Phosphatidylinositol-4-kinase IIα licenses phagosomes for TLR4 signaling and MHC-II presentation in dendritic cells

Cynthia López-Habera,b,c, Roni Levin-Konigsbergd,e,1, Yueyao Zhua,b,c, Jing Bi-Karchina,b,c, Tamas Ballaf, Sergio Grinsteind,e, Michael S. Marksa,b,c, and Adriana R. Mantegazzaa,b,c,2 aDepartment of Pathology and Laboratory Medicine, Children’s Hospital of Philadelphia, Philadelphia, PA 19104; bDepartment of Pathology and Laboratory Medicine, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA 19104; cDepartment of Physiology, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA 19104; dDivision of Cell Biology, Hospital for Sick Children, Toronto, ON M5G 1X8, Canada; eDepartment of Biochemistry, University of Toronto, Toronto, ON M5S 1A8, Canada; and fSection on Molecular , Eunice Kennedy Shriver National Institute of Child Health and Human Development, NIH, Bethesda, MD 20892

Edited by Peter Cresswell, Yale University, New Haven, CT, and approved September 22, 2020 (received for review February 1, 2020) Toll-like (TLR) recruitment to phagosomes in dendritic While TLR signaling pathways have been extensively studied, cells (DCs) and downstream TLR signaling are essential to initiate and TLR4 is recruited to phagosomes from a recycling endo- antimicrobial immune responses. However, the mechanisms un- some pool (9), less is known about how TLR localization to derlying TLR localization to phagosomes are poorly characterized. phagosomes in DCs is regulated to optimize the immune re- We show herein that phosphatidylinositol-4-kinase IIα (PI4KIIα) sponse. Our current understanding of TLR recruitment to phag- plays a key role in initiating phagosomal TLR4 responses in murine osomes in DCs was largely shaped by our analyses of a mouse DCs by generating a phosphatidylinositol-4-phosphate (PtdIns4P) model of Hermansky–Pudlak syndrome type 2 (HPS2) (10). HPS2 platform conducive to the binding of the TLR sorting adaptor Toll- is caused by inactivating mutations in the β3A subunit of adaptor IL1 receptor (TIR) domain-containing adaptor protein (TIRAP). protein-3 (AP-3), an adaptor protein that binds to cytoplasmic PI4KIIα is recruited to maturing lipopolysaccharide (LPS)-containing signals in transmembrane proteins and mediates their trafficking phagosomes in an adaptor protein-3 (AP-3)-dependent manner, and from early endosomes to lysosomes or lysosome-related organelles both PI4KIIα and PtdIns4P are detected on phagosomal membrane (11). HPS2 is characterized by immunodeficiency, among other IMMUNOLOGY AND INFLAMMATION tubules. Knockdown of PI4KIIα—but not the related PI4KIIβ— symptoms (4, 12, 13). While the role of AP-3 in T cell and plas- impairs TIRAP and TLR4 localization to phagosomes, reduces proin- macytoid DC function (14–16) may explain recurrent viral infec- flammatory cytokine secretion, abolishes phagosomal tubule forma- tions in HPS2, defects in lipid antigen presentation and granulopoiesis tion, and impairs major histocompatibility complex II (MHC-II) (17–19), together with our observations that AP-3 is required for presentation. Phagosomal TLR responses in PI4KIIα-deficient DCs optimal secretion of proinflammatory cytokines, inflammasome are restored by reexpression of wild-type PI4KIIα,butnotofvari- activity, and MHC-II presentation of phagocytosed antigen to ants lacking kinase activity or AP-3 binding. Our data indicate that T cells (10, 20), may explain defective antibacterial immunity in PI4KIIα is an essential regulator of phagosomal TLR signaling in DCs by ensuring optimal TIRAP recruitment to phagosomes. Significance

