SNX17 regulates antigen internalization and phagosomal maturation by dendritic cells

Cross-presentation is the process whereby antigenic peptides derived from exogenous antigens are associated to MHC class I molecules triggering the activation of CD8+ T lymphocytes. The endocytic route of dendritic cells (DCs) is strongly specialized to achieve antigen cross-presentation eciently, which is crucial to initiate cytotoxic immune responses against many pathogens (i.e. Toxoplasma gondii) and tumors. Nevertheless, the endosomal molecular effectors involved in this process are not completely understood. In particular, the role of sorting nexin (SNX) proteins in cross-presentation has never been addressed. In this work, we identify the endosomal protein SNX17 as a key regulator of antigen internalization and cross-presentation by DCs. Our results demonstrate that SNX17 expression in DCs is essential to guarantee a normal cross-presentation of soluble, particulate and T. gondii-associated antigens. The silencing of SNX17 expression in DCs signicantly affected the uptake of exogenous antigens by uid-phase endocytosis and phagocytosis, but not by receptor-mediated endocytosis. Moreover, the knock-down of SNX17 impaired T. gondii invasion, CD11b integrin recycling and hampered the organization of the actin cytoskeleton. Finally, we show that SNX17 controls the proper maturation of DC phagosomes. Our ndings provide compelling evidence that SNX17 plays a central role in the modulation of DC endocytic network, which is crucial for competent antigen internalization and crosspresentation.


SNX17 regulates antigen internalization and phagosomal maturation by dendritic cells
Introduction Dendritic cells (DCs) are the most adapted antigen-presenting cells (APCs) of the immune system for e cient cross-presentation. This implies the internalization of exogenous antigens, their intracellular processing and loading onto MHC-I molecules to present antigenic peptides to CD8+ T lymphocytes. 1 To do this, DCs have developed important specializations of their endocytic network that include a tight regulation of endo/phagosomal maturation, 2 a special connection between endosomes and phagosomes with the endoplasmic reticulum (ER) 3,4 and a recycling pathway for MHC-I molecules. 5 Furthermore, the process of endosomal tubulation is starting to emerge as a key modulator of antigen cross-presentation by DCs. 6,7 In this sense, sorting nexins (SNXs) are master regulators of intracellular transport by leading elongation and tubulation of endosomal membranes. 8 This family of effectors is constituted of both cytoplasmic and membrane-associated proteins that control important features of the endocytic pathway, including endocytosis, endosomal sorting, and signaling. 9 SNXs are characterized by the presence of a particular type of phox-homology (PX) domain, the SNX-PX domain. 10 As PX domains function mostly by binding phosphatidylinositol-3-monophosphate (PtdIns3P), SNXs are associated with PtdIns3P-enriched elements of the early endocytic network. 10,11 Although SNX proteins ensure critical functions that are relevant for antigen cross-presentation, their potential role in this immunological context has never been addressed.
We focused our study on the role of SNX17 within the endocytic network of DCs, with particular interest in antigen cross-presentation. In addition to the PX domain, SNX17 has an atypical FERM domain able to interact with many cargoes that contain an NPxY/NxxY motif. 12 Therefore, SNX17 is involved in the recycling of integrins in different biological models, such as cell adhesion and migration, 13,14 in association with the retriever complex 15 or contributing to the establishment of the immune synapse by T cells. 16 SNX17 was also reported to facilitate papillomavirus infection through the interaction with the L2 capsid protein and to allow viral escape from late endosomal compartments. 17 Here, we show that SNX17 controls the process of antigen cross-presentation by DCs. Indeed, we demonstrate that SNX17 regulates integrin recycling also in this cell type, but is not implicated in MHC-I recycling. Upon silencing the expression of SNX17 in DCs, uid-phase endocytosis and phagocytosis are remarkably inhibited. Strikingly, Toxoplasma gondii invasion is hampered in SNX17-de cient DCs, reporting the rst host endocytic effector that supports infection by this parasite. We also show that SNX17 is required for the proper organization of the actin cytoskeleton and for phagosomal maturation, highlighting this molecule as a critical regulator of DC endocytic network.

