The pore-forming apolipoprotein APOL7C drives phagosomal rupture and antigen cross-presentation by dendritic cells

Type I conventional dendritic cells (cDC1s) are essential for the generation of protective cytotoxic T lymphocyte (CTL) responses against many types of viruses and tumours. They do so by internalizing antigens from virally infected or tumour cells and presenting them to CD8+ T cells in a process known as cross-presentation (XP). Despite the obvious biological importance of XP, the molecular mechanism(s) driving this process remain unclear. Here, we show that a cDC-specific pore-forming protein called apolipoprotein 7C (APOL7C) is upregulated in response to innate immune stimuli and is recruited to phagosomes. Strikingly, the association of APOL7C with phagosomes leads to phagosomal rupture, which in turn allows for the escape of engulfed protein antigens to the cytosol where they can be processed via the endogenous major histocompatibility complex (MHC) class I antigen processing pathway. We show that APOL7C recruitment to phagosomes is voltage-dependent and occurs in response to NADPH oxidase-induced depolarization of the phagosomal membrane. Our data indicate the presence of dedicated pore-forming apolipoproteins that mediate the delivery of phagocytosed proteins to the cytosol of activated cDC1s to facilitate MHC class I presentation of exogenous antigen and to regulate adaptive immunity.


Introduction
Antigen-presenting cells (APCs), such as macrophages and dendritic cells, can present exogenous antigens on major histocompatibility complex class I (MHC-I) to naïve CD8 + T cells to elicit cytotoxic T lymphocyte (CTL) responses. This process, referred to as cross-presentation (XP), is essential to the clearance of many types of viruses and tumours [1][2][3][4][5][6][7][8] . While several APCs are capable of XP, experiments using Batf3-deficient (Batf3 -/-) or Irf8 + 32 -/mice, which both lack cDC1s, have demonstrated that cDC1s play a seemingly nonredundant role in the generation of CTLs [8][9][10] . Indeed, cDC1s harbor unique specializations that facilitate their role in XP. For example, they express the receptor DNGR-1 (also known as CLEC9A) which facilitates the XP of antigens extracted from dead or damaged cells 5,7,[11][12][13][14] . Similarly, cDC1s uniquely express high levels of WDFY4, a protein that is required for XP by an unknown mechanism 1 . Importantly, the molecular mechanisms by which XP is achieved likely vary depending on the environmental context and the nature of the antigen.
Two pathways for XP are believed to exist. In the first, known as the "vacuolar" pathway, proteins are processed by endosomal proteases to liberate peptides which are loaded onto MHC-I all within endocytic organelles [15][16][17] . The second is known as the "cytosolic" or the "phagosome-to-cytosol (P2C)" pathway [15][16][17] . In this pathway, endocytosed or phagocytosed proteins are released into the cytosol where they are processed into peptides by the proteasome. The peptides are then loaded onto MHC-I in the ER for XP. The mechanism by which P2C occurs is ill-defined and multiple hypotheses exist 19 . One hypothesis, the "indigestion model", proposes that under certain conditions the membrane integrity of phagosomes is lost allowing for the release of engulfed proteins into the cytosol 20,21 . Recently, evidence in support of the "indigestion model" has emerged. For example, DNGR-1 has been found to signal via SYK and the NADPH oxidase for phagosomal rupture, which in turn releases phagocytosed proteins into the cytosol 11,22 . Also, a pore-forming protein present in the lumen of endocytic organelles in APCs called perforin-2 has been found to instigate the formation of nonselective pores in endocytic organelle membranes, which allows for the escape of endocytosed proteins to the cytosol 23 . These exciting new findings have led to entirely new questions. For example, do pore-forming proteins such as perforin-2 contribute to DNGR-1 dependent phagosomal rupture, what signals regulate rupture or pore formation in the endocytic pathway of APCs, does it occur on all endocytic organelles, and are there other, yet undefined, molecular drivers of this process?
In this study, we present new evidence in support of the "indigestion model" for P2C. We show that innate immune stimuli, including agonists of Toll-like receptor 3 (TLR3) which is highly expressed by cDC1s, lead to the expression of apolipoprotein L 7C (APOL7C), a cytosolic apolipoprotein L containing a putative colicin-like pore-forming domain. We find that APOL7C inserts into phagosomes in an NADPH oxidase-dependent manner, leading to pore formation and causing phagosomal rupture and the release of phagocytosed proteins to the cytosol for XP. Our data indicate the presence of dedicated cDC-specific pore-forming proteins that insert into phagosomes in a regulated manner to drive XP.