AP-3 | dendritic cells | PI4KIIα | TIRAP | MHC-II Dendritic cells (DCs) play a key role at the interface between innate and adaptive immunity. DCs continuously sample their ignaling by pattern-recognition receptors, such as Toll-like microenvironment, respond to microbial cues by signaling Sreceptors (TLRs), is essential to initiate immune responses through pattern-recognition receptors such as Toll-like recep- (1). In addition, TLR signaling is compartmentalized to allow tors (TLRs), and present bacterial antigens to adaptive immune discrimination between myriad self and foreign stimuli that may cells. We show that the lipid kinase phosphatidylinositol-4- pose different levels of threat (2). One of the cellular compart- kinase IIα (PI4KIIα) is required to generate a phosphatidylinosi- ments where TLR signaling is particularly important is the tol-4-phosphate pool on DC phagosomes that allows binding of phagosome, a lysosome-related organelle formed in phagocytes the TLR sorting adaptor TIRAP and promotes TLR4 phagosomal such as dendritic cells (DCs) upon the capture of a particu- signaling to proinflammatory cytokine production, phagosomal late target, such as a bacterium (3, 4). Particularly in DCs, membrane tubule formation, and presentation of phagocytosed which serve as the main interface between innate and adaptive antigens. PI4KIIα therefore ensures phagosomal identity and immunity, phagosomes become autonomous TLR-sensing and autonomous signaling to initiate antimicrobial immune re- -signaling platforms that contain all of the machinery required sponses in DCs. to process the captured material, load resulting peptides into major histocompatibility complex (MHC) molecules, and form Author contributions: M.S.M. and A.R.M. designed research; C.L.-H., R.L.-K., Y.Z., J.B.-K., phagosomal tubules (phagotubules) that favor antigen presentation and A.R.M. performed research; T.B. and S.G. contributed new reagents/analytic tools; C.L.-H., R.L.-K., Y.Z., J.B.-K., T.B., S.G., M.S.M., and A.R.M. analyzed data; A.R.M. wrote the to T cells (5–7). In addition, phagosomes extend proinflammatory paper; and C.L.-H., R.L.-K., Y.Z., J.B.-K., T.B., S.G., M.S.M., and A.R.M. edited versions of responses initiated by plasma-membrane TLRs by the acquisi- the paper. tion of an additional pool of intracellular TLRs. This second The authors declare no competing interest. wave of TLR signaling from phagosomes is focused on a single This article is a PNAS Direct Submission. potentially harmful particulate entity. It is therefore essential to Published under the PNAS license. keep phagosomal identity different from the plasma membrane, 1Present address: Department of Genetics, Stanford University School of Medicine, where sensing reflects a broader spectrum of stimuli (8), and also Stanford, CA 94305. from endosomes, which can bear soluble cargo that may be less 2To whom correspondence may be addressed. Email: [email protected]. harmful. In addition, preserving phagosomal identity and au- This article contains supporting information online at https://www.pnas.org/lookup/suppl/ tonomy is essential in DCs for the initiation of adaptive immune doi:10.1073/pnas.2001948117/-/DCSupplemental. responses. www.pnas.org/cgi/doi/10.1073/pnas.2001948117 PNAS Latest Articles | 1of12 HPS2. The impaired immune responses in AP-3–deficient con- expected (42, 43) (SI Appendix, Fig. S1). Although these data do ventional DCs are, at least in part, due to reduced TLR4 re- not exclude a direct interaction through other subunits, the ab- cruitment from an endosomal pool or reduced retention of sence of the most likely μ3–TLR4 interaction led us to investigate TLR4 and other TLRs on maturing phagosomes (10). However, whether additional effectors might favor AP-3–dependent TLR4 it is not known whether AP-3 supports TLR4 localization to recruitment to phagosomes in DCs. phagosomes by direct binding and incorporation into TLR4 signaling complexes are found on lipid microdomains phagosome-bound vesicles or by an indirect mechanism. enriched in phosphoinositides. Because PtdIns4P and PI4KIIα We and others showed that TLR4 signaling to a set of down- are present on maturing phagosomes in macrophages (36, 37), stream immune responses in DCs requires the engagement of the and AP-3 directly interacts with PI4KIIα in neurons and other TLR signaling adaptor MyD88 (7, 21). MyD88 binding to TLR4 is cell types (33, 34), we investigated the role of PI4KIIα and its facilitated by the TLR sorting adaptor Toll-IL1 receptor (TIR) product PtdIns4P in TLR4 recruitment to DC phagosomes. We domain-containing adaptor protein (TIRAP; also known as MAL) first asked if PI4KIIα and PtdIns4P were present on phagosomes (22), which also binds TLR4 to form the multimolecular signaling in DCs and whether PI4KIIα recruitment was dependent on AP- platform known as the MyDdosome (23). Therefore, TLR4 requires 3. Bone-marrow-derived DCs (BMDCs) from wild-type (WT) TIRAP to dictate the cellular location of MyD88-dependent sig- and Ap3bpe/pe mice that lack AP-3 in nonneuronal cell types − − naling. Importantly, TIRAP binding to cellular membranes depends (AP-3 / ) were pulsed with magnetic beads coated with the on polyphosphorylated inositide-enriched (or phosphoinositide- TLR4 ligand lipopolysaccharide (LPS) and chased over time enriched) domains, highlighting the importance of phosphoinosi- after phagocytosis. Phagosomes were then isolated and analyzed tides as essential mediators of TLR4 positioning at sites of future by immunoblotting for PI4KIIα (Fig. 1 A and B). Phagosomes signal transduction (24, 25). from WT DCs showed an increased recruitment of endogenous Phosphoinositides play an essential role in cellular membrane PI4KIIα over 2 h, concomitant with the acquisition of markers of physiology by regulating membrane curvature, protein recruitment, phagosome maturation, such as the late endosomal/lysosomal and vesicular trafficking and, therefore, defining membrane identity proteins LAMP-1 and Rab7 (SI Appendix, Fig. S2 A and B), and (26–29). The type II lipid kinases phosphatidylinositol-4-kinase IIα of proteolytic activity as assessed by flow-cytometry analysis of (PI4KIIα)andIIβ (PI4KIIβ) (29, 30) are distinctly encoded enzymes ovalbumin (OVA) degradation on phagocytosed beads (SI Ap- that phosphorylate phosphatidylinositol (PtdIns) at position 4, gen- pendix, Fig. S2C). In contrast, whereas cellular expression of − − erating PtdIns4P at the plasma membrane, Golgi, trans-Golgi net- PI4KIIα in WT and AP-3 / DCs was similar, PI4KIIα recruit- − − work (TGN), and endolysosomes (30). In particular, PI4KIIα is ment to phagosomes in AP-3 / DCs was significantly reduced abundant on membranes of the endolysosomal network (31, 32), (Fig. 1 A and B). PI4KIIα recruitment to mature phagosomes, − − notablyinavesicularpoolrichinAP-3(33).Importantly,PI4KIIα measured 2 h after engulfment, was also impaired in AP-3 / binds AP-3 directly via a classical dileucine-based sorting signal and is tissue-resident DCs isolated from spleen compared to WT therebydeliveredtolysosomesandlysosome-relatedorganelles(34). splenic DCs (SI Appendix, Fig. S3), as determined both on iso- In macrophages, PI4KIIα is required for TIR-domain-containing lated phagosomes (SI Appendix, Fig. S3 A and B) or in cells adapter-inducing interferon-β ()-dependent signaling in re- pulsed with LPS-coated polystyrene beads (88 ± 3% in WT DCs − − sponse to soluble stimuli (35), and both PI4KIIα and its product vs. 6 ± 2% in AP-3 / DCs; SI Appendix, Fig. S3 C and D). PtdIns4P are also required for phagosome maturation and resolution To confirm these results, we expressed PI4KIIα–green fluo- (36–38). However, a possible link between PI4KIIα and phagosomal rescent protein (GFP) in DCs by recombinant retroviral trans- TLR signaling has not been investigated. duction of bone marrow (transduction efficiency was similar in − − Considering the connection between PI4KIIα and AP-3, and WT and AP-3 / DCs; SI Appendix, Fig. S4A), pulsed differen- the requirement for phosphoinositides in TIRAP/MyD88- tiated DCs with LPS-coated polystyrene beads, and tested for dependent TLR4 signaling, we investigated whether phagosomal PI4KIIα–GFP recruitment to DC phagosomes by flow cytometry TLR4 signaling in mouse DCs is dependent on the lipid kinase of isolated phagosomes (Fig. 1 C and D and SI Appendix, Fig. PI4KIIα and its product PtdIns4P. We show herein that PI4KIIα, S4B) or by live cell imaging (Fig. 1 E–G). Like endogenous but not the related PI4KIIβ, is required for the production of PI4KIIα, PI4KIIα–GFP was increasingly recruited to phag- PtdIns4P on phagosomes in DCs, for optimal TIRAP binding to osomes over 2 h after phagocytosis in WT DCs (Fig. 1 C–F), phagosomes and for downstream TLR4 responses. PI4KIIα re- concomitant with LAMP-1 (SI Appendix, Fig. S4C). Additionally, cruitment to phagosomes is, in turn, dependent on AP-3 func- endogenous LAMP-1 and expressed PI4KIIα–GFP were also tion, which may, therefore, at least partly explain the defective detected on phagosomes by immunofluorescence microscopy 60 antibacterial responses in AP-3–deficient mouse DCs and HPS2 and 120 min after pulsing WT DCs with LPS-coated polystyrene patients. Finally, we show that PI4KIIα plays a major role in DC beads (SI Appendix, Fig. S4D). In contrast to WT DCs, − − phagosomes that reflects DC specialization in antigen presen- PI4KIIα–GFP recruitment to phagosomes in AP-3 / DCs was tation and ensures the preservation of phagosome identity and significantly impaired (Fig. 1 C and D); after normalizing to the autonomous signaling. percent of PI4KIIα-GFP–positive DCs (SI Appendix, Fig. S4A; note that GFP-positive cells were not previously sorted), Results PI4KIIα-GFP was recruited to 90% of phagosomes in WT DCs, − − PI4KIIα Is Recruited to DC Phagosomes in an AP-3–Dependent Manner. but only 20% in AP-3 / DCs at 120 min (Fig. 1D). Consistent AP-3 recognizes accessible cytosolic targeting sequences on with these results, PI4KIIα–GFP recruitment to phagosomes transmembrane proteins and mediates their trafficking to lyso- containing polystyrene beads coated with LPS and Texas Red- somes or lysosome-related organelles. The consensus sequences conjugated OVA (LPS/OVA-TxR) was significantly impaired in − − YXXϕ or [DE]XXXL[LI] are recognized by the μ3andσ3–δ AP-3 / DCs, as observed by live cell imaging on transduced DCs − − subunits pf AP-3, respectively (39, 40). TLR4 bears a YDAF se- (97 ± 3% in WT DCs vs. 12 ± 3% in AP-3 / DCs after 120 min; quence in the cytoplasmic TIR domain. However, the TLR4 Fig. 1 E–G and Movies S1 and S2; note the absence of a con- − − crystal structure suggests that the tyrosine is not accessible for tinuous fluorescent pattern in AP-3 / DC phagosomes). intermolecular interactions (41). Consistent with this, we could not Together, these data indicate that AP-3 is required for optimal detect direct binding of the TLR cytoplasmic domain to the AP-3 recruitment of PI4KIIα to phagosomes in both BMDCs and μ3 subunit by yeast two-hybrid analysis (SI Appendix,Fig.S1) splenic DCs. Noteworthy, PI4KIIα–GFP was also recruited to under the same conditions that allowed binding of the cytoplasmic OVA-containing phagotubules in maturing phagosomes from tail of human TGN38 to the μ subunits of AP-1, -2, and -3, as WT DCs (arrows, Fig. 1E, time 120 min; Movies S3 and S4).

2of12 | www.pnas.org/cgi/doi/10.1073/pnas.2001948117 López-Haber et al. Fig. 1. PI4KIIα is recruited to BMDC phagosomes in − − an AP-3–dependent manner. WT and AP-3 / BMDCs that had been nontransduced (A and B) or trans- duced with retroviruses encoding PI4KIIα-GFP (C–G) were pulsed with LPS-coated magnetic beads (A and B), LPS-coated polystyrene beads (C and D), or LPS/ OVA-TxR–coated polystyrene beads (E–G) and chased as indicated. (A) Purified phagosomes (Left)orwhole- cell lysates (WCL) (Right)wereanalyzedbySDS/PAGE 10% and immunoblotting for endogenous PI4KIIα and actin. Shown are the relevant bands for each blot. (B) Quantification of band intensities for phagosomal PI4KIIα from three independent experiments, showing fold change relative to WT time 0 and normalized to actin (mean ± SD). (C and D) Phagosomes from non- − − transduced or PI4KIIα–GFP-expressing WT and AP-3 / BMDCs were isolated and analyzed by flow cytometry. GFP-positive cells were not previously sorted. (C) Shown are histogram plots of a representative ex- periment with the percentages of gated GFP-positive phagosomes (signal over phagosomes from non- transduced cells) indicated. Solid black lines, WT; solid − − orange lines, AP-3 / ; dashed lines, nontransduced controls. (D)Data(mean± SD) from three indepen- dent experiments performed in duplicate were nor- malized to the percent of transduced BMDCs (SI IMMUNOLOGY AND INFLAMMATION − − Appendix,Fig.S4A). (E–G)WTandAP-3/ BMDCs expressing PI4KIIα-GFP were analyzed by live cell im- aging. (E) Representative images at indicated times after phagocytosis. Dotted white lines, cell outlines; arrowheads, PI4KIIα–GFP-positive phagosomes; aster- isks, PI4KIIα–GFP-negative phagosomes; arrows, phagotubules. (Scale bar, 9 μm.) (F)Datafromthree independent experiments, 20 cells per experiment, are presented as percent of PI4KIIα–GFP-positive phag- osomes per cell. Black dots, WT; gray dots, AP-3−/−;solid red lines, means. (G) Quantification was performed as shown on representative images, drawing a line across the phagosomes (Left) and analyzing the line plot with ImageJ (Right; +, positive signal). Phagosomes contain- ing GFP-positive puncta, even partially in the total phagosomal membrane surface (mostly in the case of AP-3−/− BMDCs) were considered positive. **P < 0.01; ***P < 0.001.