Results
The silencing of SNX17 in DCs impairs antigen cross-presentation As part of a larger screen aiming to nd novel regulators of T. gondii antigen presentation by MHC-I molecules, we identi ed SNX17 as a key player in this process. Upon SNX17 silencing by lentiviral shRNAs, we found that MutuDCs, a murine CD8a + DC line, 18 were unable to cross-present antigens from the two T. gondii strains we tested, as compared to control MutuDCs (Scramble shRNA). These type I T. gondii strains express model MHC-I antigens that both contain an antigenic peptide, the SIINFEKL epitope from the ovalbumin (OVA) protein, and are secreted in the vacuole. However, these antigens exhibit different intra-vacuolar distribution, leading to distinct presentation pathways. 19 In the SAG1-OVA strain, the model antigen is a soluble luminal protein, 20 while in the GRA6-OVA strain, the antigenic construct is a transmembrane protein. 21 We used two different shRNAs to silence SNX17 expression in DCs and we observed a severe reduction of T. gondii presentation on MHC-I molecules with both parasite strains ( Fig.  1a and 1b). Moreover, the cross-presentation ability of the OVA protein incorporated in the MutuDC by endocytosis was strongly impaired in SNX17 knockdown (KD) DCs (Fig. 1c), while the presentation of the short SIINFEKL peptide was affected as well, but to a minimal extent (Fig. 1d). We validated these data by qPCR analysis, and for all antigens tested, the decrease in cross-presentation capacity correlated with the KD e ciency (Fig. 1e).
We next generated two stable SNX17 KD cell lines by using the same shRNAs in JAWS-II DCs. In contrast to MutuDCs, we obtained the highest KD e cacy with the shRNA #2, as evidenced by Western blot analysis of the transduced JAWS-II DCs (Fig. 1f). We examined MHC-I presentation in these cells and, again, a signi cant inhibition of cross-presentation was evidenced in both SNX17 KD DCs, either with soluble OVA (Fig. 1g) or OVA associated to 3µm latex beads (Fig. 1h). As above, the level of crosspresentation reduction was in line with the e ciency of SNX17 silencing, although in this case, we did not observe any effect on the control SIINFEKL peptide presentation by SNX17 KD DCs, as compared to Scramble DCs (Fig. 1i). Given that the shRNA #2 exhibited a higher KD and more pronounced differences in antigen cross-presentation, we used this stable line (from now on SNX17 KD cells) to continue our study.
Integrin recycling is blocked in SNX17 KD DCs SNX17 was shown to regulate integrin recycling to the cell surface in other biological systems. [13][14][15][16] Thus, we aimed to evaluate if SNX17 also performs a similar function in DCs. First, we analyzed by ow cytometry the cell surface expression level of the αMβ2 integrin CD11b, which is highly expressed in DCs. As shown in Fig. 2a and 2b, SNX17 KD cells display lower levels of CD11b, as compared to control DCs, suggesting a defect of recycling to the plasma membrane. In order to test this hypothesis, we assessed the endocytic recycling of CD11b by ow cytometry, as described and performed in previous studies. 16,22 Brie y, the decrease in the mean uorescent intensity (MFI) of PE-conjugated CD11b antibody was measured to determine the recycling to the cell surface of this integrin after binding, internalization and acid stripping. Indeed, we observed a drastic defect of CD11b recycling by SNX17 KD DCs, as compared to Scramble DCs ( Fig. 2c and 2d).
Based on the importance of SNX17 for CD11b integrin recycling and given that MHC-I recycling is a key aspect to guarantee a proper antigen cross-presentation process, 5,22-24 we hypothesized that SNX17 may also control the recycling of MHC-I molecules. Nevertheless, when we analyzed both the expression level of MHC-I at the cell surface ( Fig. 2e and 2f) and the recycling of MHC-I ( Fig. 2g and 2h), we did not nd signi cant differences between SNX17 KD and Scramble DCs. So far, these data indicate that SNX17 is required for optimal antigen cross-presentation and CD11b integrin recycling by DCs, but is dispensable for the recycling of MHC-I molecules.