Poly(I:C) leads to a greater incidence of phagosomal membrane rupture in cDC1s.
Previous studies have shown that XP is a tuneable process 24-26 that can be enhanced by innate immune stimuli 24,25 and both the TLR4 agonist LPS and the TLR9 agonist CpG DNA can result in enhanced damage to endosomes in dendritic cells 27,28 .
To investigate the effect of cDC1 activation on phagosomal rupture in cDC1s, we employed the murine cDC1 cell line MuTuDC1940 (henceforth called MuTuDCs).
We established several independent assays to assess the incidence of phagosomal rupture in MuTuDCs upon poly(I:C) activation. In the first, OVA-beads were covalently labeled with a fluorophore (Alexa Fluor 488; AF488) and incubated with MuTuDCs in the presence of the fluid-phase, membrane-impermeant dye sulforhodamine B (SRB). The uptake of the OVA-beads via phagocytosis results in the simultaneous loading of phagosomes with SRB. Unlike latex beads, the 3 µm silica beads used here are porous allowing for the entry of fluorescent molecules like SRB 43 ; as a result, the fluidphase SRB is visible throughout the phagosome. Upon phagosomal rupture, fluid-phase SRB escapes into the cytosol, but covalently attached AF488 remains within the phagosome. Using the phagolysosome-damaging compound L-leucyl-L-leucine methyl ester (LLOMe), we confirmed that phagosomal rupture results in the progressive loss of SRB, but not the AF488, from phagosomes ( Figure 1E-F). Next, we assessed the effect of poly(I:C) activation on the loss of SRB from phagosomes. Interestingly, MuTuDCs treated with poly(I:C) displayed significantly fewer SRB-positive phagosomes relative to untreated cells (Figure 1G-H) suggestive of phagosomal rupture. Therefore, we next looked at the incidence of established phagosomal damage biomarkers in poly(I:C) treated cells.
Damage to endocytic organelle or phagosome membranes initiates cellular membrane repair processes characterized by the rapid phosphorylation of RAB8A by the kinase LRRK2 which subsequently recruits components of the endosomal sorting complex required for transport (ESCRT)-III including CHMP4B and ALIX [44][45][46][47][48][49][50] . In the case of extensive damage, members of the galectin family of proteins are also recruited to phagosomes to facilitate removal of the damaged endocytic organelles by autophagy [46][47][48]50 . As a result, recruitment of components of these membrane repair/disposal pathways to endocytic organelles or phagosomes are frequently used as indirect biomarkers of membrane damage 11,28,34,46,49,[51][52][53][54] . Indeed, we found that treatment of MuTuDCs with the phagosome-damaging compound LLOMe resulted in a dose-dependent increase in the  (Figure 1S-U). Taken together, our data suggest that poly(I:C) treatment results in a greater incidence of phagosome rupture in MuTuDCs and primary FLT3L-cDCs.

Poly(I:C) increases expression of the pore-forming apolipoprotein APOL7C which is recruited to phagosomes.
To gain insight into the molecular pathways that drive phagosomal rupture upon poly(I:C) treatment of cDC1s, we treated MuTuDCs with or without poly(I:C) and performed bulk RNA sequencing (RNAseq). Among others, we found a significant increase in a group of proteins known as the apolipoprotein L (APOL) family (Figure 2A).
While several members of the family were upregulated upon poly(I:C) treatment, apolipoprotein L 7C (APOL7C) demonstrated a greater than 20-fold increase in expression ( Figure 2B-C). We confirmed that this induction also occurred in splenic cDCs in vivo by injecting mice with poly(I:C) via the tail vein, isolating splenocytes, enriching for cDCs and performing qRT-PCR. Here again we found a significant increase in Apol7c expression ( Figure 2D).
We next assessed the immune cell-specificity of Apol7c expression by single-cell RNA-sequencing (scRNAseq). To this end, we treated mice with poly(I:C) or PBS by tail vein injection and after 24 hours processed splenocytes for scRNAseq. An unsupervised uniform manifold approximation and projection (UMAP) was performed and the annotation of cell clusters was based on the expression of curated and data-driven genes as described by Bosteels et al 59 Table 1). Interestingly, we found that Apol7c expression was restricted to cDCs, with the highest expression in migratory cDC1s ( Figure 2E-H). In addition to migratory cDC1s, Apol7c expression was also found in non-migratory cDC1s, migratory cDC2s and non-migratory cDC2s (Figure 2E-H).
