Complexes containing VPS34/ATG14 or ATG16L1 are recruited to LAPosomes

Regulation of autophagy requires the sequential involvement of several protein complexes starting with the ULK1/2 complex, followed by the class III PI3K complex and then by the LC3 conjugation complex¹¹. However, the molecular mechanisms involved in LAP regulation remain to be further elucidated. There is evidence from murine macrophages that LAP regulation is ULK1/2-independent but requires the involvement of the PI3K complex and the LC3 conjugation complex to successfully form single-membrane LC3 positive vesicles, termed LAPosomes¹⁵. By analogy with phagocytosis, where it has been reported that the early endosome marker Rab5 directly interacts with VPS34 to sustain phagosome maturation^24,25, we examined whether early endosome markers were associated with LAPosomes. Human monocyte-derived macrophages were stimulated with the well-known LAP-trigger zymosan for 1 h¹⁴. We first observed that zymosan-containing phagosomes were surrounded by the autophagic protein LC3B, confirming LAPosome formation and that the majority were also decorated with EEA1 (Fig. 1A). The histogram of the phagosome cross-section revealed an increase in fluorescent intensities associated with EEA1 or LC3B on either side of the zymosan particle (Fig. 1A). We also observed that a small portion of EEA1-decorated zymosan vesicles was also positive for VPS34 (Supplementary Fig. 1a). Furthermore, the recruitment of the class III PI3K complex formed by VPS34 and ATG14 to LAPosomes was examined. After 6 h of stimulation with zymosan, almost 60% of LAPosomes were also positive for VPS34 (Fig. 1B and Supplementary Fig. 1b). Inhibition of VPS34 with the SAR405 inhibitor²⁶ decreased the ability of macrophages to form LAPosomes, confirming the involvement of VPS34 in LAPosome formation (Supplementary Fig. 1c). VPS34 is the core protein of several class III PI(3) kinase complexes and is associated with VPS15, Beclin-1, Ambra1, and ATG14 for autophagosome formation. To investigate if this class III PI3K complex is recruited to LAPosomes, we investigated the presence of ATG14 upon zymosan stimulation in human macrophages and observed that a small fraction of zymosan-containing phagosomes displayed ATG14 (Fig. 1C). Moreover, the results suggest that the VPS34/Beclin-1/ATG14 complex is recruited to early LAPosomes and at the same time as the early endosome protein EEA1/Rab5. VPS34 generates PI(3)P, which serves as docking site to recruit the LC3 conjugation complex containing ATG12, ATG5, and ATG16L1 to form autophagosomes. Therefore, the recruitment of the LC3 conjugation complex during LAP was investigated. After 6 h of stimulation, we observed that two members of the LC3 conjugation complex: ATG16L1 and ATG12 were preferentially recruited to zymosan-containing phagosomes compared to the negative control, inert beads (Fig. 1D and Supplementary Fig. 2a, b). ATG16L1 was found four times more frequently associated with zymosan than with inert beads-containing phagosomes (Fig. 1D). To assess whether these zymosan-containing phagosomes are LAPosomes, we investigate the colocalization of ATG16L1 or ATG12 with LC3B. After 6 h of LAP stimulation, we observed colocalization of LC3B-decorated zymosan-containing phagosomes with ATG16L1 (Fig. 1E). Quantitative analysis showed that 53.2% of LAPosomes were positive for ATG16L1 (Fig. 1E and Supplementary Fig. 2d). Similar results were observed for ATG12 (Supplementary Fig. 2a–c). In line with our previous findings, these results showed that the ATG12–ATG16L1–ATG5 complex is recruited and required for the LAP pathway²⁰. Altogether, these results suggest that the LC3 conjugation complex colocalizes with the VPS34/ATG14 complex upon LAP stimulation.

