The WD and linker domains of ATG16L1 required for non-canonical autophagy limit lethal respiratory infection by influenza A virus at epithelial surfaces

Respiratory viruses such as influenza A virus (IAV) and SARS-CoV-2 (Covid-19) cause pandemic infections where cytokine storm syndrome, lung inflammation and pneumonia lead to high mortality. Given the high social and economic cost of these viruses, there is an urgent need for a comprehensive understanding of how the airways defend against virus infection. Viruses entering cells by endocytosis are killed when delivered to lysosomes for degradation. Lysosome delivery is facilitated by non-canonical autophagy pathways that conjugate LC3 to endo-lysosome compartments to enhance lysosome fusion. Here we use mice lacking the WD and linker domains of ATG16L1 to demonstrate that non-canonical autophagy protects mice from lethal IAV infection of the airways. Mice with systemic loss of non-canonical autophagy are exquisitely sensitive to low-pathogenicity murine-adapted IAV where extensive viral replication throughout the lungs, coupled with cytokine amplification mediated by plasmacytoid dendritic cells, leads to fulminant pneumonia, lung inflammation and high mortality. IAV infection was controlled within epithelial barriers where non-canonical autophagy slowed fusion of IAV with endosomes and reduced activation of interferon signalling. This was consistent with conditional mouse models and ex vivo analysis showing that protection against IAV infection of lung was independent of phagocytes and other leukocytes. This establishes non-canonical autophagy pathways in airway epithelial cells as a novel innate defence mechanism that can restrict IAV infection and lethal inflammation at respiratory surfaces.