PI4KIIα Is Required for the Accumulation of PtdIns4P on Early and P4Mx2 probe detected PtdIns4P on the plasma membrane and in Maturing Phagosomes and Phagotubules. To test whether PtdIns4P an intracellular pool [as described in other cells (44)] (Fig. 2A). In was produced on DC phagosomes concomitantly with PI4KIIα addition, after engulfment of LPS/OVA-TxR beads, PtdIns4P was recruitment, we expressed GFP-tagged P4M-SidMx2 (P4Mx2)—a found on nascent phagosomes in all DC types. In addition, probe for PtdIns4P derived from the SidM domain of Legionella PtdIns4P was present in early and maturing phagosomes (arrow- pneumophila (44)—in DCs and analyzed cells over time after heads) and in phagotubules (arrows) in shRNA control DCs phagocytosis of LPS/OVA-TxR beads using live imaging. In ad- (Fig. 2A). However, DCs knocked down for PI4KIIα showed dition, we assessed the contribution of PI4KIIα and the related significantly reduced accumulation of PtdIns4P on early and ma- PI4KIIβ—which does not bind AP-3—to the production of turing phagosomes (98% in control DCs vs. 0% in PI4KIIα PtdIns4P on phagosomes by transducing bone marrow precursors knockdown DCs between 30 and 120 min; Fig. 2B), while the with short hairpin RNA (shRNA) targeted to each PI4KII isoform plasma membrane signal was not affected (Fig. 2A). Conversely, or a nontarget control shRNA (SI Appendix,Fig.S5A and DCs knocked down for PI4KIIβ showed reduced binding of Fig. 2A). Knockdown was specific for each PI4KII isoform. GFP–P4Mx2 to the plasma membrane, while accumulation of PI4KIIα shRNA #2 was more effective than #1 and was used in PtdIns4P on phagosomes (99 ± 1% between 30 and 120 min) and subsequent experiments. Retroviral transduction did not signifi- phagotubules was unaffected (Fig. 2A). Moreover, and consistent cantly affect DC differentiation (SI Appendix,Fig.S5B), and with the reduced recruitment of PI4KIIα to phagosomes in AP- − − phagocytic capacity was not affected by the protein knockdowns 3 / DCs (Fig. 1), PtdIns4P on phagosomes was significantly re- (SI Appendix,Fig.S5C). DC maturation in response to LPS was duced in AP-3–deficient DCs (2 ± 2% between 30 and 120 min; reduced by shRNA transduction, but comparable between the Fig. 2B). These data suggest that AP-3 and its cargo protein different shRNA treatments (SI Appendix,Fig.S5D). The GFP– PI4KIIα are required for PtdIns4P formation on phagosomes,

López-Haber et al. PNAS Latest Articles | 3of12 Fig. 2. PI4KIIα is needed to accumulate PtdIns4P on early and maturing phagosomes. WT BMDCs were transduced with retroviruses encoding GFP–P4Mx2 and with lentiviruses encoding nontarget (control), − − PI4KIIβ or PI4KIIα shRNAs, and AP-3 / BMDCs were transduced only with retroviruses encoding GFP– P4Mx2. DCs were pulsed with LPS/OVA-TxR–coated beads, chased as indicated, and analyzed by live cell imaging. (A) Representative images. Dotted white lines, cell outlines; arrowheads, GFP–P4Mx2-positive phagosomes; asterisks, GFP–P4Mx2-negative phag- osomes; arrows, phagotubules. (B) Data from three independent experiments, 20 cells per experiment, are presented as percent of GFP–P4Mx2-positive phagosomes per cell. Black dots, nontarget control shRNA; blue dots, PI4KIIβ shRNA; red dots, PI4KIIα shRNA; gray dots, AP-3−/−; solid color lines, means. (Scale bar, 9 μm.) ***P < 0.001; n.s., not significant.

while PI4KIIβ primarily contributes to the plasma membrane knockdown DCs or DCs treated with control shRNAs exhibited PtdIns4P pool in DCs. robust phagotubule formation, PI4KIIα knockdown severely impaired the formation of phagotubules, as also observed in AP- − − PI4KIIα Is Required for Optimal Phagosomal TLR4 Signaling and 3 / DCs (Fig. 3 B and C and Movies S5–S7; see also Figs. 1E MHC-II Presentation of Phagocytosed Antigen in DCs. We and others and 2A, time 120 min). These data show that optimal phag- showed that TLR4 activation via its signaling adaptor MyD88 osomal TLR-induced proinflammatory cytokine secretion and on phagosomes triggers a signaling cascade that leads to proin- phagotubule formation require PI4KIIα. Noteworthy, like OVA- flammatory cytokine production, phagotubule formation, and containing phagotubules (7), PI4KIIα-positive phagotubules MHC-II presentation of phagocytosed cargo in DCs (7, 10, 45). were also positive for MHC-II (SI Appendix, Fig. S6). To test whether these responses require PI4KIIα,weprobedfor Finally, we assessed MHC-II presentation of phagocytosed TLR4/MyD88 responses in DCs derived from bone marrow antigen using beads coated with the Eα52–68 peptide; subsequent transduced with nontarget shRNA or shRNA to PI4KIIα, presentation of this peptide by the MHC-II molecule I-Ab at the PI4KIIβ, or the SNARE protein Sec22b (46) as an additional cell surface can be detected by the YAe antibody (47). DCs were control (SI Appendix, Fig. S5A). We first analyzed the production pulsed with EαGFP-coated beads or with soluble EαGFP (to of the proinflammatory interleukin 6 (IL-6) by DCs pulsed with monitor presentation of soluble antigen) or Eα52–68 peptide LPS-coated beads (relative to uncoated beads as a negative alone (a control that binds to surface I-Ab); all preparations control) by enzyme-linked immunosorbent assay (ELISA) anal- contained LPS to stimulate TLR4. As expected from the reduced ysis of cell supernatants. LPS bead-induced IL-6 levels were DC maturation observed in shRNA-transduced cells (SI Ap- significantly reduced in supernatants of PI4KIIα knockdown DCs pendix, Fig. S5D), LPS-bead induced MHC-II expression on compared to those from PI4KIIβ knockdown DCs or DCs transduced DCs was reduced compared to nontransduced DCs, transduced with control shRNAs (Fig. 3A). To test for phag- but similar between the shRNA treatments (except for Sec22b otubule formation, we analyzed DCs by live cell imaging 2.5 h shRNA, which was reduced more and, thus, not included in these after exposure to LPS/OVA-TxR beads. Whereas PI4KIIβ assays; SI Appendix, Fig. S7A). Of note, the LPS-bead-induced

4of12 | www.pnas.org/cgi/doi/10.1073/pnas.2001948117 López-Haber et al. Fig. 3. PI4KIIα is required for optimal phagosomal TLR4 signaling. WT BMDCs were nontransduced (−) or transduced with lentiviruses encoding nontarget (control), PI4KIIβ, Sec22b, or either of two PI4KIIα shRNAs, and AP-3−/− BMDCs were untransduced. DCs were pulsed with uncoated or LPS-coated polysty- rene beads (A) or LPS/OVA-TxR–coated beads (B and C). (A) IL-6 released into the supernatants was measured by ELISA after a 3-h chase. Results repre- sent mean ± SD of three experiments, each per- formed in triplicate. (B and C) BMDCs were analyzed by live cell imaging after a 2.5-h chase. (B) Arrows IMMUNOLOGY AND INFLAMMATION indicate phagotubules. (Scale bar, 9 μm.) (C) The percentage of BMDCs presenting phagotubules (tu- bules ≥ 1 μm emerging from phagosomes) in three independent experiments, 20 cells per experiment, is shown. ***P < 0.001. Lack of statistical significance is not indicated. expression of MHC-II and of the costimulatory molecules CD40 and kinase activity, we tested whether PI4KIIα mutants could and CD86 (SI Appendix, Fig. S7B) and phagosomal degradation rescue these phenotypes in PI4KIIα knockdown cells. Bone marrow capacity (decreased OVA labeling from phagocytosed OVA- cells were transduced with retroviruses encoding either human WT beads; SI Appendix, Fig. S7C) did not differ between DCs PI4KIIα-GFP, the AP-3 sorting mutant (L61,62A—which does not treated with PI4KIIα and control shRNAs. As also observed for bind AP-3), or the kinase-inactive mutant (D308A), all previously − − AP-3 / DCs (9), whereas formation of cell-surface Eα:I-Ab characterized (34), and subsequently transduced with lentiviruses complexes 6 h following exposure to Eα52–68 peptide or soluble encoding murine PI4KIIα shRNA; control cells were transduced EαGFP was similar among cells treated with either shRNA with nontarget shRNA lentiviruses (Fig. 5A). Transduced DCs were (Fig. 4 A and B), surface Eα:I-Ab complex levels after phago- then pulsed with LPS/OVA-TxR beads and analyzed by live cell cytosis of EαGFP-coated beads were significantly reduced by imaging after 2.5 h. Phagosomal recruitment of both the AP-3 expression of PI4KIIα shRNA relative to PI4KIIβ or nontarget sorting mutant and the kinase-inactive mutant was significantly control shRNAs (Fig. 4 C and D). To test if MHC-II presenta- impaired compared to WT PI4KIIα–GFP (WT, 98 ± 1%; D308A, tion correlated with CD4+ T cell responses, DCs pulsed with the 24 ± 3%; L61,62A, 10 ± 2%; Fig. 5 B and C), consistent with OVA-derived OVA323–329 peptide, soluble OVA/LPS, or OVA/ previous reports for PI4KIIα–GFP recruitment to lysosomes and LPS-coated beads were cocultured with OVA323–329-specific, lysosome-related organelles in other cells (33, 34). Moreover, only I-Ab–restricted OT-II T cells. OT-II cell activation, measured by WT PI4KIIα–GFP restored phagotubule formation in knockdown CD69 expression and IL-2 production, by DCs pulsed with DCs and was itself recruited to phagosomal tubules (Fig. 5B). WT peptide alone or soluble OVA was similar whether DCs PI4KIIα–GFP, but not the kinase-inactive or AP-3 binding mutants, expressed PI4KIIα, PI4KIIβ, or control shRNAs (Fig. 4 E–H). In also restored control levels of proinflammatory cytokine production contrast, OT-II cell activation by DCs pulsed with OVA/LPS (Fig. 5D) and surface MHC-II presentation of phagocytosed Eα beads was significantly reduced in PI4KIIα shRNA-expressing (Fig. 5E). Thus, both AP-3 binding and kinase activity are required DCs relative to cells expressing other shRNAs (Fig. 4 I and J). for PI4KIIα recruitment to phagosomes and consequent phag- These data indicate that PI4KIIα promotes MHC-II presenta- osomal function in TLR4 signaling. tion of antigen following phagocytosis, but not endocytosis, and are consistent with our observations that PI4KIIα is required for PI4KIIα Promotes TLR4 Accumulation on Phagosomes in DCs. To test PtdIns4P formation on phagosomes, but not on the plasma whether TLR4 localization to phagosomes required PI4KIIα,we membrane (Fig. 2). pulsed DCs with LPS-coated beads and analyzed TLR4 presence on isolated phagosomes by immunoblotting and flow cytometry PI4KIIα Phagosomal Function Requires the AP-3–Sorting Motif and using two different antibodies. Immunoblotting showed that Kinase Activity. In order to assess if PI4KIIα recruitment and cellular TLR4 expression was similar in DCs treated with con- consequent phagosomal TLR signaling required AP-3 binding trol, PI4KIIα, or PI4KIIβ shRNAs (Fig. 6 A, Left). In both assays,