A Human macrophages stimulated with zymosan for 1 h and costained for EEA1 (green) and LC3B (red). Bar graph shows the percentage of LAPosomes colocalizing with EEA1; n = 5; each symbol represents an individual experiment. B mRFP-LC3B transduced macrophages stimulated with zymosan for 6 h and stained for VPS34 (green). Bar graph shows the percentage of LAPosomes in mRFP/GST-LC3B cells colocalizing with VPS34. Bar represents mean ± SD of data pooled from three independent experiments and each symbol represents an individual experiment. C Macrophages stimulated with zymosan-Texas Red for 1 h and stained for ATG14 (green). Graph shows the percentage of zymosan-containing phagosomes displaying ATG14, each symbol represents an individual experiment. Bar represents the mean ± SD of three independent experiments quantified by two independent investigators. D Macrophages stimulated with zymosan-Texas Red (red) or with inert beads (blue) for 6 h and stained for ATG16L1 (green). Quantification of the percentage of zymosan or beads positive for ATG16L1 is shown in the bar graph. Bars represent the mean ± SD; data pooled from n = 4 (zymosan) and n = 3 (beads) independent experiments, quantified by two independent investigators and each symbol represents an individual experiment; unpaired Student’s t test, two-tailed. E GFP-LC3B transduced macrophages stimulated with zymosan for 6 h and stained for ATG16L1 (red). Bar graph shows the quantification of the percentage of LAPosomes in GFP/GST-LC3B cells colocalizing with ATG16L1; n = 5 and each symbol represents an individual experiment. A–E Fluorescence intensities were quantified along the white segment as shown in the merged panel and plotted as a histogram. F Macrophages stimulated with zymosan for 6 h and costained for LAMP1 (green) and ATG16L1 (red). Left-bottom inserts show a zymosan-containing phagosome decorated with LAMP1 but not ATG16L1. Right-bottom inserts show a zymosan-containing phagosome negative for LAMP1 but positive for ATG16L1. Bar graph shows the percentage of zymosan-containing phagosomes displaying single staining for LAMP1 or ATG16L1, or both. The bar graph represents the mean ± SD of three independent experiments, quantified by two investigators blinded to the experimental conditions. All confocal images are representative images of three different independent experiments. Scale bar is 5 µm, and inserts are zoomed 5× from white-framed regions of the immune fluorescence micrographs. Source data are provided as a Source Data file.

LAPosomes display NOX2 and an elevated oxidation required for LAP

In this study, we confirmed that LAPosomes seem to have a delay in maturation in human macrophages. Indeed, ATG16L1-decorated zymosan-containing phagosomes were rarely positive for the lysosomal marker LAMP1. After 6 h of zymosan stimulation, only 0.5% were decorated with both ATG16L1 and LAMP1, while 12.3% and 8.9% were decorated with only ATG16L1 or LAMP1, respectively (Fig. 1F). Furthermore, the percentage of LAMP1-positive zymosan-phagosomes decreased over time to undetectable levels at 24 h after stimulation, while ATG16L1-positive and zymosan-containing phagosomes were still observed at this time point (Supplementary Fig. 2e). Therefore, we addressed the mechanism of LAPosome stabilization. Our group had previously demonstrated that macrophages produced high levels of ROS after LAP stimulation and that LC3B accumulates on phagosomal membranes in a NOX2-dependent manner²⁰. NOX2 is a protein complex formed by transmembrane and cytosolic proteins, which is only active when all the subunits come together at the phagosomal membrane²⁷. We wondered if the delay of LAPosome maturation could be orchestrated by the recruitment of NOX2 and tested whether NOX2 subunits can be found associated with LAPosomes. The transmembrane subunit gp91 was found associated with LC3B positive and zymosan-containing phagosomes (Fig. 2A–C). Overall, 30% of gp91 positive phagosomes contained LC3B, whereas 55% of LAPosomes colocalized with gp91, indicating that NOX2 is preferentially recruited to LAPosomes (Fig. 2D). The recruitment of a second transmembrane subunit of NOX2, p22-phox, was also observed and p22-phox was detected at zymosan-containing phagosomes up to 24 h after LAP stimulation (Supplementary Fig. 3a). Interestingly, the levels of the cytosolic subunit p40-phox were significantly increased upon LAP stimulation and p40-phox was recruited to zymosan-containing phagosomes, indicating that the active NOX2 complex is present at LAPosomes (Fig. 2E and Supplementary Fig. 3b). In view of NOX2 recruitment to LAPosomes, its ability to produce ROS toward the inside of LAPosomes was investigated. To address this point, we used the fluorescent probe OxyBURST, whose fluorescence intensity increases with oxidation. Macrophages were stimulated with OxyBURST-coupled zymosan, and we confirmed that OxyBURST fluorescence intensity could be increased by hydrogen peroxide (H₂O₂) exposure and this was not affected by coupling to zymosan (Supplementary Fig. 3c). Human macrophages stimulated with Candida albicans extract or OxyBURST-coated zymosan showed a significant increase in LC3B-II protein levels, compared to the unstimulated condition (Supplementary Fig. 3d, e). Furthermore, OxyBURST-coated zymosan was also observed in LAPosomes, indicating that OxyBURST coupling did not affect the ability of zymosan to trigger LAP (Fig. 2F and Supplementary Fig. 3f). The levels of oxidation measured within LAPosomes were significantly higher than the level detected inside LC3-negative phagosomes (Fig. 2F, G). This observation was further confirmed by measuring the OxyBURST fluorescence intensity from z-stack image projections (Supplementary Fig. 3f, g). Interestingly, the oxidation levels in phagosomes were significantly reduced overtime, while they remained elevated and stable within LAPosomes (Fig. 2F, G). To go one step further in the understanding of the role of NOX2 in LAPosome formation, we chemically inhibited ROS by pretreating macrophages with two NOX inhibitors: apocynin or diphenyleneiodonium chloride (DPI), for 1 h before LAP stimulation. Macrophages pretreated with NOX inhibitors displayed less LC3B-II accumulation upon zymosan stimulation, compared to untreated conditions and NOX inhibitors did not affect the basal LC3B-II levels measured in unstimulated and inert-bead stimulated conditions (Fig. 3A, B and Supplementary Fig. 3h, i). The ability of macrophages to form LAPosomes was fourfold reduced by NOX inhibition (Fig. 3C, D). In addition, treatment with NOX inhibitors was associated with a decreased oxidation level within phagosomes (Fig. 2G). Together with our previous work, these results demonstrate that NOX2 is required for LAPosome formation and maintenance²⁰. Its activity seems to sustain elevated oxidation within LAPosomes.