Summary
Respiratory viruses such as influenza A virus (IAV) and SARS-CoV-2 (Covid-19) cause pandemic infections where cytokine storm syndrome, lung inflammation and pneumonia lead to high mortality. Given the high social and economic cost of these 5 viruses, there is an urgent need for a comprehensive understanding of how the airways defend against virus infection. Viruses entering cells by endocytosis are killed when delivered to lysosomes for degradation. Lysosome delivery is facilitated by noncanonical autophagy pathways that conjugate LC3 to endo-lysosome compartments to enhance lysosome fusion. Here we use mice lacking the WD and linker domains of 10 ATG16L1 to demonstrate that non-canonical autophagy protects mice from lethal IAV infection of the airways. Mice with systemic loss of non-canonical autophagy are exquisitely sensitive to low-pathogenicity murine-adapted IAV where extensive viral replication throughout the lungs, coupled with cytokine amplification mediated by plasmacytoid dendritic cells, leads to fulminant pneumonia, lung inflammation and high Introduction Influenza A virus (IAV) is a respiratory pathogen of major global public health concern (Yamayoshi & Kawaoka, 2019). As with SARS CoV-2, animal reservoirs of IAV can contribute to zoonotic infection leading to pandemics with a high incidence of viral pneumonia, morbidity and mortality. IAV infects airway and alveolar epithelium and 5 damage results from a combination of the intrinsic pathogenicity of individual virus strains as well as the strength and timing of the host innate/inflammatory responses. Optimal cytokine levels protect from IAV replication and disease but excessive cytokine production and inflammation worsens the severity of lung injury (Davidson et al., 2014, Herold et al., 2015, Iwasaki & Pillai, 2014, Ramos & Fernandez-Sesma, 2015, Teijaro et al., 2014. 10 Even though infection of the lower respiratory tract can result in inflammation, flooding of alveolar spaces, acute respiratory distress syndrome and respiratory failure the factors that control IAV replication at epithelial surfaces and limit lethal lung inflammation remain largely unknown.
The transport of viruses to lysosomes for degradation provides an important 15 barrier against infection. Transport to lysosomes can be enhanced non-canonical autophagy pathways which conjugate autophagy marker protein LC3 to endo-lysosome compartments to increase lysosome fusion.
In phagocytes LC3-associated phagocytosis (LAP) conjugates LC3 to phagosomes and enhances phagosome maturation (Delgado et al., 2008, Fletcher et al., 2018, Lamprinaki et al., 2017, Martinez 20 et al., 2015, Sanjuan et al., 2007. In non-phagocytic cells LC3 is conjugated to endolysosome compartments during the uptake of particulate material such as apoptotic cells and aggregated β-amyloid, and following membrane damage during pathogen entry or osmotic imbalance induced by lysosomotropic drugs (Heckmann et al., 2019, Tan et al., 2018) (Florey et al., 2015, Florey et al., 2011, Roberts et al., 2013. It is known from in vitro studies that LC3 can be recruited to endo-lysosome compartments during the uptake of pathogens, but the roles played by non-canonical autophagy during viral infection in vivo are largely unknown. 5 A role for non-canonical autophagy in host defence has been implied from in vitro studies of LAP in phagocytes infected with free living microbes with a tropism for macrophages such as bacteria (Listeria monocytogenes (Gluschko et al., 2018), Legionella dumoffii (Hubber et al., 2017)), protozoa (Leishmania major) and fungi (Aspergillus fumigatus (Akoumianaki et al., 2016, Kyrmizi et al., 2018, Matte et al., 2016). 10 It is also known that IAV induces non-canonical autophagy during infection of cells in culture (Fletcher et al., 2018), however, the role played by non-canonical autophagy in controlling IAV infection and lung inflammation in vivo are currently unknown. It is not known for example if non-canonical autophagy is important in the control of IAV infection by epithelial cells at sites of infection, or if it plays a predominant role within phagocytes 15 and antigen-presenting cells during development of an immune response. Herein we use mice with specific loss of non-canonical autophagy to determine the role played by noncanonical autophagy in host defence against IAV infection of the respiratory tract. The mice (WD) lack the WD and linker domains of ATG16L1 that are required for conjugation of LC3 to endo-lysosome membranes (Rai et al., 2019) but express the N-terminal ATG5-20 binding domain and the CCD of ATG16L1 that are required for WIPI2 binding and autophagy (Dooley et al., 2014). Importantly, the WD mice grow normally and maintain tissue homeostasis (Rai et al., 2019), and unlike autophagy-defective mice, the WD mice do not have pro-inflammatory phenotype.
We show that loss of non-canonical autophagy from all tissues renders mice highly sensitive to low-pathogenicity murine-adapted IAV (A/X-31) leading to extensive viral replication throughout the lungs, cytokine dysregulation and high mortality typically seen 5 after infection with highly pathogenic IAV. Conditional mouse models and ex vivo analysis showed that protection against IAV infection of lung was independent of phagocytes and other leukocytes, and that infection was controlled within epithelial barriers where noncanonical autophagy slowed fusion of IAV with endosomes and reduced interferon signalling. This establishes non-canonical autophagy pathways in airway epithelial cells 10 as a novel innate defence mechanism that restricts IAV infection at respiratory surfaces.

Results
Mice with systemic loss of the WD and linker domains of ATG16L1 are highly sensitive to IAV infection 15 The consequences of loss of the WD and linker domains of ATG16L1 on conventional autophagy and non-canonical autophagy were confirmed using cell lines taken from controls and WD mice. Mouse embryo fibroblasts (MEFs) from littermate control mice expressed full-length  and  forms of ATG16L1 at 70kDa (Fig. 1B), and generated PE-conjugated LC3II during recruitment of LC3 to autophagosomes following 20 starvation in HBSS. The MEFs also recruited LC3 to endo-lysosome compartments swollen by monensin and control bone marrow-derived macrophages (BMDM) activated LAP to recruit LC3 to phagosomes containing zymosan (Fig. 1B). MEFs from δWD mice expressed a truncated ATG16L1 at 30 kDa (Fig. 1C). Cells from δWD mice generated LC3II and autophagosomes in response to starvation but failed to recruit LC3 to swollen endo-lysosome compartments or phagosomes containing zymosan. These data confirm defects in non-canonical autophagy and LAP in the δWD mice.
IAV enters airway and lung epithelial cells by endocytosis, and in tissue culture 5 IAV induces non-canonical autophagy leading to ATG16L1-WD domain-dependent conjugation of LC3 to the plasma membrane and peri-nuclear structures (Fletcher et al., 2018). To test whether non-canonical autophagy has a host defence function in vivo, δWD mice were infected with IAV. We used a low-pathogenicity murine-adapted IAV (A/X31) that does not normally lead to extensive viral replication throughout the lungs, or 10 cause the cytokine storm syndrome and death typically seen after infection with highly pathogenic viral strains. The results (Fig. 2) showed that δWD mice became moribund and showed severe signs of clinical illness (rapid breathing, piloerection). They also displayed rapid weight loss compared to littermate controls ( Fig