Poly(I:C) treatment did not change this pattern, as Apol7c expression remained restricted to cDCs (Figure 2H).
We next turned our attention to the subcellular distribution of APOL7C. For this we generated a stable MuTuDC cell line expressing a doxycycline(dox)-inducible APOL7C::mCherry fusion protein (henceforth referred to as MuTuDC.APOL7C::mCherry cells). We found APOL7C::mCherry to be largely cytosolic in otherwise untreated cells.
Remarkably, upon phagocytosis of OVA beads, APOL7C::mCherry redistributed to phagosomes (Figure 3A-B). Interestingly, we saw similar phagosomal recruitment of APOL7C::mCherry to phagosomes in another phagocyte cell line Raw 264.7, suggesting that while APOL7C is selectively expressed by cDCs, its ability to be recruited to phagosomes can be reconstituted in other cells ( Figure 3E). Phagosomal recruitment of APOL7C::mCherry was not limited to OVA-bead-containing phagosomes, as we observed similar recruitment to zymosan-, IgG-opsonized sheep red blood cell (IgG-sRBC)-, Leishmania majorand Staphylococcus aureus-containing phagosomes (Supplementary Figure 3A-D). Recruitment of APOL7C::mCherry to phagosomes appeared to peak several hours after phagocytosis suggesting that it was recruited to late phagosomes that have likely already fused with the lysosomal compartment ( Figure 3B).
PtdIns(3)P is only present on newly formed or early phagosomes. We found no phagosomes that were positive for both APOL7C::mCherry and p40PX::GFP (Supplementary Figure 3E-G). Next, we stained cells expressing APOL7C::mCherry for a marker of late phagosomes, LAMP1. We found that nearly all phagosomes that were positive for APOL7C::mCherry were also positive for LAMP1 (Supplementary Figure  3H-J). Together, these data demonstrate that APOL7C::mCherry accumulates on late phagosomes that have already fused with lysosomes (i.e., phagolysosomes). Given that APOL7C harbors a putative pore-forming domain and that other APOL family members have been shown to permeabilize biological membranes, we next assessed whether APOL7C contributes to phagosomal rupture.
APOL7C recruitment to phagosomes results in phagosome rupture.

8
To investigate the effect of APOL7C on phagosome membrane integrity we first analyzed the distribution of the membrane damage biomarker galectin 3 in MuTuDC.APOL7C::mCherry cells three hours after phagocytosis. Strikingly, we found that the majority of phagosomes that were positive for APOL7C::mCherry was also positive for galectin 3 (Figure 3C-D). This was not the case for APOL7C::mCherrynegative phagosomes, which were rarely positive for galectin 3 (Figure 3C-D  To investigate whether in addition to a small membrane impermeant dye such as LY, proteins could also move freely across APOL7C::mCherry-positive phagosome membranes, we generated a cell line that stably expressed cytosolic GFP (cGFP) (henceforth RawKb.cGFP.APOL7C::mCherry cells). In these cells, we readily observed the movement of cGFP from the cytosol to the lumen of APOL7C::mCherry-positive phagosomes but not APOL7C::mCherry negative phagosomes (Figure 3M-N). This movement was continuous as we readily observed the continued influx of cGFP in the APOL7C::mCherry positive phagosomes after photobleaching, which was not due to damage from photobleaching as it was not observed with photobleached APOL7C::mCherry-negative phagosomes ( Figure 3O). We next assessed whether we Finally, we examined the ultrastructure of APOL7C::mCherry positive phagosomes by correlative confocal and focused ion beam scanning electron microscopy (FIB-SEM).
This revealed discontinuity in the phagosomal membrane that would allow for the escape of the luminal contents of phagosomes to the cytosol, as observed above ( Figure 3P and  Figure 4A). Doxycycline also had no effect on the ability of RawKb cells pulsed with SIINFEKL peptide to elicit B3Z cell activation ( Figure 4B). However, RawKb.APOL7C::mCherry cells incubated with OVA-beads for 4 hours showed a large increase in XP when treated with doxycycline ( Figure 4C). Direct presentation of SIINFEKL peptide by RawKb.APOL7C::mCherry cells was unaffected by doxycycline treatment ( Figure 4D). Importantly, doxycycline did not affect the ability of RawKb.APOL7C::mCherry cells to internalize OVA-beads and did not affect the surface expression of H-2K b (Supplementary Figure 5A-D). These data indicate that APOL7C can promote XP in non-cDC1 cells.