A Human macrophages were stimulated with zymosan for 6 h, fixed and costained for gp91 (green) and LC3B (red). B Histogram shows the fluorescence intensity of gp91 (green) and LC3 (red) along the white segment in the merged fluorescence panel of A. Confocal images are representative of three different independent experiments. Scale bar is 5 µm, and inserts are zoomed 5× from white-framed regions of the immune fluorescence micrographs. C Quantitative analysis of the percentage of zymosan-containing vesicles displaying only LC3 or gp91, both or none. Bar graph represents the mean ± SD of three independent experiments, quantified by two independent investigators. One-way ANOVA test: ***p < 0.005. D The first pie chart shows the percentage of gp91⁺ phagosomes displaying LC3B. Second pie chart on the left shows the percentage of LAPosome (LC3B⁺) colocalizing with gp91, Mann–Whitney test, two-tailed: **p < 0.005. E Macrophages were stimulated with zymosan, Candida albicans extract, or beads for 1 up to 24 h, lysed and the change on p40-phox protein level were assessed by western blotting. One representative experiment of three is shown. Bars represent the mean of band intensity of p40-phox normalized to the loading control vinculin from three independent experiments (symbols represent one experiment). All conditions are normalized to the unstimulated condition (US), which is set to 1. One-way ANOVA test. F Macrophages were stimulated with OxyBURST-coated zymosan for 6 h, fixed and stained for LC3 (red). OxyBURST intensity levels were measured inside LAPosomes and LC3-negative phagosomes. Confocal images are representative of three different independent experiments. Scale bar is 5 µm, and inserts are zoomed 5× from white-framed regions of the immune fluorescence micrographs. G Bar graph represents OxyBURST intensity increase after 3 or 6 h of OxyBURST-coated zymosan stimulation inside phagosomes or LAPosomes. All conditions are normalized to the condition of phagosome after 3 h of stimulation, which is set to 1. Bars represent the mean ± SD of three independent experiments (symbols represent one experiment) and for each, more than 50 LAPosomes and phagosomes were analyzed. One-way ANOVA test. Source data are provided as a Source Data file.