Non-canonical autophagy controls lung inflammation after IAV infection
Innate protection against IAV is provided by type 1 (α, β) and III (λ) interferon (IFN) with severe IAV infection causing excessive airway inflammation and pulmonary pathology attributable in part to IFNαβ and TNF-α (Davidson et al., 2014, Szretter et al., 2007. Measurement of cytokine expression at 2 d.p.i showed that IAV induced a transient increase in transcripts for interferon-stimulated genes (ISGs), ISG15 and IFIT1 (Iwasaki & Pillai, 2014) and pro-inflammatory cytokines ) in the lungs of both control and δWD mice ( Fig 3A). This increase in cytokine 5 expression was resolved by 3 d.p.i. before a second wave of increased cytokine expression at 5 d.p.i. This second wave of cytokine expression was resolved by 7 d.p.i in control mice, but δWD mice showed sustained increases in ISG15, IFIT1, IL-1β, TNFα and CCL2 transcripts, co-incident with exacerbated weight loss. At 3 d.p.i lungs of δWD mice showed increased expression of neutrophil chemotaxis factor CXCL1 mRNA (Fig. 10 3A), coincident with increased neutrophil infiltration of airways and parenchyma, and extensive neutrophil extracellular traps (NETs) as a consequence of neutrophil degeneration as shown by IH ( Fig. 3B and S1). Increased neutrophil infiltration of airways in δWD mice at 2 d.p.i. was confirmed and quantified using flow cytometric analysis of broncho-alveolar lavage (BAL; Fig. 3C). At 5 -7 d.p.i. increased expression of CCL2 15 mRNA in δWD mice was coincident with extensive macrophage/monocyte infiltration into lung parenchyma observed by IH ( Fig. 3B and S2) which was not seen in controls. This increased macrophage/monocyte infiltration in δWD mice was confirmed and quantified using flow cytometric analysis of single cell suspensions from lung tissue (Fig. 3D). It is known that, in severe IAV infection, a cytokine storm occurs that is amplified by 20 plasmacytoid dendritic cells pDCs (Davidson et al., 2014). pDCs detect virus-infected cells and produce large amounts of cytokines, in particular IFNαβ, that in severe infections can enhance disease. In these cases, depletion of pDCs can decrease morbidity (Davidson et al., 2014). Depletion of pDCs in IAV-infected δWD mice using anti-PDCA-1 led to markedly decreased weight loss as compared with isotype control-treated mice and that was similar to that seen in littermate controls (Fig3E). This indicates that excessive cytokine production amplified by pDCs is responsible for the increased morbidity seen in the δWD mice. 5 Thus, mice with systemic loss of non-canonical autophagy failed to control lung virus replication and inflammation, leading to increased cytokine production, morbidity and mortality.

