably expressing the indicated shRNA constructs. Cells were unstimulated (Media) or treated with MDP (20 μg/ml, 4 h). Cells with two or more LC3⁺ puncta were considered positive, with four fields measured per condition. A representative confocal micrograph for each condition is shown. Green, LC3 immunofluorescence; blue, DAPI-stained nuclei. Data are representative of three independent experiments, average ± S.D. shown. **, p < 0.01. E, quantification of cells with LC3⁺ puncta as performed in D. *, p < 0.05. F, quantification of cells with LC3⁺ puncta as performed in D, except HCT116 cells were pretreated with DMSO (Mock) or 100 nm erlotinib for 1 h prior to MDP stimulation. **, p < 0.01. G, immunoblot of LC3-II levels in HCT116 cell lysates from cells stably expressing the indicated shRNA stimulated with MDP (10 μg/ml) for the indicated times. Samples were also blotted for GAPDH as a loading control. H, immunoblot of LC3-II levels as performed in G. I, immunoblot of LC3-II levels in HCT116 cell lysates from cells pretreated with DMSO (Mock) or 100 nm erlotinib for 1 h and then stimulated with MDP (10 μg/ml) for the indicated times. Samples were also blotted for GAPDH as a loading control. J, confirmation of target gene knockdown by immunoblot or quantitative real-time PCR of the stable shRNA-expressing HCT116 cell lines used in these studies. ***, p < 0.001.

RIP2 has been described to promote NLR-dependent signaling, acting as both a kinase and a protein scaffold (7, 14). To investigate whether RIP2 kinase activity is required for NOD2-dependent autophagy induction, we treated HCT116 cells with erlotinib, a newly described RIP2 tyrosine kinase inhibitor (14). We observed that erlotinib pretreatment of HCT116 cells blocked MDP-enhanced Salmonella killing in gentamycin protection assays (Fig. 1C). Additionally, erlotinib pretreatment impaired both MDP-stimulated LC3⁺ puncta formation and LC3-II accumulation in HCT116 cells (Fig. 1, F and I). These findings demonstrate a requirement for RIP2 tyrosine kinase activity in MDP-induced autophagy.

NOD2 Induces Autophagy in a p38 MAPK-dependent Manner

In response to bacterial infection or exposure to MDP, NOD2 recruits RIP2 and activates two major signaling pathways, NFκB and the MAPKs: p38, JNK, and ERK1/2 (7). NFκB signaling involves activation of IκB kinases (IKKα and IKKβ), which phosphorylate the NFκB-inhibitory molecule, IκBα, initiating its degradation and subsequent nuclear translocation of NFκB. However, NOD2-dependent autophagy was not altered when dominant negative mutants of IKKα and -β (IKKα DN and IKKβ DN) or a non-phosphorylatable mutant of IκBα (IκBα-SR) were expressed in HCT116 cells (Fig. 2A). Although these dominant negative molecules were effective at blocking NFκB activity in reporter gene assays (supplemental Fig. 2, A–D), the lack of effect of these constructs in either gentamycin protection assays (Fig. 2A) or LC3-II immunoblots (Fig. 2C) clearly demonstrates that NFκB activation is dispensable for NOD2-mediated autophagy.

FIGURE 2.