Antigen internalization is inhibited in SNX17 KD DCs
In order to explain the defect in the cross-presentation ability of SNX17 KD DCs, we decided to investigate the rst step of this process, which is antigen uptake. To this aim, we rst analyzed uid-phase endocytosis of a uorescent version of the OVA model antigen by ow cytometry. As shown in Fig. 3a and 3b, a signi cant impairment of OVA uptake at different concentrations compatible with macropinocytosis was observed in SNX17 KD DCs, as compared to control cells. Since this defect could be also explained by an enhancement of OVA degradation inside SNX17 KD cells, we next evaluated the uid-phase endocytosis of a non-degradable uorescent antigen (dextran). Again in this experimental setup, the macropinocytic activity of SNX17 KD DCs was signi cantly altered ( Fig. 3c and 3d).
Because there is also an impairment of particulate antigen cross-presentation in SNX17 KD DCs, we then examined the phagocytic activity of these cells. By using uorescent 3 µm latex beads coated with OVA and ow cytometry analysis, we were able to distinguish between cells that completely incorporate these particles from those that have beads attached to the surface but not fully internalized. We used this approach to measure the percentage of effective phagocytosis and we determined that SNX17 KD DCs exhibit a marked de ciency in this process at different time points of uptake (1, 3 and 5 hours), as compared to control cells ( Fig. 3e and 3f).
Aiming to understand the relevance of SNX17 in the presentation of T. gondii antigens, we decided to evaluate whether parasite invasion was affected in the KD cells. We used a uorescent strain of T. gondii (TgRH YFP SAG1-OVA) and an antibody against SAG1 (a parasite surface protein). In this way, we could determine the percentage of infected DCs (YFP+/SAG1-cells) and the percentage of DCs simply bound to extracellular parasites (YFP+/SAG1+ cells). We performed this assay at 3 hours post-infection, a time point that precedes the rst parasite replication cycle inside the DCs. Interestingly, while the proportion of DC surface-bound parasites was similar in the two conditions ( Fig. S1a), SNX17 KD DCs were much less infected by T. gondii ( Fig. 3g and 3h).
Since all the internalization pathways relevant in our cross-presentation assays (macropinocytosis, phagocytosis and T. gondii invasion) were found to be defective in the SNX17 KD cells, we wondered whether receptor-mediated endocytosis would also be affected by SNX17 silencing. To do this, we measured the internalization of two antigens by different receptors; MHC-I for the uptake of anti-H2K b coupled to FITC (Fig. S1b), and Fc receptor for the uptake of the immune complex uorescent OVA/anti-OVA (Fig. S1c). After the steps of antigen binding to receptor at 4°C, washing, internalization at 37°C and acidic stripping of the cell surface, we determined by ow cytometry that the effective uptake of these antigens was not affected by the silencing of SNX17, indicating that receptor-mediated endocytosis is independent of this endosomal molecule in DCs. Overall, this set of experiments points to SNX17 as a novel and critical regulator of macropinocytosis, phagocytosis and T. gondii infection in DCs.

SNX17 modulates the organization of the actin cytoskeleton
Since the antigen internalization pathways found to be defective in SNX17 KD cells have in common to be dependent on actin remodeling, we asked whether the actin cytoskeleton may be altered in the KD cells. To test this, we used phalloidin and an antibody against actin to label polymerized lamentous actin (F-actin) and total actin, respectively, and we analyzed the cells by confocal microscopy. In control DCs, we observed that most of the F-actin is around the cell surface with a clear distribution within long dendrites (Fig. 4a, upper panels). By contrast, these long dendrites were mostly absent in SNX17 KD DCs, although F-actin still remained around the cell surface (Fig. 4b, upper panels). We better documented this difference by quantifying the number of dendrites of 3 µm or more in Scramble and SNX17 KD cells (Fig.  4c). As expected, the continuous phalloidin staining was lost upon DC treatment with latrunculin B, which inhibits actin polymerization ( Fig. 4a and 4b, lower panels), con rming that the dendrites are F-actinbased.
In order to determine whether the phenotype observed in SNX17 KD cells was due to a defect in actin polymerization or an inaccurate organization of actin remodeling, we quanti ed F-actin and total actin by ow cytometry. As shown in Fig. 4d and 4e, no signi cant difference was found between KD and control cells, suggesting that SNX17 drives lopodial elongation relevant for macropinocytosis and phagocytosis rather than actin polymerization in DCs. We observed that SNX17 KD DCs were slightly bigger in size than Scramble DCs, exhibiting higher uorescent background and speci c staining, but the phalloidin/actin ratio was similar between both DC types.