Systemic loss of the WD and linker domains of ATG16L1 does not lead to gross
10 changes in inflammatory threshold or immunological homeostasis Macrophages cultured from embryonic livers from mice with complete loss of ATG16L1 secrete high levels of IL1- (Saitoh et al., 2008), and LysMcre-mediated deletion of genes essential for conventional autophagy (eg: Atg5, Atg7, Atg14, Atg16L1, FIP200) in mice leads to raised pro-inflammatory cytokine expression in the lung. This 15 has been reported to increase resistance to IAV infection (Lu et al., 2016), and this was also observed in mice used in our study (Fig. S3) where LysMcre-mediated loss of Atg16L1 prevented rapid weight loss and reduced virus titre. This led us to test the possibility that the δWD mutation to ATG16L1 could also increase IL-1β secretion, and cause the increased inflammation observed during IAV infection. This was tested by 20 incubating BMDM with LPS and purine receptor agonist, BzATP (Fig S4A), or by challenging mice with LPS ( Fig S4B). Mice with a complete loss of ATG16L1 in myeloid cells (Atg16L1 fl/fl -lysMcre) showed three-fold increases in IL-1β in both serum and of secretion IL-1β from BMDM in vitro. In contrast IL-1β secretion in δWD mice did not differ significantly from littermate controls ( Fig S4A&B). This was consistent with lack of elevated cytokines in lungs prior to infection (see day 0 in Fig. 3A), and our previous work showing that serum levels of IL-1β, IL-12p70, IL-13, and TNF-α in δWD mice are the same as in littermate controls at 8-12 and 20-24 weeks (Rai et al., 2019). The exaggerated 5 inflammatory response to IAV in δWD mice did not therefore result from a raised proinflammatory threshold or dysregulated IL-1β responses in the lung. Also, the frequencies of T-cell, B-cell and macrophages were similar in δWD mice to littermate controls (Fig.   S5). These data suggest that the exaggerated responses of δWD mice to IAV do not occur because the mice have a raised inflammatory threshold or abnormal immunological 10 homeostasis.

Non-canonical autophagy limits IAV infection independently of phagocytic cells
The link between non-canonical autophagy/LAP, TLR signalling, NADPH oxidase activation and ROS production (Delgado et al., 2008, Martinez et al., 2015, Sanjuan et 15 al., 2007 provides phagocytes with a powerful mechanism to limit infections in vivo. To test whether wild-type bone marrow-derived cells could protect susceptible δWD mice from lethal IAV infection, we generated radiation chimeras (Fig. S6). When challenged with IAV, δWD mice reconstituted with either wild-type or δWD bone marrow remained highly sensitive to IAV ( Fig. 4A & B) with body weight reduced by up to 25% and 20 decreased survival by 5 d.p.i. As seen for δWD mice, weight loss was associated with a 10-fold increase in lung viral titre (Fig. 4C), fulminant pneumonia and inflammatory infiltration into the lung (Fig. 4D). This increased susceptibility to IAV was not observed for control mice reconstituted with wild-type marrow, showing that non-canonical autophagy pathways in phagocytes and other leukocytes from control mice were not able to protect δWD mice against lethal IAV infection. In a reciprocal experiment ( delivery of viral ribonuclear proteins (RNPs) into the cytoplasm (Skehel & Wiley, 2000, Wharton et al., 1994. RNPs are then imported into the nucleus for genome replication (Boulo et al., 2007). The effect of non-canonical autophagy on IAV entry was tested using a fluorescence de-quenching assay where the envelope of purified IAV was labelled with green (DiOC18) and red (R18) lipophilic dyes. MEFs incubated with IAV for increasing times were analyzed by FACs to detect the green fluorescence signal generated when the dyes are diluted following IAV fusion with the endosome membrane ( Fig. 6A-C). The percentage of cells emitting a green signal was greater in MEFs from δWD mice (60% 5 compared to 40% for controls at 30 min p.i.) and increased with time (73% versus 56% for controls; Fig. 6B). Likewise, the median fluorescence intensity was 1.6 -1.4-fold higher in MEFs from δWD mice ( Fig 6C). This showed that non-canonical autophagy slowed fusion of IAV with endosome membranes. The recognition of viral RNA by interferon sensors following delivery of RNPs into the cytoplasm was used as a second 10 assay for IAV entry. MEFs from δWD mice showed between 3 and 5-fold increases in expression of IFN responsive genes, ISG15 and IFIT1 ( Fig. 6D & E), and this was also observed in the lung in vivo (Fig. 3A). Taken together the results demonstrate for the first time that the WD and linker domains of ATG16L1 allow non-canonical autophagy to provide a novel innate defence mechanism against lethal IAV infection within the epithelial 15 barrier in vivo. showed profound sensitivity to infection by a low-pathogenicity murine-adapted IAV (A/X31) leading to extensive viral replication throughout the lungs, dysregulated cytokine production, fulminant pneumonia and lung inflammation leading to high mortality and death usually seen after infection with virulent strains (Belser et al., 2011). These signs 5 mirror the cytokine storms and mortality seen in humans infected with highly pathogenic strains of IAV such as the 1918 'Spanish' Influenza (Belser et al., 2011).