NOD2-mediated autophagy requires p38 MAPK activity and is independent of NFκB signaling. A, gentamycin protection assay in HCT116 cells transfected with expression plasmids of dominant negative NFκB signaling pathway molecules for 48 h. Cells were treated with medium or MDP (10 μg/ml) during the course of Salmonella infection and assayed in triplicate, with averages ± S.D. (error bars) graphed. *, p < 0.05; **, p < 0.01. B, gentamycin protection assay performed in HCT116 cells stably expressing the indicated shRNA construct. The assay was performed as in A. C, immunoblot of LC3-II expression in lysates from HCT116 cells stably expressing the indicated shRNA construct and stimulated with MDP (10 μg/ml) for the indicated times. Membranes were also probed for expression of targeted proteins and GAPDH as a loading control. D, gentamycin protection assay performed in HCT116 cells transfected with expression plasmids of kinase-dead p38 MAPK (p38KD) for 48 h. The assay was performed as in A. *, p < 0.05; **, p < 0.01. E, MAPK p38 luciferase reporter gene assay performed in HCT116 cells transfected with reporter and the indicated expression constructs as well as β-galactosidase transfection control. Cells were stimulated with 0.1 μg/ml MDP for 18 h. Luciferase values were normalized to transfection efficiency measured by β-galactosidase activity (nLuc). Averages ± S.D. are graphed. F, immunoblot of LC3-II levels in HCT116 cell lysates from cells transfected with the indicated expression plasmids for 48 h. Cells were stimulated with MDP (10 μg/ml) for the indicated times. Samples were also blotted for expression of the transfected constructs and GAPDH as a loading control. G, gentamycin protection assay in HCT116 cells transfected with RIP2 or RIP2 T95M for 48 h and then pretreated with DMSO, 100 nm erlotinib, or 10 nm SB203580 for 1 h prior to Salmonella infection. The assay was performed in triplicate, with averages ± S.D. graphed. **, p < 0.01; ***, p < 0.001. H, gentamycin protection assay in HCT116 cells transfected with vector or p38 expression construct alongside control or ULK1 shRNA. After 48 h, Salmonella infection was performed in triplicate, with averages ± S.D. graphed. **, p < 0.01; ***, p < 0.001.

Next, we examined components of the MAPK signaling cascade for their role in NOD2-mediated autophagy. The MAPK pathway involves the cell type-specific activation of several mitogen-activated protein kinase kinase kinases (MEKKs), which stimulate signaling cascades, resulting in activation of p38, JNK, and ERK1/2. MDP-enhanced Salmonella killing in HCT116 cells was blocked by RNAi-mediated knockdown of MEKK4 and p38 MAPK but unaffected by knockdown of JNK, ERK1, or NFκB p65 (Fig. 2B and supplemental Fig. 2, E–G). The requirement of p38 MAPK expression for MDP-stimulated autophagy was confirmed in LC3-II immunoblots (Fig. 2C). These results suggest that autophagy induction by NOD2 in intestinal epithelial cells is specifically dependent on p38 MAPK.

Additionally, we examined whether the kinase activity of p38 MAPK was required for NOD2-dependent autophagy. In HCT116 cells, MDP-stimulated Salmonella killing and p38 activity was suppressed by expression of a dominant negative kinase-dead mutant of p38 in gentamycin protection and reporter gene assays (Fig. 2, D and E). The dominant negative kinase-dead p38 mutant also suppressed MDP-induced LC3-II in immunoblots (Fig. 2F).

To further confirm the requirement of p38 kinase activity in NOD2-dependent autophagy, we treated cells with the kinase inhibitor SB203580. SB203580 was originally described as a specific p38 MAPK inhibitor but has now been demonstrated to inhibit both p38 MAPK and RIP2 kinase activities (19). To differentiate between the two inhibitory actions of this drug, we assessed Salmonella killing in HEK293T cells expressing either RIP2 or a RIP2 mutant (T95M) that contains a point mutation in the ATP binding pocket that interferes with the binding of both SB203580 and erlotinib but does not affect RIP2 kinase activity (14, 19). We observed that expression of either RIP2 or RIP2 T95M increased Salmonella killing (Fig. 2G and supplemental Fig. 2H). Treatment of RIP2-expressing cells with either SB203580 or erlotinib inhibited this effect. However, in RIP2 T95M-expressing cells, only SB203580 inhibited Salmonella killing, indicating that p38 kinase activity is required in addition to RIP2 kinase activity.

We further linked p38 activation in our system to autophagy through assessing whether p38 expression in itself is sufficient to induce Salmonella killing via autophagy. In a manner similar to other kinases, overexpression of p38 results in autophos phorylation and activation (20). When p38 was overexpressed in HCT116 cells, we observed an increase in Salmonella killing similar to the effects of MDP treatment (Fig. 2H). Knockdown of ULK1 expression by RNAi blocked both MDP-enhanced killing and the increased killing by p38 overexpression (Fig. 2H), indicating that p38 expression alone in HCT116 cells enhances bacterial killing in a process dependent on autophagy. These findings demonstrate that NOD2 activates autophagy in a process dependent on RIP2 tyrosine kinase activity and mediated by the kinases MEKK4 and p38 but independent of NFκB signaling.