Phagosomal maturation is delayed in SNX17 KD DCs
In addition to perturbing the antigen uptake, a dysregulation of the actin cytoskeleton organization may impact on other important features of DCs during the intracellular transport of exogenous antigens.
Hence, we decided to address the role of SNX17 on phagosomal maturation by ow cytometry. After coating 3 µm latex beads with OVA and inducing phagocytosis at different time periods (15, 60 and 180 min), we labeled isolated phagosomes with anti-OVA and anti-Lamp1 to measure intra-phagosomal antigen degradation and phago-lysosomal fusion, respectively. While phagosomes recovered from Scramble cells displayed OVA degradation over time, antigen degradation was less e cient in the SNX17 KD-derived phagosomes at 3 hours post-internalization ( Fig. 5a and 5b, left axis). This suggests that phago-lysosomal fusion may be impaired in SNX17 KD DCs. Indeed, phagosomal acquisition of the lysosomal marker Lamp1 was inhibited in SNX17 KD DCs already from 15 min post-internalization, as compared to Scramble phagosomes ( Fig. 5c and 5b, right axis).
Altogether, our results demonstrate that SNX17 is in charge of key functions within the endocytic pathway of DCs, mainly by regulating antigen internalization, actin remodeling and phagosomal maturation, all critical events needed to ensure an e cient antigen cross-presentation process.

