General Event in the Execution of Apoptosis a Proinflammatory Signal

Cell Death Triggered by Yersinia enterocolitica Identifies Processing of the Proinflammatory Signal Adapter MyD88 as a General Event in the Execution of Apoptosis This information is current as of September 30, 2021. Lena Novikova, Nicole Czymmeck, Anne Deuretzbacher, Friedrich Buck, Kathleen Richter, Alexander N. R. Weber, Martin Aepfelbacher and Klaus Ruckdeschel J Immunol published online 20 December 2013 http://www.jimmunol.org/content/early/2013/12/20/jimmun Downloaded from ol.1203464

References This article cites 40 articles, 15 of which you can access for free at: http://www.jimmunol.org/ http://www.jimmunol.org/content/early/2013/12/20/jimmunol.1203464.ful l#ref-list-1

Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists by guest on September 30, 2021

• Fast Publication! 4 weeks from acceptance to publication

*average

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2013 by The American Association of Immunologists, Inc. All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. Published December 20, 2013, doi:10.4049/jimmunol.1203464 The Journal of Immunology

Cell Death Triggered by Yersinia enterocolitica Identiﬁes Processing of the Proinﬂammatory Signal Adapter MyD88 as a General Event in the Execution of Apoptosis

Lena Novikova,* Nicole Czymmeck,* Anne Deuretzbacher,*,† Friedrich Buck,‡ Kathleen Richter,x,{ Alexander N. R. Weber,‖ Martin Aepfelbacher,* and Klaus Ruckdeschel*,x

Many pathogenic microorganisms have evolved tactics to modulate host cell death or survival pathways for establishing infection. The enteropathogenic bacterium Yersinia enterocolitica deactivates TLR-induced signaling pathways, which triggers apoptosis in macrophages. In this article, we show that Yersinia-induced apoptosis of human macrophages involves caspase-dependent cleavage of the TLR adapter protein MyD88. MyD88 was also cleaved when apoptosis was mediated by overexpression of the Toll–IL-1R Downloaded from domain–containing adapter inducing IFN-b in epithelial cells. The caspase-processing site was mapped to aspartate-135 in the central region of MyD88. MyD88 is consequently split by caspases in two fragments, one harboring the death domain and the other the Toll–IL-1R domain. Caspase-3 was identified as the protease that conferred the cleavage of MyD88 in in vitro caspase assays. In line with a broad role of caspase-3 in the execution of apoptosis, the processing of MyD88 was not restricted to Yersinia infection and to proapoptotic Toll–IL-1R domain–containing adapter inducing IFN-b signaling, but was also triggered by staurosporine treatment. The cleavage of MyD88 therefore seems to be a common event in the advanced stages of apoptosis, when http://www.jimmunol.org/ caspase-3 is active. We propose that the processing of MyD88 disrupts its scaffolding function and uncouples the activation of TLR and IL-1Rs from the initiation of proinflammatory signaling events. The disruption of MyD88 may consequently render dying cells less sensitive to proinflammatory stimuli in the execution phase of apoptosis. The cleavage of MyD88 could therefore be a means of conferring immunogenic tolerance to apoptotic cells to ensure silent, noninflammatory cell demise. The Journal of Immunology, 2014, 192: 000–000.

oll-like receptors are transmembrane receptors that play typic TIR–TIR domain interaction (3). MyD88 is a central TIR an essential role in the recognition of foreign, invading adapter that is implicated in TLR and IL-1R signaling (4). It is by guest on September 30, 2021 T microbial pathogens. They are activated by conserved, activated by the IL-1R family and by all TLRs with the exception pathogen-associated molecular patterns and trigger the induction of of TLR3. In contrast, TLR3, stimulated by microbial dsRNA, protective proinflammatory signaling pathways (1, 2). The human signals via the Toll–IL-1R domain–containing adapter inducing TLR family comprises . 10 TLRs, each detecting a specific set of IFN-b (TRIF). TRIF is recruited also by TLR4, which is the sole ligands. LPS from Gram-negative bacteria, for instance, activates TLR that can engage both the MyD88 and TRIF pathways. The TLR4, whereas bacterial lipoproteins signal through TLR2. The TIR adapter protein–dependent signals converge on NF-kB and TLR-triggered effects depend on the activation of signaling cas- MAPK activation, which mounts the inflammatory response. Mean- cades that are initiated by the recruitment of intracytoplasmic while, it is well established that the TLRs, in addition to promoting adapter proteins to the Toll–IL-1R (TIR) domain through homo- inflammation, are also implicated in controlling cellular life and death. Hence, the engagement of TLRs may enable cell survival, but may also elicit cell death through either apoptosis or necrosis *Institute for Medical Microbiology, Virology and Hygiene, University Medical (5, 6). The signaling networks that control the differential out- † Center Eppendorf, 20246 Hamburg, Germany; Division of Hematology, Department come are not yet completely understood. However, it seems that of Internal Medicine II, University Hospital of Wu¨rzburg, 97080 Wu¨rzburg, Ger- many; ‡Institute of Clinical Chemistry, University Medical Center Eppendorf, 20246 the type of cell, as well as the composition of the signals induced x Hamburg, Germany; Max von Pettenkofer-Institute for Hygiene and Medical Mi- by the microbial challenge, determines the fate of the stimulated crobiology, 80336 Munich, Germany; {Stiefel/GlaxoSmithKline Early Development, ‖ Durham, NC 27709; and Department of Immunology, University of Tu¨bingen, cell. 72076 Tu¨bingen, Germany Several pathogenic bacteria have evolved mechanisms to mod- Received for publication December 19, 2012. Accepted for publication November ify TLR signaling pathways for establishing infection (7, 8). One 20, 2013. example is given by the Gram-negative bacterial pathogen Yersi- This work was supported by Deutsche Forschungsgemeinschaft Grants Ru788/3-1, 3-2. nia, which deregulates TLR-induced immune responses by en- Address correspondence and reprint requests to Dr. Klaus Ruckdeschel, Institute for gaging Yersinia outer protein (Yop) P in Y. enterocolitica, or the Medical Microbiology, Virology and Hygiene, University Medical Center Eppendorf, homologous YopJ protein in Y. pestis and Y. pseudotuberculosis. Martinistrasse 52, 20246 Hamburg, Germany. E-mail address: [email protected] YopP/YopJ is injected with several other Yops by Yersinia type III Abbreviations used in this article: DD, death domain; HEK, human embryonic kidney; IKK, inhibitory kB kinase; IRAK, IL-1R–associated kinase; MOI, multiplicity protein secretion into host cells to modify critical signaling path- of infection; PVDF, polyvinylidene difluoride; TAK1, TGF-b–activated kinase-1; ways of innate immunity (9, 10). YopP/YopJ acts as an acetyltrans- TIR, Toll–IL-1R; TRIF, Toll–IL-1R domain–containing adapter inducing IFN-b; ferase on the NF-kB–activating IkB kinase (IKK) b and on mem- Yop, Yersinia outer protein. bers of the MAPK kinase superfamily (11, 12). The acetylation of Copyright Ó 2013 by The American Association of Immunologists, Inc. 0022-1767/13/$16.00 these kinases deactivates the NF-kB and MAPK pathways and