A Macrophages pretreated with NOX inhibitors were stimulated with zymosan for 6 h and then lysed. Cell lysates were subjected to SDS-PAGE gel electrophoresis and immunoblotted for LC3B and the loading control vinculin. One representative experiment of three is shown and US represents unstimulated conditions. B Bar graph shows the level of LC3B-II protein expression normalized to vinculin. All conditions are normalized to the DMSO unstimulated condition, which is set to 1. Bars represent the mean ± SD: DMSO (n = 6), apocynin (n = 5), and DPI (n = 7) independent experiments, and each symbol represents a single experiment, ordinary one-way ANOVA. C Macrophages were pretreated with NOX inhibitors (apocynin 50 nM and DPI 50 nM) and stimulated for 6 h with zymosan, fixed and stained for LC3B (green). Representative images from three independent experiments are shown, scale bar is 5 µm, and inserts are 5× zoomed from white-framed regions in the original images. D Bar graph displays the percentage of cells with LAPosomes in each condition. Mean of three independent experiments and each symbol represents a single experiment. Unpaired t test two-tailed. E, F Macrophages transduced with lentiviruses encoding Flag, Flag-ATG4Bwt, or Flag-ATG4BC78S were stimulated with zymosan for 1–24 h, lysed, and LC3B protein levels were assessed by western blot. Image from one representative of three independent experiments. Bar graph shows the level of LC3B-II expression normalized to vinculin. All conditions are normalized to the respective unstimulated conditions (US), which are set to 1, and bars represent mean ± SD of three independent experiments, and each symbol represents a single experiment. Mann–Whitney test, two-tailed. G Macrophages transduced with lentiviruses encoding Flag-ATG4Bwt or Flag-ATG4BC78S were stimulated with zymosan for 6 h, fixed and stained for LC3B (red). White arrows indicate the recruitment of LC3B to zymosan-containing phagosomes. Representative images from three independent experiments are shown with scale bar of 5 µM. H Bar graph shows the percentage of LAPosomes formed per cell upon zymosan stimulation (1 h up to 24 h) in macrophages overexpressing Flag-ATG4Bwt or Flag-ATG4BC78S. Bars represent the fraction of total mean of five independent experiments. One-way ANOVA test. Source data are provided as a Source Data file.

ATG4B-delipidation activity is inhibited by ROS during LAP

ROS have been reported to regulate autophagy^28,29. Subsequent studies have shown that ROS are essential for starvation induced autophagy by regulating key ATGs proteins, such as ATG3, ATG7, or ATG4^30,31,32. Scherz-Shouval et al. recently reported that autophagosome formation is blocked by antioxidants and that cysteine proteases ATG4A and ATG4B are sensitive to oxidation by H₂O₂. They identified cysteine residue Cys-81 in ATG4A, corresponding to Cys-78 in ATG4B, as essential for oxidative inhibition of LC3 delipidation as catalyzed by these proteases³⁰. Based on our findings, we examined whether NOX2-generated ROS regulate ATG4B-delipidation activity during LAP. To address this question we used the ATG4BC78S mutant, previously shown to be oxidation insensitive³⁰. The osteosarcoma cell line U2OS, either overexpressing Flag-ATG4Bwt or Flag-ATG4BC78S, was used to confirm that Cys-78 of ATG4B facilitates redox regulation of its delipidation activity. The level of LC3B lipidation in U2OS cells overexpressing Flag-ATG4Bwt was increased upon H₂O₂ exposure with drugs stimulating macroautophagy (rapamycin) or inhibiting the autophagosome degradation (bafilomycin A1) (Supplementary Fig. 4a). Accumulation of lipidated LC3B was not observed in cells overexpressing Flag-ATG4BC78S, suggesting that this mutant is insensitive to oxidative inactivation (Supplementary Fig. 4a). To investigate whether redox regulation of ATG4B plays a role during LAP, human macrophages were transduced with lentiviruses encoding Flag-ATG4Bwt or Flag-ATG4BC78S, and stimulated with zymosan (Fig. 3E–H and Supplementary Fig. 4b). No significant differences of expression between the oxidation-insensitive mutant and ATG4Bwt were observed. However, LC3B-II protein levels were significantly elevated upon zymosan stimulation in controls and Flag-ATG4Bwt transduced macrophages, but not upon Flag-ATG4C78S expression (Fig. 3E, F and Supplementary Fig. 4b). Furthermore, the expression of Flag-ATG4Bwt in macrophages did not affect the recruitment of LC3B to zymosan-containing phagosomes, whereas LAPosome accumulation was compromised upon Flag-ATG4BC78S expression (Fig. 3G). The majority of Flag-ATG4Bwt expressing macrophages were able to form one or more LAPosomes upon zymosan stimulation, while only up to 20% of macrophages overexpressing Flag-ATG4BC78S were able to form mostly only one LAPosome, suggesting that delipidation by the oxidation-insensitive ATG4B mutant compromises LAP (Fig. 3H). Expression of the Flag tag alone in macrophages did not inhibit LAPosome formation (Supplementary Fig. 4c). Furthermore, the fraction of zymosan in LAPosomes per cell was stable upon Flag-ATG4BC78S expression, but not in Flag-ATG4Bwt transduced macrophages, which increased with time (Fig. 3H and Supplementary Fig. 4d). These results show that ATG4B plays a role during LAP and suggest that its oxidative inhibition by NOX2-produced ROS stabilizes LC3B conjugation to phagosomes during LAP.