Discussion
Different molecular regulators of the endocytic pathway that contribute to the process of antigen crosspresentation have been characterized in the past fteen years, 25 but our study is the rst to address the role of SNXs in this immunological context. From our original T. gondii presentation screening, we observed that this family of proteins play a key role to achieve e cient MHC-I presentation of parasitederived antigens. Among other interesting candidates that modulate antigen cross-presentation, we identi ed SNX17 as a positive regulator of this process. Upon DC silencing (we used two different DC lines) of SNX17 expression, we observed a signi cant reduction of cross-presentation for all exogenous antigens tested derived either from soluble OVA, latex beads or T. gondii. SNX17 has been described to control important aspects of endocytic recycling, [13][14][15][16] which is a determinant feature that guarantees proper intracellular transport of MHC-I molecules required for crosspresentation. [22][23][24] Based on this, our rst hypothesis was that SNX17 was implicated in MHC-I recycling. Nevertheless, we did not observe a signi cant difference between SNX17 KD and control DCs, although MHC-I recycling capacity of KD cells was slightly affected. By contrast, the recycling of the αMβ2 integrin CD11b in SNX17 KD DCs was drastically impaired, as reported before for several integrins in other cell types. [13][14][15][16] Instead of a de cient MHC-I recycling, we detected a strong inhibition of antigen internalization in SNX17 KD DCs. With the exception of receptor-mediated endocytosis, all the internalization pathways tested turned out to be highly dependent on SNX17 in DCs. This discovery was striking, not only because we observed that SNX17 controls the uptake of exogenous antigens by uidphase endocytosis and phagocytosis, but also because T. gondii infection was surprisingly abrogated in SNX17 KD cells.
T. gondii is an obligate intracellular protozoa that establishes life-long chronic infections in up to a third of the world's population and is the causative agent of toxoplasmosis, an opportunistic disease of potential deadly consequences for immunocompromised people and congenitally infected newborns. 26 This parasite replicates in a specialized parasitophorous vacuole (PV), surrounded by a limiting membrane that restricts access to the nutrient-rich cytoplasm 27 and interacts with different organelles of the host cell, including ER, mitochondria and endosomal vesicles. 28 As it grows, the vacuole also has an enormous impact on the microtubule network, which reorganizes around the PV. 29 Despite the description of these multiple interactions at the PV border, little is known about the host molecular factors controlling them or required for T. gondii entry. Although parasite entry is mainly achieved by a parasite-driven mechanism that involves the parasite acto-myosin system, 30 a normal host actin organization is also required for e cient T. gondii penetration. 31,32 A previous study in human cells, identi ed six proteins involved in the modulation of host actin dynamics that facilitate T. gondii entry. 33 More recently, a genome-wide CRISPR-based screen revealed 1183 human genes that are signi cantly involved in T.
gondii infection, 53 of them were linked to the regulation of host actin cytoskeleton. 34 But our study is original because it describes a host cell endosomal protein capable to strongly control the process of T. gondii entry, which to our knowledge has not been reported before. This is particularly interesting since at present, there is not effective vaccine or treatment to prevent T. gondii infection, thus SNX17 could represent a unique target in order to potentially block the parasite entry. As a rst step, it will be important to evaluate the role of SNX17 during T. gondii infection of other cell types, including human DCs.
At least two members of the SNX family, SNX9 [35][36][37] and SNX33, 38 have been reported to interact with the Wiskott-Aldrich syndrome protein (WASp) inducing changes in actin dynamics and regulating endocytosis. So far, SNX17 has not been implicated to bind with WASp, but the SNX17-retriever complex associates with the WASH complex, 15 which may impact on vesicular transport through actin remodeling. 39 Indeed, the drastic disorganization of F-actin observed in our SNX17 KD DCs provides new evidence to support this hypothesis.
A clear demonstration that intracellular actin-mediated transport strongly depends on SNX17 is the signi cant delay of phagosome maturation in SNX17 KD cells. The ine cient acquisition of lysosomal vesicles by SNX17 KD phagosomes also inhibits intra-phagosomal antigen degradation. In this sense, it was reported that SNX10 promotes phagosomal maturation in macrophages by recruiting the complex Mon1-Ccz1 to endosomes and phagosomes. As consequence, SNX10 de cient macrophages exhibit less acidic phagosomes and a reduced ability to kill the bacteria Listeria monocytogenes. 40 Also SNX3 is e ciently recruited to DC phagosomes and negatively regulates phagocytic uptake. 41 Moreover, it was shown in macrophages that SNX3 is transported with Rab5a positive vesicles towards Borrelia burgdorferi-containing phagosomes to promote maturation of these organelles and the spirochete compaction. 42 Also in macrophages, SNX5 plays an important role during macropinosome biogenesis controlling the uptake and processing of soluble exogenous antigens. 43 We still do not know whether the difference of phagosomal maturation observed in SNX17 KD DCs enhances or partially restores the phenotype of cross-presentation found in these cells.
Taking into account that SNX17 actively participates in antigen and T. gondii internalization by DCs to trigger CD8+ T cell-mediated responses, we consider that this molecule could represent a key target for future immunotherapy approaches based on DC stimulation and for new therapeutic strategies against T. gondii infection.

Cells
MutuDCs were maintained with IMDM medium supplemented with 8% FBS and were provided by H. Acha-Orbea through an MTA with N. Blanchard. JAWS-II DCs were maintained in culture by using GM-CSF-containing medium (IMDM) with 10% FBS. The cytokine GM-CSF was produced with the cell line J558, also maintained in IMDM medium. Both, JAWS-II and J558 cells, were kindly provided by S. Amigorena (INSERM U932, Institute Curie, France). MutuDCs and JAWS-II DCs were expanded and diluted every 2-4 days, depending on their growth rates. T. gondii parasites were grown through continuous infection of Human Foreskin Fibroblasts (HFF), which were maintained in DMEM complete medium (10% FBS). B3Z T cells were cultured with RPMI complete medium (10% FBS).