Respiratory viruses such as influenza A virus (IAV) and SARS
The observation that bone marrow transfers from wild-type mice were unable to protect δWD mice from IAV suggested that protection against IAV infection in vivo was independent of leukocytes and did not require non-canonical autophagy in leukocyte 10 populations (e.g. macrophages, dendritic cells, neutrophils, granulocytes, lymphocytes).
In a reciprocal experiment the linker and WD domains of ATG16L1 were deleted specifically from myeloid cells. These mice, which lack non-canonical autophagy in phagocytic cells ( δWD mice infected with IAV appeared to be unable to resolve inflammatory responses resulting in sustained expression of pro-inflammatory cytokines, morbidity and 10 a striking lung pathology characterized by profuse migration of neutrophils into the airway at day 3 followed by macrophages on day 7. pDCs detect IAV-infected cells and produce large amounts of cytokines, in particular IFNαβ, that in severe infections can enhance disease (Davidson et al., 2014). The fact that morbidity in δWD mice could be decreased by depleting pDCs indicates that excessive cytokine production, amplified by pDCs was 15 a major factor. This is not due to a lack of non-canonical autophagy/LAP in pDC as bonemarrow chimaeras of δWD mice with wild-type leukocytes have the same phenotype as δWD mice. IAV is recognized by endosomal TLR3 in respiratory epithelial cells and RIG-I detects virus replicating in the cytosol leading to activation of IRF3 and NFkB with subsequent induction of interferon, ISG and proinflammatory cytokine production (Iwasaki 20 & Pillai, 2014). Increased inflammation may result directly from increased virus in the lungs, but the increased fusion of IAV envelope with endosomes in WD mice may increase delivery of viral RNA to the cytoplasm resulting in the sustained pro-inflammatory cytokine signalling. A similar pro-inflammatory phenotype resulting from decreased trafficking of inflammatory cargoes is observed following disruption of non-canonical autophagy by LysMcre-mediated loss of Rubicon from macrophages or microglia (Heckmann et al., 2019, Martinez et al., 2016. We have dissected the roles played by conventional autophagy and non-canonical Several non-canonical pathways leading to recruitment of LC3 to endo-lysosomal compartments, rather than phagosomes, are beginning to emerge. Non-canonical autophagy in microglia facilitates endocytosis of amyloid and TLR receptors to reduce β-amyloid deposition and inflammation in mouse models of Alzheimer's disease (Heckmann et al., 2019). This may involve and interaction between the WD domain and TMEM59 which is required for -amyloid glycosylation (Ullrich et al., 2010).
Lysosomotropic drugs, which stimulate direct recruitment of LC3 to endosomes, create  The δWD MEFs do not recruit LC3 (green) to endo-lysosomes following incubation with monensin or to bone marrow-derived macrophage phagosomes containing zymosan.

Fig. 2. Systemic loss of non-canonical autophagy increases susceptibility to IAV infection.
Littermate control and WD mice were challenged intranasally with IAV strain X31 (10 3 pfu).  Mann-Whitney U test was used to determine significance. (D) Precision-cut lung slices from control and δWD mice were infected with IAV. Virus titres were determined at indicated time points. Comparisons were made using two-way ANOVA with Bonferroni post-tests.

Cell culture and virus.
Influenza virus A/HKx31 (X31, H3N2) was propagated in the allantoic cavity of 9-day-old embryonated chicken eggs at 35°C for 72 h. Titres were determined by plaque assay using MDCK cells with an Avicel overlay. 5