www.jimmunol.org/cgi/doi/10.4049/jimmunol.1203464 2 TARGETING AND REPROGRAMMING OF MyD88 IN APOPTOSIS downregulates the proinflammatory host cell response. Because micin (100 mg/ml) after 90 min of infection to prevent bacterial over- these pathways also promote cell survival, the action of YopP/YopJ growth. Staurosporine (Sigma-Aldrich, Munich, Germany) was applied sensitizes macrophages to undergo apoptosis following Yersinia to the cells at a concentration of 5 mM. Where indicated, the cells were pretreated with 40 mM of the caspase inhibitor zVAD-fmk (Bachem, infection (13, 14). The proapoptotic response thereby is accelerated Bubendorf, Switzerland) 30 min prior to infection. by the activation of TLR4 and TRIF, whereas the MyD88 pathway Eukaryotic expression plasmids, cell transfection, and seems to be dispensable for Yersinia-induced cell death (15, 16). measurement of NF-kB activation TRIF is connected to the cell death machinery by the adapter’s receptor-interacting protein 1 and Fas-associated death domain Transfections of HEK293 cells were performed by the calcium phosphate (DD) to initiate apoptosis through caspase-8 (16–18). The signal- method using ProFection according to the manufacturer’s instructions (Promega, Mannheim, Germany). For the detection of transiently over- ing of apoptosis following TLR4 or TLR3 activation seems to be expressed Myc-tagged MyD88 (24), the cells were transfected with 3 mg a prominent function of TRIF, which is substantiated by the finding expression plasmid in 6-cm cell culture dishes. Where indicated, plas- that the overexpression of TRIF significantly compels transfected mids for TRIF (25), the caspase-8 inhibitor CrmA (26), or IL-1R–as- cells to undergo apoptosis (16–18). sociated kinase (IRAK) 4 (27) were included. Expression plasmids for In the current study, we report on a feedback control mechanism MyD88, IRAK4, TRIF, and CrmA, were kindly provided by Tularik (South San Francisco, CA), Shizuo Akira (Research Institute for Microbial Dis- that links apoptosis execution with the modulation of proinflam- eases, Osaka University, Osaka, Japan), and David Vaux (Walter and Eliza matory signaling. It is shown that human MyD88 is cleaved when Hall Institute of Medical Research, Melbourne, Australia), respectively. apoptosis is induced by Yersinia infection or TRIF overexpres- The total amount of DNA was kept constant with empty vector for each sion. The processing of MyD88 is mediated by caspase-3 and transfection condition (6 mg). To insert point mutations into Myc-tagged MyD88, the wild-type myd88 gene was subjected to site-directed muta- appears to be a general event downstream from effector caspase genesis using the QuikChange Mutagenesis Kit, as recommended by the Downloaded from activation. Caspase-3 cleaves MyD88 within the MyD88 inter- manufacturer (Stratagene, Agilent Technologies, Santa Clara, CA). Trun- mediate domain at aspartate-135. This cleaving separates the cated versions of MyD88 were generated by PCR amplification and by MyD88 N-terminal DD from the C-terminal TIR domain and subcloning of the DNA products into the vector backbone of the wild-type disrupts the ability of MyD88 to function as proinflammatory MyD88-myc expression plasmid pRK5 (24). All PCR products were checked for sequence accuracy by sequencing. adapter that connects the activation of TLR and IL1-R members To reconstitute the MyD88-deficient, bone marrow–derived immortal- to downstream signaling events. The cleavage of MyD88 there- ized macrophages with either wild-type or D135A-mutagenized MyD88, http://www.jimmunol.org/ fore may help to uncouple the engagement of these transmem- the cells were transfected with the respective MyD88 expression plasmids brane receptors from the initiation of proinflammatory signaling using Lipofectamine LTX & Plus Reagent in line with the manufacturer’s instructions (Invitrogen; by Life Technologies, Darmstadt, Germany). The in apoptotic cells. Several proinflammatory signal transmitters cells were seeded 1 d before transfection in 6-well cell culture vessels have been reported to be cleaved during apoptosis, such as NF- (0.5 3 106 cells) for immunoblotting experiments and in 24-well plates kB p65, components of the IKK complex, receptor-interacting (0.2 3 106 cells) for analysis of apoptosis. A total of 2.5 mg plasmid DNA protein kinases, and TRIF itself (19–22). The degradation and was combined with 2.5 ml Plus Reagent and 3.5 ml Lipid Reagent per well processing of MyD88 could complementarily reduce the sensi- for transfection of cells in 6-well plates, whereas 1 mg plasmid DNA, 1 ml Plus Reagent, and 1.5 ml Lipid Reagent were applied for 24-well plate tivity of apoptotic cells to proinflammatory stimuli and thereby transfection according to the manufacturer’s protocol. The cells were an- counteract inflammation triggered by the TLR and IL-1R fam- alyzed 1 d later. The transfection efficiency usually varied in a range be- by guest on September 30, 2021 ily. This activity may fine-tune the apoptotic process and pre- tween 30 and 50%, as determined by transfection of pEGFP expression serve apoptosis as a discrete, noninflammatory cell death mech- plasmid and analysis by fluorescence microscopy. To measure NF-kB activation following MyD88 overexpression, HEK293 anism. cells were seeded in 24-well cell culture plates and transfected with the MyD88 expression constructs (40 ng of each) together with endothelial leukocyte adhesion molecule-1–NF-kB firefly luciferase (200 ng) and Materials and Methods phRL-null Renilla luciferase reporter vectors (20 ng of each; Promega, Cell culture, stimulation, and infection conditions Madison, WI). Cells were lysed 20 h after transfection and luciferase activities were measured with a microtiter plate chemiluminometer (Tecan, The human embryonic kidney (HEK) 293 cell line was cultured in DMEM Ma¨nnedorf, Switzerland) according to the manufacturer’s instructions (Dual- cell growth medium containing 10% heat-inactivated FCS (Life Technol- Luciferase Reporter System, Promega; Ref. 15). The NF-kB–directed firefly ogies, Darmstadt, Germany). Peripheral blood monocytes were isolated luciferase activities were normalized to Renilla luciferase activities to form buffy coats of healthy donors by gradient centrifugation using Ficoll compensate for differences in transfection efficiencies. To assess the ef- cell separation medium (PAA Laboratories, Pasching, Austria). The cells fect of the TIR domain–harboring, truncated MyD88 expression construct were resuspended in cell growth medium (RPMI 1640 medium supple- DMyD88_C-136 on TLR-dependent NF-kB activation, HEK293 cells were mented with 10% heat-inactivated FCS and 5 mM L-glutamine) and dif- transfected with TLR4 (20 ng), CD14 (20 ng), and MD2 (2 ng) expression ferentiated to macrophages by incubation with 10 ng ($100 U) of M-CSF plasmids, together with the luciferase reporter vectors. The plasmids for (PeproTech, Hamburg, Germany) per milliliter for 7 d. Nonadherent, nondif- TLR4, CD14, and MD2 were kindly provided by Tularik (South San Fran- ferentiated cells were thoroughly removed by washing. MyD88-deficient, cisco, CA; Ref. 28) and by Kensuke Miyake (Division of Infectious Genetics, murine bone marrow–derived macrophages immortalized by retrovirus (23) University of Tokyo, Tokyo, Japan; Ref. 29), respectively. At 24 h after were kindly provided by Katherine A. Fitzgerald (Division of Infectious transfection, the cells were stimulated with Yersinia or LPS treatment Disease and Immunology, University of Massachusetts Medical School, (5 mg/ml) for an additional 20 h before luciferase activities were measured, Worcester, MA). The immortalized macrophages were cultured in DMEM as described above. Data on NF-kB activities are the means 6 SEM of containing 10% heat-inactivated FCS, 5% heat-inactivated horse serum, four to six independent experiments, and p values were calculated using a 1 mM sodium pyruvate, 10 mM Hepes, 1% nonessential amino acids, and two-tailed, unpaired Student t test. 2mML-glutamine (Life Technologies). Cells were always incubated at 37˚C and with 5% CO2 in a humidified atmosphere. Western blotting The Y. enterocolitica strains used in this study were the serotype O8 wild-type strain WA, and its isogenic yopP-knockout mutant WA- For assessment of cellular MyD88 levels, cells were solubilized with lysis DyopP (13). For infection, overnight bacterial cultures grown at 27˚C buffer containing 150 mM NaCl, 1% Nonidet P-40, 0.5% deoxycholic acid, were diluted 1:20 in fresh Luria–Bertani broth and grown for another 0.1% SDS, 50 mM Tris (pH 8.0), and phosphatase and protease inhibitors 2 h at 37˚C. Shift of the growth temperature to 37˚C initializes activation (Life Technologies). The lysates were cleared by centrifugation, separated of the Yersinia type III secretion machinery for efficient translocation by SDS-PAGE, and transferred to polyvinylidene difluoride (PVDF) mem- of Yops into the host cell upon cellular contact. To equalize and syn- brane. For the detection of MyD88, the membrane was probed with rabbit chronize infection, bacteria were seeded on the cells by centrifugation polyclonal Ab directed against the C terminus of human MyD88 (recog- at 200 3 g for 5 min at a ratio of 20 bacteria per cell. For incubation nizing amino acids 279–296; Sigma-Aldrich, Munich, Germany; or a re- times . 90 min, bacteria were routinely killed by addition of genta- gion around amino acid 233; Cell Signaling Technology, Danvers, MA). The Journal of Immunology 3