López-Haber et al. PNAS Latest Articles | 5of12 Fig. 4. PI4KIIα is required for optimal MHC-II pre- sentation of phagocytosed antigen. WT BMDCs transduced with lentiviruses encoding nontarget (control), PI4KIIβ, or PI4KIIα shRNAs were pulsed with

Eα52–68 peptide (A), soluble EαGFP fusion protein (B), EαGFP-coated beads (C and D), OVA323–329 peptide (E and F), soluble OVA (G and H), or OVA:BSA-coated beads (I and J) and chased for 6 h. (A–D) Surface b expression of Eα52–68:I-A complexes was analyzed by flow cytometry using YAe antibody. (A–C) Shown are the percentages of CD11c+ BMDCs that were b+ Eα52–68:I-A .(D) Shown is the YAe mean fluores- cence intensity (MFI) normalized to MHC-II MFI (SI Appendix, Fig. S5D). (E–J) Pulsed BMDCs were fixed and cocultured with preactivated OT-II cells for 18 h. (E, G, and I) CD69 expression by OT-II cells was assessed by flow cytometry. Shown are the per- centages of vb5+ T cells that expressed CD69. (F, H, and J) IL-2 production by OT-II cells was measured by ELISA on the coculture supernatants. Data in all panels represent mean ± SD of three experiments performed in duplicate. *P < 0.05; **P < 0.01; ***P < 0.001. Lack of significance is not indicated.

TLR4 increasingly accumulated on phagosomes over time after membrane and on nascent phagosomes (arrowheads, Fig. 7 A phagocytosis in control [as we had shown before (10)] and PI4KIIβ and B), consistent with its ability to bind PtdIns(4,5)P2 and knockdown DCs, but not in cells knocked down for PI4KIIα similar to the localization of the PtdIns(4,5)P2-sensing probe (Fig. 6 A–D). This was true whether data were analyzed for total GFP–PH-PLCδ (pleckstrin homology domain of phospholipase TLR4 content on phagosomes by immunoblotting (Fig. 6 A and Cδ; SI Appendix, Fig. S8 A and B). Recruitment of TIRAP–GFP C) or by percentage of phagosomes harboring TLR4 by flow and GFP–PH-PLCδ to the plasma membrane was modestly af- cytometry (Fig. 6 B and D). These data indicate that efficient fected by knockdown of PI4KIIβ, but not PI4KIIα, whereas TLR4 localization to phagosomes requires PI4KIIα. TIRAP–GFP (but not GFP–PH-PLCδ) was detected on many fewer phagosomes in PI4KIIα knockdown cells after the pulse Sorting Adaptor TIRAP Recruitment to Phagosomes Requires PI4KIIα. (Fig. 7 A and B, time 0, and SI Appendix, Fig. S8 A and B). Over Formation of the MyDdosome complex and downstream TLR4 time, TIRAP–GFP was recruited to phagosomes in control and signaling is favored by TLR4 binding to its sorting adaptor PI4KIIβ knockdown cells, as quantified both by fluorescence TIRAP (23, 48). TIRAP contains an N-terminal lysine-rich microscopy (Fig. 7B and Movie S8) and flow cytometry on iso- polybasic motif that promiscuously binds to different phosphoi- lated phagosomes (Fig. 8A), normalizing to the percentage of nositide species, including PtdIns(4,5)P2 and PtdIns4P (25). TIRAP–GFP-positive cells (Fig. 8C). TIRAP–GFP was also While PtdIns(4,5)P2 mostly localizes to the plasma membrane detected on phagotubules at 120 min (arrows, Fig. 7A and Movie and to nascent phagosomes, PtdIns4P—which is present on late S9). However, TIRAP recruitment to phagosomes was severely endocytic compartments—was a strong candidate for TIRAP reduced in PI4KIIα knockdown cells between 30 and 120 min binding to maturing phagosomal membranes. To test whether (on 90 ± 10% of phagosomes in control and PI4KIIβ knockdown TIRAP is recruited to phagosomes and whether recruitment cells vs. 1 ± 5% of phagosomes in PI4KIIα knockdown DCs; requires PI4KIIα, we followed the kinetics of TIRAP–GFP re- Fig. 7B and Movie S10), with no detectable phagosomal increase, cruitment to phagosomes by live cell imaging (Fig. 7) and flow as measured by flow cytometry on isolated phagosomes (Fig. 8 A cytometry on isolated phagosomes (Fig. 8) from TIRAP–GFP- and C). Consistent with the role of AP-3 in PI4KIIα recruitment transduced DCs (SI Appendix, Fig. S7 D, Upper; note that GFP- to phagosomes, TIRAP recruitment to phagosomes was also positive cells were not previously sorted). In cells pulsed with impaired in AP-3–deficient DCs (on 1 ± 5% of phagosomes OVA-TxR beads, TIRAP–GFP was detected on the plasma between 30 and 120 min; Fig. 7B). In contrast to these

6of12 | www.pnas.org/cgi/doi/10.1073/pnas.2001948117 López-Haber et al. Fig. 5. PI4KIIα function on phagosomes requires the AP-3 sorting motif and kinase activity. WT BMDCs that were nontransduced (−) or transduced first with retroviruses encoding GFP, human WT PI4KIIα–GFP, PI4KIIα(D308A)–GFP, or PI4KIIα(L61,62A)–GFP were then transduced with lentiviruses encoding mouse nontarget (ctrl) or PI4KIIα shRNAs. Cells were left untreated (A) or pulsed with LPS/OVA-TxR–coated beads (B and C), uncoated or LPS-coated polystyrene beads (D), or EαGFP-coated beads (E). (A) Whole-cell lysates were analyzed by SDS/PAGE and immuno- blotting for PI4KIIα, GFP, and tubulin. Shown are the relevant bands for each blot. (B and C) BMDCs were analyzed by live cell imaging after a 2.5-h chase. (B) Representative images. Dotted white lines, cell out- lines; arrowheads, PI4KIIα–GFP-positive phagosomes; asterisks, PI4KIIα–GFP-negative phagosomes; arrows, phagotubules. (Scale bar, 9 μm.) (C) Data from three independent experiments, 20 cells per experiment, are presented as percent of PI4KIIα–GFP-positive phagosomes per cell. Black dots, WT; purple dots, PI4KIIα(D308A)–GFP; tidal dots, PI4KIIα(L61,62A)– GFP; solid color lines, means. (D) IL-6 released into the supernatants after 3 h was measured by ELISA. (D, Left) Representative experiment performed in triplicate. (D, Right) IL-6 values from three indepen- dent experiments performed in triplicate are shown