Next, we performed loss-of-function studies in FLT3L-cDCs derived from wild-type Finally, to assess the importance of APOL7C for XP in vivo, we immunized Apol7c +/and Apol7c -/mice with OVA-beds + poly(I:C) and measured OVA-specific CD8 + T cell responses by H-2K b -OVA-pentamer staining. Apol7c +/mice mounted a robust OVA-specific CD8 + T cell response that was significantly reduced in Apol7c -/littermates ( Figure 4H). Together, these data indicate that APOL7C can promote the XP of OVAbeads both in vivo and in vitro by primary FLT3L-cDCs and when ectopically expressed in other phagocytes.
Apol7C recruitment is independent of SYK but requires NADPH oxidase activity on phagosomes.
Phagosome rupture can be modulated by dedicated cDC1 receptors, such as DNGR-1, through a SYK-and NADPH oxidase-dependent pathway 11,22,50,63 . While the OVA beads used here do not harbor DNGR-1 ligands, we wondered whether the recruitment of APOL7C to phagosomes also relied on SYK-and NADPH oxidasedependent signaling. We stained RawKb.APOL7C::mCherry cells that had phagocytosed OVA beads for phospho-SYK and while we could detect the presence of phospho-SYK on APOL7C::mCherry positive phagosomes ( Figure 5A-B), inhibition of SYK with the inhibitor R406 had no effect on the recruitment of APOL7C::mCherry ( Figure 5C-D). We also generated SYK CRISPR knockouts from the RawKb.APOL7C::mCherry cells and observed no difference in the recruitment of APOL7C::mCherry to phagosomes (Supplementary Figure 8A-C).
Next we checked whether the NADPH oxidase was present on APOL7C::mCherry positive phagosomes. Staining for the p22 phox subunit of the oxidase revealed that it was also present on APOL7C::mCherry positive phagosomes ( Figure 5E-F). We could also detect oxidase activity in phagosomes by covalently attaching the reactive oxygen species (ROS)-sensitive dye H2DFFDA along with a ROS-insensitive fluorophore Alexa Fluor 633 (AF633) directly to the beads. Oxidation of H2DFFDA in the phagosomes was sensitive to the NADPH oxidase inhibitor diphenylene iodonium (DPI) (Supplementary Figure 8D-E). As previously reported 64 , we observed heterogeneity in the ability of individual phagosomes to generate superoxide ( Figure 5G). Interestingly, we found a very strong correlation between oxidase activity and the recruitment of APOL7C::mCherry to phagosomes suggesting a potential role for the NADPH oxidase in the recruitment of APOL7C::mCherry ( Figure 5G-H). DPI treatment of the RawKb.APOL7C::mCherry cells revealed a dose-dependent reduction in APOL7C::mCherry recruitment, without any effect on OVA bead uptake ( Figure 5I-K). We also generated p22 phox CRISPR knockouts from the RawKb.APOL7CmCherry line which were incapable of producing phagosomal ROS ( Figure 5L-N). The knockout cell lines also showed a significant decrease in the number of APOL7C::mCherry positive phagosomes (Figure 5O-P). Altogether our data indicate that APOL7C recruitment to OVA bead-containing phagosomes is independent of SYK but does require the activity of the NADPH oxidase.

NADPH oxidase activity is required for APOL7C-dependent XP
As discussed earlier, expression of APOL7C::mCherry in cells that normally lack APOL7C results in enhanced XP. We next investigated whether NADPH oxidase activity, which appears to be a pre-requisite for APOL7C recruitment to phagosomes, is necessary Together these findings indicate that APOL7C-dependent XP is dependent upon NADPH oxidase activity.

APOL7C insertion into phagosomes is a voltage-dependent process that relies on
NADPH oxidase-driven depolarization of the phagosome membrane.