Mice.
All experiments were performed in accordance with UK Home Office guidelines and under the UK Animals (Scientific procedures) Act1986.
The generation of WD mice (Atg16L1 δWD/δWD ) has been described previously (Rai et al., 2019). Generation of WD phag and Atg16L1 fl/fl -LysMCre mice is described in detail 10 in Fig. S7 and separate cohorts inoculated intra-nasally with 10 3 PFU IAV strain X31 in 50 µl sterile PBS. Mice were infected between 9 and 11 AM. Animals were sacrificed at variable timepoints after infection by cervical dislocation. Tissues were removed immediately for 5 downstream processing. Sample sizes of n = 6 were used as determined using power calculations and previous experience of experimental infection with these viruses. For survival analysis, a humane endpoint was determined using a scoring matrix that included excessive (>20%) weight loss.
To specifically deplete plasmacytoid dendritic cells (pDCs), mice were treated with 10 anti-PDCA-1 (Cambridge Bioscience) or IgG2b isotype-matched control, using a dose of 500 mg per 200 ml via the i.p. route on day 1 of infection with IAV and every 48 h thereafter (Davidson et al., 2014).

15
The general strategy is shown in Fig S6A. Mice were subjected to whole body irradiation with 11 Gy in two doses 4 h apart using a 137 Cs source in a rotating closed chamber.
Bone marrow was collected from male wild-type C57BL/6-Ly5.1 (B6.SJL-Ptprc a Pepc b /BoyCrl; Atg16L1 +/+ ) mice that are congenic for the CD45.1 allele or from δWD mice (that are congenic for CD45.2). The C57BL/6 CD45.1 marrows were used to 20 enable confirmation of chimaerism by FACS analysis of bon-marrow-derived cells as littermate control and δWD mice are CD45.2 (Fig S6B). The femur and tibia of the donor mouse was collected and sterilised for 2 min in 70% ethanol. The ends of the bones were removed and PBS was used to flush out the bone marrow through a 40 μm cell sieve.
Red blood cell lysis was performed using 0.83% ammonium chloride and the cells were washed twice in PBS and re-suspended at a concentration of 10 7 cells/ml. T cell depletion was performed prior to transfusion by using a commercial mouse hematopoietic progenitor cell isolation kit (EasySep, STEMCELL™ Technologies, #19856).

5
After depletion, 10 6 donor bone marrow cells were injected into each irradiated mouse by tail vein injection 3 h following irradiation. Mice were then allowed to recover for 12 weeks with daily monitoring of mouse weights and general condition for at least the first two weeks to monitor for any severe radiation sickness or illness due to being immunocompromised.

Flow Cytometric analysis of Cells.
Brocho-alvolear lavage fluid (BAL) was obtained by lavage of mice via the trachea using 1 ml ice-cold RPMI containing 5% FCS. For lung tissue, single-cell suspensions were made from minced lung and subjected to collagenase and DNase I digestion, then treated with ACK buffer to remove red blood cells. In both cases, approximately10 6 cells were

Precision-cut lung slices
Infection of ex vivo lung slices was used to examine the responses of lungs without any contribution from recruited leukocytes, which could not be present. Mouse lungs were inflated with 2% low melting point agarose in HBSS and then sliced into 300 µm sections 5 using a vibrating microtome. They were then cultured overnight in DMEM/F12 medium (Thermofisher 21331020) prior to infection with IAV.
IAV Endosome fusion assay. The envelope of purified IAV (0.1mg protein mL -1 ) was labelled using an ethanol solution containing 33µM 3,3'-dioctadecyloxacarbocyanine 10 (DIOC18) and 67µM octadecyl rhodamine B (R18  S1. Increased neutrophilia and NETosis in IAV-infected mice deficient in noncanonical autophagy δWD and littermate control mice were infected i.n. with 10 3 pfu IAV X31. Lung tissues were harvested at 3d p.i. Neutrophils and H3 (marker of NETosis) were detected by IH using anti-Ly6G and anti-H3, visualized with DAB and counter-stained with hematoxylin. Micrographs of representative areas from lungs of six mice are shown. Scale bars represent 500 µm (upper panels), 50 µm (middle panels) or 250 µm (lower panels). There are dramatically increased numbers of neutrophils in airways (bronchi and bronchioles) and lung parenchyma of δWD mice, accompanied by markedly-increased NETosis, indicating significant neutrophil degeneration.