Oxidation of ATG4B during LAP

Protein thiols can serve as functional switches by undergoing rapid and reversible oxidative posttranslational modifications in response to redox changes in the environment³³. ATG4 is a cysteine protease that can be inhibited by thiol oxidation. Interestingly, its oxidation sensitivity does not seem to pertain to its active site cysteine (Cys-74 in ATG4B), but instead to another cysteine residue, located in the vicinity of the active site (Cys-78), suggesting that oxidative modification of Cys-78, the nature of which is unknown (e.g., sulfenic acid formation, glutathionylation, intramolecular disulfide bond formation), leads to a structural change that prevents the enzyme from operating³⁰. Surprisingly, in human macrophages overexpressing Flag-ATG4Bwt, an increase of Flag-ATG4Bwt protein levels were observed after LAP stimulation, while this protein accumulation was not observed in macrophages with the oxidation-insensitive mutant Flag-ATG4BC78S (Fig. 4A and Supplementary Fig. 5a). This result indicated that ATG4B accumulates during LAP and the associated ROS production. A study in the green alga Chlamydomonas reinhardtii showed that ATG4 oxidation leads to its aggregation³⁴. Based on this study and our previous observations, we investigated if ATG4B aggregates during LAP by using immune fluorescence microscopy. Upon zymosan stimulation, Flag-ATG4Bwt formed more and bigger protein dots in macrophages compared to Flag-ATG4BC78S (Fig. 4B, C and Supplementary Fig. 5b). Indeed, oxidation-sensitive, but not oxidation-insensitive ATG4B formed protein aggregates in significantly more cells upon LAP stimulation (Fig. 4C). These results suggest that ROS production during LAP triggers ATG4B aggregation, inhibiting its LC3-delipidation activity. A role for Cys-78 in the oxidation of ATG4B in U2OS cells and macrophages was further suggested by the PEG-switch assay, a semiquantitative method to detect reversible cysteine oxidation^31,35. In U2OS cells overexpressing Flag-ATG4Bwt and treated with H₂O₂, a shift in the molecular weight of Flag-ATG4Bwt was observed. Exposure to 500 µM H₂O₂ primarily induced one band with a molecular weight corresponding to 1X-PEGylated ATG4B, suggesting reversible oxidation of one cysteine (Fig. 4D). Formation of this band was significantly decreased upon mutation of Cys-78 (Fig. 4D). We speculate that the pattern of lower molecular weight bands was caused by degradation of Flag-ATG4B. ATG4B oxidation was also confirmed in macrophages upon H₂O₂ exposure and zymosan stimulation (Fig. 4E). Overall, these results supported the notion that LAP induces ATG4B oxidation and aggregation mainly in a Cys-78 dependent manner.