The rupture of endocytic organelles, including phagosomes, in cDC1s has been proposed to occur as a result of peroxidation of their limiting lipid bilayer by NADPH oxidase-derived ROS 27,65 . It was conceivable, however, that lipid peroxidation of the phagosomal membrane may result in the insertion of APOL7C. For this, we employed the radical-trapping antioxidant liproxstatin-1 (LPX) which potently inhibits the accumulation of lipid peroxides in cellular membranes 66,67 . LPX treatment had no effect on the uptake of OVA beads, nor did it affect the oxidation of H2DFFDA on the OVA beads ( Figure 6A-C). It did however potently inhibit the oxidation of BODIPY 581/591 C11, a redox-sensitive lipid oxidation probe (Figure 6D-E). This contrasts with DPI which inhibits the enzymatic activity of the NADPH oxidase and therefore blocks oxidation of both H2DFFDA and BODIPY 581/591 C11 (Figure 6A-E). Notably, unlike DPI, LPX treatment had no effect on either the recruitment of APOL7C::mCherry to phagosomes or on APOL7C-dependent XP ( Figure 6F-J). This indicates that lipid peroxidation is unlikely to be the signal for APOL7C insertion into phagosome membranes.
We next investigated other consequences of NADPH oxidase activity that may mark phagosomes for APOL7C insertion. In addition to generating ROS, the NADPH oxidase has electrogenic activity [68][69][70] . This is a consequence of the transfer of electrons  Figure 6B and Figure 7A).
These residues are also present in APOL7C and correspond to A178, A186 and G190 of the APOL7C MID ( Figure 7A). It is therefore conceivable that NADPH oxidase-driven depolarization of the phagosome membrane may be the trigger for the insertion of APOL7C into phagosome membranes. To investigate this possibility, we first measured the membrane potential of phagosomes in untreated cells and in cells treated with DPI.
Phagosomal membrane potential measurements were performed using a previously established FRET-based approach 73 . OVA beads were covalently labeled with the succinimidyl ester of the FRET donor 7-diethlyaminocoumarin-3-carboxylic acid (DACCA). The potential-sensitive, lipid-soluble probe bis-(1,3-dibutylbarbituric acid)pentamethine oxonol [DiBAC4 (5)], which partitions across membranes in accordance with their transmembrane potential was used as the FRET acceptor. As FRET relies on the close apposition of the donor:acceptor pair (< 10 nm), the physical attachment of the donor fluorophore to the phagocytic target restricts the FRET signal to the phagosome. Bathing DACCA-OVA-beads in increasing concentrations of DiBAC4 (5) revealed that the corrected FRET (cFRET) signal was restricted to the OVA-bead surface and that sensitive cFRET measurements could be attained over a broad range of DiBAC4(5) concentrations ( Figure 6K). Next we fed DACCA-OVA-beads to RawKb.APOL7C::mCherry cells for 1 hour, then thoroughly washed away external beads, and bathed the cells in 250 nM DiBAC4 (5). We found that the DiBAC4(5) was, as expected, present throughout the cell but the cFRET signal was restricted to the phagosome ( Figure 6L). As the NADPH oxidase transfers electrons from the cytosol to the lumen of the phagosome, oxidase activity would result in the exclusion of DiBAC4(5), which has a negative charge, from the phagosome and therefore a reduced cFRET signal. Interestingly, we found a progressive depolarization (reduced phagosomal cFRET) of the phagosome membrane over the first 3 hours post-phagocytosis ( Figure  6M). This depolarization was DPI-sensitive, indicating that it was likely driven by NADPH oxidase activity ( Figure 6M). It is therefore possible that NADPH oxidase-driven depolarization of the phagosomal membrane can instigate insertion of APOL7C.
To next test the voltage-dependence of APOL7C insertion into lipid bilayers we employed both cellular-and giant unilamellar vesicle (GUV)-based approaches. First, as we could not manipulate phagosome membrane potential without affecting other cellular membranes, we attempted to clamp the plasma membrane (PM) potential of RawKb.mCherry and RawKb.APOL7C::mCherry cells such that the cytosolic side became positive by imposing an inward K + gradient in the presence of a K + -selective ionophore. We reasoned that this would expose APOL7C::mCherry to positive voltages on the cytosolic side and therefore drive its insertion into the PM. To achieve this, we loaded the cytosol of the cells with Na + and depleted them of K + by inhibiting Na + /K + -ATPase activity with ouabain for three hours. Then we transferred the cells to a K + -rich, Na + -free medium in the presence of the K + -selective ionophore valinomycin. This imposes an inward K + gradient and is predicted to clamp the plasma membrane potential (jpm) at Faraday's constant, respectively, and z is the charge of K + . This treatment had no effect on the distribution of mCherry but, remarkably, resulted in the near immediate insertion of APOL7C::mCherry into the plasma membrane of the RawKb.APOL7C::mCherry cells ( Figure 6N, 6O and 6Q). Moreover, insertion of APOL7C::mCherry into the PM resulted in its permeabilization to propidium iodide (PI) (Figure 6P and 6R).