as percent of values for BMDCs treated with non- IMMUNOLOGY AND INFLAMMATION target (ctrl) shRNA, as a representation of pheno- typic rescue (mean ± SD). (E) Surface expression of b Eα52–68:I-A complexes was analyzed by flow cytom- etry using YAe antibody. (E, Left) Shown are the + b+ percentages of CD11c BMDCs that were Eα52–68:I-A in a representative experiment. (E, Right)YAevalues from two independent experiments performed in duplicate are shown as percent of values for BMDCs treated with nontarget (ctrl) shRNA, as a representa- tion of phenotypic rescue (mean ± SD). (D and E) Significance relative to nontarget shRNA-treated WT control (Left)orPI4KIIα shRNA-treated DCs (−)(Right) is indicated. ***P < 0.001. Lack of significance is not indicated. observations, phagosomal recruitment of the p40-phox domain TIR–GFP. TIRAP knockdown did not affect DC differentia- containing, PX-TIR–GFP construct (SI Appendix, Fig. S7 D, tion compared to shRNA control-treated DCs (SI Appendix, Fig. Lower), the TIRAP variant that exclusively binds PtdIns3P S9B), and knocked-down DCs were similarly transduced with (25)—a lipid enriched on early endosomes and early phag- TIRAP or PX-TIR constructs (SI Appendix, Fig. S9C). Consis- osomes (49)—was not affected by PI4KIIα knockdown or AP-3 tent with our observations in WT DCs (Figs. 7 and 8), deficiency (Fig. 8 B and D and SI Appendix, Fig. S8 C and D). TIRAP–GFP was efficiently recruited to phagosomes at all times PX-TIR–GFP was mainly detected on phagosomes between 30 and to phagosomal tubules in mature phagosomes, while PX- and 60 min after the pulse (Fig. 8B and SI Appendix, Fig. S8 C TIR–GFP was recruited to early phagosomes and was no longer and D), at times when the PtdIns4P probe GFP–P4Mx2 and detected at 120 min (asterisks, SI Appendix, Fig. S9D), as TIRAP–GFP were also detected in control and PI4KIIβ expected, since PtdIns3P is limited to early phagosomes (49). knockdown DCs (Figs. 2 and 7). The observation that PI4KIIα TIRAP knockdown significantly reduced DC production of IL-6 knockdown or AP-3 knockout severely impairs TIRAP, but not after LPS-bead stimulation, as expected (Fig. 8E). Remarkably, PX-TIR, recruitment to phagosomes suggests that TIRAP expression of PX-TIR–GFP did not improve IL-6 production, preferentially binds PtdIns4P on DC phagosomes. Note that whereas TIRAP–GFP significantly increased proinflammatory phagotubules are labeled by OVA-TxR, but not by PX-TIR-GFP cytokine production (Fig. 8E). This result supports that TIRAP (SI Appendix, Fig. S8C), suggesting that phagotubules, which binding to PtdIns3P is not sufficient to promote TLR4 signaling emanate from mature phagosomes, do not contain PtdIns3P. from phagosomes and that binding to PtdIns4P is necessary. These results show that PI4KIIα is necessary and sufficient for TIRAP recruitment to phagosomes and suggest that, in contrast Discussion to its recruitment to early endosomes, PtdIns3P is not sufficient Intracellular trafficking pathways play a crucial role in preserving to recruit TIRAP to early phagosomes in DCs. organelle identity. We previously showed that a key mediator of To further investigate if TIRAP binding to PtdIns4P is re- TLR4 localization to phagosomes in DCs is the trafficking quired for TLR4 signaling from maturing phagosomes, we adaptor protein AP-3. However, loss of AP-3 expression did not knocked down the endogenous TIRAP in BMDCs (SI Appendix, completely abrogate phagosomal TLR accumulation or signal- Fig. S9A) and expressed either human TIRAP–GFP or PX- ing, suggesting that AP-3 might play a regulatory role in this

López-Haber et al. PNAS Latest Articles | 7of12 Fig. 6. PI4KIIα promotes TLR4 recruitment to phag- osomes in DCs. WT BMDCs transduced with lentiviruses encoding nontarget (control), PI4KIIβ,orPI4KIIα shRNAs were pulsed with LPS-coated magnetic beads (A and C) or LPS-coated polystyrene beads (B and D)andchased as indicated. (A) Purified phagosomes (Right) or whole- cell lysates (WCL) (Left) were analyzed by SDS/PAGE and immunoblotting for TLR4 and actin. Shown are the relevant bands for each blot. (C) Quantification of band intensities for phagosomal TLR4 from three indepen- dent experiments, showing fold change relative to each shRNA treatment at time 0 and normalized to actin (mean ± SD). (B and D) Phagosomes were pu- rified and analyzed by flow cytometry using a fluores- cein isothiocyanate-conjugated anti-TLR4 antibody. (B) Shown are histogram plots of a representative experi- ment with the percentages of gated TLR4-positive phagosomes indicated. Solid black lines, nontarget (control) shRNA; solid blue lines, PI4KIIβ shRNA; solid red lines, PI4KIIα shRNA; dashed lines, nontransduced controls. (D) Data from three independent experi- ments performed in duplicate are shown as fold change of percent of TLR4-positive phagosomes rela- tive to each shRNA treatment at time 0 (mean ± SD). *P < 0.05; **P < 0.01; ***P < 0.001.

process. We have now identified the lipid kinase PI4KIIα as an TIRAP was shown to promiscuously bind to different phos- AP-3 cargo that is required to promote TLR4 signaling from phoinositides in macrophages and DCs, including PtdIns(4,5)P2 phagosomes. PI4KIIα trafficking to phagosomes in DCs requires on the plasma membrane and PtdIns3P on early endosomes (25). its kinase activity and binding to AP-3, consistent with the re- Here, we show that on DC-maturing phagosomes, TIRAP quirements for PI4KIIα delivery to lysosomes, lysosome-related binding is highly dependent on PtdIns4P and that knockdown of organelles, and synaptic vesicles in other cell types (34). PI4KIIα PI4KIIα, but not of PI4KIIβ, significantly reduces TIRAP re- recruitment to phagosomes and its dependence on AP-3 binding cruitment to phagosomes. In contrast, TIRAP association with were observed both in BMDCs—which, despite their heteroge- the plasma membrane and nascent phagosomes is PI4KIIα- neity, may mirror monocyte-derived DCs in vivo (50, 51)—and in independent—similarly to PH-PLCδ, a probe for PtdIns(4,5) tissue-resident DCs isolated from spleen. PI4KIIα, in turn, P2—suggesting that TIRAP binding to the plasma membrane is generates PtdIns4P on phagosomes, and knockdown of PI4KIIα not absolutely dependent on PtdIns4P and requires PtdIns(4,5) leads to reduced phagosomal PtdIns4P detected by the P4Mx2 P2. Consistent with observations in macrophages, PtdIns3P (de- probe. The phagosomal effects are specific for PI4KIIα,as tected with the TIRAP construct PX-TIR) accumulates on early knockdown of the genetically distinct PI4KIIβ—which does not phagosomes in DCs, and its presence is not dependent on bind AP-3—does not impact phagosomal PtdIns4P or TLR4 PI4KIIα or PI4KIIβ. However, WT TIRAP is not recruited to signaling. Finally, we show that PI4KIIα activity is required to phagosomes at this stage in the absence of PI4KIIα, suggesting recruit the TLR sorting adaptor TIRAP to phagosomes and that that TIRAP is preferentially recruited to early phagosomes by TIRAP binding to PtdIns4P is necessary for optimal proin- PtdIns4P. Furthermore, unlike exogenous WT TIRAP, PX-TIR flammatory TLR4 signaling. Together, the data suggest a model recruitment to early phagosomes failed to promote proin- in which AP-3–dependent trafficking of PI4KIIα to phagosomes flammatory signaling in the absence of endogenous TIRAP. Our generates a pool of PtdIns4P that is necessary to recruit TIRAP observation that TIRAP is preferentially recruited to PtdIns4P- and sustain TLR4 proinflammatory signaling and MHC-II rich domains on phagosomes resembles TIRAP preferential presentation from phagosomes. binding to PtdIns(4,5)P2 on the plasma membrane to promote Our data show that PI4KIIα builds a PtdIns4P platform on the TLR4 signaling (23). This also suggests that additional organelle- maturing phagosome that allows the binding of the TLR sorting specific determinants might limit or promote phosphoinositide- adaptor TIRAP. TIRAP serves as a landmark for the assembly of dependent TIRAP recruitment to membranes and highlights the MyDdosome complex on phosphoinositide-enriched do- differences between the early endosomal and phagosomal sys- mains, promoting the initiation of TLR signal transduction. tems in DCs. Moreover, these data support the importance of

8of12 | www.pnas.org/cgi/doi/10.1073/pnas.2001948117 López-Haber et al. Fig. 7. Sorting adaptor TIRAP recruitment to phag- osomes is severely impaired by PI4KIIα knockdown. WT BMDCs were transduced with retroviruses encod- –

ing TIRAP GFP, and lentiviruses encoding nontarget IMMUNOLOGY AND INFLAMMATION (control), PI4KIIβ or PI4KIIα shRNAs, and AP-3−/− BMDCs were transduced only with retroviruses encoding TIRAP–GFP. DCs were pulsed with LPS/OVA-TxR–coated beads, chased as indicated and analyzed by live cell imaging. (A) Representative images. Dotted white lines, cell outlines; arrowheads, TIRAP–GFP-positive phagosomes; asterisks, TIRAP–GFP-negative phagosomes; arrows, phagotubules. (B) Data from three indepen- dent experiments, 20 cells per experiment, are pre- sented as percent of TIRAP–GFP-positive phagosomes per cell. Black dots, nontarget control shRNA; blue dots, PI4KIIβ shRNA; red dots, PI4KIIα shRNA; gray − − dots, AP-3 / ; solid color lines, means. (Scale bar, 9 μm.) ***P < 0.001; n.s., not significant. distinguishing TLR signaling from distinct subcellular locations responses were not completely abrogated. The persistence of to reflect differences in cargo characteristics or origins—in this TLR4 signaling in the absence of TIRAP phagosomal recruitment case, soluble vs. particulate—consistent with our observations may result from residual signaling from plasma membrane- that AP-3 is required for TLR signaling induced by phagosomal, engaged TLRs, which are not regulated by PI4KIIα or AP-3. In- but not endosomal, cargoes (10). deed, we show that PI4KIIα specifically contributes to the phag- This distinction between the endosomal and phagosomal sys- osomal PtdIns4P pool. In contrast, the plasma-membrane pool of tems in DCs could also reflect variations between DCs and other PtdIns4P—while somewhat reduced by knockdown of PI4KIIβ, phagocyte endolysosomal systems. Consistent with this, while but not of PI4KIIα—is primarily maintained by PI4KIIIα (56), PtdIns(4,5)P2 and PtdIns3P are present on nascent phagosomes supporting—together with the large pool of PtdIns(4,5)P2— and early phagosomes, respectively, in macrophages (37), both TLR4/MyDdosome signaling from the plasma membrane in cells persist longer on phagosomes in DCs than in macrophages. By depleted of PI4KIIα. In addition, it is possible that TLR4 could contrast, PtdIns4P is generated earlier in the phagosomal mat- signal on the phagosome independently of TIRAP. Indeed, the uration process in DCs compared to macrophages. Based on TIRAP requirement for TLR signaling could be bypassed by high differences in the progression of phosphoinositide content shown ligand concentrations that may not reflect physiological conditions here for DCs and previously for macrophages (36, 37), and on (25, 57, 58). the acquisition of lysosomal proteins and low pH, the phag- In live cell-imaging experiments, PI4KIIα, PtdIns4P, and osomal maturation process is slowed in DCs compared to that in TIRAP were also detected on phagosomal tubules that we had macrophages, supporting DC specialization in antigen presen- previously shown to promote MHC-II presentation from phag- tation (52, 53). The earlier formation of PtdIns4P on DC osomes. Moreover, knockdown of PI4KIIα or loss of AP-3 ex- phagosomes with the consequent recruitment of TIRAP and pression abrogated phagosomal tubule formation. This could assembly of the MyDdosome licenses DC phagosomes for anti- reflect reduced binding of the Rab7 adaptor FYCO1, which gen presentation, an activity that is not a primary function in modulates kinesin microtubule motor protein required for tu- macrophages (54, 55). bule formation (59). Indeed, Rab7 is present on PtdIns4P Of note, even though knockdown of PI4KIIα abolished enriched compartments (60). Of note, kinesin recruitment to TIRAP binding to maturing phagosomes, TLR4 downstream phagosomal tubules in macrophages is mediated by the GTPase