Alternatively, myc-tagged overexpressed MyD88 was labeled by probing and subjected to N-terminal Edman sequencing using an automated protein with a monoclonal mouse anti-myc epitope tag Ab (Life Technologies). sequencer (473A; Applied Biosystems, Foster City, CA). Immunoreactive bands were visualized using appropriate secondary Abs and ECL detection reagents (GE Healthcare, Bucks, U.K.). When required, the membranes were stripped by successive incubation with 0.1 N NaOH Results and 0.1 N HCl for 30 s and reused for a second immunoblotting procedure. MyD88 is processed during the execution of apoptosis The stripping method was also applied to recycle the membranes for controlling equal protein loading of the gels with cellular lysates by detecting Our previous studies have shown that MyD88 relays the Yersinia- actin or HSC-70 with appropriate Abs (mouse anti-actin; Merck-Millipore, induced, proinflammatory signals downstream from TLR2 and Darmstadt, Germany; and mouse anti–HSC-70; Santa Cruz Biotechnology, TLR4, but it does not decisively contribute to the proapoptotic Dallas, TX). Phospho-specific immunoblotting against IRAK4 and TGF-b– activated kinase-1 (TAK1) was performed as described above, using mAbs response following Yersinia infection (15, 16). In this study, directed against phosphorylated threonine-345 and serine-346 of IRAK4, or we assessed the protein levels of MyD88 in human monocyte- phosphorylated threonine-184 and threonine-187 of TAK1 (Cell Signaling derived macrophages infected with Yersinia. Infections were Technology, Danvers, MA). The total cellular pool of transfected IRAK4 performed with the Y. enterocolitica wild-type strain WA, or its was labeled by stripping the membrane and successive immunoblotting with isogenic YopP-negative mutant WA-DyopP, which is impaired in anti-HA (Cell Signaling Technology). Because the recycled membranes showed reduced sensitivity to successive TAK1 immunostaining with the apoptosis induction (13). The cellular MyD88 levels were ana- available Abs, the cellular levels of TAK1 were analyzed in parallel control lyzed 20 h later. In these experiments, it was noticed that the samples by immunoblotting with anti-TAK1 Ab (Santa Cruz Biotechnology, amount of MyD88 was diminished when cells were infected with Dallas, TX). The levels of MyD88 or of the phosphorylation of IRAK4 and the wild-type strain WA (Fig. 1A, lane 2), whereas no significant TAK1 were quantified in relation to actin or the total cellular pool of the respective protein by densitometry using Adobe Photoshop CS5.1 software. alteration was observed post infection with the YopP-negative Downloaded from All experiments were repeated several times, and representative results of mutant WA-DyopP (lane 3). MyD88 disappeared in correlation at least four independent experiments are shown. with the extent of apoptosis triggered by Yersinia and was nearly complete after 20 h, a time point at which . 80% of the cells Detection and quantification of apoptotic cells infected with the wild-type strain were apoptotic (Fig. 1B, lane Apoptotic macrophages were detected by labeling of cells with fluorescein- 6). Importantly, coadministration of the broad-spectrum caspase conjugated annexin V (Roche, Mannheim, Germany), which binds to phos- inhibitor zVAD-fmk substantially stabilizedMyD88 levels in the phatidylserine exposedontheouterleafletofapoptoticcells.Thesimultaneous WA-infected cells (Fig. 1A, lane 4). This finding suggests that http://www.jimmunol.org/ application of the red DNA stain propidium iodide (Sigma-Aldrich) allowed the discrimination of apoptotic from necrotic nuclei in cells that had al- caspases, activated in the course of Yersinia-induced apoptosis, ready lost their membrane integrity. The percentages of apoptotic cells may target the cellular levels of MyD88. Of interest, the amount were determined by scoring $ 200 cells per sample by fluorescence micros- of MyD88 was also affected by staurosporine treatment. Staur- copy (13). Results are expressed as the mean percentage 6 SD of apoptotic osporine triggers apoptosis through activation of the intrinsic cells versus the total number of cells from four independent experiments, mitochondrial pathway and conferred 89 6 6% apoptosis to the and p values were calculated using two-tailed, unpaired Student t test. macrophages within 20 h. At that time point, MyD88 was sub- Caspase cleavage assay stantially decreased (Fig. 1A, right panel), which was compa-

rable to the MyD88 levels in infection with wild-type Yersinia. by guest on September 30, 2021 The caspase in vitro cleavage experiments were conducted on MyD88 using recombinant caspases purchased from BioVision (Milpitas, CA). Thus, the MyD88 levels seemed be subjected to negative regu- MyD88 was either precipitated from HEK293 cells after overexpression or lation by caspases once apoptosis was induced, irrespective of purified from Escherichia coli as GST-fusion protein. For immunopre- the apoptosis trigger. A more detailed view on the MyD88 levels cipitation, myc-tagged MyD88 was precipitated with anti-myc Ab from in macrophages infected with Yersinia at different multiplicities lysates of transfected HEK293 cells 24 h after transfection. The cell lysates were generated by resuspending the transfected cells in caspase of infection (MOIs) confirmed that the disappearance of MyD88 cleavage buffer (50 mM HEPES pH 7.2, 0.1% CHAPS, 50 nm NaCl, 10 correlated closely with the extent of apoptosis triggered by the mM EDTA, 0.1% Triton X-100, 10% Saccharose, 10 mM DTT, 10 mM Yersinia wild-type strain. Accordingly, WA infection gradually Leupeptin, 0.3 mM Aprotinin, 1 mM PMSF) and repeated freezing and reduced the levels of MyD88 with increased bacterial loads (Fig. thawing. The immune complexes were collected with protein A/G-agarose 1C). Apoptosis in the WA-infected macrophages rose from 57 6 (Santa Cruz Biotechnology, Heidelberg, Germany), washed five times with 6 caspase cleavage buffer containing 1% Triton X-100, and subjected to in 6% to 90 4% at the given MOIs ranging from 5 to 40. WA-DyopP vitro caspase cleavage assay. Accordingly, the immunoprecipitated, puri- less severely affected the viability of the macrophages (12 6 4% fied MyD88 complexes were incubated with 2 U caspase at 37˚C for 2–3 h and 30 6 4% apoptosis at MOIs of 5 and 40, respectively), as in 80 ml reaction buffer. The reaction was stopped by the addition of SDS expected. Of note, along with the reduction of full-length MyD88 sample buffer and by boiling for 5 min. The samples were then subjected to immunoblotting, as described above. In a related approach, the caspase by WA infection, a 17-kDa protein was recognized by the Ab di- cleavage experiments were conducted on recombinant purified MyD88 rected against the C terminus of MyD88 (Fig. 1C). A fragment proteins. For this issue, cDNAs for human and mouse MyD88 were am- of comparable size was detected also in lysates of HEK293 cells plified by PCR and subcloned into plasmid pGEX-4T-1 to express the MyD88 when MyD88 was overexpressed together with TRIF (Fig. 1D). constructs as GST-fusion proteins in E. coli BL21. The mouse myD88 In these experiments, the overexpression of N-terminally myc- sequence was derived from Addgene plasmid 13093, which was kindly provided by Ruslan Medzhitov (Yale University, New Haven, CT). The tagged MyD88, appearing at a molecular mass of 37 kDa, resulted GST–MyD88 fusion proteins were purified from E. coli by sonification in the appearance of two additional protein bands upon cotrans- in caspase cleavage buffer and by incubation of the bacterial lysates with fection with TRIF (Fig. 1D). The N-terminal fragment was ∼ 20 glutathione Sepharose (GE Healthcare, Freiburg, Germany). Proteins bound kDa, as detected by the anti-myc Ab (lane 3), whereas the 17-kDa to the Sepharose beads were pulled down by centrifugation, washed, and subjected to in vitro caspase cleavage, as indicated above. Equal amounts C-terminal protein band was recognized by the Ab directed against of bacterial lysates were used for GST-fusion protein pull-down and for the MyD88 C terminus (lane 7). The overexpression of TRIF trig- processing in the caspase cleavage assays. gers 70–90% apoptosis in the transfected cells under these condi- To determine the caspase-3 cleavage site by Edman degradation, hu- tions (16–18). This result tempted us to speculate that TRIF-related man GST–MyD88 was incubated with 10 U caspase-3 at 37˚C for 16 h. apoptosis, comparable to apoptosis conferred by Yersinia infection The sample was then subjected to SDS-PAGE, transferred to PVDF membrane, and stained with Coomassie Blue to detect the cleavage products of or staurosporine treatment, could be responsible for the fragmen- MyD88. The protein band that corresponded to the C-terminal fragment of tation of MyD88. In line with this idea, the coexpression of CrmA, MyD88 was excised from the PVDF membrane, destained with methanol, an inhibitor of caspase-8 that counteracts TRIF-related apoptosis 4 TARGETING AND REPROGRAMMING OF MyD88 IN APOPTOSIS Downloaded from http://www.jimmunol.org/ by guest on September 30, 2021