Identification of PPP2R1A as a Novel NOD2-interacting Protein

Having established that NOD2 mediates autophagy through p38, we wished to identify specific regulators of this process. We had previously performed an immunoprecipitation-coupled mass spectrometry screen to identify NOD2-interacting proteins from MDP-stimulated HEK293 cells stably expressing low levels of FLAG-tagged NOD2 protein (17). Analysis of the screen results for enzymes described to modulate both p38 and autophagic activity resulted in the identification of a candidate regulator. This candidate was the protein phosphatase 2 regulatory subunit A α protein (PPP2R1A), a scaffolding component of the protein phosphatase 2A (PP2A) complex (supplemental Fig. 3). PP2A is a ubiquitous serine/threonine phosphatase with a myriad of functions in growth, cell death, and cancer (21). PP2A primarily exists as a heterotrimer consisting of a catalytic subunit (PP2Ac), a scaffolding A subunit (PPP2R1A or PPP2R1B), and a regulatory B subunit, which confers specificity. A smaller fraction of PP2A exists as a heterodimer of PP2Ac and the α4 subunit. PP2A has demonstrated roles in both starvation-induced autophagy (22–24) and p38 modulation (25–29), suggesting that PP2A may also regulate anti-bacterial autophagy mediated by NOD2.

Initially, we confirmed the interaction between NOD2 and PPP2R1A in co-immunoprecipitation assays using epitope-tagged expression constructs in HEK293T cells. To preserve signaling complex stoichiometry, we expressed HA-RIP2 in addition to combinations of Omni-NOD2, FLAG-PPP2R1A, and HA-PP2Ac. When cell lysates were immunoprecipitated for NOD2 using an Omni antibody, complexes containing NOD2, RIP2, PPP2R1A, and PP2Ac were observed (Fig. 3A). We confirmed an endogenous interaction of PPP2R1A with RIP2 by co-immunoprecipitation from HCT116 cells (Fig. 3B) as well as an endogenous interaction of RIP2 with PP2Ac (Fig. 3C). These endogenous interactions were not significantly altered in immunoprecipitation assays from MDP-stimulated cells. These findings demonstrate a novel interaction between the NOD2-RIP2 complex and a PP2A phosphatase heterocomplex mediated through the scaffold protein PPP2R1A.

FIGURE 3.

The NOD2-RIP2 complex interacts with the PPP2R1A-PP2A phosphatase. A, co-immunoprecipitation of a PP2Ac-PPP2R1A complex with the NOD2-RIP2 complex. HEK293T cells were transfected with indicated combinations of Omni-NOD2, HA-RIP2, HA-PP2Ac, and FLAG-PPP2R1A, and cell lysates were immunoprecipitated with Omni antibody. Proteins were detected by immunoblot. B, co-immunoprecipitation of endogenous PPP2R1A by endogenous RIP2 in lysates of HCT116 cells. Cells were either unstimulated or stimulated with MDP (10 μg/ml, 1 h) prior to lysis and immunoprecipitation with rabbit IgG (lane C) or RIP2 antibodies. Proteins detected by immunoblot. C, co-immunoprecipitation of endogenous RIP2 by endogenous PP2Ac in lysates of HCT116 cells. Cells were either unstimulated or stimulated with MDP (10 μg/ml, 1 h) prior to lysis and immunoprecipitation with rabbit IgG (lane C) or PP2Ac antibodies. Proteins were detected by immunoblot.

Characterization of PPP2R1A as a Negative Regulator of NOD2-mediated Autophagy

The role of PPP2R1A in the PP2A phosphatase complex is to provide a scaffold for recruitment of the catalytic and regulatory subunits and to direct PP2Ac to specific targets (21, 30). We assessed if alterations in PPP2R1A expression or function affected MDP-stimulated autophagy in HCT116 cells. MDP-stimulated autophagy was increased in cells when PPP2R1A expression was knocked down by RNAi, as measured by LC3-II immunoblots (Fig. 4A). Reduction of PPP2R1A expression also increased bacterial killing in gentamycin protection assays with decreased Salmonella survival in cells treated with PPP2R1A RNAi but not with control RNAi (Fig. 4B). This enhancement of killing occurred in the absence of exogenous MDP stimulation and was not further increased by MDP treatment. Because MDP is released during the natural course of Salmonella infection, we examined whether the enhancement of Salmonella killing by PPP2R1A RNAi knockdown in the absence of exogenous MDP stimulation was due to enhanced NOD2 activity. When NOD2 activity was blocked by expression of a dominant negative NOD2 construct (NOD2 D291N), the PPP2R1A RNAi-enhanced Salmonella killing was ablated (Fig. 4C). These results indicate that NOD2-dependent killing of Salmonella is negatively regulated by PPP2R1A.