Material and buffers
The following reagents were used in this study: Ovalbumin, lyophilized powder (Worthington Biochemical Corporation). Bovine Serum Albumin (BSA) was purchased from Santa Cruz.

Lentiviral shRNA knockdown of SNX17
High titer lentiviral particles (~ 10 7 CFU/ml) were purchased from Sigma-Aldrich. SNX17 #1 shRNA (TRCN0000190245, target sequence CCGGCCTCTACCTGAGAGGAGATTTC TCGAGAAATCTCCTCTCAGGTAGAGGTTTTTTG), SNX17 #2 shRNA (TRCN0000190340, target sequence CCGGCCAGATGACTTGATCGGATATCTCGAG ATATCCGATCAAGTCATCTGGTTTTTTG) and the control hairpin (a scramble sequence against GFP) were used to infect Mutu or JAWS-II DCs. Brie y, 5x10 4 MutuDCs or 10 5 JAWS-II DCs were plated on a 96-well plate with 200 µl of their corresponding complete medium. After 48 h, the medium was carefully removed and 10 µl of virus was added. Cell pellet was mixed by resuspending 3-5 times with a pipette and 40 µl of 8 µg/ml polyB in complete medium was added. Cells were centrifuged during 90 min at 800 g and 37°C, all media were removed and replaced by 200 µl of fresh complete medium. Cells were incubated for 48 h, selected with puromycin at 5 µg/ml or 20 µg/ml for MutuDCs and JAWS-II DCs, respectively. Mutu DCs were transiently infected and collected for experiments 72 h after puromycin selection. In the case of JAWS-II cells, they were expanded until the generation of stable KD cell lines.
qPCR Cell homogenization and RNA extraction were performed by using the TRIzol Reagent protocol (Invitrogen). After determining RNA concentration, cDNA was produced with the iScript cDNA Synthesis Kit (Bio-Rad) and PCR ampli cation was performed by using 5x HOT FIREPol EvaGreen qPCR Mix (Solis BioDyne) during 40 cycles. GAPDH/SNX17 normalization was done by midpoint slope determination.

Immunoblotting
Total cell lysates from JAWS-II DCs (10 5 cells, around 50 µg of proteins) were subjected to SDS-PAGE on 10% gel. After transferring, the membranes were blocked in 10% Milk/PBS during 1 hour and incubated with anti-SNX17, anti-β-actin and then with peroxidase-conjugated antibodies. Bound antibodies were revealed using the kit Chemiluminescent Peroxidase Substrate-3 (Sigma-Aldrich), according to the manufacturers' instructions. The intensity of the bands was quanti ed by densitometry using Quantity One 4.6.6 software (Bio-Rad) and was expressed as arbitrary units.