FIGURE 1. MyD88 is processed in cells undergoing apoptosis. (A) MyD88 levels in Yersinia-infected and staurosporine-treated macrophages. Human monocyte–derived macrophages were infected with wild-type Y. enterocolitica WA, with the YopP-negative mutant WA-DyopP at a ratio of 20 bacteria per cell (left panel) or treated with staurosporine (right panel). In lane 4, zVAD was coadministered during infection. Bacterial replication was stopped by gentamicin treatment 90 min after onset of infection. After a total incubation period of 20 h, cellular lysates were prepared and subjected to immunoblotting using anti- MyD88 Ab. (B) Reduction of cellular MyD88 is dependent on the extent of Yersinia-mediated apoptosis. Human monocyte–derived macrophages were infected with wild-type Y. enterocolitica WA or the YopP-negative mutant WA-DyopP. Cell lysates were prepared at different time points post infection and subjected to immunoblotting using anti-MyD88 Ab (left panel), as in (A). The rates of apoptosis were determined in parallel by labeling apoptotic cells with annexin V and propidium iodide and counting apoptotic cells by fluorescence microscopy (right panel). Results are expressed as mean percentages 6 SD from four independent experiments. Apoptosis induction was significant for the wild-type strain compared with untreated cells at each investigated time point (p , 0.01). (C) Cleavage of MyD88 in Yersinia-induced apoptosis. Human monocyte–derived macrophages were infected at increasing MOIs (ratio of bacteria per cell) with the indicated Yersinia strains. Cells were processed and subjected to immunoblotting as in (A). *, a nonspecific band. (D) Cleavage of MyD88 in TRIF- overexpressing cells. HEK293 cells were transfected with expression plasmids encoding MyD88-myc, TRIF, and CrmA, as indicated. Cellular extracts were prepared 20 h after transfection. In lane 4, the caspase inhibitor zVAD was applied to the cells 5 h before lysis. The cell lysates were subjected to immunoblotting using anti-myc Ab, which recognizes the MyD88-myc N terminus (left panel). The membrane was subsequently stripped and reprobed with an anti- MyD88 Ab directed against the MyD88 C terminus (right panel). Equal loading of the gels with cellular lysates was controlled by immunoblotting of the membranes against actin or HSC-70. Molecular masses of standard marker proteins are indicated in kilodaltons.

(16–18), as well as the caspase inhibitor zVAD-fmk prevented the Caspase-3 cleaves human MyD88 at D135 appearance of the two MyD88 fragments (Fig. 1D). Together, these Caspases are cysteine proteases cleaving their client molecules at data suggest that MyD88 is cleaved and processed by caspases in aspartate residues within dedicated caspase recognition sequences two smaller fragments of 20 kDa (N-terminal) and 17 kDa (C- (30). To pinpoint a potential role of caspases in the fragmentation terminal) during the execution of apoptosis. of MyD88, we mutagenized codons for several aspartate residues The Journal of Immunology 5 at which the cleavage might occur and replaced them with alanine Next, wild-type and D135A-mutated MyD88 were analyzed in (Fig. 2A). The mutagenized MyD88 constructs were then compared parallel on their sensitivities to become processed by caspase-3. with wild-type MyD88 to become processed upon cotransfection Only wild-type MyD88, but not the D135A mutant, was cleaved, with the TRIF expression plasmid. Fig. 2B demonstrates that the leading to the formation of the two fragmentation products (Fig. substitution mutation of aspartate at position 135 prevented the 3B, lanes 5 and 8). This finding substantiates the idea that capase- cleavage of MyD88 (lane 5), whereas the other MyD88 mutants 3 mediates the processing of MyD88 at D135. D135 is located were still processed. It may be concluded that D135 represents the in the quadripartite sequence AAVD, a motif that is well described caspase cleavage site of MyD88. We subcloned the sequences for as recognized and processed by caspase-3 in vivo (30, 31). Of amino acids 1–135 and 136–296 of MyD88 into a eukaryotic ex- interest, the AAVD motif of human MyD88 (GenBank acces- pression vector in fusion with an N-terminal myc-tag epitope. Com- sion number Q99836.1) is absent in murine MyD88 (GenBank paring the electrophoretic mobilities of these constructs with those accession number P22366.3, www.ncbi.nlm.nih.gov/genbank/), in of processed, N-terminally myc-tagged wild-type MyD88, we found which ARVE replaces the AAVD sequence. In confirmation of the that the myc-tagged N terminus of MyD88 (DMyD88_N-135) dis- importance of D135 in enabling the processing of human MyD88, played exactly the same mobility as the 20-kDa myc-tagged frag- we expressed GST-tagged versions of human and murine MyD88 ment generated upon MyD88 cleavage (Fig. 2C, lanes 3 and 4). The and subjected the purified proteins to cleavage by caspase-3 in C-terminal part of MyD88 (DMyD88_C-136) displayed a higher vitro. As expected, human, but not murine, MyD88 underwent molecular mass than the original 17 kDa C-terminal fragment be- caspase-3–dependent protein fragmentation (Fig. 3C, compare cause of the fusion of DMyD88_C-136 with the myc-tag (data not lanes 2 and 4). We additionally performed Edman sequencing of shown). However, the identical migration pattern of DMyD88_N-135 the C-terminal MyD88 fragment generated by caspase-3 to con- Downloaded from and the N terminus of processed wild-type MyD88 support the clusively determine the MyD88 cleavage site. The Edman deg- idea that MyD88 is cleaved at position D135. radation reaction sequentially removes one residue at a time from To gain further insights into a potential role of caspases in the the N terminus of a protein. By this method we mapped a pep- fragmentation of MyD88, we subjected immunoprecipitated myc- tide characterized by the sequence X-X-V-P-R-(T)-A-E-L-(A). tagged MyD88 to an in vitro caspase assay using recombinant The first two amino acids (“X”) could not be identified, and de-

caspases. The application of caspase-3 thereby led to the generation termination of the amino acids in brackets was not unambiguous. http://www.jimmunol.org/ of a 17-kDa protein band that was recognized by the C-terminal However, from the sequence in total it became apparent that the anti-MyD88 Ab (Fig. 3A, lane 2). This protein was slightly visi- peptide matched amino acids 136–145 of human MyD88 (Fig. ble upon caspase-7 application, and was not detected after treat- 2A). This result conﬁrms that MyD88 is cleaved by caspase-3 ment with caspase-6 or caspase-8. Consequently, it appears that immediately after D135. Thus, D135 speciﬁcally determines the caspases—in particular, caspase-3—mediate the cleavage of MyD88. sensitivity of human MyD88 to caspase-3–related processing. by guest on September 30, 2021

FIGURE 2. Mutation of D135 prevents the fragmentation of MyD88. (A) Mapping of aspartates in the intermediate region of MyD88. Down- stream from the DD, human MyD88 harbors eight aspartates (bold) that may serve as potential caspase-processing sites. The deduced sequence of human MyD88 (GenBank accession No. Q99836.1) is shown between amino acids 111 and 200. The motif identiﬁed as being cleaved by caspase-3 is underlined (amino acids 132–135, aavd). (B) Processing of MyD88 aspartate mutants in TRIF-related apoptosis. HEK293 cells were cotransfected with TRIF expression plasmid, and either empty vector control (vector) or vectors encoding wild-type MyD88-myc (WT) or MyD88-myc mutants in which single aspartates were replaced by alanines as indicated. Cellular extracts were prepared 20 h after transfection and subjected to immunoblotting using anti-myc Ab. (C) Comparison of the electrophoretic mobilities of fragmented wild-type MyD88 and DMyD88_N-135. HEK293 cells were transfected with expression plasmids for TRIF, MyD88-myc, a combina- tion of both plasmids (TRIF + MyD88), or a MyD88-myc fragment encoding the N-terminal amino acids 1 to 135 (DMyD88_N-135). Cell extracts were prepared and processed for anti-myc Ab immunoblotting, as in (B). Equal loading of the gels with cellular lysates was controlled in (B) and (C) by immunoblotting of the membranes against actin. Molecular masses of standard marker proteins are indicated in kilodaltons. 6 TARGETING AND REPROGRAMMING OF MyD88 IN APOPTOSIS

FIGURE 3. Caspase-3 cleaves human MyD88 at D135. (A) In vitro processing of immunoprecipitated human MyD88 by different caspases. Human MyD88- myc was immunoprecipitated from lysates of transfected HEK293 cells with anti-myc Ab, and either left untreated (Ø) or subjected to in vitro caspase cleavage assay with the indicated caspases at 37˚C for 2 h. (B) In vitro processing of human wild-type MyD88 versus the D135A mutant. Wild-type MyD88 (WT) and the corresponding D135A MyD88 mutant (D135A) were immunoprecipitated from lysates of transfected HEK293 cells, and either left untreated (Ø) or subjected to in vitro cleavage assay with caspase-3, as described in

(A). (C) In vitro processing of recombinant human ver- Downloaded from sus mouse MyD88. Human and mouse MyD88 were puriﬁed as recombinant GST-fusion protein and subjected to in vitro caspase assay with caspase-3, as in described in (A). The samples of the caspase assays were separated by SDS-PAGE and processed for immunoblotting using anti-MyD88 Ab (A; B, left panel; C). In