López-Haber et al. PNAS Latest Articles | 9of12 Fig. 8. Impaired IL-6 production upon TIRAP knock- down is restored by expression and phagosomal re- cruitment of TIRAP–GFP, but not of the PtdIns3P- binding PX-TIR. (A–D) WT BMDCs transduced with retroviruses encoding TIRAP–GFP or PX-TIR–GFP and lentiviruses encoding nontarget (control), PI4KIIβ,or PI4KIIα shRNAs, were pulsed with LPS-coated polysty- rene beads and chased as indicated. Phagosomes were purified and analyzed by flow cytometry. GFP- positive cells were not previously sorted. (A and B) Shown are histogram plots of a representative ex- periment with the percentages of gated phagosomes that were GFP-positive indicated. Solid black lines, nontarget (control) shRNA; solid blue lines, PI4KIIβ shRNA; solid red lines, PI4KIIα shRNA; dashed lines, nontransduced controls. (C and D)Data(mean± SD) from three independent experiments performed in duplicate were normalized to the percent of trans- duced BMDCs (SI Appendix,Fig.S7D). (E)WTBMDCs were pulsed with LPS-coated polystyrene beads, and IL-6 released into the supernatants after 5 h was measured by ELISA. (E, Left) Representative experi- ment performed in triplicates. (E, Right) IL-6 values from three independent experiments performed in triplicate are shown as percent of values (mean ± SD) for BMDCs treated with nontarget (ctrl) shRNA, as a representation of phenotypic rescue. Significance rel- ative to TIRAP shRNA-treated DCs (−)(E, Right)isin- dicated. **P < 0.01; ***P < 0.001; n.s., not significant.

Arl8b (38), which is also required for MHC-II presentation in federal regulations set forth in the recommendations in the Public Health DCs (61). In addition, PtdIns4P may be required for the re- Service Policy on the Humane Care and Use of Laboratory Animals (64), the ’ cruitment of membrane curvature stabilizing proteins (62). Ex- National Research Council s Guide for the Care and Use of Laboratory Ani- pression of WT PI4KIIα, but not the kinase-inactive or AP-3 mals (65), the NIH Office of Laboratory Animal Welfare, the American Vet- erinary Medical Association Guidelines on Euthanasia, and the guidelines of sorting mutants, rescued DC defects in phagosomal TLR sig- the Institutional Animal Care and Use Committees of the Children’s Hospital naling and MHC-II presentation. These results are in agreement of Philadelphia. All protocols used in this study were approved by the In- with previous reports showing that both the dileucine sorting stitutional Animal Care and Use Committee at the Children’s Hospital of motif and the kinase active site are required for PI4KIIα re- Philadelphia. cruitment to lysosomes in neuronal cells and that PI4KIIα acts See SI Appendix for details on mouse strains, reagents, antibodies, DNA both as a cargo and as a regulator of AP-3 function by promoting constructs, shRNAs, lentiviral and retroviral production, and transduction the formation of additional AP-3–recruiting PtdIns4P patches of DCs. (34). Together, these observations support the conclusion that AP- ± 3–dependent recruitment of PI4KIIα to phagosomes generates a Cell Culture. Bone-marrow cells were isolated and cultured for 7 9din pool of PtdIns4P that is required for the formation of phagosomal Roswell Park Memorial Institute (RPMI) 1640 medium (Gibco, ThermoFisher) supplemented with 10% low endotoxin-defined fetal bovine serum tubules, TIRAP recruitment, and concomitant enhancement of (HyClone), 2 mM L-Gln, 50 μM 2-mercaptoethanol (Invitrogen), and 30% MHC-II presentation. Our data also suggest that impaired phag- granulocyte-macrophage colony-stimulating factor-containing conditioned osomal PI4KIIα recruitment may at least, in part, explain defec- medium from J558L cells (kindly provided by former Ralph Steinman labo- tive antibacterial immune responses in AP-3–deficient mice and ratory, The Rockefeller University, New York, and Maria Paula Longhi, HPS2 patients. Queen Mary University of London, London) for differentiation to DCs, In summary, our data indicate that AP-3–mediated recruitment as described (66, 67). Maturation was induced by 18-h treatment with LPS of PI4KIIα early in the life cycle of the DC phagosome is a pre- (0.1 μg/mL). Splenic DCs were isolated from single-cell suspensions with anti- requisite for the binding of TIRAP and the promotion of TLR4 CD11c (N418) microbeads after depletion of T, B, and natural killer cells with signaling and is a key determinant of the fate of the phagosome as a mixture of biotin-conjugated antibodies (to CD90.2, CD45R, and CD49b) and anti-biotin microbeads (Miltenyi Biotec Inc.). CD4+ T cells were isolated an autonomous signaling organelle. Further studies will be re- from single-cell suspensions of lymph nodes from OT-II transgenic mice, by quired to elucidate the signals that drive TIRAP and TLR4 to- positive selection with anti-CD4 microbeads (Miltenyi Biotec Inc.). gether on the phagosome to allow MyDdosome formation. Phagosome Purification and Protein Recruitment. Phagosomes were isolated Experimental Procedures essentially as described (68). Briefly, BMDCs were incubated for 15 min with Mice. Mice were bred under pathogen-free conditions in the Department of LPS-coated latex beads or 3-μm magnetic beads (Dynabeads M-280 strep- Veterinary Resources at the Children’s Hospital of Philadelphia and were tavidin; Invitrogen) and then chased. Magnetic and nonmagnetic phag- euthanized by carbon dioxide narcosis according to guidelines of the osomes were purified after different chase times by means of a magnet or American Veterinary Medical Association Guidelines for the Euthanasia of differential centrifugation, respectively, as described (69, 70). Purified LPS- Animals (63). All animal studies were performed in compliance with the bead phagosomes were fixed and stained with antibodies to TLR-4, lamp-1,