Antigen cross-presentation assays
MutuDCs were infected with TgRH GRA6-SL8 or TgRH YFP SAG1-OVA parasites during 8 h at 37°C. Previously, T. gondii parasites were released from the HFF cells by forcing them through a 23-G needle and added to Scramble and SNX17 KD1 and KD2 MutuDCs at the indicated multiplicity of infection (MOI). These cells were also incubated during 5 h with 1.5, 3 or 6 mg/ml of soluble OVA, or during 1 h with different concentrations of the SIINFEKL peptide at 37°C. In the case of JAWS-II DCs, Scramble and SNX17 KD cells were incubated during 5 h with 3 or 6 mg/ml of soluble OVA, or with 3 µm latex beads coated with different ratios of OVA and BSA (OVA 10 mg/ml alone; OVA 3 mg/ml-BSA 7 mg/ml; and BSA 10 mg/ml alone), or during 1 h with the indicated doses of SIINFEKL at 37°C. Then, MutuDCs and JAWS-II DCs were washed with 0.5% of PBS/BSA, xed with 0.008% glutaraldehyde during 3 min at 4°C, and quenched with 0.2 M glycine. After one nal wash with PBS, B3Z hybrid T cells were added during 16 h at 37°C. T cell activation was measured detecting β-galactosidase activity by optical density (absorbance at 595-655 nm) using CPRG as substrate for the reaction. Antigen uptake and T. gondii infection assays Scramble and SNX17 KD JAWS-II DCs were incubated during 1 h at 37°C either with OVA or dextran conjugated to Alexa 647 in complete medium at the indicated concentrations. The incubation at 4°C with the highest concentration of these soluble antigens was performed as negative control for endocytosis. Then, cells were washed with 0.5% PBS/BSA and the MFI for every condition was calculated by ow cytometry analysis.
To determine the phagocytic capacity of Scramble and SNX17 KD DCs, JAWS-II cells were incubated with OVA-coated 3 µm blue latex beads in complete medium during 1, 3 and 5 h at 37°C or during 5 h at 4°C. After these internalization time periods, cells were extensively washed with 0.5% PBS/BSA, labeled rst with anti-OVA and then with a secondary anti-rabbit conjugated to Alexa 488. Finally, DCs were analyzed by ow cytometry and the percentage of cells with uorescent beads fully internalized was determined, as shown in Fig. 3e and 3f.
For T. gondii infection experiments, Scramble and SNX17 KD JAWS-II DCs were incubated with TgRH YFP SAG1-OVA parasites at the indicated MOIs during 8 h at 37°C. Then, DCs were washed with 0.5% PBS/BSA, xed with 1% PFA, labeled with anti-SAG1 and an anti-mouse conjugated to Alexa 647 and analyzed by ow cytometry. In this way, the percentages of YFP+/SAG1-(genuinely infected DCs) and YFP+/SAG1+ (parasite bound to the plasma membrane) cells were determined as shown in Fig. 3g, 3h and Sup. 1a.
Receptor-mediated endocytosis was measured by incubating Scramble and SNX17 KD JAWS-II DCs with 5 µg/ml of anti-H-2K b -FITC or anti-OVA/OVA-Alexa 488 conjugates for 30 min at 4°C. Then, cells were extensively washed with 0.5% PBS/BSA and incubated in complete medium for 30 min at 37°C. After this, DCs were treated with acid buffer stripping solution (0.5 M NaCl and 0.5% acetic acid, pH 3) for 10 min on ice, as described before. Finally, cells were washed twice with cold PBS and twice with complete medium, and the MFI of FITC and Alexa 488 were evaluated by ow cytometry.

Immuno uorescence
Scramble and SNX17 KD JAWS-II DCs were placed on poly-L-lysine-coated glass coverslips during 30 min at room temperature (RT). After one wash with PBS, cells were incubated in complete IMDM medium (supplemented with GM-CSF) alone or with 10 µM of latrunculin B for 1 h at 37°C in an atmosphere of 5% CO 2 . After extensive washing with PBS, JAWS-II DCs were rst xed with 2% PFA during 15 min at 37°C and then quenched with 0.2 M glycine. After this, cells were permeabilized with PBS/0.05% saponin/0.2% BSA for 20 min at RT, washed and incubated with mouse anti-β-actin overnight at 4°C. The next day, cells were washed with permeabilization buffer and incubated with a secondary anti-mouse coupled to Alexa 568 for 45 min at 4°C. Cells were washed again three times with permeabilization buffer and twice with PBS, and 60 nM of phalloidin coupled to Alexa 488 diluted in PBS was added for 1 h at 37°C. After extensive washing with PBS, coverslips were mounted with ProLong Diamond Anti-Fade (with DAPI). Image acquisition was performed on an Olympus FV-1000 confocal microscope with a 63x/1.4 NA oil immersion objective. One z-stack plane is shown from the acquired images and they were processed with the ImageJ software (Wayne Rasband, National Institutes of Health). Image deconvolution was performed with the Parallel Spectral Deconvolution plugin (Piotr Wendykier) using a theoretical PSF generated by the Diffraction PSF 3D plugin (Robert Dougherty).