(B), the membrane was additionally stripped and repro- http://www.jimmunol.org/ bed with anti-myc Ab (right panel). by guest on September 30, 2021

The processing of MyD88 helps to deregulate proinflammatory only at the receptor level. It was unable to counteract the consti- signaling events in the execution of apoptosis tutive NF-kB response conferred by the DD-harboring construct The cleavage of MyD88 by capase-3 in the intermediate domain DMyD88_N-136 (Fig. 4C). Thus, the simultaneous generation of splits MyD88 into two fragments and separates the MyD88 DD DMyD88_N-135 and DMyD88_C-136 by caspase-3 could exert from the TIR domain. This activity may disrupt the ability of MyD88 two distinct effects on the cell: DMyD88_C-136 may disrupt the to couple receptor stimulation to the activation of downstream signaling downstream from TLR4 by blocking the intracellular proinflammatory signaling events. To gain more insight into the receptor domain; DMyD88_N-135 may, on the other hand, mediate functional consequences of the cleavage of human MyD88 for constitutive NF-kB activation insensitive to inhibition by DMyD88_ the proinflammatory response, we analyzed the influence of the C-136. The simultaneous presence of both DMyD88_N-135 and different MyD88 constructs on NF-kB activation. Wild-type DMyD88_C-136 could therefore disconnect TLR activation from MyD88, D135A-mutagenized MyD88, or the DMyD88_N-135 or the induction of proinflammatory signaling and concomitantly elicit DMyD88_C-136 constructs were cotransfected with a NF-kB lu- a receptor-independent, cell-autonomous intracellular signal that ciferase reporter vector into HEK293 cells, and NF-kB–driven may fine-tune the apoptotic process. reporter activation was monitored. Wild-type MyD88, the D135A To evaluate a potential role for the processing of MyD88 in mutant, and DMyD88_N-135 triggered constitutive activation of the silencing of inflammatory signaling events in the course of NF-kB, whereas DMyD88_C-136 did not induce NF-kB (Fig. 4A). apoptosis, we overexpressed noncleavable MyD88-D135A and The induction of NF-kB by the MyD88 constructs consequently wild-type MyD88 in HEK293 cells and induced apoptosis by staur- correlates with the presence of the MyD88 DD, which is known osporine treatment. In these experiments, we were unable to pinpoint to signal NF-kB activation through recruiting downstream signal a clear MyD88-related effect on NF-kB–driven luciferase reporter transmitters (4). The C-terminal part of MyD88 (DMyD88_C-136), activities in the dying cells (data not shown). This finding could be in contrast, harbors the TIR domain that interacts with the cyto- due to the inactivation of several signal transmitters that control the plasmic tails of TLRs and IL-1Rs. When the DD is absent, binding activation of NF-kB downstream from MyD88 by caspases during of the isolated TIR domain to the intracellular receptor site may the execution of apoptosis (19). We therefore sought to analyze a prevent the induction of proinflammatory signals after receptor more upstream proinflammatory event dependent on MyD88. Pre- stimulation (24). In line with these previous observations, the vious studies have shown that MyD88 is organized in a postreceptor overexpression of DMyD88_C-136 acted as dominant negative and complex named “Myddosome” (32). The Myddossome is formed reduced the activation of NF-kB mediated by WA-DyopP or LPS in by the DDs of MyD88, IRAK4, and IRAK2, with IRAK molecules cells transfected with TLR4 and its coreceptors, CD14 and MD2 being subjected to auto- and cross-phosphorylation once the com- (Fig. 4B). However, DMyD88_C-136 blocked NF-kB activation plex has formed. We wondered whether the cleavage of MyD88 The Journal of Immunology 7

FIGURE 4. The fragmentation of MyD88 deregulates the cellular NF-kB response. (A) NF-kB activation owing to the different MyD88 constructs. HEK293 cells were cotransfected with a NF-kB–driven reporter plasmid and either empty control vector (vector) or expression vectors encoding wild-type MyD88 (WT), D135A-mutagenized MyD88 (D135A), or the N-terminal (DMyD88_N-135; N-135) or C-terminal (DMyD88_C-136; C-136) MyD88 fragments. After 20 h, cells were lysed, and luciferase activities were measured. (B) Down- regulation of the TLR4-derived NF-kB signal by cotransfection of the C-terminal MyD88 fragment. HEK293 cells were cotransfected with expression vectors for TLR4/CD14/MD2, and either empty control plasmid (vector) or the MyD88 C terminus (DMyD88_C-136; C-136), Downloaded from as indicated. At 24 h after transfection, the cells were left untreated (Ø) or stimulated with the YopP-negative Yersinia strain (WA-DyopP)or with LPS. After an additional 20 h, cells were lysed, and luciferase activities of the cotransfected reporter plasmids were measured. (C) Divergent NF-kB responses triggered by the two MyD88 fragments. HEK293 cells were cotransfected with empty control vector (vector), or expression plasmids encoding the N-terminal (N-135), or C-terminal http://www.jimmunol.org/ (C-136) MyD88 fragments, or with both plasmids together (N-135 + C- 136). After 20 h, cells were lysed, and luciferase activities of the cotransfected reporter plasmids were measured. The total amount of DNA was kept constant, with an empty vector for each of the transfection conditions. The NF-kB activation levels in (A–C) are expressed as folds of the NF-kB activity measured in cells transfected with the control vector (control). The means 6 SEM of four to six independent experiments are shown. Asterisks denote statistically signiﬁcant differences (*p , 0.05, **p , 0.01, ***p , 0.001) in the NF-kB acti- by guest on September 30, 2021 vation levels measured in MyD88- versus control vector–transfected cells (A and C), or in TLR4/CD14/MD2–expressing cells that were cotransfected either with empty vector control or with the C-terminal MyD88 fragment C-136 (B).

may affect the phosphorylation of IRAK4 within the Myddosome, the cleavage of MyD88 in the course of apoptosis can impair the which assembles when MyD88 is overexpressed together with ability of MyD88 to trigger IRAK4 activation. The processing of IRAKs. We cotransfected HEK293 cells with MyD88 and IRAK4 MyD88 by caspase-3 could therefore be a means that disrupts ef- expression plasmids and monitored the phosphorylation of IRAK4 ficient clustering of MyD88 and concomitant nucleation of proin- in cell extracts. The used Ab recognizes the phosphorylation of flammatory signaling complexes. This mechanism may uncouple the threonine-345 and serine-346, which are two autophosphorylation activation of proinflammatory cell surface receptors from intra- sites required for IRAK4 activation (33). Fig. 5 shows that the cellular signaling. overexpression of both wild-type and D135A-mutagenized MyD88 We next assessed whether the cleavage of MyD88 could also triggered the phosphorylation of coexpressed IRAK4 (lanes 2 and have an effect on the signaling in cells undergoing Yersinia-induced 3). When apoptosis was induced by treatment with staurosporine, apoptosis. We aimed to perform these experiments in MyD88- the phosphorylation of IRAK4 was reduced in cells cotransfected negative cells that were reconstituted with either wild-type MyD88 with wild-type MyD88 (lane 4), whereas cells expressing D135A- or the noncleavable MyD88-D135A mutant. Because Yersinia mutagenized MyD88 retained stronger IRAK4 phosphorylation triggers apoptosis in macrophages, but not in epithelial cells or (lane 5). The decrease of IRAK4 phosphorylation correlated with monocytic cell types (34), we relied in these experiments on im- a reduced level of wild-type MyD88, which was not observed, as mortalized bone marrow macrophages that were derived from expected, for the D135A mutant. These experiments indicate that MyD88-deficient mice. The cells were transiently transfected with 8 TARGETING AND REPROGRAMMING OF MyD88 IN APOPTOSIS

phosphorylated threonine-184 and threonine-187, we analyzed the phosphorylation status of TAK1 in lysates of Yersinia-infected cells. It is shown in Fig. 6C that Yersinia infection for 2 h initially triggered the phosphorylation of TAK1. This phosphorylation event largely depended on the presence of MyD88 because cells solely transfected with empty control plasmid did not exhibit substantial phosphorylation (Fig. 6C, lower panel). The TAK1 phosphorylation status decreased in the course of Yersinia-induced apoptosis triggered by the wild-type strain WA (Fig. 6C, upper panel). This decrease was particularly obvious in macrophages expressing wild-type MyD88, in which the TAK1 phosphorylation status was impaired after 3.5 and 5 h of infection (lanes 4 and 5). TAK1 itself seemed to be destabilized during apoptosis at 5 h (lane 5). Importantly, the phosphorylation of TAK1 was conserved remarkably longer over time in cells expressing the MyD88- D135A mutant, in which the phosphorylation of TAK1 was still FIGURE 5. The stability of MyD88 controls the phosphorylation of obvious after 3.5 h (lane 8). This finding suggests that the stability IRAK4 during apoptosis execution. HEK293 cells were transfected with of MyD88 has influence on the phosphorylation and activation empty vector control (vector) or cotransfected with expression plasmids Downloaded from for HA-tagged IRAK4, and either wild-type (WT) or D135A-mutagenized status of TAK1 in cells undergoing apoptosis. The cleavage of (D135A) MyD88-myc. Cells were left untreated (Ø) or subjected for 12 h MyD88 by caspase-3 may consequently contribute to shutting to staurosporine-induced apoptosis. The levels of transfected MyD88 were down the signaling of inflammation through TAK1 in apoptotic determined by anti-myc immunoblotting (upper panel). The phosphory- cells. In total, these results indicate that the processing of MyD88 lation status of IRAK4 was assessed in the cellular lysates by immuno- disrupts the transmission of intracellular signals induced by the blotting with an Ab that recognizes IRAK4 phosphorylated at T345/S346 activation of proinflammatory cell surface receptors, which ren-