A Macrophages transduced with lentiviruses encoding Flag-ATG4Bwt or Flag-ATG4BC78S were stimulated with zymosan for 1–24 h, lysed, and the level of Flag tagged proteins assessed by western blot. Bar graph shows the accumulation of Flag-ATG4Bwt upon LAP stimulation while no accumulation of the Flag-ATG4BC78S is observed. Bar graph represents mean ± SD of n = 6 independent experiments for the US to 6 h conditions and n = 4 independent experiments for the 24 h condition. Each symbol represents a single experiment. Mann–Whitney test, two-tailed. B Macrophages transduced with lentivirus expressing Flag-ATG4Bwt or Flag-ATG4BC78S were stimulated or not with zymosan for 6 h, fixed, and then stained for Flag (green) or LC3 (red). Representative images from three independent experiments are shown with scale bars of 5 µM. White arrows show Flag-ATG4B dots and US represents unstimulated conditions. C Graph shows the number of Flag-ATG4B dots per cells in unstimulated or zymosan stimulated conditions. Each dot represents a single cell and more than 20 cells per experiment were analyzed from three independent experiments. Unpaired Student’s t test, two-tailed. D U20S cells transduced with lentiviruses encoding Flag-ATG4Bwt or Flag-ATG4BC78S were treated with H₂O₂ and assessed for oxidation of ATG4B by PEG-switch assays. Bar graph shows quantification of Flag-ATG4B oxidation from treated U20S and bars represent mean ± SD of three independent experiments, and each symbol represents a single experiment. One-way ANOVA test. E Human macrophages transduced with lentiviruses encoding Flag-ATG4Bwt or Flag-ATG4BC78S were treated with H₂O₂ and assessed for oxidation of ATG4B by PEG-switch assays. Bar graph shows quantification of Flag-ATG4B oxidation from treated human macrophages and bars represent mean ± SD of three independent experiments, and each symbol represents a single experiment. N.D stands for not detected. Source data are provided as a Source Data file.

Inhibition of ATG4B by ROS prolongs antigen presentation via MHC class II

We previously reported that LAP is involved in prolonged MHC class II antigen presentation of phagocytosed C. albicans antigens²⁰. Accordingly, we wondered whether the oxidative inhibition of ATG4B would support this prolonged MHC class II antigen presentation. For this purpose, human macrophages were transduced with lentiviruses encoding Flag, Flag-ATG4Bwt, or Flag-ATG4BC78S and were stimulated with C. albicans extract, also known to be a LAP stimulus. This stimulation indeed accumulated LC3B-II in human macrophages (Fig. 5A). However, Flag-ATG4BC78S expressing macrophages formed LAPosomes less frequently upon stimulation with C. albicans extract-coated beads, compared to their Flag-ATG4Bwt expressing counterparts (Fig. 5B). These data demonstrated that both C. albicans extract-coated beads and zymosan (Fig. 3G, H) require oxidation sensitivity of ATG4B to induce LAP. Therefore, we investigated the dependence of prolonged MHC class II antigen presentation after LAP, and especially the importance of oxidation sensitivity of ATG4B for this process. Along these lines, MHC class II antigen-presentation assays were performed according to the scheme in Fig. 5C. The IFNγ production of clonal C. albicans-specific CD4⁺ T cells of one donor in response to antigen-pulsed HLA-DR4 matched macrophages that expressed Flag, Flag-ATG4Bwt, or Flag-ATG4BC78S was assessed (Fig. 5D and Supplementary Fig. 6a). Oxidation-insensitive ATG4B expressing macrophages were significantly less able to present extracellular C. albicans antigens on MHC class II molecules to the specific CD4⁺ T-cell clone than their Flag or Flag-ATG4Bwt transduced counterparts (Fig. 5D). We also observed a nonsignificant decrease of antigen presentation upon Flag-ATG4Bwt expression, possibly due to saturation of the ROS regulation or a delipidation independent function of overexpressed ATG4B. In addition, the ability of whole blood memory CD4⁺ T cells from five independent donors, cocultured with their autologous macrophages pulsed with C. albicans extract, to secrete interleukin 17A (IL-17A) was also assessed (Fig. 5E and Supplementary Fig. 6b–e). The cytokine IL-17A is preferentially secreted by CD4⁺ Th17 cells in response to fungal infections, such as C. albicans³⁶. Flag-ATG4BC78S expression was comparable to the expression levels of Flag-ATG4Bwt (Supplementary Fig. 6b, c). Transduced and antigen-pulsed macrophages expressing Flag or Flag-ATG4Bwt constructs were able to activate autologous C. albicans-specific whole blood memory CD4⁺ T cells to a similar extent. This was demonstrated by the secretion of IL-17A, as well as the production of the inflammatory Th17 cytokine IL-31 (Fig. 5E, Supplementary Fig. 6d, e). The differences in IL-17A production were established by ELISA and multiarray technology, while the latter was also used to detect the variation in IL-31. In the oxidation-insensitive ATG4B condition, with the compromised LAP pathway, the transduced and pulsed macrophages were less able to activate their autologous memory C. albicans-specific CD4⁺ T cells. Indeed, less IL-17A and IL-31 were secreted by the memory CD4⁺ T cells in coculture with the ATG4C78S transduced macrophages (Fig. 5E and Supplementary Fig. 6d, e). However, the overexpression of Flag, Flag-ATG4Bwt, and Flag-ATG4BC78S by macrophages only specifically influenced their ability to activate Th17 cells, as there were no significant differences in the secretion of promiscuously expressed human macrophage inflammatory protein-3α (MIP-3α or CCL20) (Supplementary Fig. 6e). Moreover, this inability of the ATG4C78S-macrophages to activate T cells was particularly pronounced at later time points (18 h) after antigen pulsing (Fig. 5D, E and Supplementary Fig. 6d, e). Altogether, these results suggest that the inhibition of ATG4B’s delipidation activity by oxidation during LAP leads to maintained MHC class II antigen presentation by human myeloid cells. Furthermore, we were wondering if this defect of presenting extracellular antigen on MHC class II molecules was associated with decreased MHC class II expression on the macrophage surface. However, we did not observe any difference in HLA-DR surface expression of macrophages after lentiviral transduction of Flag, Flag-ATG4Bwt, or Flag-ATG4BC78S (Fig. 5F and Supplementary Fig. 7a–d). Similarly, the expression of the two costimulatory molecules CD86 and CD80 was unaltered (Fig. 5F and Supplementary Fig. 7a–d), suggesting that the impairment of ATG4BC78S expressing macrophages to sustain MHC class II antigen presentation might be rather due to a defect in antigen loading during compromised LAP. Thereby, ROS production by NOX2 seems to inhibit LC3B deconjugation from LAPosomes via ATG4B oxidation and the resulting LAPosomes maintain endocytosed antigens for prolonged presentation on MHC class II molecules.