10 of 12 | www.pnas.org/cgi/doi/10.1073/pnas.2001948117 López-Haber et al. or negative controls, or left unstained in the case of the GFP-expressing DCs, on day 7 of culture. On day 8, DCs were pulsed for 30 min with 1 mg/mL TxR- and analyzed concurrently by flow cytometry, gating on the bead pop- conjugated OVA (Invitrogen, ThermoFisher Scientific) and LPS (100 μg/mL) ulation (68), and normalizing to the total percentage of GFP-positive cells in covalently coupled to 3-μm amino polystyrene beads, as described (67). DCs the case of the transduced DCs. Flow cytometry was performed by using were then washed with RPMI, chased for 0 to 2.5 h, and visualized by FACSCalibur and CellQuest software (BD Biosciences). Protein extracts from spinning-disk confocal microscopy using an Olympus inverted microscope purified magnetic phagosomes were analyzed by immunoblotting. equipped with an environmental chamber at 37 °C and 5% CO2 at the University of Pennsylvania’s Confocal Microscopy core or a BioVision and Protein-Degradation Assays. To evaluate intraphagosomal degradation, DMi8 Leica spinning-disk system using an Andor 888 cooled electron- BMDCs were pulsed with OVA-coated latex beads in a 1:5 ratio of DC:beads multiplying charge-coupled device camera equipped with temperature for 15 min and chased for the indicated times. Cells were then disrupted in and CO2 control units and associated VisiView software for image and video × detergent-containing lysis buffer and pelleted by centrifugation at 150 g capture. Time-lapse microscopy was performed by capturing image streams for 4 min as described (68). Supernatants containing the latex beads were over 1 to 5 min at 1 frame per s and analyzed by using ImageJ. Protein re- collected and stained with anti-OVA antibodies (Sigma), followed by cruitment to phagosomes was visualized and quantified as explained above. phycoerythrin-conjugated anti-rabbit antibodies (Jackson ImmunoResearch) in 96-well V-bottom microplates. Labeling was analyzed by flow cytometry, Antigen Presentation Assays. DCs were exposed to OVA, OVA:BSA-coated gating on the latex bead population by forward scatter and side scatter (68). 3-μm latex beads (Polysciences), or OVA-specific MHC-II peptides for 15 to 30 min at 37 °C, then washed in PBS and chased in complete medium at Immunoblotting. Immunoblotting was performed essentially as described 37 °C. DCs were then fixed with 0.005% glutaraldehyde in PBS for 1 min, (10). Briefly, Laemmli sample buffer with 2-mercaptoethanol was added to washed with 0.2 M glycine in PBS, and cocultured with CD4+ OT-II T cells that protein lysates from phagosome purification and whole-cell lysates. Samples had been prestimulated with anti-CD3 and anti-CD8 antibodies (72). T cell were then fractionated by 8%, 10%, or 12% sodium dodecyl sulfate (SDS)/ activation was monitored 18 h later as CD69 expression by flow cytometry polyacrylamide gel electrophoresis (PAGE) on polyacrylamide gels, trans- (FACSCalibur, BD Biosciences) and IL-2 secretion in coculture supernatants by ferred to polyvinylidene difluoride membranes (Immobilon-FL, Millipore), b ELISA (BD Biosciences). For presentation of Eα – peptide on I-A , BMDCs and analyzed by using Alexa Fluor 680- or 790-conjugated or horseradish 52 68 were incubated with 0.5 mg of soluble EαGFP, Eα – peptide, or EαGFP- peroxidase-conjugated secondary antibodies (Jackson ImmunoResearch) and 52 68 Odyssey (LI-COR) or iBright (Invitrogen) imaging systems. Densitometric coated (1 mg/mL) beads for 30 min, washed with PBS, and chased. Cells were analyses of band intensity was performed by using NIH ImageJ software, then fixed with 3% paraformaldehyde in PBS, stained with biotinylated YAe – normalizing to control protein levels. and allophycocyanin streptavidin (Invitrogen), and analyzed by flow cytometry. YAe labeling was quantified on CD11c+ cells that had taken up Cytokine Secretion after TLR4 Stimulation. BMDCs were incubated with LPS- one bead as gated on a forward-scatter vs. side-scatter plot (73). coated beads (1 μg/mL) for 3 or 5 h as described (10). IL-6 concentration in culture supernatants was measured by ELISA (BD Biosciences). Statistical Analyses. Statistical analyses and data plots were performed by IMMUNOLOGY AND INFLAMMATION using Microsoft Excel and GraphPad Prism software. Significance for ex- Immunofluorescence Microscopy. BMDCs on day 7 of culture or freshly isolated perimental samples relative to untreated or nontarget shRNA-treated WT splenic DCs were seeded on poly-L-lysine–coated glass-bottom 35-mm culture control (unless otherwise stated) was determined by using the unpaired dishes (MatTek), pulsed with OVA (1 mg/mL) and LPS (100 μg/mL) coupled to Student’s t test and ANOVA. Mean ± SEM values are indicated in Results. 3-μm amino polystyrene beads as described (67), fixed with 3% formalde- Error bars in figures represent mean ± SD. hyde in phosphate-buffered saline (PBS), permeabilized with Permwash (BD), and labeled with the indicated antibodies. Cells were analyzed by Data Availability. All study data are included in the article and SI Appendix. fluorescence confocal microscopy with a DMi8 Leica microscope and Visi- View software (Visitron Systems GmbH) for image capture and analyzed by ACKNOWLEDGMENTS. We thank Pietro De Camilli, Victor Faundez, Juan using ImageJ (NIH). Protein recruitment to phagosomes was visualized with Bonifacino, Warren Pear, Marion Pepper Pew, Mark Jenkins, Paula Oliver, the ImageJ plugin three-dimensional (3D) viewer and quantified by using Susan Ross, Maria Paula Longhi, and the former Ralph Steinman laboratory Analyze/Plot profile and Analyze/3D surface plot. for the generous gifts of reagents; Anand Sitaram and Shuixing Li for For the detection of phagosomal tubules, BMDCs were seeded and pulsed experimental assistance; Andrea Stout and the Microscopy core and David as indicated, chased for 2 h, fixed for 10 min at 37 °C in a periodate– Schultz and the High-Throughput Screening core at the University of Pennsylvania for expert technical assistance; and the Flow Cytometry core lysine–paraformaldehyde fixative (71) at 37 °C, washed with Hank’s bal- at the Children’s Hospital of Philadelphia. This work was supported by NIH anced salt solution, permeabilized with Permwash, and stained with the Grants R01 AI137173 (to C.L.-H. and A.R.M.) and R01 HL121323 (to J.B.-K., indicated antibodies. Y.Z., and M.S.M.); Canadian Institutes of Health Research Grant FDN-143202 (to R.L.-K. and S.G.); and the intramural research program of the NIH Eunice Live Cell Imaging. BMDCs expressing retroviral and/or lentiviral constructs Kennedy Shriver National Institute of Child Health and Human Development were seeded on poly-L-lysine–coated glass-bottom 35-mm culture dishes (to T.B.).

1. R. Medzhitov, C. A. Janeway Jr, Decoding the patterns of self and nonself by the 11. E. C. Dell’Angelica, V. Shotelersuk, R. C. Aguilar, W. A. Gahl, J. S. Bonifacino, Altered innate immune system. Science 296, 298–300 (2002). trafficking of lysosomal proteins in Hermansky-Pudlak syndrome due to mutations in 2. G. M. Barton, J. C. Kagan, A cell biological view of toll-like receptor function: Regu- the beta 3A subunit of the AP-3 adaptor. Mol. Cell 3,11–21 (1999). lation through compartmentalization. Nat. Rev. Immunol. 9, 535–542 (2009). 12. E. C. Dell’Angelica, J. S. Bonifacino, Coatopathies: Genetic disorders of protein coats. 3. R. S. Flannagan, V. Jaumouillé, S. Grinstein, The cell biology of phagocytosis. Annu. Annu. Rev. Cell Dev. Biol. 35, 131–168 (2019). Rev. Pathol. 7,61–98 (2012). 13.S.L.Bowman,J.Bi-Karchin,L.Le,M.S.Marks,Theroadtolysosome-relatedorganelles:In- 4. A. R. Mantegazza, M. S. Marks, “Lysosome-related organelles: Modifications of the sights from Hermansky-Pudlak syndrome and other rare diseases. Traffic 20,404–435 (2019). lysosome paradigm” in Lysosomes: Biology, Diseases, and Therapeutics, F. R. Maxfield, 14. R. H. Clark et al., Adaptor protein 3-dependent microtubule-mediated movement of J. M. Willard, S. Lu, Eds. (John Wiley and Sons, New York, NY, 2016), pp. 239–278. lytic granules to the immunological synapse. Nat. Immunol. 4, 1111–1120 (2003). 5. D. M. Underhill, H. S. Goodridge, Information processing during phagocytosis. Nat. 15. M. Sasai, M. M. Linehan, A. Iwasaki, Bifurcation of Toll-like receptor 9 signaling by Rev. Immunol. 12, 492–502 (2012). adaptor protein 3. Science 329, 1530–1534 (2010). 6. E. Hoffmann et al., Autonomous phagosomal degradation and antigen presentation 16. A. L. Blasius et al., Slc15a4, AP-3, and Hermansky-Pudlak syndrome proteins are re- in dendritic cells. Proc. Natl. Acad. Sci. U.S.A. 109, 14556–14561 (2012). quired for Toll-like receptor signaling in plasmacytoid dendritic cells. Proc. Natl. Acad. 7. A. R. Mantegazza et al., TLR-dependent phagosome tubulation in dendritic cells Sci. U.S.A. 107, 19973–19978 (2010). promotes phagosome cross-talk to optimize MHC-II antigen presentation. Proc. Natl. 17. S. Fontana et al., Innate immunity defects in Hermansky-Pudlak type 2 syndrome. Acad. Sci. U.S.A. 111, 15508–15513 (2014). Blood 107, 4857–4864 (2006). 8. J. M. Blander, L. E. Sander, Beyond pattern recognition: Five immune checkpoints for 18. M. Sugita et al., Failure of trafficking and antigen presentation by CD1 in AP-3-deficient scaling the microbial threat. Nat. Rev. Immunol. 12, 215–225 (2012). cells. Immunity 16,697–706 (2002). 9. H. Husebye et al., The Rab11a GTPase controls Toll-like receptor 4-induced activation 19. V. Briken, R. M. Jackman, S. Dasgupta, S. Hoening, S. A. Porcelli, Intracellular traf- of interferon regulatory factor-3 on phagosomes. Immunity 33, 583–596 (2010). ficking pathway of newly synthesized CD1b molecules. EMBO J. 21, 825–834 (2002). 10. A. R. Mantegazza et al., Adaptor protein-3 in dendritic cells facilitates phagosomal 20. A. R. Mantegazza et al., Increased autophagic sequestration in adaptor protein-3 toll-like receptor signaling and antigen presentation to CD4(+) T cells. Immunity 36, deficient dendritic cells limits inflammasome activity and impairs antibacterial im- 782–794 (2012). munity. PLoS Pathog. 13, e1006785 (2017).