(middle panel). The total pool of IRAK4 overexpressed in the cells was ders the apoptotic cell unable to respond adequately to inflam- http://www.jimmunol.org/ controlled by reprobing of the membrane with anti-HA Ab (lower panel). matory stimuli. The phosphorylation levels of IRAK4 were quantified in relation to the total levels of IRAK4 in the respective sample. The values obtained for Discussion untreated cells were set to 100% and the results following staurosporine Apoptosis is a regulated cell death mechanism that ensures the treatment are expressed as percentages thereof. removal of unwanted cells from the multicellular organism without inducing considerable inflammation (36). This form of cell death also plays a role in infectious diseases (37, 38). Apoptosis can me- empty control plasmid or expression vectors encoding wild-type diate the destruction of infected cells and the concomitant elimi- or D135A-mutagenized MyD88. Fig. 6 shows that apoptosis in- nation of intracellular pathogens. The signaling of apoptosis through by guest on September 30, 2021 duction by the Yersinia wild-type strain WA (Fig. 6B) resulted in TLRs is understood to be part of this host innate immune reaction a time-dependent decrease in the protein levels of wild-type that enables the silent, noninflammatory clearance of infected cells MyD88 (Fig. 6A, lane 6). The levels of MyD88-D135A, in con- (5, 16, 18, 39, 40). Apoptosis triggered by TLR4 and TRIF in host trast, were less severely destabilized (lane 9). which likely resulted immune cells is, furthermore, a physiological response in systemic from the inability of caspase-3 to cleave the MyD88-D135A Gram-negative bacterial infection (41). This pathway is involved mutant. Longer exposition of the immunoblot enabled detection in mediating the maturation of dendritic cells and in shaping adap- of the C-terminal cleavage product of wild-type MyD88 (Fig. 6A, tive immunity. Our previous studies have demonstrated that TLR4 lanes 5 and 6). The levels of the MyD88 fragment increased in the and TRIF also contribute to Yersinia-induced apoptosis. Y. enter- course of Yersinia-induced apoptosis. Relatively fair amounts of ocolitica injects YopP in infected cells to inhibit the anti-apoptotic wild-type MyD88 and a long exposure time of the immunoblot NF-kB pathway. This activity sensitizes macrophages to undergo were required to detect the MyD88 fragments, which may suggest apoptosis following activation of TLR4 and TRIF (16). In pur- that the truncated MyD88 proteins were not stable inside the cell. suing the exploration of the life and death signaling circuits in A certain degree of processing of MyD88 became apparent also in Yersinia-infected macrophages, we show in this article that the cells infected with the YopP-negative mutant WA-DyopP (Fig. 6A, execution of apoptosis vitally affects the MyD88 pathway. Acti- lanes 10 and 11). This processing was likely due to a low level of vation of the apoptotic cascade through TRIF was accompanied YopP-independent apoptosis that occurred in a fraction of the by the cleavage of MyD88. Similarly, MyD88 was processed fol- infected cells (Fig. 6B). It should be noted that the time course of lowing induction of the intrinsic, mitochondrial apoptotic pathway Yersinia-mediated apoptosis substantially differs between human through staurosporine treatment. Our data provide evidence that (Fig. 1B) and murine macrophages (Fig. 6B), which could result MyD88 is cleaved and processed by caspase-3 in the execution from differences in their origin and maturation (34). We subse- phase of apoptosis, which critically affects its activity in the dying quently analyzed whether the MyD88 constructs can reconstitute cell. the MyD88-negative macrophages and confer a functional MyD88- A plethora of cellular proteins have been reported to be cleaved dependent proinflammatory phenotype. The Ab that recognized the by caspases (19, 30). For most of the proteins, however, the bio- phosphorylation of IRAK4 in transfected HEK293 cells (Fig. 5) logical function of the cleavage is not defined. Recent studies have was, unfortunately, inefficient in detecting a comparable signal on indicated that caspases do not dismantle cells by indiscriminate endogenous IRAK4 in macrophages (data not shown). We there- proteolysis, but rather selectively target certain protein complexes fore characterized a second phosphorylation event that occurs down- and networks (30, 36, 42). Many proteins proteolyzed during stream from IRAK members on TAK1. TAK1 requires phosphor- apoptosis may, furthermore, yield fragments that correspond to ylation at threonine-184 and threonine-187 to be activated upon discrete protein domains that could fulfill temporary but specific TLR or IL-1R stimulation (35). Using an Ab directed against functions as active or dominant negative effector proteins in the The Journal of Immunology 9 Downloaded from http://www.jimmunol.org/ by guest on September 30, 2021

FIGURE 6. The phosphorylation of TAK1 is influenced by the stability of MyD88 in the course of Yersinia-induced apoptosis. (A) Reconstitution of MyD88-knockout macrophages with wild-type or D135A-mutagenized MyD88 and analysis of the MyD88 protein levels during Yersinia infection. Im- mortalized MyD88-deficient, murine macrophages were transiently transfected with either empty control vector (vector) or expression plasmids for human wild-type (WT) or D135A-mutagenized (D135A) MyD88-myc. Cells were left untreated (Ø) or infected with wild-type Y. enterocolitica WA or the YopP- negative mutant WA-DyopP at an MOI of 20 bacteria per cell for the indicated time points. Then, cell lysates were prepared and subjected to immunoblotting using anti–C-terminal MyD88 Ab and ECL reaction. Short exposure of the immunoblot (upper panel) allowed the assessment of the MyD88 levels in the course of Yersinia infection. The MyD88 protein content was quantified in relation to cellular actin, and the values obtained for untreated cells were set to 100% for WT- as well as D135A-mutagenized MyD88. Results are expressed as percentages thereof. Longer exposure of the immunoblot enabled the detection of the C-terminal MyD88 cleavage product (lower panel). (B) Time course of Yersinia-induced apoptosis. Cells were transfected and infected as in (A). At 2 and 3.5 h after onset of Yersinia infection, the rates of apoptosis were determined by labeling apoptotic cells with annexin V and propidium iodide and counting apoptotic cells by fluorescence microscopy. Results are expressed as mean percentages 6 SD from four independent experiments. Apoptosis induction was significant for the wild-type strain compared with noninfected cells at each investigated time point (p , 0.001). (C) MyD88-dependent phosphorylation of TAK1 in the course of Yersinia infection. MyD88-deficient macrophages were transfected with empty control vector (vector), or expression plasmids for wild-type (WT) or D135A-mutagenized (D135A) MyD88-myc, and then either left untreated (Ø) or infected with wild-type WA or YopP-negative WA-DyopP at an MOI of 15 bacteria per cell. The differential effects of the MyD88 constructs on the phosphorylation of TAK1 became most apparent at this MOI. At the indicated time points, cell lysates were prepared and subjected to immunoblotting with an Ab (Figure legend continues) 10 TARGETING AND REPROGRAMMING OF MyD88 IN APOPTOSIS course of apoptosis (31, 43). The targeting of MyD88 by caspase- TLR and IL-1R members to the initiation of intracellular sig- 3 may fit into this concept. By cleaving MyD88, the proteolytic naling (4). The proteolysis of MyD88 is consequently predicted caspase cascade affects a central and essential proinflammatory to render the dying cell less susceptible to TLR and IL-1R adapter molecule. Caspase-3 cuts MyD88 into two fragments, ligands, which disrupts the cellular response to these proin- each harboring a specific domain that could fulfill a particular flammatory receptors. This could also be advantageous for an function. The C-terminal MyD88 fragment that comprises the TIR extracellular pathogen, such as Yersinia, that may benefit from domain can act as dominant negative inhibitor on TLR signaling. impaired TLR signaling of infected host cells. Of interest, the The MyD88 N terminus with the DD, in contrast, has some ca- MyD88 caspase-processing site D135ishighlyconservedacross pacity to autonomously recruit binding partners for initiating several vertebral species, including human, dog, zebrafish, zebra downstream signaling events. The concerted actions of both MyD88 finch, clawed frog, and chinese softshell turtle (Ref. 45 and data fragments may fine-tune the apoptotic process, which could govern not shown). This observation supports the idea that the MyD88 specific cell death–related events in a temporary or spatial manner. caspase-processing site plays a physiological role in the course Similar features have been reported for other signal transmitters of apoptosis. In the mouse MyD88 sequence, D135 is replaced of the NF-kB pathway that are cleaved by caspases, including by E135. However, our data indicate that murine MyD88 is TRAF1, receptor-interacting protein 1, IKKb, IKKg/NEMO, and also diminished in Yersinia-induced apoptosis (data not shown). NF-kB p65 (19, 20, 22). The processing of these molecules Alternative proteases other than caspases activated during apo- hampers their prosurvival functions and accelerates apoptosis. ptosis, such as serine proteases or cathepsins, could be respon- These events are apparently precisely controlled over time. The sible for the processing of mouse MyD88 (46). The removal of cleavage products of many of these signal transmitters are short MyD88 in apoptosis execution may consequently fulfill a par- Downloaded from lived and shut down in their activities by regulated degradation ticular, conserved function. It is characteristic of apoptotic cells (44). This activity tends to balance excessive effects of the trun- that plasma membrane integrity is long preserved. This feature cated proteins on the apoptotic cell, to fine-tune the progression of ensures that no proinflammatory intracellular constituents are cell death. The MyD88 fragments characterized in our studies released until the apoptotic cells are internalized by bystander could be subjected to a similar regulatory pathway. Fair amounts cells (36). As long as the cell membrane of the dying cell is