FIGURE 4.

The PPP2R1A-PP2A phosphatase negatively regulates NOD2-dependent autophagy. A, LC3-II immunoblot of lysates from HCT116 cells transfected with control or PPP2R1A siRNA for 48 h and stimulated with MDP (10 μg/ml) for the indicated times. Samples were also blotted for knockdown of PPP2R1A and GAPDH as a loading control. B, gentamycin protection assay performed in HCT116 cells transfected with control or PPP2R1A siRNA for 48 h. Cells were treated with medium or MDP (10 μg/ml) during Salmonella infection and assayed in triplicate, with averages ± S.D. graphed. **, p < 0.01. C, gentamycin protection assay performed in HCT116 cells transfected with vector or a dominant negative NOD2 (NOD2 DN) and control or PPP2R1A siRNA for 48 h. Cells were treated with medium or MDP (10 μg/ml) during Salmonella infection and assayed in triplicate, with averages ± S.D. (error bars) graphed. **, p < 0.01; ***, p < 0.001. D, gentamycin protection assay performed in HCT116 cells transfected with vector or PPP2R1AΔN for 48 h. Cells were treated with medium or MDP (10 μg/ml) during Salmonella infection and assessed in triplicate, with averages ± S.D. graphed. **, p < 0.01; ***, p < 0.001. E, immunoblot of transfected HCT116 cells to confirm expression of FLAG-tagged PPP2R1A and PPP2R1AΔN constructs. GAPDH was probed as a loading control. F, gentamycin protection assay was performed in HCT116 cells transfected with empty vector or PPP2R1AΔN and control, NOD2, p38, or ATG16L1 shRNA for 48 h. Cells were treated with medium or MDP (10 μg/ml) during Salmonella infection and assayed in triplicate, with averages ± S.D. graphed. **, p < 0.01; ***, p < 0.001.

These findings are complemented by the results obtained with a dominant negative mutant of PPP2R1A that lacks the amino-terminal PP2Ac interaction domain (PPP2R1AΔN) (16). We observed a significant increase in Salmonella killing in the presence and absence of exogenous MDP treatment when PPP2R1AΔN was expressed in HCT116 cells (Fig. 4, D and E). These results are similar to what was observed with PPP2R1A knockdown (Fig. 4B). The increase in Salmonella killing by PPP2R1AΔN expression required the expression of NOD2, p38, and ATG16L1 because RNAi-mediated knockdown of these genes blocked this effect in gentamycin protection assays (Fig. 4F). These findings identify PPP2R1A as a novel regulatory component of MDP-induced autophagy.

The PP2Ac-PPP2R1A Phosphatase Complex Regulates NOD2-mediated Autophagy

Previous studies have shown that PP2A modulates autophagy both positively and negatively, depending on the cell type and stimulus used (22–24). Our data indicate that the PPP2R1A subunit of PP2A regulates MDP-stimulated autophagy, so we assessed the requirement of the catalytic subunit (PP2Ac) in this process. Overexpression of PP2Ac in HCT116 cells blocked MDP-stimulated bacterial killing but not enhanced killing stimulated by rapamycin in gentamycin protection assays (Fig. 5A). PPP2R1A expression was also required for PP2Ac inhibition of NOD2-mediated Salmonella killing because depletion of PPP2R1A by RNAi knockdown prevented PP2Ac blockade of MDP-enhanced killing (Fig. 5B). PP2Ac was also demonstrated to inhibit MDP-stimulated LC3-II levels in immunoblots (Fig. 5C).

FIGURE 5.