We next employed a GUV-based approach to investigate the voltage-gated insertion of APOL7C in lipid bilayers. To this end, we generated GUVs containing the monovalent cation-selective ionophore gramicidin 74,75 . By generating these gramicidin- To further investigate the voltage-dependence of APOL7C insertion into phagosome membranes, we next assessed whether mutation of the 3 key residues that dictate voltage or pH-dependent insertion of APOLs would impact APOL7C insertion into phagosome membranes. As discussed above, APOLs insert into bilayers in either a pHor voltage-dependent manner. This is largely determined by three residues in the MID that align with E201, E209 and E213 of APOL1 71,72 ( Figure 7A). The pH-dependent APOLs, APOL1 and APOL2, have conserved negatively charged amino acids at these positions, whereas all other APOLs, including APOL7C, have neutral amino acids at these residues. Importantly, negatively charged amino acids at these residues not only dictate pH-dependent insertion, but render APOLs voltage-insensitive 71,72 . We therefore reasoned that converting the neutral amino acid residues at positions A178, A186 and G190 of APOL7C (corresponding to E201, E209 and E213 in APOL1) to negatively charged glutamic acid residues would block the voltage-dependent insertion of APOL7C into phagosome membranes. Indeed, we found that APOL7C ( when the cytosolic pH was clamped at acidic pH; whereas, wild-type APOL7C::mCherry did not (Figure 7B-I). Therefore, converting APOL7C to a voltage-insensitive APOL, akin to APOL1 and APOL2, completely blocked both its recruitment to phagosomes and its ability to drive XP.
Altogether our data indicate that lipid peroxidation is unlikely to affect the insertion of APOL7C into the phagosome membrane. Instead, APOL7C insertion into membranes appears to be voltage-dependent and insertion into phagosomes is likely driven by NADPH oxidase-induced depolarization of the phagosome membrane.

Discussion
XP is important for the generation of many virus-and tumour-specific CTL responses 1-8 . Despite its obvious biological importance, the mechanism(s) by which XP is achieved remain unclear 10,19,63,[76][77][78][79] . Two recent observations from the XP literature informed the present study. First, while XP has been observed in a variety of APCs in vitro, only cDC1s appear to serve a non-redundant role in cross-priming to many types of viruses and tumours in vivo 1,9,10,[80][81][82][83] . Second, the XP of cell-associated antigens by cDC1s often involves the delivery of antigens to the cytosol 11,20,21,23,28,34,50,63 . In line with this, several groups have recently shown that, in cDC1s, endocytic organelles, including phagosomes, appear to be "damage prone" and to display signs of breached or ruptured membranes 11,23,27,28,65 . This is believed to facilitate the delivery of endocytosed or phagocytosed antigens to the cytosol where they gain access to the endogenous pathway of MHC class I processing and presentation 63,76,78 . Given the specialized role of cDC1s in XP in vivo and the observation that endocytic organelles in cDC1s often display signs of compromised membrane integrity, we hypothesized that cDC1s harbor adaptations that facilitate the delivery of phagocytosed antigens to the cytosol where they can be processed for XP.
Our data indicate that cDC1s express the cytosolic, pore-forming protein APOL7C.
APOL7C is recruited to phagosomes in a regulated manner that requires the NADPH oxidase-dependent depolarization of the phagosomal membrane. Recruitment of APOL7C to phagosomes instigates the formation of large, nonselective pores or breaches in the phagosomal membrane. This in turn allows for the release of phagosomal contents to the cytosol for XP. The unique expression pattern of APOL7C and its role in the delivery of phagocytosed material to the cytosol supports our hypothesis and supports other reports 23 indicating the existence of dedicated cDC-specific pore-forming proteins that can facilitate XP.