López-Haber et al. PNAS Latest Articles | 11 of 12 21. J. M. Blander, R. Medzhitov, Regulation of phagosome maturation by signals from 48. K. C. Barnett, J. C. Kagan, Lipids that directly regulate innate immune signal trans- toll-like receptors. Science 304, 1014–1018 (2004). duction. Innate Immun. 26,4–14 (2019). 22. L. A. O’Neill, A. G. Bowie, The family of five: TIR-domain-containing adaptors in toll- 49. O. V. Vieira et al., Distinct roles of class I and class III phosphatidylinositol 3-kinases in like receptor signalling. Nat. Rev. Immunol. 7, 353–364 (2007). phagosome formation and maturation. J. Cell Biol. 155,19–25 (2001). 23. N. J. Gay, M. Gangloff, L. A. O’Neill, What the Myddosome structure tells us about the 50. J. Helft et al., GM-CSF mouse bone marrow cultures comprise a heterogeneous initiation of innate immunity. Trends Immunol. 32, 104–109 (2011). population of CD11c(+)MHCII(+) macrophages and dendritic cells. Immunity 42, 24. J. C. Kagan, R. Medzhitov, Phosphoinositide-mediated adaptor recruitment controls 1197–1211 (2015). – Toll-like receptor signaling. Cell 125, 943 955 (2006). 51. M. B. Lutz, K. Inaba, G. Schuler, N. Romani, Still alive and kicking: In-vitro-generated 25. K. S. Bonham et al., A promiscuous lipid-binding protein diversifies the subcellular GM-CSF dendritic cells! Immunity 44,1–2 (2016). – sites of toll-like receptor signal transduction. Cell 156, 705 716 (2014). 52. A. Savina et al., NOX2 controls phagosomal pH to regulate antigen processing during 26. G. R. Hammond et al., PI4P and PI(4,5)P2 are essential but independent lipid deter- crosspresentation by dendritic cells. Cell 126, 205–218 (2006). – minants of membrane identity. Science 337, 727 730 (2012). 53. E. S. Trombetta, M. Ebersold, W. Garrett, M. Pypaert, I. Mellman, Activation of lyso- 27. J. Tan, J. A. Brill, Cinderella story: PI4P goes from precursor to key signaling molecule. somal function during dendritic cell maturation. Science 299, 1400–1403 (2003). Crit. Rev. Biochem. Mol. Biol. 49,33–58 (2014). 54. A. Savina, S. Amigorena, Phagocytosis and antigen presentation in dendritic cells. 28. M. Krauss, V. Haucke, Phosphoinositide-metabolizing enzymes at the interface be- Immunol. Rev. 219, 143–156 (2007). tween membrane traffic and cell signalling. EMBO Rep. 8, 241–246 (2007). 55. A. R. Mantegazza, J. G. Magalhaes, S. Amigorena, M. S. Marks, Presentation of 29. P. A. Janmey, R. Bucki, R. Radhakrishnan, Regulation of actin assembly by PI(4,5)P2 phagocytosed antigens by MHC class I and II. Traffic 14, 135–152 (2013). and other inositol phospholipids: An update on possible mechanisms. Biochem. Bio- 56. A. Balla, G. Tuymetova, A. Tsiomenko, P. Várnai, T. Balla, A plasma membrane pool of phys. Res. Commun. 506, 307–314 (2018). phosphatidylinositol 4-phosphate is generated by phosphatidylinositol 4-kinase type- 30. S. Minogue, The many roles of type II phosphatidylinositol 4-kinases in membrane trafficking: New tricks for old dogs. BioEssays 40, (2018). III alpha: Studies with the PH domains of the oxysterol binding protein and FAPP1. – 31. A. Balla, G. Tuymetova, M. Barshishat, M. Geiszt, T. Balla, Characterization of type II Mol. Biol. Cell 16, 1282 1295 (2005). phosphatidylinositol 4-kinase isoforms reveals association of the enzymes with en- 57. M. Yamamoto et al., Essential role for TIRAP in activation of the signalling cascade – dosomal vesicular compartments. J. Biol. Chem. 277, 20041–20050 (2002). shared by TLR2 and TLR4. Nature 420, 324 329 (2002). 32. A. Balla, T. Balla, Phosphatidylinositol 4-kinases: Old enzymes with emerging func- 58. T. Horng, G. M. Barton, R. A. Flavell, R. Medzhitov, The adaptor molecule TIRAP tions. Trends Cell Biol. 16, 351–361 (2006). provides signalling specificity for Toll-like receptors. Nature 420, 329–333 (2002). 33. G. Salazar et al., Phosphatidylinositol-4-kinase type II a is a component of AP-3-de- 59. A. Mrakovic, J. G. Kay, W. Furuya, J. H. Brumell, R. J. Botelho, Rab7 and Arl8 GTPases rived vesicles. Mol. Biol. Cell 16, 3692–3704 (2005). are necessary for lysosome tubulation in macrophages. Traffic 13, 1667–1679 (2012). 34. B. Craige, G. Salazar, V. Faundez, Phosphatidylinositol-4-kinase type II alpha contains 60. T. Baba, D. J. Toth, N. Sengupta, Y. J. Kim, T. Balla, Phosphatidylinositol 4,5-bi- an AP-3-sorting motif and a kinase domain that are both required for endosome sphosphate controls Rab7 and PLEKHM1 membrane cycling during autophagosome- traffic. Mol. Biol. Cell 19, 1415–1426 (2008). lysosome fusion. EMBO J. 38, e100312 (2019). 35. J. Sobocinska et al., Lipopolysaccharide upregulates palmitoylated enzymes of the 61. X. Michelet et al., MHC class II presentation is controlled by the lysosomal small phosphatidylinositol cycle: An insight from proteomic studies. Mol. Cell. Proteomics GTPase, Arl8b. J. Immunol. 194, 2079–2088 (2015). 17, 233–254 (2018). 62. S. Suetsugu, S. Kurisu, T. Takenawa, Dynamic shaping of cellular membranes by 36. A. Jeschke et al., Phosphatidylinositol 4-phosphate and phosphatidylinositol phospholipids and membrane-deforming proteins. Physiol. Rev. 94, 1219–1248 (2014). 3-phosphate regulate phagolysosome biogenesis. Proc. Natl. Acad. Sci. U.S.A. 112, 63. American Veterinary Medical Association, AVMA Guidelines for the Euthanasia of – 4636 4641 (2015). Animals (American Veterinary Medical Association, Schaumburg, IL, 2020). 37. R. Levin et al., Multiphasic dynamics of phosphatidylinositol 4-phosphate during 64. Office of Laboratory Animal Welfare, Public Health Service Policy on Humane Care – phagocytosis. Mol. Biol. Cell 28, 128 140 (2017). and Use of Laboratory Animals (NIH Publication 15-8013, Office of Laboratory Animal 38. R. Levin-Konigsberg et al., Phagolysosome resolution requires contacts with the en- Welfare, Bethesda, MD, 2015). doplasmic reticulum and phosphatidylinositol-4-phosphate signalling. Nat. Cell Biol. 65. National Research Council, Guide for the Care and Use of Laboratory Animals (Na- 21, 1234–1247 (2019). tional Academies Press, Washington, DC, 2011) ed. 8. 39. J. S. Bonifacino, L. M. Traub, Signals for sorting of transmembrane proteins to en- 66. C. Winzler et al., Maturation stages of mouse dendritic cells in growth factor- dosomes and lysosomes. Annu. Rev. Biochem. 72, 395–447 (2003). dependent long-term cultures. J. Exp. Med. 185, 317–328 (1997). 40. D. J. Owen, J. P. Luzio, Structural insights into clathrin-mediated endocytosis. Curr. 67. A. R. Mantegazza, M. S. Marks, Visualizing toll-like receptor-dependent phagosomal Opin. Cell Biol. 12, 467–474 (2000). dynamics in murine dendritic cells using live cell microscopy. Methods Mol. Biol. 1270, 41. Y. Wang et al., TLR4/MD-2 activation by a synthetic agonist with no similarity to LPS. 191–203 (2015). Proc. Natl. Acad. Sci. U.S.A. 113, E884–E893 (2016). 68. A. Savina, P. Vargas, P. Guermonprez, A. M. Lennon, S. Amigorena, Measuring pH, 42. H. Ohno et al., Interaction of tyrosine-based sorting signals with clathrin-associated proteins. Science 269, 1872–1875 (1995). ROS production, maturation, and degradation in dendritic cell phagosomes using – 43. H. Ohno et al., The medium subunits of adaptor complexes recognize distinct but cytofluorometry-based assays. Methods Mol. Biol. 595, 383 402 (2010). overlapping sets of tyrosine-based sorting signals. J. Biol. Chem. 273, 25915–25921 69. A. R. Mantegazza et al., NADPH oxidase controls phagosomal pH and antigen cross- – (1998). presentation in human dendritic cells. Blood 112, 4712 4722 (2008). 44. G. R. Hammond, M. P. Machner, T. Balla, A novel probe for phosphatidylinositol 70. P. Guermonprez et al., ER-phagosome fusion defines an MHC class I cross- 4-phosphate reveals multiple pools beyond the Golgi. J. Cell Biol. 205, 113–126 (2014). presentation compartment in dendritic cells. Nature 425, 397–402 (2003). 45. J. M. Blander, R. Medzhitov, Toll-dependent selection of microbial antigens for pre- 71. I. W. McLean, P. K. Nakane, Periodate-lysine-paraformaldehyde fixative. A new fix- sentation by dendritic cells. Nature 440, 808–812 (2006). ation for immunoelectron microscopy. J. Histochem. Cytochem. 22, 1077–1083 (1974). 46. I. Cebrian et al., Sec22b regulates phagosomal maturation and antigen cross- 72. F. W. Fitch, T. F. Gajewski, J. Hu-Li, Production of TH1 and TH2 cell lines and clones. presentation by dendritic cells. Cell 147, 1355–1368 (2011). Curr. Protoc. Immunol. Chapter 3, Unit 3 13 (2006). 47. Rudensky AYu, S. Rath, P. Preston-Hurlburt, D. B. Murphy, C. A. Janeway Jr, On the 73. H. K. Lee et al., In vivo requirement for Atg5 in antigen presentation by dendritic complexity of self. Nature 353, 660–662 (1991). cells. Immunity 32, 227–239 (2010).

12 of 12 | www.pnas.org/cgi/doi/10.1073/pnas.2001948117 López-Haber et al.