of the protein and an appropriate time frame were required to intact, the processing of MyD88 could aid in deactivating proin- http://www.jimmunol.org/ detect the MyD88 fragments. This requirement could reflect suc- flammatory signals that could possibly derive from the TLR and cessive destabilization of the liberated MyD88 domains following IL-1R family in a permissive environment. The cleavage of MyD88 caspase-3–dependent cleavage. Removal of the MyD88 cleavage may help to extinguish the proinflammatory responses and prop- products may ensure a timed and restricted activity of the truncated erties of dying cells. This cleavage could also be a strategy to proteins on the dying cell. Importantly, the cleavage of MyD88 counteract oncogenesis. MyD88 has been shown to be implicated seemed to target predominantly its proinflammatory function, in carcinogenesis through the induction of inflammation as well but not the execution of cell death itself. Neither overexpression as through cell-autonomous mechanisms of cell transformation of the noncleavable MyD88 mutant nor overexpression of the (47, 48). The processing of MyD88 could therefore be a regulatory resulting N- or C-terminal MyD88 fragments significantly in- event protecting against diverse disorders of tissue homeostasis by guest on September 30, 2021 fluenced the initiation or progression of apoptosis in several that could arise from malfunctions in apoptosis, such as the devel- investigated conditions (data not shown). This finding is in line opment of aberrant inflammatory responses or eventually cancer. with previous studies in which TLR-related apoptosis signaling was prominent for the TRIF, but not for the MyD88, pathway Acknowledgments (16, 18). Our results suggest, rather, that the processing of MyD88 We thank Shizuo Akira, David Vaux, Kensuke Miyake, Ruslan Medzhitov, primarily affects its capacity to signal inflammation. The over- and Tularik for kindly providing expression plasmids; Katherine A. expression of MyD88 under conditions that facilitate formation of Fitzgerald for supplying immortalized MyD88-deficient bone mar- the Myddosome triggered the phosphorylation of IRAK4, which row–derived macrophages; Rudolf Haase for fruitful scientific input; was diminished when apoptosis was executed. On the contrary, and Gudrun Pfaffinger for providing the first clues to these experiments. overexpression of the noncleavable MyD88 mutant remarkably preserved the phosphorylation of IRAK4 in this setting. In the Disclosures same manner, the phosphorylation of TAK1 was largely controlled The authors have no financial conflicts of interest. by the stability of MyD88 in macrophages that underwent apoptosis in response to Yersinia infection. Thus, the levels of full- References length MyD88 can regulate the phosphorylation status of IRAK4 1. Kawai, T., and S. Akira. 2006. TLR signaling. Cell Death Differ. 13: 816–825. and TAK1 in dying cells. This observation suggests conversely 2. Takeuchi, O., and S. Akira. 2010. Pattern recognition receptors and inflamma- that the cleavage of MyD88 by the apoptosis machinery could tion. Cell 140: 805–820. 3. O’Neill, L. A., and A. G. Bowie. 2007. The family of five: TIR-domain- specifically impair the initiation of proinflammatory signaling by containing adaptors in Toll-like receptor signalling. Nat. Rev. Immunol. 7: preventing the phosphorylation and activation of IRAK molecules 353–364. and of other downstream signal transmitters. 4. Janssens, S., and R. Beyaert. 2002. A universal role for MyD88 in TLR/IL-1R- mediated signaling. Trends Biochem. Sci. 27: 474–482. By being fragmented between the death and the TIR domain, 5. Salaun, B., P. Romero, and S. Lebecque. 2007. Toll-like receptors’ two-edged MyD88 loses its adaptor function that connects the activation of sword: when immunity meets apoptosis. Eur. J. Immunol. 37: 3311–3318.

that recognized TAK1 phosphorylated at T184/187 (upper panel each). Equal loading of the gels with cell lysates was controlled by stripping of the membrane and reprobing with anti-actin Ab (middle panel each). Because the recycled membranes showed reduced sensitivity to successive TAK1 immunostaining with the available Abs, the cellular levels of TAK1 were analyzed in parallel control samples that were treated under the same conditions (lower panel each). The phosphorylation of TAK1 was quantiﬁed in relation to actin. The values obtained upon infection with WA-DyopP for 3.5 h were set to 100% for MyD88-WT– and MyD88-D135A–expressing cells. The results are expressed as percentages thereof. Molecular masses of standard marker proteins are indicated in kilodaltons. The Journal of Immunology 11