Two main pathways have been described for the escape of phagosomal contents from the phagosomes of APCs during XP -the "transporter" hypothesis and the "indigestion" or phagosomal rupture hypothesis 18,78 . Our findings support the "indigestion" hypothesis. Interestingly, we show that cDC1s not only possess dedicated pore-forming proteins that are recruited to phagosomes to elicit phagosomal rupture, but that the process is tightly regulated. We propose that APOL7C-dependent phagosomal rupture is regulated at least at three different levels. First, we found that the TLR3 agonist poly(I:C) significantly increased the expression of APOL7C. This implies a priming effect of innate immune stimuli and is consistent with the concept of cDC activation whereby cDCs undergo gene transcription and translation changes in response to stimulation to increase their antigen processing and presentation capacity [24][25][26]36,40 . Second, we found that not all phagosomes recruit APOL7C. Instead, APOL7C is selectively recruited to phagosomes on which the NADPH oxidase is active. Although the heterogeneity of NADPH oxidase activity at the level of individual phagosomes has been previously documented 64 , the basis of this heterogeneity is poorly understood. We speculate that receptors present in the phagosome survey its contents and trigger oxidase activity. This notion is supported by the recent finding that DNGR-1, a cDC1-specific receptor that signals for phagosome rupture, can drive NADPH oxidase activity in phagosomes where DNGR-1 ligands are present 11,22 . We cannot, however, rule out that other, non-cargo dependent pathways contribute to the observed heterogeneity in phagosomal oxidase activity and the subsequent recruitment of APOL7C. Finally, cellular membrane repair pathways are also likely to regulate APOL7C-dependent phagosome rupture. We observed the preferential recruitment of the ESCRT-III machinery and galectin 3 to APOL7C-positive phagosomes.
Both have been described to repair damaged endocytic organelles including phagosomes [46][47][48][49]51,54 . Notably, genetic ablation of components of the ESCRT-III machinery has been shown to increase XP in cDC1s 28 . Altogether our findings, and others, support the notion that APOL7C-dependent phagosomal rupture is a carefully regulated process. Altogether, our data identify dedicated pore-forming apolipoproteins that operate in the endocytic pathway of cDCs. We find that one of these apolipoproteins, APOL7C, supports XP by mediating the delivery of exogenous antigen to the cytosol. While we show that the expression of APOL7C is inducible and that its recruitment to phagosomes is dependent on the NADPH oxidase, future studies on what signals drive oxidase activity and therefore "mark" individual phagosomes for APOL7C recruitment and phagosome rupture are of considerable interest and may inform immunotherapeutic strategies for priming CD8 + T cells in the context of cancer treatment and vaccines.  Significance determined using student's t test. n.s. = no significance; *P £ 0.05; **P £ 0.01; ***P £ 0.001; ****P £ 0.0001.    Significance determined using student's t test. n.s. = no significance; **P £ 0.01; ****P £ 0.0001. Significance determined using student's t test. n.s. = no significance; *P £ 0.05; **P £ 0.01; ****P £ 0.0001. Segmentation was performed as in methods and movie was prepared with Imaris.

Methods and Materials
General materials used in this study.  Giant unilamellar vesicle (GUV) assays. GUVs were prepared on agarose films overlayed on glass slides. Specifically, 1 % ultra-low gelling temperature agarose was spread over glass coverslips and dried in a 37°C incubator until the agarose layer was near invisible.

Product
Lipid mixes were prepared (according to the recipe below) in glass vials using chloroform as the solvent.  640 ± 30 nm. Corrected FRET (cFRET) was determined as previously described 73 . Donor and acceptor bleed-through coefficients were determined by acquiring donor alone and acceptor alone control images in all three channels described above. This correction was performed for each independent experiment. Next, cells were incubated with DACCAlabeled OVA beads in serum-free RPMI for the indicated time and then the medium was removed and replaced with fresh serum-free RPMI containing 250 nM DiBAC4(5) for 7 minutes. cFRET in cells was characterized by imaging as described above immediately afterwards.
Ouabain and valinomycin treatment. Cells were incubated in serum-free RPMI containing 1 mM ouabain for 3 hours at 37°C. Next cells were washed 3 times with a K + -rich medium UMIs, and greater than 7% of mitochondrial reads were excluded from subsequent analysis. Cell identity was annotated based on previously published markers [59][60][61][62] .

Correlative Focused Ion Beam Scanning Electron Microscopy (FIB-SEM).
RawKb.APOL7C::mCherry cells were seeded into a gridded glass ibidi µdish at 20%  Data and Code Availability. The bulk and single-cell RNA-seq will be deposited in the NCBI Genome Expression Omnibus and Sequence Read Archive once the paper is accepted for publication. Prior to this, it will be made available upon reasonable request.
This paper does not report original code.