The phosphatase activity of the PPP2R1A-PP2A complex is required for regulation of MDP-enhanced autophagy. A, gentamycin protection assay performed in HCT116 cells transfected with vector or PP2Ac expression plasmids for 48 h. Cells were treated with medium, MDP (10 μg/ml), or rapamycin (25 μg/ml) during Salmonella infection and assayed in triplicate, with averages ± S.D. (error bars) graphed. ***, p < 0.001. B, gentamycin protection assay performed in HCT116 cells transfected with vector or PP2Ac expression plasmids and control or PPP2R1A siRNA for 48 h. Cells were treated with medium or MDP (10 μg/ml) during Salmonella infection and assayed in triplicate, with averages ± S.D. graphed. **, p < 0.01. C, immunoblot of LC3-II levels from HCT116 cells transfected with empty vector, wild-type PP2Ac (PP2Ac Wt), and phosphatase-dead PP2Ac (PP2Ac Mt) expression constructs and treated with MDP (10 μg/ml) for the indicated times. Membranes were also probed with HA antibody to confirm expression of PP2Ac constructs and GAPDH as a loading control. D, gentamycin protection assay performed in HCT116 cells transfected with vector, PP2Ac Wt, or PP2Ac Mt expression plasmids for 48 h. Cells were treated with medium or MDP (10 μg/ml) during Salmonella infection and assayed in triplicate, with averages ± S.D. graphed. **, p < 0.01. E, gentamycin protection assay performed in HCT116 cells transfected with vector or PP2Ac Mt expression plasmids and control, NOD2, p38, or ATG16L1 shRNA for 48 h. Cells were treated with medium or MDP (10 μg/ml) during Salmonella infection and assayed in triplicate, with averages ± S.D. graphed. ***, p < 0.001.

Next, we investigated the requirement of PP2A phosphatase activity for modulation of NOD2-dependent autophagy. Expression of a phosphatase-dead PP2Ac mutant (PP2Ac Mt) enhanced MDP-stimulated LC3-II levels and increased Salmonella killing in the presence and absence of MDP stimulation (Fig. 5, C and D). This increase in Salmonella killing stimulated by the PP2Ac mutant was dependent on NOD2, p38, and ATG16L1 expression because RNAi-mediated knockdown of these genes blocked this effect (Fig. 5E). These results indicate that a PP2A phosphatase complex consisting of PP2Ac and PPP2R1A is a selective inhibitor of NOD2-triggered anti-bacterial autophagy.

PP2A Inhibits MDP-stimulated Autophagy Downstream of the MAPK p38

Next, we determined if PP2A inhibited NOD2 signaling globally or specifically targeted processes required for autophagy induction. Therefore, we examined whether overexpression of PP2Ac affected NOD2-dependent activation of NFκB or p38 MAPK by reporter gene assays in NOD2-expressing HEK293T cells. In contrast to the dramatic effects on MDP-induced autophagy (Fig. 5, A and C), PP2Ac overexpression had no effect upon either MDP-stimulated NFκB or p38 activity (Fig. 6, A and B), suggesting a role downstream of p38 activation. We confirmed the effect of PP2Ac on events downstream of p38 through measurement of IL-8 secretion in response to MDP. The MAPK p38 has been demonstrated to be critical for IL-8 production in response to inflammatory stimuli (31). Overexpression of PP2Ac blocked IL-8 secretion in response to MDP stimulation (Fig. 6C), suggesting that PP2Ac does not affect NOD2 activation of NFκB and MAPK pathways but rather modulates a target of p38 activity involved in autophagy induction and IL-8 secretion.

FIGURE 6.

PP2A targets NOD2 signaling downstream of NFκB and p38 MAPK activation. A, NFκB luciferase reporter assay performed in HEK293T cells transfected with NOD2 and vector or PP2Ac constructs for 24 h. Cells were treated with medium or MDP (10 ng/ml, 18 h) and assayed in triplicate. Samples were normalized to β-galactosidase activity, and averages ± S.D. (error bars) are graphed. B, MAPK p38 luciferase reporter assay performed in HEK293T cells transfected with NOD2 and vector or PP2Ac constructs for 24 h. Cells were treated with medium or MDP (100 ng/ml, 18 h) and assayed in triplicate. Samples were normalized to β-galactosidase activity, and averages ± S.D. are graphed. C, IL-8 secretion was assessed by ELISA from cells used in A. Samples were assayed in triplicate, with averages ± S.D. graphed.