6. Vanlangenakker, N., T. Vanden Berghe, and P. Vandenabeele. 2012. Many inhibit apoptosis and/or bind tumor necrosis factor receptor-associated factors. stimuli pull the necrotic trigger, an overview. Cell Death Differ. 19: 75–86. Proc. Natl. Acad. Sci. USA 93: 4974–4978. 7. Brodsky, I. E., and R. Medzhitov. 2009. Targeting of immune signalling net- 27. Li, S., A. Strelow, E. J. Fontana, and H. Wesche. 2002. IRAK-4: a novel member works by bacterial pathogens. Nat. Cell Biol. 11: 521–526. of the IRAK family with the properties of an IRAK-kinase. Proc. Natl. Acad. 8. Finlay, B. B., and G. McFadden. 2006. Anti-immunology: evasion of the host Sci. USA 99: 5567–5572. immune system by bacterial and viral pathogens. Cell 124: 767–782. 28. Kirschning, C. J., H. Wesche, T. Merrill Ayres, and M. Rothe. 1998. Human toll- 9. Aepfelbacher, M., C. Trasak, and K. Ruckdeschel. 2007. Effector functions of like receptor 2 confers responsiveness to bacterial lipopolysaccharide. J. Exp. pathogenic Yersinia species. Thromb. Haemost. 98: 521–529. Med. 188: 2091–2097. 10. Viboud, G. I., and J. B. Bliska. 2005. Yersinia outer proteins: role in modulation 29. Nagai, Y., S. Akashi, M. Nagafuku, M. Ogata, Y. Iwakura, S. Akira, T. Kitamura, of host cell signaling responses and pathogenesis. Annu. Rev. Microbiol. 59: 69– A. Kosugi, M. Kimoto, and K. Miyake. 2002. Essential role of MD-2 in LPS 89. responsiveness and TLR4 distribution. Nat. Immunol. 3: 667–672. 11. Mittal, R., S. Y. Peak-Chew, and H. T. McMahon. 2006. Acetylation of MEK2 30. Crawford, E. D., and J. A. Wells. 2011. Caspase substrates and cellular and I kappa B kinase (IKK) activation loop residues by YopJ inhibits signaling. remodeling. Annu. Rev. Biochem. 80: 1055–1087. Proc. Natl. Acad. Sci. USA 103: 18574–18579. 31. Mahrus, S., J. C. Trinidad, D. T. Barkan, A. Sali, A. L. Burlingame, and 12. Mukherjee, S., G. Keitany, Y. Li, Y. Wang, H. L. Ball, E. J. Goldsmith, and J. A. Wells. 2008. Global sequencing of proteolytic cleavage sites in apoptosis by K. Orth. 2006. Yersinia YopJ acetylates and inhibits kinase activation by speciﬁc labeling of protein N termini. Cell 134: 866–876. blocking phosphorylation. Science 312: 1211–1214. 32. Gay, N. J., M. Gangloff, and L. A. O’Neill. 2011. What the Myddosome 13. Ruckdeschel, K., O. Mannel, K. Richter, C. A. Jacobi, K. Tru¨lzsch, B. Rouot, structure tells us about the initiation of innate immunity. Trends Immunol. 32: and J. Heesemann. 2001. Yersinia outer protein P of Yersinia enterocolitica si- 104–109. multaneously blocks the nuclear factor-kappa B pathway and exploits lipo- 33. Cheng, H., T. Addona, H. Keshishian, E. Dahlstrand, C. Lu, M. Dorsch, Z. Li, polysaccharide signaling to trigger apoptosis in macrophages. J. Immunol. 166: A. Wang, T. D. Ocain, P. Li, et al. 2007. Regulation of IRAK-4 kinase activity 1823–1831. via autophosphorylation within its activation loop. Biochem. Biophys. Res. 14. Zhang, Y., A. T. Ting, K. B. Marcu, and J. B. Bliska. 2005. Inhibition of MAPK Commun. 352: 609–616. and NF-kappa B pathways is necessary for rapid apoptosis in macrophages 34. Ruckdeschel, K., A. Roggenkamp, V. Lafont, P. Mangeat, J. Heesemann, and infected with Yersinia. J. Immunol. 174: 7939–7949. B. Rouot. 1997. Interaction of Yersinia enterocolitica with macrophages leads to Downloaded from 15. Haase, R., C. J. Kirschning, A. Sing, P. Schro¨ttner, K. Fukase, S. Kusumoto, macrophage cell death through apoptosis. Infect. Immun. 65: 4813–4821. H. Wagner, J. Heesemann, and K. Ruckdeschel. 2003. A dominant role of Toll- 35. Sakurai, H., H. Miyoshi, J. Mizukami, and T. Sugita. 2000. Phosphorylation- like receptor 4 in the signaling of apoptosis in bacteria-faced macrophages. J. dependent activation of TAK1 mitogen-activated protein kinase kinase kinase by FEBS Lett. Immunol. 171: 4294–4303. TAB1. 474: 141–145. 36. He, B., N. Lu, and Z. Zhou. 2009. Cellular and nuclear degradation during ap- 16. Ruckdeschel, K., G. Pfafﬁnger, R. Haase, A. Sing, H. Weighardt, G. Ha¨cker, optosis. Curr. Opin. Cell Biol. 21: 900–912. B. Holzmann, and J. Heesemann. 2004. Signaling of apoptosis through TLRs 37. Labbe´, K., and M. Saleh. 2008. Cell death in the host response to infection. Cell critically involves toll/IL-1 receptor domain-containing adapter inducing IFN- Death Differ. 15: 1339–1349. beta, but not MyD88, in bacteria-infected murine macrophages. J. Immunol. 173:

38. Lamkanfi, M., and V. M. Dixit. 2010. Manipulation of host cell death pathways http://www.jimmunol.org/ 3320–3328. during microbial infections. Cell Host Microbe 8: 44–54. 17. Han, K. J., X. Su, L. G. Xu, L. H. Bin, J. Zhang, and H. B. Shu. 2004. Mech- 39. Hitomi, J., D. E. Christofferson, A. Ng, J. Yao, A. Degterev, R. J. Xavier, and anisms of the TRIF-induced interferon-stimulated response element and NF- J. Yuan. 2008. Identification of a molecular signaling network that regulates J. Biol. Chem. kappaB activation and apoptosis pathways. 279: 15652–15661. a cellular necrotic cell death pathway. Cell 135: 1311–1323. 18. Kaiser, W. J., and M. K. Offermann. 2005. Apoptosis induced by the toll-like 40. Mocarski, E. S., J. W. Upton, and W. J. Kaiser. 2012. Viral infection and the receptor adaptor TRIF is dependent on its receptor interacting protein homotypic evolution of caspase 8-regulated apoptotic and necrotic death pathways. Nat. interaction motif. J. Immunol. 174: 4942–4952. Rev. Immunol. 12: 79–88. 19. Fischer, U., R. U. Ja¨nicke, and K. Schulze-Osthoff. 2003. Many cuts to ruin: 41. De Trez, C., B. Pajak, M. Brait, N. Glaichenhaus, J. Urbain, M. Moser, a comprehensive update of caspase substrates. Cell Death Differ. 10: 76–100. G. Lauvau, and E. Muraille. 2005. TLR4 and Toll-IL-1 receptor domain- 20. Frelin, C., V. Imbert, V. Bottero, N. Gonthier, A. K. Samraj, K. Schulze-Osthoff, containing adapter-inducing IFN-beta, but not MyD88, regulate Escherichia P. Auberger, G. Courtois, and J. F. Peyron. 2008. Inhibition of the NF-kappaB coli-induced dendritic cell maturation and apoptosis in vivo. J. Immunol. 175: survival pathway via caspase-dependent cleavage of the IKK complex scaffold 839–846. by guest on September 30, 2021 protein and NF-kappaB essential modulator NEMO. Cell Death Differ. 15: 152– 42. Johnson, C. E., and S. Kornbluth. 2008. Caspase cleavage is not for everyone. 160. Cell 134: 720–721. 21. Rebsamen, M., E. Meylan, J. Curran, and J. Tschopp. 2008. The antiviral adaptor 43. Dix, M. M., G. M. Simon, and B. F. Cravatt. 2008. Global mapping of the to- proteins Cardif and Trif are processed and inactivated by caspases. Cell Death pography and magnitude of proteolytic events in apoptosis. Cell 134: 679–691. Differ. 15: 1804–1811. 44. Piatkov, K. I., C. S. Brower, and A. Varshavsky. 2012. The N-end rule pathway 22. Tang, G., J. Yang, Y. Minemoto, and A. Lin. 2001. Blocking caspase-3-mediated counteracts cell death by destroying proapoptotic protein fragments. Proc. Natl. proteolysis of IKKbeta suppresses TNF-alpha-induced apoptosis. Mol. Cell 8: Acad. Sci. USA 109: E1839–E1847. 1005–1016. 45. Avbelj, M., S. Horvat, and R. Jerala. 2011. The role of intermediary domain of 23. Youn, H. S., J. Y. Lee, K. A. Fitzgerald, H. A. Young, S. Akira, and MyD88 in cell activation and therapeutic inhibition of TLRs. J. Immunol. 187: D. H. Hwang. 2005. Specific inhibition of MyD88-independent signaling path- 2394–2404. ways of TLR3 and TLR4 by resveratrol: molecular targets are TBK1 and RIP1 in 46. Schrader, K., J. Huai, L. Jo¨ckel, C. Oberle, and C. Borner. 2010. Non-caspase TRIF complex. J. Immunol. 175: 3339–3346. proteases: triggers or amplifiers of apoptosis? Cell. Mol. Life Sci. 67: 1607–1618. 24. Wesche, H., W. J. Henzel, W. Shillinglaw, S. Li, and Z. Cao. 1997. MyD88: an 47. Cataisson, C., R. Salcedo, S. Hakim, B. A. Moffitt, L. Wright, M. Yi, adapter that recruits IRAK to the IL-1 receptor complex. Immunity 7: 837–847. R. Stephens, R. M. Dai, L. Lyakh, D. Schenten, et al. 2012. IL-1R-MyD88 25. Yamamoto, M., S. Sato, K. Mori, K. Hoshino, O. Takeuchi, K. Takeda, and signaling in keratinocyte transformation and carcinogenesis. J. Exp. Med. 209: S. Akira. 2002. Cutting edge: a novel Toll/IL-1 receptor domain-containing 1689–1702. adapter that preferentially activates the IFN-beta promoter in the Toll-like re- 48. Coste, I., K. Le Corf, A. Kfoury, I. Hmitou, S. Druillennec, P. Hainaut, ceptor signaling. J. Immunol. 169: 6668–6672. A. Eychene, S. Lebecque, and T. Renno. 2010. Dual function of MyD88 in RAS 26. Uren, A. G., M. Pakusch, C. J. Hawkins, K. L. Puls, and D. L. Vaux. 1996. signaling and inflammation, leading to mouse and human cell transformation. J. Cloning and expression of apoptosis inhibitory protein homologs that function to Clin. Invest. 120: 3663–3667.