Macrophages transduced with the indicated lentiviruses encoding Flag, Flag-ATG4Bwt, or Flag-ATG4BC78S were investigated. A Transduced macrophages were stimulated with Candida (C.) albicans extract for 6 h and the expression of LC3B was analyzed by western blot. Bar graph shows the level of LC3B-II expression normalized to vinculin. All conditions are normalized to each unstimulated condition (US), which is set to 1. Bars represent mean ± SD of three independent experiments, Mann–Whitney test, two-tailed. B Transduced macrophages were stimulated with C. albicans extract-coated beads for 6 h and then stained for LC3 (red). Representative images from three independent experiments, scale bars 5 µm, and inserts are zoomed 5× from white-framed regions of the original images. Bar graph shows the percentage of cells with or without LAPosomes. Bars represent the mean ± SD from three independent experiments, and each symbol represents a single experiment. One-way ANOVA test. C Experimental outline of MHC class II antigen-presentation assay. We created this diagram using the free online platform Smart Servier Medical Art. D ELISA assays on supernatants of a C. albicans-specific CD4⁺ T-cell clone cocultured with MHC class II matched macrophages transduced with the indicated Flag tagged constructs. Transduced macrophages were pulsed with C. albicans extract. Time indicated below the graph corresponds to the resting time of the pulsed macrophages before coculture. Scatter plot represents the IFNγ production normalized to unstimulated conditions (US) from five independent experiments and only values above the detection limit are displayed. Each symbol represents a single experiment. Ratio paired Student’s t test, two-tailed. E ELISA on supernatants of whole blood memory CD4⁺ T cells with autologous macrophages transduced with the indicated Flag tagged constructs. Transduced macrophages were pulsed with C. albicans extract. Time indicated below the graph corresponds to the resting time of the pulsed macrophages before coculture. Scatter plot represents the IL-17A production normalized to unstimulated conditions (US) from five independent experiments with five different donors and each symbol represents a biological replicate. Kruskal–Wallis test. F Macrophages transduced with the indicated lentiviruses were stimulated with C. albicans extract for 4 h. The macrophages were washed and then incubated for 18 h without stimuli or directly analyzed by FACS. Scatter plot from four independent experiments and each symbol represents a different experiment. Source data are provided as a Source Data file.