PP2Ac Is Inactivated upon Activation of RIP2 Kinase by MDP Treatment

Because PP2Ac phosphatase activity inhibits MDP-induced autophagy, we investigated whether PP2A activity was altered by MDP treatment. PP2Ac activity is tightly controlled by post-translational modifications of the carboxyl terminus, which include phosphorylation on threonine 304 and tyrosine 307 as well as carboxymethylation of leucine 309 (30, 32, 33). The best characterized regulatory site is phosphorylation of Tyr-307, which has been reported to reduce PP2Ac activity by up to 90% (32). We assessed whether MDP treatment altered tyrosine phosphorylation of PP2Ac on Tyr-307 by immunoblot using a phosphospecific antibody to PP2Ac Tyr-307. We observed a significant increase in Tyr-307 phosphorylation upon MDP treatment of HCT116 cells (Fig. 7A). Induction of PP2Ac phosphorylation was dependent on RIP2 expression because RNAi-mediated knockdown of RIP2 ablated MDP stimulated phosphorylation of PP2Ac Tyr-307 (Fig. 7A). RIP2 kinase activity was also found to be required for this modification because pretreatment of HCT116 with erlotinib similarly blocked MDP-stimulated phosphorylation of Tyr-307 (Fig. 7B). However, RIP2 does not appear to directly phosphorylate PP2Ac because results from in vitro kinase assays were negative (supplemental Fig. 4). We also tested whether PPP2R1A expression was required for MDP-induced phosphorylation of PP2A on Tyr-307. We found that MDP-induced phosphorylation of Tyr-307 was prevented by RNAi knockdown of PPP2R1A (Fig. 7C). These results indicate that upon MDP stimulation, the phosphatase activity of the PP2Ac-PPP2R1A complex is inhibited through the phosphorylation of PP2Ac on Tyr-307 in a process dependent on the kinase activity of RIP2.

FIGURE 7.

The catalytic subunit of PP2A is modified in response to MDP in a process dependent on RIP2 and PPP2R1A. A, immunoblot analysis of PP2A Tyr-307 phosphorylation in lysates from HCT116 stably expressing control or RIP2 shRNA and stimulated with MDP (10 μg/ml) for the indicated times. Samples were also immunoblotted for levels of total PP2Ac, RIP2, and GAPDH as a loading control. B, immunoblot analysis of PP2A Tyr-307 phosphorylation in lysates from HCT116 cells pretreated with DMSO or 100 nm erlotinib for 1 h prior to stimulation with MDP (10 μg/ml) for the indicated times. Samples were also immunoblotted for GAPDH as a loading control. C, immunoblot analysis of PP2A Tyr-307 phosphorylation in lysates from HCT116 cells transfected with either control or PPP2R1A siRNA for 48 h and then stimulated with MDP (10 μg/ml) for the indicated times. Samples were also immunoblotted for PPP2R1A and GAPDH as a loading control.

DISCUSSION

Autophagy is a crucial innate immune response for the clearance of intracellular pathogens (2). Previous studies have linked genetic variants of autophagy genes and NOD2 to defects in autophagy and an increased susceptibility to CD (3–5, 9–11). Additionally, NOD2 function is involved in clearance of several bacteria targeted by autophagy, such as M. tuberculosis, M. leprae, S. typhimurium, and L. monocytogenes (7, 34). A more complete understanding of the mechanisms underlying the induction of autophagy may have a significant effect on development of new therapeutics for CD and chronic bacterial diseases, such as tuberculosis and leprosy.

The exact mechanism behind how NOD2 directs autophagosome formation is still under examination, and there has been some discrepancy regarding whether this process requires NOD2 signaling (9, 10, 12) or is mediated solely by direct recruitment of autophagic machinery to sites of bacterial invasion (11). Our data as well as data from other groups (9, 10, 12) support a requirement for RIP2- and NOD2-dependent signaling for autophagy induction (Fig. 8). This contrasts with the findings of Travassos et al. (11), who found that autophagy is still activated in RIP2-deficient mouse embryo fibroblasts in response to Shigella infection (11). The difference in our results is most likely due to the cell type (fibroblasts versus epithelial cells) and the strain of bacteria (Shigella versus Salmonella) examined and highlights the need to study autophagy in a cell- and stimulus-dependent context.

FIGURE 8.

Model of the molecular pathway required for NOD2-dependent autophagy. Essential activating components identified in this study are outlined in black and filled in with dark gray, whereas negative regulators are shown in light gray. Non-essential NOD2 signaling molecules are shown in light gray outlines. Connecting lines ending in balls indicate an inhibitory effect. Dashed lines indicate an indirect link between components.