Fig. S2. Increased macrophage rich inflammation in IAV-infected mice deficient in non-canonical autophagy
δWD and littermate control mice were infected i.n. with 10 3 pfu IAV X31. Lung tissues were harvested at 7d p.i. Macrophages were detected by IH using anti-Iba-1, visualized with DAB and counter-stained with hematoxylin. Micrographs of representative areas from lungs of six mice are shown. Scale bars represent 500 µm (upper panels) and 50 µm (lower panels). Lower panels are the same as in Fig. 2B. Upper panels show the lower magnification images of the lung to illustrate the general nature of the observations. There is clearly increased inflammation in δWD mice with higher numbers of macrophages in the lung parenchyma.

Fig. S3. LysMcre-mediated loss of canonical macro-autophagy from phagocytes decreases sensitivity to IAV infection
Atg16L1 fl/fl -LysMcre mice and littermate controls; n = 5 or 6 per group) were infected i.n. with 10 3 pfu IAV X31. Panel A. Mice were weighed daily and the weights presented as a percentage of the starting weight. Panel B. Lung tissues were taken at 5 d.p.i. and virus titer determined by plaque assay. Data represent the mean value ± SEM. Analysis using the Mann-Whitney U test showed a significant difference (* p < 0.05). Thus, Atg16L1 fl/fl -LysMcre mice that are deficient in canonical autophagy in phagocytes lose weight at the same rate as littermate controls but are more resistant to virus replication as they have lower lung virus titres. LysMcre-mediated deletion of autophagy genes from mice leads to increased inflammatory threshold characterised by raised secretion of IL-1β from macrophages (15), and in the lung this can increase resistance to IAV infection (16). The possibility that the δWD mutation could affect IL-1β secretion was tested by stimulating BMDM with bacterial lipopolysaccharide (LPS) and BzATP (P2X7 receptor agonist) or challenging mice with LPS. Panel A. Bone marrow-derived macrophages (BMDM) from mice strains as indicated were incubated with 100 ng/ml of LPS for 4 h and 150 μM of BzATP for 30 min. Supernatants were assayed for IL-1β by ELISA. (Mock group: untreated, BzATP controls only received BzATP). Representative data are shown as the means ± SD of readings from 20 wells per group and were analyzed using one-way ANOVA with Tukey's post-hoc analysis (**** p<0.0001). Approximately three-fold increases in IL-1β secretion were seen for BMDM from Atg16L1 fl/fl -LysMCre mice. However, IL-1β secretion from δWD BMDM did not differ significantly from littermate controls. Panel B. Mouse strains (as indicated) were injected with 20 mg/kg of LPS via the IP route. Serum collected 90 min post injection was assayed for IL-1β by ELISA. In nontreated mice IL-1β was below the detection limit in all 3 strains (not shown). Data are shown as the means ± SD of duplicate assays from 4 mice per group and were analyzed using one-way ANOVA with Tukey's post-hoc analysis (** p < 0.01). Approximately threefold increases in IL-1β secretion were seen for Atg16L1 fl/fl -LysMCre, however IL-1β secretion for δWD mice did not differ significantly from littermate controls.

Fig. S8. Mice deficient in non-canonical autophagy/LAP in phagocytes control IAV infection and do not show increased lung inflammation
δWD phag that are deficient in non-canonical autophagy in phagocytes and littermate control mice were infected i.n. with 10 3 pfu IAV X31. Lung tissues were harvested at 5 d p.i. Macrophages were detected by IH using anti-Iba-1, visualized with DAB and counterstained with hematoxylin. Scale bars represent 500 µm (upper panels) and 50 µm (lower panels). Micrographs of representative areas from lungs of six mice are shown. Both δWD phag and control mice show little inflammatory response in the parenchyma and mild, macrophage-rich peri-bronchiolar infiltration (black arrows).

Fig. S9 Fluorescent signal in cells infected with SP-DiOC18 labelled IAV is pHdependent fusion.
Sub-confluent monolayers of MEFs were incubated with dual-labelled (SP-DiOC18/R18) IAV at 4 0 C for 45 min and warmed to 37°C for 90 min. pH-dependent fusion was assessed by adding bafilomycin A1 (BAF) to some wells. Cells were harvested by trypsinisation, fixed in PFA and analysed by flow cytometry.