Phagosomal isolation
Scramble and SNX17 KD JAWS-II DCs were pulsed for 20 min at 18°C and 15 min at 37°C with 3 µm latex beads. Then, cells were washed once with cold 2% PBS/BSA and twice with 1 ml of FBS, and chased for the indicated time points with complete GM-CSF-containing medium at 37°C. After each chase period, DCs were washed once with cold 2% PBS/BSA to stop phagocytosis and then they were disrupted with a syringe (22G needle) in homogenization buffer (PBS 8% sucrose, 3 mM imidazole, 1 mM DTT, and 1X protease inhibitor cocktail) as we have described before in other studies. 4,22 Samples were centrifuged at 150 g during 5 min at 4°C and supernatants were placed on a 96-well plate (round bottom). Isolated beads, representing the phagosomal fraction, were washed with cold 2% PBS/BSA by centrifugation at 1,000 g during 5 min at 4°C and xed with 1% PBS/PFA during 10 min on ice. After two washes with 0.2 M glycine and one wash with 2% PBS/BSA, samples were labeled ON at 4°C with rabbit anti-OVA and rat anti-Lamp1 antibodies. The next day, samples were incubated with the secondary antibodies (anti-rabbit coupled to Alexa 647 and anti-rat coupled to Alexa 488) during 45 min at 4°C. Finally, stained phagosomal preparations were analyzed after gating on a particular FSC/SSC region corresponding to single beads in solution.

Statistical analysis
The Kyplot 6.0 software was used to perform a one-way ANOVA with Dunnett post-test for the qPCR analysis. The two-tailed Student's paired and unpaired t-tests, and the two-way ANOVA with Bonferroni post-test were performed at the indicated Figures by using the GraphPad Prism 5 software. The ImageJ software was used for imaging processing and dendrite size analysis. Figure 1 SNX17 silencing inhibits antigen cross-presentation by DCs. a-b) The cross-presentation of the SIINFEKL peptide (a) appended at the C-terminus of the GRA6 antigen expressed by T. gondii (TgRH GRA6-OVA) or (b) incorporated within the SAG1-OVA antigen expressed by the parasite (TgRH YFP SAG1-OVA) by MutuDCs previously transduced with Scramble, SNX17 KD1 or SNX17 KD2 shRNA, was evaluated 8 h post-infection at the indicated MOI with the B3Z T cell hybridoma. Data represent mean ± SEM of triplicate values from one experiment (a larger screening). P > 0.05 (ns) and ***P < 0.001. A two-way ANOVA and the Bonferroni post-test were performed. c-d) The cross-presentation ability after incubation with (c) soluble OVA protein or (d) the SIINFEKL control peptide at the indicated concentrations by MutuDCs previously transduced with Scramble, SNX17 KD1 or SNX17 KD2 shRNA, was evaluated with the B3Z T cell hybridoma. Data represent mean ± SEM of triplicate values from one experiment (a larger screening). **P < 0.01 and ***P < 0.001. A two-way ANOVA and the Bonferroni post-test were performed. e) RT-qPCR quanti cation of SNX17 mRNA in MutuDCs transduced with lentiviruses encoding a random sequence (Scramble) and two shRNA targeting SNX17 (KD1 and KD2). Data show mean ± SEM of duplicate values from one experiment (a larger screening). **P < 0.01. A one-way ANOVA and the Dunnett post-test were performed. f) Immunoblotting and densitometry quanti cation of SNX17 in JAWS-II DCs infected with the same lentivirus used in (e). Data show mean ± SEM of three independent experiments. **P = 0.0020 and ***P < 0.0001. The two-tailed Student's paired t-test was performed. g-i) The crosspresentation ability of JAWS-II DCs transduced with Scramble, SNX17 KD1 or SNX17 KD2 shRNA, after incubation with (g) soluble OVA, (h) OVA/BSA-coated beads, and (i) the SIINFEKL control peptide at the indicated concentrations was evaluated with the B3Z T cell hybridoma. Data represent mean ± SEM of three independent experiments (n = 5). P > 0.05 (ns), *P < 0.05, **P < 0.01 and ***P < 0.001. A two-way ANOVA and the Bonferroni post-test were performed.