Several reports examining NOD2-dependent autophagy demonstrate a requirement for RIP2 expression (9, 10, 12), but the role of RIP2 kinase activity in this process was unknown. Early analyses of NOD1 and NOD2 signaling indicated that RIP2 functioned primarily as a scaffolding protein to induce the proximity of signaling molecules for the initiation of NFκB signaling (35). This was further supported by reports demonstrating that the polyubiquitination of RIP2 required for the recruitment of TAK1 is independent of RIP2 kinase activity (36, 37). Later studies also suggested that the only role for RIP2 kinase activity was in stability of RIP2 expression (38, 39). However, recent reports indicate that RIP2 tyrosine kinase activity is required for MDP-induced signaling and cytokine secretion (14). Our data demonstrate that NOD2 signaling is required for autophagy induction, and RIP2 tyrosine kinase activity plays an essential dual-faceted role in this process. RIP2 tyrosine kinase activity both provides an activating signal through stimulation of MAPK p38 and relieves PP2Ac-PPP2R1A-mediated repression of autophagy (Fig. 8).

The MAPK p38 has been reported to either negatively or positively regulate autophagy dependent on stimulus as well as cell context. Mechanistically, the process by which p38 regulates autophagy is not well understood, with both enzymatic and transcriptional roles proposed. For example, TLR4 (Toll-like receptor 4)-stimulated autophagy is mediated by p38-dependent expression of the autophagy gene, LRG47 (40). In other reports, it has been proposed that p38 activation and production of reactive oxygen species results in autophagy induction; however, these reports differ on whether autophagy is stimulated by p38-dependent reactive oxygen species generation or activation of p38 by reactive oxygen species (41–43). Alternatively, p38 activity has been suggested to suppress autophagy by mechanisms targeting ATG9 trafficking and activation of mTOR (44, 45). Our studies demonstrate a positive role for p38 in the induction of autophagy in a ULK1-, Beclin-1-, and ATG16L1-dependent manner and independent of NFκB signaling in human intestinal epithelial cells (Fig. 8). This contrasts with recently published results showing an essential role for ERK1/2 in Listeria-stimulated autophagy in mouse macrophages and may reflect differences in anti-bacterial responses of professional immune cells versus epithelial cells. Future studies will focus on defining the mechanism by which p38 kinase activity promotes NOD2-stimulated autophagy in epithelial cells.

Activation of p38 as well as other MAPKs has been shown to be regulated by PP2A (46). This has been most clearly demonstrated in the context of tumorigenesis and cellular transformation, where inhibition of PP2A leads to enhanced MAPK activity and cellular proliferation (29). PP2A targets multiple steps in the MAPK cascade, depending on the subunit composition of the PP2A complex. For example, PP2Ac-α4 inhibits TNFα-stimulated p38 MAPK activation through dephosphorylation of the upstream MAPK kinase, MEK3 (27). This contrasts with the results of our studies, where we did not observe an effect of PP2A on the activation of p38 MAPK in response to MDP (Fig. 6). Conversely, the PP2Ac-PPP2R1A complex we analyzed appears to specifically target a substrate downstream of p38 involved in both NOD2-dependent autophagy and IL-8 secretion in epithelial cells.

PP2A has been reported to play both positive and negative roles in autophagy induction. This may be a reflection of multiple mechanisms for autophagy induction, which vary for different stimuli, as well as the combinatorial nature of the PP2A complex, where numerous regulatory subunits can combine with the PP2Ac subunit to form dimeric or trimeric complexes (21). For example, a dimeric PP2Ac-α4 complex has been demonstrated to negatively regulate starvation-induced autophagy in a process dependent on ULK1 (24). Likewise, we found that NOD2-triggered autophagy is also dependent on ULK1 and repressed by PP2A, although the complex we identified in our cells was composed of PP2Ac-PPP2R1A. We determined that this PP2A complex interacts with NOD2-RIP2 and becomes tyrosine-phosphorylated on a negative regulatory site (Tyr-307) upon MDP stimulation. Although RIP2 tyrosine kinase activity is required for phosphorylation of PP2Ac, our in vitro kinase assay results suggest that PP2A is not a direct target of RIP2 kinase activity (supplemental Fig. 4). Future studies, including the identification of the RIP2-activated tyrosine kinase responsible for inhibitory phosphorylation of PP2A and the downstream targets of the PP2A phosphatase, will provide further insight into the molecular signaling mechanisms controlling anti-bacterial autophagy.